Argentina : poor people in a rich country (Vol. 2 of 2) : Background papers

Report No. 19992-AR Argentina Poor People in a Rich Country (In Two Volumes) Volume 11: Background Papers April 14, 2000 Poverty Reduction and Economic Management Latin America and Caribbean Region Document of the World Bank Task Team This report is a product of a team, led by Norman Hicks, which included World Bank staff members Alexandre Abrantes (health), Sandra Cesilini (poverty perceptions), Michael Cohen (urban),Wendy Cunningham (gender), William Experton (education), Luis Guasch (labor), Haeduck Lee (poverty data and analysis), William Maloney (labor), Vicente Paqueo (social protection), and Tom Wiens (rural). Contributions were also incorporated from Myrna Alexander, Carlos Arango, Omar Arias, Dariush Akhavan, Ana Maria Arriagada, Omar Arias, Mercedes Avellaneda, Katherine Bain,Walter Sosa Escudero, Estanislao Gacitua-Mario, Paul K. Freeman, Gillete Hall, Kihoon Lee, Mariana Marchioni, Pia Peeters, Steven Schonberger, Carlos Velez, Quentin Wodon, and Enrique Zuleta. The peer reviewers were Martin Ravallion, Ariel Fiszbein and Miguel Szekely. Valuable comments and advice were received from Myrna Alexander (country director), Guillermo Perry (chief economist), and Paul Levy (lead economist). Editing done by Marta A. Cervantes-Miguel, Tania Gomez and Cecilia Zavaleta. POOR PEOPLE IN A RICH COUNTRY A Poverty Report for Argentina Background Papers Volume II Table of Contents Background Paper 1 Poverty and Income Distribution in Argentina, Patterns and Changes Author: Haeduck Lee Background Paper 2: Perceptions and Self-Perceptions of Poverty in Argentina Authors: Sandra Cesilini and Enrique Zuleta Background Paper 3: Employment, Informality and Poverty in the Argentine Labor Market Authors: Carlos Arango and William F. Maloney Background Paper 4: Protecting the Poor and Improving Investment in Their Human Capital Authors: Vicente Paqueo and Kihoon Lee Background Paper 5: Poverty and Health Authors: Alexandre V. Abrantes and Dariush Akhavan Background Paper 6: An Analysis of Rural Poverty in Argentina Author: Tom Wiens Background Paper 7: Household Structure, Gender, and the Economic Determinants of School Attendance in Argentina Authors: Walter Sosa Escudero and Mariana Marchionni Background Paper 8: Marginalization of Men in Argentina: Evidence from Changes in Wage Distribution Author: Omar Arias Background Paper 9: Does the Gender Recipient of Social Programs Matter in Argentina? Authors: Pia Peeters & Wendy Cunningham r Argentina Poverty Assessment Background Paper No. 1 Poor People in a Rich Country A Poverty Report for Argentina Background Paper No. I Poverty and Income Distribution in Argentina Patterns and Changes By Haeduck Lee The World Bank LCSPR January 2000 The report was prepared as a contribution to the Argentina Poverty Assessments. The author wishes to thank Mr. Norman L. Hicks, Task Manager, for his guidance throughout the project. He is also grateful to Messrs../Mmes. Hector Montero, Clyde Trabuchi, Carmen Dopico, Miriam Fischman (INDEC), Irene Novacovsky, Luis Beccaria (SIEMPRO) for providing the data and background materials, and to the workshop participants at the Buenos Aires Bank Office on March 23, 1999 and at INDEC on March 29, 1999 for their insightful comments and valuable suggestions. Contributions were incorporated from Mr. Carlos Eduardo Velez (inequality decomposition) and Quentin T. Wodon (growth and poverty). Argentina Poverty Assessment Background Paper No. 1 Poverty and Income Distribution in Argentina Patterns and Changes Table of Contents Page Number I. INTRODUCTION ............................................ 3 II. DEFINING POVERTY ........................................ 4 A. Argentina's PovertyLine ................................. 4 1. Current PovertyLines ............................... 4 2. Poverty Lines for This Study .......................... 6 B. Counting thePoor........................................ 12 C. Limits of the AvailableData................................ 13 III. DIMENSIONS OF POVERTY...................................17 A. Snapshot ofPoverty....................................... 17 B. Trends in Poverty ......................................... 18 IV. POVERTY PROFILE .......................................... 24 V. DETERMINANTS OF POVERTY................................ 27 VI. INCOME DISTRIBUTION ...................................... 31 A. Trends in IncomeDistribution............................... 31 B. Regional IncomeDisparity.................................. 40 VII. SUMMARY AND CONCLUSIONS............................... 44 1 REFERENCE Annex Table 1: A Profile of the EPH Data Sets, 1990-1998 Annex Table 2: Adult Equivalent Basic Food Basket for the Current Poverty Line, MBA Annex Table 3: A Matrix of Adult Equivalence Coefficients Annex Table 4: CPIs for the Metropolitan Buenos Aires (MBA), 1980-1998 Annex Table 5: Adult Equivalent Poverty Lines for the Metropolitan Buenos Aires (MBA), 1980-1998 Annex Table 6: Adult Equivalent Indigent Lines by Region, 1980-1998 Annex Table 7: Non-Food Consumption in Adult Equivalent Poverty Lines, Indexed by Non-Food Necessities CPI, 1980-1998 Annex Table 8: Adult Equivalent Poverty Lines by Region, 1980-1998 Annex Table 9: Implicit Engel Coefficients by Region, 1980-1998 Annex Table 10: Earnings Equation Estimation to Predict Earnings Among the Unemployed, 1990-1998 Annex Table 11: Distribution of Adult Equivalent Per Capita Income in Argentina, 1990-1998 u Argentina Poverty Assessment Background Paper No. 1 Poverty and Income Distribution in Argentina Patterns and Changes I. INTRODUCTION It has been five years since the Bank issued the Poverty Assessments report for Argentina. Poverty trend and snapshot contained in the report drew upon the available information on household living conditions up to May 1993. Argentina just emerged from a traumatic experience of hyperinflation of 1989-90, during which almost half the population of Metropolitan Buenos Aires (MBA) fell below the poverty threshold. Economic stability brought about by the Convertibility Plan (April 1992) had produced significant poverty reduction. By April 1993, poverty rate fell below 20% for the Metropolitan Buenos Aires (MBA), and the prospects for continued poverty reduction was good. Why Poverty Assessments again? A lot has happened since then. Following Mexico's Tequila Peso Crisis in 1994, the jobless rate rose sharply and reached more than 18% in May 1995. In spite of buoyant performance of the economy during 1996-97, unemployment rate stayed at more than 13%, and there is a widespread perception that the situation for the poor and the vulnerable did not improve much even during this brief upturn of the economy. Recently even though the spread and contagion of the Asian crisis seem to have been stemmed at least, it still had the effect of putting a brake on the Argentine economy since 1998. The 1995 report heavily depended on the data from the Metropolitan Buenos Aires. Poverty estimates for other parts of the country was sketchy relying on partial data. As home for 33% of the Argentines, the MBA is the most prosperous part of the country. However, omission of other parts of the country seriously have compromised the validity and implications of the report findings. Argentina still does not have a national household survey with the representative sample for both urban and rural areas. Still there has been marked improvement in the availability of survey data at the household level with which to document and analyze the living standards and poverty situation of Argentine population. The Permanent Household Survey (EPH) progressively expanded its list of urban agglomerations from which they collect information. Now it covers about 60% of Argentine population, or 70% of all urban population. The National Income and Expenditure Survey (ENGHO) was carried out during February 1996 and February 1997 has become available. And the Social Development Survey supported by the World Bank's 1st Social Protection Project and implemented by SIEMPRO/INDEC in August 1997 contains a wealth of information on the living conditions and access to public social programs in Argentina. This makes it opportune that the Bank and the Government of Argentina make concerted efforts to take stock of where Argentine is in terms of poverty situation, so that it could formulate effective strategy for poverty alleviation for the next millennium. 3 Argentina Poverty Assessment Background Paper No. 1 II. DEFINING POVERTY Poverty is a multi-dimensional phenomenon which can be characterized by the lack of control over monetary resources (income or consumption), satisfaction of basic needs (consumption to some extent, and NBI indicators), capabilities to be self-sufficient, or any combination thereof. In this chapter, the focus will be on how to define and measure household welfare and poverty in Argentina by income or consumption. Suppose that poverty is defined as the inability to command the necessary material resources for satisfying the minimum level of living standards. Such thresholds which would guarantee satisfaction of the minimum needs are the so-called poverty lines. And those segments of population who cannot reach this income thresholds would be deemed as poor. Once the poverty line is defined for a particular place at one point in time, the problem of poverty measurement is one of finding such welfare yardsticks which will be held constant in real terms over time and at different locations, so that meaningful comparison of household welfare and poverty is possible. A. Argentina's Poverty Lines Current poverty lines in Argentina were the product of active poverty work program during the late 1980s, when the 1985-86 Income and Expenditure Survey for the Metropolitan Buenos Aires (MBA) became available. Poverty lines have been calculated by applying the internationally accepted standards and norms to the Argentine situation. As these poverty lines have been employed in research papers and official bulletins on poverty by academics and policymakers, they have gained more acceptance as the semi-official and/or official reference poverty thresholds. Since its elaboration in 1988, the current poverty lines have been in effect for the past 11 years. However, Argentina has experienced an episode of hyper-inflation during this period, where cumulative changes in prices of goods and services must have brought about fundamental shifts in consumption pattern, in particular that of foods and non- foods consumed by the poor. This calls for a critical review of the way how poverty lines have been updated for poverty analysis over time, since poverty comparison should be carried out by applying absolute poverty lines which are consistent with a constant level of household welfare.' 1. Current Poverty Lines Current poverty lines were constructed by analyzing the consumption patterns of low income households in the data from 1985-86 Income and Expenditure Survey for the MBA (B.A. de Morales, 1988; Epszteyn and Orsatti, 1989). 1 Absolute poverty lines and consistent poverty lines will be used interchangeably from now on. They refer to that poverty threshold which is compatible with guaranteeing the same level of welfare for individuals no matter when and no matter where they live. 4 Argentina Poverty Assessment Background Paper No. 1 The first step was the choice of food bundle consumed by the poor. The basic food basket was taken from the list of food items recommended by the 1981 Consultative Meeting of FAO/OMS/UNU (1985). Once the basic food basket was defined, food consumption behavior by households in 2-4 income deciles in the 1985-86 National Income and Expenditure Survey was carried out to find out how much of each of these food items was consumed. The nutritional contents (grams) have been converted to the energy intakes (calories) and aggregated to obtain the overall daily caloric intakes corresponding to the food bundle. The nutritional contents of the basic food basket were then adjusted proportionally to allow a representative adult male of age 30-59 with moderate activity level to achieve a daily intake of 2,700 Calories (B.A. de Morales, 1988; CEPA, 1993). (See Annex Table 1 for details). The monetary value of the basic food basket thus defined is the so-called adult equivalent indigent line. In March 1988, the indigent line corresponding to 2,700 Calories was given by 149.2 Australes per month for an adult. As the Engel coefficient for the 2-4 decile households was approximately 0.483 (the inverse Engel coefficient = 2.07), the adult equivalent poverty line was about 308.9 Australes per month for the MBA. In obtaining the current poverty lines for the purpose of poverty comparison over time, their two constituent components have been updated by different factors: (i) the indigent lines by the price index of basic food basket, and (ii) the non-food component by the non-food CPI using the consumption weights of households in the 2nd income quintile.2 As of September 1996, the adult equivalent indigent was 67.4 Pesos per month and the corresponding poverty line was 174.0 Pesos per month for the MBA. Table 1: Current Poverty Lines for the Metropolitan Buenos Aires Implicit When the non-food Indigent Non-food Poverty Engel component was indexed line component line Coefficient by: Official poverty lines March 1988 (Australes) 149.2 159.7 308.9 0.483 -- September 1996 (Pesos) 67.4 106.6 174.0 0.387 Non-food CPI with weights of the 2nd quintile Alternative indexing of non-food component September 1996 (Pesos) 67.4 71.9 139.3 0.483 Basic Food Basket CPI September 1996 (Pesos) 67.4 100.1 167.5 0.402 General CPI September 1996 (Pesos) 67.4 106.8 174.2 0.387 Non-food CPI Sources: E. B.A. de Morales (1988), Canasta basica de alimentos: Gran Buenos Aires, Documento de Trabajo No. 3, IPA, INDEC, Buenos Aires. E. Epszteyn and A. Orsatti (1989), Caracteristicas de una linea de pobreza para Argentina, 1985, Documento de Trabajo No. 8, IPA, INDEC, Buenos Aires. INDEC, Indice de precios al consumidor, contained in Estadistica mensual, 1988-1998. The official poverty lines given in Table 1 have been widely utilized Argentina to produce the government's official poverty rates. However, the current poverty lines 2 Since the weights for price indices are fixed in both cases, the resulting poverty lines are the Paasche index for February 1988 and backward, and the Laspeyres index from April 1988 to present. 5 Argentina Poverty Assessment Background Paper No. I possess two weaknesses as a measure of consistent poverty lines, i.e., as anchor of constant welfare level. First, the method of indexing the non-food component has the tendency to overstate non-food needs of the poor, thereby allowing poverty lines to drift upward in real value over time. Even though INDEC utilizes the consumption weights of the households in the 2nd income quintile to index the non-food component of poverty lines, the constituent items in the non-food bundle come from the commodity list for the general CPI which includes luxury items as well as basic necessities. While the food component is indexed by the basic food basket , i.e., a subset of food items that are more likely to be purchased by the poor, the non-food commodity bundle is not indexed in a consistent manner. According to the 1996-97 National Household Expenditure Survey data, the food share of consumption expenditure among households in the second income quintile has come down from about 48% to 43% in the MBA.3 If poverty lines were indexed correctly, one would expect to find the resulting Engel coefficient in the neighborhood of 0.43 circa 1996. In contrast, a significant change of Engel coefficient from 0.483 in March 1988 to 0.387 in September 1996 implicit in the current poverty lines is evidence of such a bias at work. One possible avenue of rectifying the situation might be to apply other types of publicly available price index to update the non-food component of poverty lines. As Table 1 indicates, use of neither the basic food basket CPI nor the general CPI produces poverty lines with the Engel coefficient around the neighborhood of 0.43 in 1996/96. Second, the current poverty lines are available only for the MBA. Since the 1985- 86 Income and Expenditure Survey was carried out in the MBA alone, there was no statistical base upon which to construct the regional price index to adjust for the cost-of- living differences across regions within Argentina. As a result the current poverty lines has limited usefulness for regional poverty comparison. In light of the short-comings of the current poverty, it appears that additional work is needed to construct absolute poverty lines for Argentina that are time-invariant and location-invariant with respect to welfare, so that they could be used to map out the overall picture of poverty in a consistent manner at different places over time. 2. Poverty Lines for This Study The poverty lines for this study are constructed by combining information from the current poverty lines with itemized price statistics from INDEC to produce time- consistent poverty lines on the one hand, and with the newly available data from the 1996-97 National Household Expenditure Survey to define regional poverty lines. A point of departure is the official poverty line defined for the MBA in March 1988. Since the current poverty line is considered as official, this poverty threshold will be used as the basic welfare anchor. First, we turn to a description of the methodology utilized to ensure time consistency of poverty lines. 3 INDEC (1999), EncuestaNacional de Gastos de los Hogares (ENGHO), 1996-1997, Resultados Definitivos, Buenos Aires. 6 Argentina Poverty Assessment Background Paper No. I Defining the Non-food CPIfor the Poor In order to preserve their time consistency, poverty lines should be updated by the price index of commodity bundles that are more likely to be consumed by the poor. Ideally, we would like to use the identical commodity bundle made up of foods and non- foods as the vector of weights to which a time series of unit prices could be applied to update poverty lines. However, it is rare that we have access to statistical information at such level of detail. Moreover, construction of poverty lines from the indigent line by applying the Engel method means that we do not have information on the distribution of different categories of non-foods in the poverty lines. INDEC continuously updates the price index of basic food basket, and the indigent lines reflect the purchasing power of the extremely poor in order for them to be able to buy just enough basic foods to survive over time. However, there does not exist any corresponding official price index for the non-food commodity bundle more likely to be consumed by the poor. As it was stated before, inclusion of all CPI non-food items in the indexing of non-food component of current poverty lines tends to overstate poverty lines, even though the consumption weights of the 2d income quintile was utilized. Therefore, it is necessary to construct a new price index of the non-food commodity bundle that would reflect the consumption pattern of the poor, namely the non-food price index for the poor, in order to define poverty lines that is consistent with a constant level of welfare over time. As Annex Table 4 shows, the general CPI went up by 627,047% between 1988- 1996, while the rate of increase was 10% lower for the food CPI and 7% higher for the non-food CPI 7% than the general CPI. During the same period, the increase in price of the basic food basket increased lagged behind the general CPI by as much as 28%. This indicates that the cost of living index for the poor did not catch up with the general inflation as far as food consumption was concerned. Then the relevant question to complete the picture would be, "What happened to the prices of non-foods that are more likely to be consumed by the poor?" In order to answer this, we have constructed the price index of a subset of non-food commodities which were obtained after removing luxury or other goods and services which are more likely to be consumed by the rich. The Price Index Division of INDEC carries out monthly updates of price indices for 139 commodity sub-groups, 58 sub-groups for food items and 81 for non-foods, where weights for these items are based on the 1985-86 Income and Expenditure for the MBA (40.1% for foods, and 59.9% for non-foods).4 Now from the list of 81 items of non-foods which represent 59.9% weights, the author, working with INDEC technical team, have identified 19 items (19.5% weights) that have the characteristics of luxury or other goods and services with higher income elasticity. They included, among other things, expenditures on formal education (1.77%), domestic service (2.13%), purchase and maintenance of automobile (7.39%), tourism and hotels (1.77%), etc. Table 2 shows that there have occurred significant shifts in consumption pattern in Argentina during the last decade. In particular, there has been noticeable drop in food share in consumption. In non-food categories, Argentines spend less on clothing, 4 The reference population of the price index is the residents of the MBA, excluding persons living in households with top 5% per capita household income and persons living alone. 7 Argentina Poverty Assessment Background Paper No. I household articles in spite of general rise in overall expenditure share. In general, consumption increased in such areas as housing and public utilities, transport, and other services. These are the areas where heavy subsidies in consumption have been either scaled back or eliminated in the 1990s. The price indices of the remaining 62 non-food items (representing 40.4% weights) have been utilized to generate a set of eight price sub-indices according to the commodity groups shown in Table 2, using their own weights. Finally, the non-food price index for the poor was obtained by calculating the weighted average over time of the eight non- food groups purged of luxuries with weights coming from the consumption pattern of households in the 2nd income quintile of the MBA, after normalizing for the non-food categories. Table 2: Changes in the Structure of Consumption Expenditure, 1986-1996 ENGHO, 1985-86 ENGHO, 1996-97 Metro Buenos Aires Metro Buenos Aires All Regions 2nd 2nd 2nd Quintile Total Quintile Total Quintile Total Overall (% 100.0 100.0 100.0 100.0 100.0 100.0 Food & Beverage 48.1 38.2 42.9 32.9 43.2 36.6 Non-Foods 51.9 61.8 57.1 67.1 56.8 63.4 1. Clothes & Shoes 9.0 9.3 5.6 5.8 7.5 7.5 2. Housing, Fuel & Electricity 9.0 9.0 12.5 13.1 12.5 12.0 3. Household Articles 6.2 8.0 5.1 6.9 4.7 5.1 4. Medical & Health Supplies 6.9 7.9 7.4 9.8 7.3 8.9 5. Transport& Communications 8.1 11.6 12.6 13.7 11.3 14.3 6. Recreation & Culture 5.0 7.5 5.6 11.9 5.3 6.6 7. Education 2.2 2.6 3.3 5.0 2.9 3.7 8. Miscellaneous Services 5.5 5.9 5.1 4.7 5.1 5.4 Sources: INDEC (1988), Encuesta de Gastos e Ingresos de los Hogare, 1985-1986, Buenos Aires. INDEC (1999), Encuesta Nacional de Gastos de los Hogares, 1996-1997, Resultados Definitivos, Buenos Aires. The last column of Table 3 gives the resulting CPI of the non-food necessities for the poor, based on the consumption structure of the MBA population in the 2nd per capita household income quintile. It indicates that price inflation of the non-food CPI for the poor during the last decade was significantly less than either the general CPI or the non- food CPI for the general population. Poverty line elaborated for the MBA using this indexing method was given by 155.6 Pesos in September 1996, and it was made up of the indigent line of 67.4 Pesos and the non-food component of 88.2 Pesos. The food share of 0.433 implied by the poverty line was close to the food share ratio of 0.429 among population in the 2nd income quintile in the MBA. 8 Argentina Poverty Assessment Background Paper No. 1 Table 3: Changes in Price Structure for the MBA, 1980-1998 General Consumer Price Index (General CPI) CPI for the Poor Year Reference Overall Food Non-food Basic Food Non-food Period Basket Necessities (weights) (100.0) (40.1) (59.9) 1 _(40.4) (March 1998=100) 1988 March 100.0 100.0 100.0 100.0 100.0 1996 September 627,047 561,795 668,543 451,511 552,374 (index) (100.0) (89.6) (106.6) (72.0) (88.1) Source: INDEC, INDEC Informa, 1980-1998. Defining Regional Poverty Lines Elaboration of regional poverty lines is based on an analysis of the 1996-97 National Household Expenditure Survey which has the national urban coverage. This survey covers all urban population residing in urban centers (namely localidad) in which more than 5,000 inhabitants live (and covers 84% of all Argentine population). A preliminary analysis of the data revealed that households with the male adult consuming 2,750 Calories of daily food intake fall within the 11-20% range of the national survey.5 In order to increase degrees of freedom, households falling in the 20% range between 16% and 35% were chosen as the reference population group. For the MBA, the indigent line was calculated by multiplying the basic food basket with the corresponding unit price vector that comes from the Price Index Division of INDEC, not with the implicit prices in the survey itself. The resulting indigent line and poverty line for the MBA are then given by 82.4 Pesos and 172.2 Pesos per month for an adult, respectively. As for regional poverty lines, the data subsets of the national reference group which correspond to respective regions were analyzed and adjusted proportionally to allow each regional reference group to achieve consumption of 2,750 Calories. Once the value of the basic food basket for each region was obtained, then coefficients of the pair-wise difference of other regional baskets from that of the MBA have been computed, and applied to the MBA' indigent line to obtain the set of indigent lines for five other regions. Engel coefficients from the regional data yield the corresponding poverty lines for the regions. There are some unexpected results. For instance, Table 4 shows that the food share is significantly higher in the MBA than in all other regions. Moreover, poverty lines for the North West and Patagonia regions are more than 5% higher than for the MBA. And the new indigent line for the MBA is substantially higher than the current indigent line than would be justified by increase in nutritional allowances from 2,700 to 2,750 Calories. 5 SIEMPRO, Avances de las estimaciones de linea de pobreza, (documento preliminar), 1999. This result contains some new features. It added wines and the value of eating out to the list of basic food basket to reflect the changing pattern of food consumption among the poor. It also presents poverty lines both with and without imputed rents for owner-occupied housing included in their non-food components. 9 Argentina Poverty Assessment Background Paper No. 1 Table 4: Preliminary Regional Poverty Lines Based on the 1996/97 ENHGO by Region, 1996 Indigent line Non-foods Poverty line Implicit Engel 1/ Coefficient (September 1996) Metro Buenos Aires 82.4 89.8 172.2 0.478 North West 80.7 100.9 181.6 0.444 North East 78.0 92.0 170.0 0.459 Cuyo 74.6 104.4 179.0 0.417 Pampeana 79.3 91.3 170.6 0.465 Patagonia 78.7 106.2 184.9 0.426 Sources: SIEMPRO, Avances de las estimaciones de linea depobreza, (docunento preliminar), 1999. INDEC, EncuestaNacional de Ingresos y Gastos (ENGHO), February 1996-February 1997. Note: 1/ The value of each food item contained in the basic food basket for the MBA was calculated by applying the unit item price that comes from INDEC's CPI unit, whereas for other regions each region's difference in value of the basic food basket with respect to that of the MBA for the reference population group was applied to obtain respective indigent lines. Consistent Poverty Lines: Synthesis Poverty lines for this study were elaborated by combining information obtained in previous sections. Food and non-food component of the poverty lines in MBA was updated by the price index of basic food basket and the non-food price index for the poor, respectively. In September 1996, the indigent line was 67.4 Pesos, and the poverty line was given by 155.6 Pesos per adult equivalent, with the food share of 0.433. And the same indexing method was applied to other survey periods. The indigent line of September 1996 then served as the welfare anchor for computing regional poverty lines. With the indigent line of 67.4 Pesos for the MBA as the anchor, the coefficients of difference of regional indigent lines from the MBA's indigent line in the ENGHO data have been applied to the MBA's indigent line to obtain the corresponding indigent line for each region. For instance, the indigent line was given by 66.0 Pesos for the North West, and 64.4 Pesos for Patagonia. Table 5 reports all the indigent lines, the food shares and the corresponding poverty lines for September 1999, where the food shares for all five regions except the MBA are taken from the results from the preliminary poverty lines. In order to carry out poverty comparison over time, it is necessary to update regional poverty lines by appropriate price indices. For this, one needs to have either regional price indices of basic foods/non-foods, or at least some implicit prices in the survey data. In Argentina, however, neither is available over time. The Income and Expenditure Survey is implemented only once in every 10 years, and there is no price index work for other areas outside the MBA. Therefore, in order to carry out poverty comparison over time for other regions, the price index of the basic food basket and price index of non-food CPI for the poor of the MBA were applied to all other regions in updating their corresponding indigent lines and the non-food component of the poverty lines, respectively. This assumes that evolution of prices in the MBA is parallel to that in other regions. There might be some temporary 10 Argentina Poverty Assessment Background Paper No. I short-term deviations in prices of some items among different regions. However, given that our objective is poverty comparison over an extended period of time, this study will adopt the working assumption that regional prices would revert back to the long-term trend as represented by the MBA prices. Table 5 presents the indigent lines and poverty lines corresponding to September 1996 for all regions. This shows that poverty lines in other regions are lower than in the Metropolitan Buenos Aires. A full sets of indigent lines and poverty lines are reported at Annex Tables 6-9, and they have been used to conduct poverty comparison throughout this study. Table 5: Adult Equivalent Poverty Lines and Average Per Capita Household Incomes, Sept. 1996 Indigent Non- Poverty Implicit Per Capita Line Foods Lines Engel Per Capita Income Consumption I I_ Coeff. (September 1996) EPH EDS ENGHO ENGHO 1996 1997 1996-97 1996-97 Overall 292.4 284.8 292.6 240.2 Metro Buenos Aires 67.4 88.2 155.6 0.433 342.2 367.9 348.8 292.4 North West 66.0 82.5 148.5 0.444 191.6 208.1 197.6 172.6 North East 63.8 75.3 139.1 0.459 183.0 201.5 199.4 159.5 Cuyo 61.0 85.4 146.4 0.417 242.0 258.2 237.3 196.7 Pampeana 64.9 74.6 139.4 0.465 254.3 260.1 280.2 222.5 Patagonia 64.4 86.9 151.3 0.426 339.4 355.1 310.5 236.0 Sources: INDEC, Encuesta Permanente de Hogares (EPH), October 1997. SIEMPRO/ INDEC, Encuesta de Desarrollo Social (EDS), August 1997. INDEC, Encuesta Nacional de Ingresos y Gastos (ENGHO), February 1996-February 1997. 11 Argentina Poverty Assessment Background Paper No. 1 B. Counting the Poor The basic unit of observation for this study is an individual. Even though households are a common unit of sharing income and consumption, measurement of welfare ultimately rests with individuals. Once poverty lines have been defined on a per capita basis for a male adult, the adult equivalent poverty lines may be applied to the household survey data set only after taking into account the composition of the population in terms of different nutritional requirements. Adult Equivalence Scale There are two ways for handling adjustment of indigent lines for an adult male to reflect the socio-demographic composition of population in terms of age and gender, and to account for differences in the type and level of activities. The first method is to estimate the adult equivalence adjustment factor for the population as a whole by computing the weighted average of adult equivalence coefficients corresponding to all sub-groups of population, where the corresponding weight is given by the population share of each sub-group. For instance, the value of adjusted food basket for the average population, C, is given by: C = CA1 Wj*Xj, (1) i where CA is the value of food basket for adult male (as in Annex Table B.1), x is adult equivalence coefficient for population subgroup j (see Annex Table B.2), and wj is its population share. In the current poverty lines, the average adult equivalence scale was estimated to be 0.8125. There is an alternative, more sophisticated method. And this study employs this method. In this approach, instead of deriving a single average poverty line for the whole population, one defines different indigent lines and poverty lines for each household, taking into account the composition and characteristics of the concerned families. Starting from the basic food basket for an adult male, the indigent line for a family with a certain composition is given by: C = CA1 Xij, (2) 1 where xj is each individual's adult equivalence coefficient, i indexes each family member in the household and j indexes the characteristics of the population subgroup. (See Annex Table 2 for the matrix of adult equivalence coefficients for Argentina; CEPA, 1993). Note that Y xj will vary among different families of the same size, depending on their composition. It is the adult equivalence scale-adjusted family size. We can express Equation (2) in per capita terms by dividing both sides by adult equivalence adjusted family size (E xj). In poverty analysis, adult equivalent indigent lines would be 12 Argentina Poverty Assessment Background Paper No. I juxtaposed with the adult equivalence per capita income, i.e., total household income divided by the adult equivalence family size, to determine who are the poor. Poverty Measure In counting the number of the poor, we utilize the concept of headcount of the indigent and the poor to measure the extent and incidence of poverty. We also use the poverty gap and the higher-order Foster-Greere-Thorbecke (FGT) indicators to measure to depth and severity of poverty. Per Capita Income for Income Distribution In carrying out analysis of income distribution, the traditional concept of per capita household income was utilized in order to maintain comparability of findings with the existing literature on income inequality.6 C. Limits of the Available Data There are three official household surveys in Argentina which can be utilized to carry out poverty analysis. Each survey is designed to achieve a certain set of specific objectives, e.g., estimating official unemployment rate for the Permanent Household Survey (EPH) and updating weights for the Consumer Price Index in the case of National Income and Expenditure Survey (ENGHO). The Social Development Survey was designed to study the living conditions and utilization of public social programs. Therefore, each has its own limitation when it is to be useful for helping estimate the incidence and profile of poverty in Argentina, as we examine in what follows. However, one of the most glaring weakness of the official program of household surveys in Argentina is that it lacks in the national sample frame, where the most updated and comprehensive sample frame still continues to exclude rural population (13%) and inhabitants of semi-urban areas of population size between 2,000 and 5,000 (2%). So Argentina will enter the next millennium without a national household survey. Permanent Household Survey (EPH) INDEC's Permanent Household Survey (Encuesta Permanente de Hogares-EPH) is a semi-annual household survey of basically urban employment which has been carried out since 1972. The survey data are readily available, and are the primary source of generating official unemployment rates. The EPH is administered in 28 major provincial cities including the Metropolitan Buenos Aires (about 60% of the Argentine population, or 70% of the urban population). With 87% of the Argentine population urban, this means that the EPH excludes not only 6 Use of either the traditional per capita household income or adult equivalent measure made immaterial difference in the estimates of Gini coefficients of income inequality, where the differences ranged anywhere between 0.2 to 1.4 percentage points (See the Annex Table 11 for the results). 7 In Argentina, urban areas are defined as the locality known by a commonly recognized name by its resident that has 2,000 or more inhabitants. 13 Argentina Poverty Assessment Background Paper No. I all rural population (13% of the overall population), but also inhabitants of small-cities as well as semi-urban areas (23%). 8 No matter how incomplete its geographic coverage is, the EPH survey is the only one of its kind in Argentina which has been implemented continuously over time. It is often alleged among researchers and policymakers that the income data in the EPH suffer from the problem of underreporting. However, underreporting is likely to be highest among the rich, and may not materially affect poverty estimates made using the EPH data set. For these reasons, the EPH data have been most heavily exploited by researchers to study poverty in Argentina. National Household Expenditure Survey. 1996-97 The National Household Survey of Income and Expenditure (Encuesta Nacional de Gastos de los Hogares-ENGHO) was carried out by INDEC during February 1996 - February 1997. The survey was administered for all urban areas with population of more than 5,000 inhabitants which covers about 84% of the Argentine population (114 urban centers). While the principal objective of the survey is to update the weights for the CPI, the data set is rich with detailed information on household characteristics, income, consumption, and utilization of public services. Therefore, the ENGHO data are an ideal source of information to produce detailed profile of poverty and households' living conditions Social Development Survey, 1997 In 1995, the Social Development Secretariat created, with the support of a World Bank operation, a project unit SIEMPRO in charge of poverty analysis, and evaluation and monitoring of social programs, among other things. In order to strengthen its capacity to carry out analytical work on poverty, SIEMPRO implemented the Household Survey of Living Conditions and Access to Social Services in 1997 (hereafter called the 1997 Social Development Survey - Encuesta de Desarrollo Social (EDS)) in collaboration with INDEC. The fieldwork was carried out in August of 1997, and the survey data contain a rich set of information on access to education, health, other public infrastructure services, and targeted social programs as well as improved income variable. 9 Comparison of the Survey Data The quality of statistical information from household surveys can vary widely according to sample size, coverage, non-response rate, the extent of non-reporting or under-reporting of income, etc. However, a majority of these differences are the results of choices and trade-off made regarding deployment of resources during various stages in the design and implementation of the surveys. 8 Rural Living Standard (LSMS) Survey Supported by the Bank's Loan entitled Argentina: Small Farmer Development Project, the Agricultural Secretariat developed and implemented an LSMS-type Rural Household Survey of Living Conditions in the two northern provinces of Salta and Misiones in April/May of 1996. A similar survey is planned for one province each in three regions, Cuyo, Pampas and Patagonia towards the end of CY1998. Even though an in-depth analysis of the 1996 data by Bank staff confirmed its high utility for quantifying living conditions among the rural population, the coverage remains a tiny fraction of rural population. 9 Recently SIEMPRO posted a sample of results from the EDS data at the Web site, "http://www.siempro.org.ar". 14 Argentina Poverty Assessment Background Paper No. 1 Table 6: Comparison of Earnings by Income Source Proportion of Family Income which Comes from EPH (Oct 1996) ENGHO 96-97 EDS 1997 Overall (100%) 100.0 100.0 100.0 Wages and Salaries 57.2 54.0 59.6 Self-Employment 15.9 17.5 17.0 Retirement Income 16.4 15.9 15.1 Capital Income 1.1 2.3 Other Incomes 9.0 10.2 6.7 - Business Owners (6.2) (6.7) (5.8) - Transfer Income (1.6) (3.5) (0.5) - Miscellaneous (1.2) (..) (0.4) Sources: INDEC, Encuesta Permanente de Hogares (EPH), October 1996. INDEC, Encuesta Nacional de Ingresos y Gastos (ENGHO), February 1996-February 1997. SIEMPRO/ INDEC, Encuesta de Desarrollo Social (EDS), August 1997. Table 6 compares the distribution of family earnings by income sources from three survey data for similar time periods. Income data from these surveys are comparable, in that they measure monetary income net of taxes and do not include market value of home production for own consumption. Given that they are urban surveys, the latter category might not be all that important. It is often alleged that the EPH data are deficient in various aspects, e.g., high non-response rate, income underreporting, etc. However, it appears that the functional distribution of income is very similar to each other. Combining the EPH Data A cursory examination of key variables presented in Table 7 suggests that the EPH is a stable source of statistical information to carry out analysis of poverty and income distribution for this study. Indeed, the Permanent Household Survey (Encuesta Permanente de Hogares-EPH) has been the principal source of information in tracing the changes in poverty which took place during the last 20 years in Argentina. However, most of previous poverty studies in Argentina have utilized the EPH data only from the MBA, since they are the most readily available on a consistent basis. This study significantly expands the population coverage from the more affluent 33% of the Argentines living in the Metropolitan Buenos Aires (MBA) to approximately 60% of the entire population, in particular during the 1990s. This has been achieved by combining the data sets from different survey periods after standardizing different data formats and heterogeneous variable definitions. Annex Table 1 shows the profile of the EPH data sets from 1990-1998 that have been utilized to construct the combined master data file for this study. 15 Argentina Poverty Assessment Background Paper No. 1 Table 7: Key Variables of the Permanent Household Survey (EPH), 1980-1998 Average Years of % Households with Average Per Capita Median Per Capita Schooling among No Income Household Income Household Income Adults Aged 18 and Above Survey Num. of Urban Metro All Metro All Urban Metro Metro Year Period Agglomerations Buenos Urban Buenos Areas Buenos All Urban Buenos All Urban (% Population) Aires Areas Aires Aires Areas Aires Areas (September 1998 Price) 1980 October 1(33.4) 22.7 -- 479.9 -- 365.4 -- 8.0 -- 1985 October 1(33.4) 21.5 -- 380.0 -- 284.6 -- 8.4 -- 1988 May 1(33.4) 16.6 -- 345.8 -- 226.6 - 8.8 -- October 1(33.4) 19.4 -- 306.1 -- 215.0 -- 8.8 -- 1989 May 1(33.4) 19.4 -- 364.3 -- 234.5 -- 8.8 -- October 1(33.4) 25.8 -- 259.8 -- 151.1 - 8.8 -- 1990 May 20(56.1) 28.2 20.3 258.1 229.4 171.2 136.9 8.9 9.0 October 24(58.4) 26.6 18.8 273.0 244.8 184.9 149.1 8.9 9.1 1991 May 25 (58.7) 27.9 19.3 284.7 248.3 188.1 151.6 9.0 9.2 October 24 (58.4) 25.9 18.5 319.3 276.9 208.0 161.8 9.0 9.2 1992 May 24 (57.9) 28.5 20.6 323.9 281.8 223.9 165.9 9.0 9.2 October 25 (58.7) 20.1 15.0 350.3 299.7 235.0 173.9 9.2 9.3 1993 May 25 (58.7) 13.4 11.5 374.1 318.6 249.5 178.2 9.4 9.4 October 25(58.7) 13.9 12.0 361.9 314.3 251.9 178.1 9.3 9.4 1994 May 25 (58.7) 9.9 9.9 381.3 330.1 266.9 181.5 9.4 9.5 October 25(58.7) 12.0 11.1 356.5 313.3 239.4 177.0 9.3 9.4 1995 May 25(58.7) 11.6 9.8 334.2 292.1 224.5 170.5 9.5 9.6 October 28(61.1) 12.1 10.3 338.9 288.0 216.5 163.0 9.6 9.6 1996 May 28(61.1) 12.2 10.9 327.0 281.6 204.5 163.6 9.7 9.7 October 27 (57.5) 14.2 13.2 329.4 282.1 205.4 155.1 9.8 9.7 1997 May 28(61.1) 8.0 8.2 335.9 288.3 215.9 165.1 9.8 9.8 October 28 (61.1) 11.1 10.6 345.9 295.7 225.8 166.8 9.8 9.8 1998 May 28(61.1) 11.1 10.5 365.2 310.4 224.0 169.1 9.8 9.8 October 27(60.9) 10.0 9.9 359.5 304.6 216.7 160.0 9.8 9.9 Sources: INDEC, Encuesta Permanente de Hogares (EPH), 1980-1998. 16 Argentina Poverty Assessment Background Paper No. I III. DIMENSIONS OF POVERTY Poverty is a persistent and recurring problem in Argentina, and poverty rate is relatively high for its income level. This chapter gives an overview of the magnitude and distribution of poverty situation in Argentina. A. Snapshot of Poverty According to the latest estimates, about 29% of Argentines (21% of households) are living in poverty (EPH 1998). There are also important regional differences in poverty, as shown in Table 8 for the 1996 and 1997 data that are most comparable. Table 8: A Snapshot of Poverty in Argentina by Region, 1996 and 1997 Per Capita % Population % Households Household Income Below Below Below Below /Consumption Indigent Lines Poverty Lines Indigent Lines Poverty Lines Income Poverty EPH, October 1996 Overall 277.4 7.8 30.7 4.8 21.8 Metro Buenos Aires 323.9 5.6 26.0 3.4 18.0 North West 173.5 15.5 48.4 11.0 39.9 North East 167.7 17.8, 49.3 11.8 38.2 Cuyo 215.3 8.4 39.2 6.1 30.5 Pampeana 235.4 8.2 28.5 4.8 20.0 Patagonia 327.8 5.5 21.9 4.3 17.2 ENGHO, Feb. 1996 - Feb. 1997 Overall 292.6 6.7 27.3 4.1 18.9 Metro Buenos Aires 348.8 4.3 21.4 2.4 13.8 North West 197.6 11.9 43.8 8.7 34.8 North East 199.4 13.6 45.6 9.4 35.2 Cuyo 237.3 7.3 34.0 5.2 35.6 Pampeana 280.2 6.3 24.1 4.0 17.1 Pata.gonia 310.5 7.6 28.3 4.9 22.1 EDS, August 1997 Overall 295.3 8.1 30.3 5.1 21.2 Metro Buenos Aires 395.3 5.6 23.4 3.3 15.1 North West 189.2 13.6 47.4 10.4 37.9 North East 184.6 13.2 48.1 9.6 38.0 Cuyo 223.5 5.0 35.7 3.3 26.2 Pampeana 260.1 8.4 27.4 5.1 19.9 Patagonia 303.6 8.3 29.5 5.9 21.9 Consumption Poverty ENGHO, Feb. 1996 - Feb. 1997 Overall 240.2 9.9 35.0 6.3 26.0 Metro Buenos Aires 292.4 6.9 30.0 4.1 21.1 North West 172.6 15.5 49.5 11.0 39.9 North East 159.5 20.4 53.1 14.2 42.2 Cuyo 196.7 10.1 41.4 6.9 32.8 Pampeana 222.5 9.1 31.5 6.0 24.2 Patagonia 236.0 13.1 36.8 9.5 29.6 Sources: INDEC, Encuesta Permanente de Hogares (EPH), October 1996. INDEC, Encuesta Nacional de Ingresos y Gastos (ENGHO), February 1996-February 1997. SIEMPRO/ INDEC, Encuesta de Desarrollo Social (EDS), August 1997. 17 Argentina Poverty Assessment Background Paper No. 1 Poverty rates are substantially higher in areas in the north, particularly in the North West (48.4% in 1996) and North East (49.3% in 1996). Conversely the Metropolitan Buenos Aires and Patagonia are lower than the national average. B. Trends in Poverty Since 1990, there has been a substantial reduction in poverty, with the overall poverty headcount ratio for all urban areas dropping from 42% in 1990 to 29% in 1998, extreme poverty from 10% to 7%. However, this longer period can be subdivided to two periods. During 1990-94 following the 1991 Convertibility Plan, poverty rate fell to almost 22%, and the extreme poverty to 4%. Things have reversed in the aftermath of the Table 9: Changes in Urban Poverty in Argentina, 1980-1998 % Pop lation % Households Below Indigent Lines Below Poverty Lines Below Indigent Lines Below Poverty Lines Survey Metro All Metro All Metro All Metro All Year Period Buenos Urban Buenos Urban Buenos Urban Buenos Urban Aires Areas Aires Areas Aires Areas Aires Areas 1980 October 1.2 -- 8.0 -- 0.6 -- 5.4 -- 1985 October 2.9 -- 16.0 -- 1.7 -- 11.7 -- 1988 May 7.7 -- 30.9 -- 4.6 -- 23.2 -- October 10.1 -- 35.4 -- 6.3 -- 26.6 -- 1989 May 7.0 -- 29.4 -- 4.5 -- 22.5 -- October 15.5 -- 46.8 -- 10.2 -- 37.3 -- 1990 May 10.7 12.6 41.4 43.6 6.7 8.3 32.3 34.3 October 10.1 10.0 41.0 39.4 8.3 7.7 33.7 31.8 1991 May 4.3 7.4 30.1 33.8 2.6 5.1 22.4 25.9 October 2.6 4.5 22.7 27.1 1.5 2.7 16.8 20.0 1992 May 2.6 4.4 19.7 25.1 1.4 2.6 14.9 18.8 October 2.2 4.4 17.6 23.3 1.3 2.6 12.9 17.1 1993 May 2.5 4.1 17.0 21.8 1.6 2.5 12.6 15.9 October 3.6 4.5 16.8 21.7 2.1 2.7 12.4 15.7 1994 May 2.3 3.7 15.5 20.3 1.5 2.2 11.1 14.4 October 2.4 3.8 18.5 22.8 1.5 2.3 13.2 16.2 1995 May 4.2 5.5 21.4 25.9 2.5 3.3 15.1 18.4 October 4.9 6.6 23.7 28.4 2.9 4.0 16.9 20.4 1996 May 5.0 7.0 25.0 29.4 3.0 4.3 17.6 20.9 October 5.6 7.8 26.0 30.7 3.4 4.8 18.0 21.8 1997 May 4.8 6.6 25.6 29.8 3.1 4.2 18.0 21.5 October 5.1 7.1 24.7 29.0 3.5 4.6 17.6 20.9 1998 May 4.5 6.5 24.2 28.8 3.0 4.1 17.2 20.6 October 6.0 7.6 25.6 29.9 3.7 4.7 17.8 21.3 Sources: INDEC, Encuesta Permanente de Hogares (EPH), 1980-1998. 18 Argentina Poverty Assessment Background Paper No. 1 Mexican Pesos crisis, with poverty rate climbing up to around 30% in urban areas. In 1998, about 29% of Argentines are living below poverty lines, and one of every five families are living in poverty. Other standard measures of poverty also show a similar trend. Incidence, Depth, and Severity ofPoverty Table 10 shows that depth and severity of poverty among the poor were moving with the poverty trend. Given that there were about 32.2 million people in urban Argentina in 1998, this means that 9.4 million were living in poverty and 2.3 million living in extreme poverty. Poverty gap index in 1998 was 11.2%, and increase in income needed by the average poor person to get out of poverty (the income gap) was about 38%. Calculating the income gap as a share of GDP, wee see that about 4.5% of GDP would have been sufficient to eradicate poverty, i.e., bring everyone up to the poverty line. Table 10: Poverty Rates, Poverty Gap, and Severity of Poverty in Urban Argentina, 1990-1998 Survey Poverty Extreme Poverty Severity of Income Income Year Period Headcount Poverty Gap Poverty Gap Gap/GDP (FGT c=1) (FGT 0x=2) 1990 May 43.6 12.6 17.9 9.9 41.1 10.5% October 39.4 10.0 15.1 7.8 38.3 9.8% 1991 May 33.8 7.4 13.4 7.7 39.6 6.2% October 27.1 4.5 8.9 4.2 32.8 5.2% 1992 May 25.1 4.4 8.1 3.8 32.3 3.7% October 23.3 4.4 7.6 3.6 32.6 3.7% 1993 May 21.8 4.1 7.3 3.5 33.5 3.3% October 21.7 4.5 7.6 3.7 35.0 3.4% 1994 May 20.3 3.7 6.6 3.2 32.5 2.9% October 22.8 3.8 7.7 3.7 33.8 3.1% 1995 May 25.9 5.5 9.3 4.7 35.9 4.3% October 28.4 6.6 10.4 5.4 36.6 4.4% 1996 May 29.4 7.0 11.0 5.8 37.4 4.9% October 30.7 7.8 11.7 6.2 38.1 5.0% 1997 May 29.8 6.6 11.3 5.9 37.9 4.5% October 29.0 7.1 11.0 5.8 37.9 4.5% 1998 May 28.8 6.5 10.7 5.6 37.2 4.4% October 29.9 7.6 11.6 6.2 38.8 4.6% Sources: INDEC, Encuesta Permanente de Hogares (EPH), 1990-1998. Employment and Poverty Table 11 shows that participation rate of the non-poor are substantially higher than that of the poor and the extremely poor. While the poor and the extremely poor are less likely to be in the labor market, they are more likely to be out of work. Even before the May 1995 sharp rise in overall unemployment, there had been high incidence of unemployment among the poor, in particular among the indigent poor. For most of the 1990s, at least one of out five indigent poor workers were out of work, and one of out two were unemployed at the height of high unemployment episode of May 1995. On the 19 Argentina Poverty Assessment Background Paper No. I other hand, the unemployment rate among the non-poor stayed below 12% almost all this period. Table 11: Labor Market Participation and Poverty in Argentina, 1990-1998 Labor Force Participation (population of age 15-64) Unemployment Rate Year Survey Indigent Moderate Non-poor All Urban Indigent Moderate Non-poor All Urban Period poor Areas - poor Areas 1990 May 55.7 53.2 65.8 61.1 23.3 13.2 4.4 8.5 October 49.6 54.1 65.4 61.2 18.4 9.8 3.7 6.1 1991 May 54.1 50.1 65.4 61.1 17.7 12.5 4.9 7.1 October 48.8 49.8 64.7 61.3 23.3 11.2 4.1 5.7 1992 May 47.7 50.3 65.0 61.8 24.7 13.6 4.9 6.6 October 51.3 51.2 65.3 62.6 23.8 16.4 5.0 7.0 1993 May 49.9 53.3 66.6 64.1 34.3 19.6 7.4 9.6 October 49.4 51.8 65.4 62.8 33.0 18.6 7.0 9.1 1994 May 53.3 53.3 65.9 63.8 37.7 22.6 7.9 10.3 October 49.2 54.4 65.8 63.5 39.2 24.5 8.7 11.5 1995 May 55.9 59.0 67.5 65.5 48.1 31.9 13.9 18.1 October 54.7 56.7 66.4 63.9 41.1 27.6 12.0 16.0 1996 May 55.8 56.5 65.5 63.1 42.4 28.8 11.8 16.4 October 56.9 57.7 66.4 64.0 43.3 27.8 12.0 16.8 1997 May 59.9 57.3 67.1 64.7 45.2 26.9 11.5 15.9 October 55.3 56.5 67.2 64.4 35.3 23.4 9.8 13.4 1998 May 54.7 56.6 67.6 64.8 34.3 23.5 9.5 13.0 October 54.9 56.2 67.1 64.2 38.5 21.7 8.4 12.2 Sources: INDEC, Encuesta Permanente de Hogares (EPH), 1990-1998. Clearly both employment and wages are important for poverty reduction. But how much would poverty fall if there were no unemployment? To answer this, we first estimated a Mincerian earnings equation for employed workers of the form: In (earnings from the primary job) = f (years of schooling, years of schooling squared, experience [=age-s-6], experience squared, female dummy, regional dummies) Separate equations were estimated with the semi-annual data for each year, 1990- 98. Using these equations, we then predict earnings for the unemployed workers."o This "full employment" income is then combined with other sources of household income to recompute average income per adult equivalent, and the numbers below the poverty line, as well as the poverty gap (FGT oL=1), and squared poverty gap (FGT ca=2), are then reestimated (See the Annex Table 10 for the regression estimates). The results are given in the tables below. 10 In order to guard against undue influence of outliers, however minor their frequency might be, we have applied the robust estimate technique. Predicted earnings of the unemployed have been compared to the distribution of the earnings variable in the semi-annual pooled time-series and cross-section data. Whenever the value of predicted earnings exceeded twice the standard errors of the earnings variable in September 1998 price, the value was set to this value. 20 Argentina Poverty Assessment Background Paper No. 1 Table 12: Escaping Poverty Through Employment, 1990-1998 From Being An Indigent From Being a Moderately Poor Non-poor Unemploy Staying Becoming Escaping Staying Escaping Year-Mo -ment rate Indigent Moderately Poverty Moderately Poverty Non-poor Poor Altogether Poor 1990-May 8.5 12.6 9.4 2.6 0.6 31.0 28.1 2.9 56.4 1990-Oct 6.1 10.0 8.0 1.5 0.5 39.4 26.7 2.7 60.6 1991-May 7.1 7.4 6.0 1.0 0.4 26.4 23.8 2.7 66.2 1991-Oct 5.7 4.5 3.4 0.7 0.4 22.6 20.3 2.3 72.9 1992-May 6.6 4.4 3.3 0.4 0.4 20.7 18.0 2.7 74.9 1992-Oct 7.0 4.4 3.3 0.7 0.4 18.9 15.8 3.1 76.7 1993-May 9.6 4.1 2.8 0.7 0.6 17.7 14.2 3.6 78.2 1993-Oct 9.1 4.5 3.0 1.0 0.5 17.2 14.1 3.2 78.3 1994-May 10.3 3.7 2.3 0.8 0.6 16.6 12.9 3.7 79.7 1994-Oct 11.5 3.8 2.4 0.8 0.6 19.0 14.6 4.4 77.2 1995-May 18.1 5.5 2.6 1.7 1.2 20.4 13.5 6.9 74.1 1995-Oct 16.0 6.6 3.5 2.0 1.1 21.8 15.7 6.0 71.7 1996-May 16.4 7.0 3.7 1.9 1.4 22.4 16.0 6.4 70.6 1996-Oct 16.8 7.8 4.1 2.4 1.3 22.9 16.5 6.5 69.3 1997-May 15.9 6.6 3.2 2.3 1.1 23.2 17.2 6.1 70.2 1997-Oct 13.4 7.1 4.3 1.9 0.9 21.9 16.8 5.1 71.0 1998-May 13.0 6.5 4.1 1.6 1.8 22.3 16.9 5.3 71.2 1998-Oct 12.2 7.6 4.5 2.1 1.0 22.3 17.4 4.9 70.1 Sources: Caculated from the EPH surveys, average of May and October. Table 12 shows how "full employment" affects both the moderately poor and the indigent, while Table 13 gives the adjusted poverty headcount, poverty gap and poverty depth (squared poverty gap) estimates. For 1998, the overall poverty rate of 29.4% would become 23.3% under fall employment, with the extreme poverty falling from 7.1% to 4.3%. The fall in extreme poverty is made up of 1.9% of the labor force moving from extreme to moderate poverty, and 0.9% moving from extreme poverty to the non-poor. These estimates make the extreme assumption of zero unemployment; however, even with a 50% drop in unemployment, poverty would drop by 3 percentage points, more than a 10% reduction in poverty. These estimates also assume that more unemployment is attained at current real wages. If full employment is attained through greater labor market flexibility, instead of growth in demand, real wages would fall, reducing the overall effect. Not only would some of the recently employed not escape poverty, but it is possible that some of the non-poor would become poor if real wages fell. While we can see that full employment, particularly if generated by rising labor demand, would have major effect in reducing the overall poverty rate, it is also true that full employment by itself is not enough. Given the characteristics of the poor, in terms of their education and skills, even fully employed they would not earn enough to raise themselves over the poverty line. 21 Argentina Poverty Assessment Background Paper No. I Table 13: Simulation of Full Employment Impacts on Poverty Measures, 1990-1998 Poverty Gap Severity of Poverty Poverty Headcount Extreme Poverty (FGT t= 1) (FGT o=2) Unemploy If Fully If Fully If Fully If Fully Year-Mo -ment rate Actual Employed Actual Employed Actual Employed Actual Employed 1990-May 8.5 43.6 40.2 12.6 9.4 17.9 15.1 9.9 7.8 1990-Oct 6.1 39.4 36.2 10.0 8.0 15.1 13.1 7.8 6.5 1991-May 7.1 33.8 30.8 7.4 6.0 13.4 11.6 7.7 6.5 1991-Oct 5.7 27.1 24.5 4.5 3.4 8.9 7.6 4.2 3.5 1992-May 6.6 25.1 22.0 4.4 3.3 8.1 6.7 3.8 3.0 1992-Oct 7.0 23.3 19.8 4.4 3.3 7.6 6.2 3.6 2.9 1993-May 9.6 21.8 17.7 4.1 2.8 7.3 5.5 3.5 2.5 1993-Oct 9.1 21.7 18.0 4.5 3.0 7.6 5.8 3.7 2.6 1994-May 10.3 20.3 16.0 3.7 2.3 6.6 4.8 3.2 2.2 1994-Oct 11.5 22.8 17.8 3.8 2.4 7.7 5.5 3.7 2.4 1995-May 18.1 25.9 17.8 5.5 2.6 9.3 5.6 4.7 2.6 1995-Oct 16.0 28.4 21.3 6.6 3.5 10.4 6.8 5.4 3.3 1996-May 16.4 29.4 21.6 7.0 3.7 11.0 7.1 5.8 3.4 1996-Oct 16.8 30.7 23.0 7.8 4.1 11.7 7.6 6.2 3.6 1997-May 15.9 29.8 22.7 6.6 3.2 11.3 7.4 5.9 3.4 1997-Oct 13.4 29.0 23.0 7.1 4.3 11.0 7.8 5.8 3.9 1998-May 13.0 28.8 22.7 6.5 4.1 10.7 7.7 5.6 3.8 1998-Oct 12.2 29.9 24.0 7.6 4.5 11.3 8.4 6.2 4.1 Sources: Caculated from the EPH surveys, average of May and October. Regional Poverty There are also important regional differences in poverty. As shown in Table 14, poverty rates are substantially higher in areas in the north, particularly in the North West (46% in 1998) and North East (49%). Conversely Buenos Aires and Patagonia are lower than the national average. The period 1990-98 can be divided into two sub periods: 1990-94, when poverty decreased, and 1994-98, when it increased. In the first period, overall poverty rates fell by 48%, but poverty in the two poorest regions fell by about half of that. In other words, the poorest areas seem to benefit less from general poverty reduction than the richer areas. However, when poverty rose during the 1994-98 period, the increases in poverty were greater in areas, outside of the Northeast and Northwest, particularly in Buenos Aires, Cuyo and Pampeana. Thus, it appears that the poorer areas are more stable with regard to poverty in times of economic changes, and the impact of macroeconomic crises seem to occur in the richer areas. It seems possible that the areas outside of Buenos Aires are more dependent on agriculture, and therefore less susceptible to economic shocks. 22 Argentina Poverty Assessment Background Paper No. 1 Table 14: Poverty Rates by Region, 1990-1998 Survey North North All Year Period MBA West East Cuyo Pampeana Patagonia Regions 1990 May 41.4 59.2 58.6 52.7 38.9 37.8 43.6 October 41.0 49.6 52.8 43.4 28.4 15.5 39.4 1991 May 30.1 57.9 51.1 36.3 27.9 23.8 33.8 October 22.7 45.0 46.7 34.3 22.8 17.9 27.1 1992 May 19.7 42.7 45.3 32.3 23.4 19.2 25.1 October 17.6 43.5 43.9 28.5 21.8 17.4 23.3 1993 May 17.0 40.1 41.6 28.9 20.0 17.3 21.8 October 16.8 41.5 40.8 28.1 19.9 15.9 21.7 1994 May 15.5 40.7 39.3 24.8 19.3 16.1 20.3 October 18.5 42.4 41.3 27.3 20.3 18.1 22.8 1995 May 21.4 41.3 43.7 33.4 25.1 18.6 25.9 October 23.7 45.8 44.7 39.2 25.5 21.3 28.4 1996 May 25.0 48.0 45.7 34.0 27.4 19.8 29.4 October 26.0 48.6 49.3 39.2 28.5 21.9 30.7 1997 May 25.6 44.6 48.4 39.4 27.6 21.8 29.8 October 24.7 46.1 47.7 34.5 26.4 21.9 29.0 1998 May 24.2 45.4 48.2 36.0 27.3 20.7 28.8 October 25.6 46.6 49.3 36.0 27.5 24.1 29.9 % Change: 1990-94 -58.7% -23.6% -27.6% -45.8% -41.2% -35.8% -48.1% 1994-98 46.5% 10.7% 21.0% 38.2% 38.4% 31.0% 36.2% Population (millions) 11.8 2.9 2.0 1.9 9.4 1.3 29.4 Poor (millions), 1998 2.9 1.3 1.0 0.7 2.6 0.3 8.6 Sources: INDEC, Encuesta Permanente de Hogares (EPH), 1990-1998. 23 Argentina Poverty Assessment Background Paper No. 1 IV. POVERTY PROFILE The poor have characteristics that distinguish them from the non-poor, and may explain part of their failure to participate in economic growth. As summarized in Tables 15 to 17, the poor: (i) have significantly larger families (4.7 vs.3.1), (ii) have younger families with a much higher dependency ratio (3.0 vs. 1.4), (iii) have much higher unemployment rates (twice the rate of the non-poor), (iv)have fewer years of schooling (about 25% less), and (v) are more likely to work in the informal sector. The situation is clearly even more critical in these respects for the indigent. Table 15: Comparison of Basic Sample Characteristics of Household Surveys by Poverty Group Poverty Group The Indigent The Moderate Poor Poor The Non-Poor Total Average Age of the Sample Population EPH, 1998 20.9 25.1 34.7 31.6 ENGHO, 1996-87 20.1 24.2 34.1 31.1 EDS, 1997 20.3 24.1 34.0 30.7 Average Family Size EPH, 1998 5.5 4.7 3.1 3.5 ENGHO, 1996-87 5.8 5.0 3.2 3.6 EDS, 1997 5.8 5.0 3.2 3.6 Average Adult Equivalence Scale (AES) EPH, 1998 0.7723 0.7877 0.8084 0.7985 ENGHO, 1996-87 0.7696 0.7926 0.8023 0.7981 EDS, 1997 0.7684 0.7891 0.8023 0.7966 AES-Adjusted Family Size EPH, 1998 4.3 3.7 2.5 2.8 ENGHO, 1996-87 4.5 4.0 2.6 2.9 EDS, 1997 4.4 3.9 2.6 2.9 Sources: INDEC, Encuesta Permanente de Hogares (EPH), May and October, 1998. INDEC, Encuesta Nacional de Ingresos y Gastos (ENGHO), February 1996-February 1997. SIEMPRO/ INDEC, Encuesta de Desarrollo Social (EDS), August 1997. However, there does not appear to be much of a relationship between female headed households and poverty. Surprisingly, the poor have lower rate of labor force participation (among those aged 15-64), and work fewer hours per week. Among the indigent, the unemployment rate is a surprising 37%, which suggests that many of those classified as "indigent" are temporarily in this situation because of loss of employment. There is evidence that job turnover rate is higher among the poor in relative terms. For instance, about 30% of the non-poor who are actively looking for job have been in job marker more than 6 months, while the corresponding figure for the indigent is about 20%. 24 Argentina Poverty Assessment Background Paper No. 1 Table 16: Characteristics of Argentina's Urban Population by Poverty Group, 1998 EPH Poverty Group The Indigent The Moderate Poor Poor The Non-Poor Total % Female-Headed Households 26.8 21.0 27.7 26.6 Dependency Ratio (number of dependents/ 4.11 3.03 1.39 1.75 number of workers) Years of schooling among individuals with 6.9 7.8 10.5 9.9 ages 18 and more Labor Force Participation (Age 15-64) 54.8 56.4 67.3 64.5 Unemployment Rate 36.5 22.6 8.9 12.6 Among the Unemployed Unemployment Spell (Months) 1/ 100.0 100.0 100.0 100.0 Under 2 months 45.5 46.8 42.8 44.3 More than 2 months - 6 months 33.6 29.4 26.3 28.2 More than 6 months - 1 year 13.3 18.8 23.0 20.5 More than 1 year - 2 years 7.6 5.0 7.9 7.0 Among the Employed Number of Jobs Held 1.03 1.04 1.10 1.09 Total Hours Worked per Week 34.9 41.7 45.1 44.2 % Employed in the Informal Sector 46.4 44.6 36.2 38.2 Sources: INDEC, Encuesta Permanente de Hogares (EPH), October of 1997, May and October of 1998. Note: 1/ The figures for unemployment spell come from the EPH data of October 1997. All others were calculated using the pooled data of the EPH surveys of May and October of 1998. Age and Poverty. Because poor families are both larger and younger than average, children are more apt to be in poverty than adults. As shown in Table 17, 45% of all children (aged 0-14) were living in poverty in 1998, compared to a 25% poverty rate for adults. Poverty among older adults is below average. Only 13% of those 65 and older were found to be below the poverty line. Table 17: Poverty Rate by Age Group and Sex, (Urban Argentina, 1998) Total Population Female Population Age Group % Indigent % Poor % Indigent % Poor Total 7.1 29.4 7.2 30.1 0-4 12.2 43.2 11.5 43.0 5-14 13.0 45.3 12.5 44.4 Subtotal, 0-14 12.7 44.6 12.1 43.9 15-24 7.0 30.6 7.1 31.1 25-39 5.6 26.0 5.1 25.1 40-64 4.5 21.0 4.8 21.7 Subtotal, 15-64 5.6 25.4 5.6 25.6 65 & older 1.4 1 3.2 1 .7 15.2 Sources: INDEC, Encuesta Permanente de Hogares, Average of the EPH for May and October, 1998. 25 Argentina Poverty Assessment Background Paper No. I The lower rate of poverty among older people probably reflects the fact that the formal sector, which covers about 60% of the labor force ensures that retired workers receive adequate pensions. Poverty among women is not significantly different from the general population, with the possible exception of the 65 and over group, which is slightly poorer Household Head and Poverty When we look at the characteristics of the main bread winner of the households, the incidence of poverty is highly correlated with household heads' low education attainment, low job skill, and employment in the construction industry. Table 18: Characteristics of Household Heads by Poverty Group, 1998 EPH Poverty Group The Indigent The Moderate Poor Poor The Non-Poor Total Marital Status of Household Head 100.0 100.0 100.0 100.0 Single 7.1 3.7 6.4 5.9 Co-Habitation 29.8 22.3 10.9 14.8 Married 46.5 58.7 65.4 62.6 Separated/Divorced 9.3 7.7 7.1 7.4 Widow/er 7.3 7.5 10.2 9.4 Education Level of Household Head 100.0 100.0 100.0 100.0 No schooling 3.8 2.1 1.0 1.4 Primary incomplete 33.0 24.8 11.3 15.9 Primary completed 40.5 43.0 29.1 33.0 Secondary incomplete 16.7 18.6 18.8 18.6 Secondary completed 4.2 8.1 17.4 14.4 Tertiary incomplete 1.1 2.6 9.2 7.2 Tertiary completed 0.7 0.8 13.2 9.6 Job Skill Level of Household Head 100.0 100.0 100.0 100.40 Unskilled 40.7 31.5 15.8 20.8 Skilled 59.1 68.0 71.6 70.0 Professional 0.3 0.4 12.6 9.1 Industry of Household Head's Job 100.0 100.0 100.0 100.0 Agriculture and mining 1.9 1.7 1.1 1.3 Manufacturing 11.1 17.4 18.0 17.3 Public utilities 0.5 0.7 1.2 1.0 Construction 32.3 23.8 9.9 14.9 Commerce, Hotels, & Restaurants 14.5 15.2 16.5 16.1 Transport 6.6 8.7 12.6 11.2 Finance 3.2 3.5 10.2 8.1 Services 29.1 28.9 30.2 29.8 Unknown 0.9 0.2 0.2 0.2 Sources: INDEC, Encuesta Permanente de Hogares, Average of the EPH for May and October, 1998. 26 Argentina Poverty Assessment Background Paper No. 1 V. DETERMINANTS OF POVERTY The poverty profile in the previous section showed that there are wide variations in poverty rates among different sub-groups of population. A detailed description of population by poverty group helps a great deal in gaining an understanding of who is more likely to be poor. However, the two-dimensional poverty profile cannot answer the relative magnitude and significance of many different intervening variables which determine one's probability of being poor. There is only so much that we can say by looking at one-way or two-way contingency tables of partial correlation between poverty and other factors such as individual and household characteristics. Since all other things are not controlled for, the picture of poverty by this method will remain partial, and a more thorough understanding of poverty calls for simultaneously taking all factors into account by using multiple regression method. This section will attempt to shed light on "why" the poor become or stay poor. The probability of poverty can be specified as a reduced-form probit model as follows. Since poverty is defined as the shortfall of income from poverty line, the specification of the probit model follows. The probability of being poor is given by Ppoor = 1, iff yi<yp, or yi-yp<O and Ppoo, = 0, iff yi!yp, or yi-yp>0, where yi is per capita income of an individual and yp is a known poverty line.11 If the error term, &i=yi-yp, is normally distributed, the probability of being poor will be given by a probit model, Prob(poor-1) = Q(Xp), and XP will be the inverse function of standard normal of the probability, XP= (1(poor-l). " There is one importance difference between this probabilistic model of poverty and the usual probit model. While the underlying variable and its associated threshold level, yi and y, in our example, are usually not observable, or, when observable, at least are not available in the data. Only the outcomes of decision are known. But in our case, we know both the level of per capita income and poverty line. This can be used to argue that a probit model is an inefficient way of estimating the probability of being poor. Indeed there is an alternative way of estimating the incidence of poverty, say, regressing the level of income itself against a set of explanatory variables, obtaining a set of predicted values conditional on a certain set of hypothetical values of individual and fimily characteristics, and seeing if that particular individual is below or above poverty line. 27 Argentina Poverty Assessment Background Paper No. 1 The probability of being poor is dependent upon per capita income of an individual conditional on exogenously determined poverty lines, whereas income level in turn is a function of a set of individual and family characteristics, and other factors. Choice of independent variables reflects our judgement as to what are important determinants of income level, and so the incidence of poverty. A detailed profile of poverty in previous section offered us a list of prior information regarding the potential candidacy of variables for inclusion in the regression analysis. Consideration given are as follows: total family income, family size, life cycle, other family attributes, etc. Table 19: Sample Means of the Variables for Probit Regression, EPH 1998 Mean Standard Deviation Dependent Variable: 0.294 0.455 Poor=1, and Non-Poor=0 Individual's Personal Characteristics Age under 15 0.271 0.445 Age 65 and Over 0.102 0.303 Retired 0.093 0.291 Female 0.519 0.500 Secondary School Completed 0.183 0.389 Higher Education Completed 0.071 0.256 Household Characteristics Female Headed Household 0.199 0.399 Household Size Larger than 7 0.088 0.283 Number of Children between Ages 0-18 1.91 1.85 Number of the Dependents 3.04 1.99 Number of the Employed 1.48 1.00 Characteristics of the Household Head Living Together w/o Marriage 0.148 0.355 Married 0.626 0.484 Divorced 0.074 0.261 Illiterate 0.018 0.133 Secondary School Completed 0.216 0.411 Higher Education Completed 0.096 0.294 Unemployed 0.046 0.208 Work More Than 45 Hours per Week 0.374 0.484 Skilled Job 0.490 0.500 Professional Job 0.056 0.230 Sources: INDEC, Encuesta Permanente de Hogares, Average of the EPH for May and October, 1998. Note: Sample means reported above are all weighted by the normalized sample weights (i.e., factor of expansion). 28 Argentina Poverty Assessment Background Paper No. 1 The explanatory variables for the probit probability equation of poverty include: (i) individuals' personal characteristics; (ii) household characteristics; (iii) characteristics of household head, and (iv) a set of regional dummy variables. Rationales for inclusion of most variables are self-explanatory. The following are some supplemental explanations for treatment of variables in order to define them in model specification. Individual's characteristics are mainly meant to capture each individual's situation in his/her life-cycle. The correlation between the variable for the Retired and that for the aged 65 and over are was fairly high (p=0.67). However, both variables were used since the Retired would indicate the formal nature of jobs held during employment, whereas the age 65 and over would indicate life-cycle accumulation of wealth, if any. Since the number of children between age 0 and 18 are most likely to be dependent members in the household, regression have been run with two model specifications, one or the other. In fact, the correlation coefficient between these are very high at p=0.87. The number of employed is an important variable that measures the extent to which household members can bring income to the family. Educational and labor market characteristics of household heads are proxies for the potential capacity to generate income, important contributors to escape from poverty. The results in Table 20 are not surprising and are consistent with findings in poverty profiles. Individuals over 65 years old are less likely to be poor, but people who are retired, presumably with pensions, are even less likely to be poor. As expected, individuals who are more educated are less likely to be poor. The number of children between the age of 0-18, or the number of dependents has large and significant effect on the incidence of poor. Evaluated at the mean, the marginal effect of these variable with respect to the probability of being poor is 12% each. If household heads are illiterate, their family members are 21% more likely to be poor than otherwise. On the other hand, they graduate from high school or receive tertiary education, the household members are 18% and 25% less likely to be poor, respectively. With household heads having unskilled job as the omitted category, professional job would reduce the incidence of poverty by 25%. 29 s Argentina Poverty Assessment Background Paper No. 1 Table 20: Probit Regression of the Probability of Being Poor, EPH 1998 Dependent Variable (Poor=1, and Non-Poor=O) Model I Model 2 3 (z-score) 5F/5x ' (z-score) 8F/8x Intercept -0.423 -1.053 (-21.31) (-49.64) Individual's Personal Characteristics Age under 15 -0.104 -0.035 0.009 0.003 (-11.64) (1.03) Age 65 and Over -0.269 -0.084 -0.307 -0.094 (-14.42) (-16.47) Retired -0.565 -0.161 -0.629 -0.173 (-30.88) (-30.30) Female -0.036 -0.012 -0.033 -0.011 (-4.84) (-4.56) Secondary School Completed -0.232 -0.075 -0.343 -0.106 (-19.10) (-28.33) Higher Education Completed -0.580 -0.163 -0.666 -0.178 (-21.21) (-23.98) Household Characteristics Female Headed Household -0.044 -0.015 0.020 0.007 (-3.15) (1.34) Household Size Larger than 7 0.136 0.047 -0.127 -0.041 (8.78) (-7.63) Number of Children between Ages 0-18 0.368 0.124 - - (124.93) Number of the Dependents - - 0.343 0.115 (120.96) Number of the Employed -0.424 -0.143 -0.279 -0.093 (-93.47) (-61.17) Characteristics of the Household Head Living Together w/o Marriage 0.339 0.121 0.361 0.128 (20.76) (22.08) Married 0.004 zj 0.001 -0.090 -0.030 (0.25) (-6.06) Divorced 0.099 0.034 0.214 0.075 (5.87) (12.72) Illiterate 0.576 0.217 0.558 0.208 (24.41) (23.75) Secondary School Completed -0.612 -0.183 -0.571 -0.170 (-58.18) (-54.49) Higher Education Completed -1.070 -0.253 -1.054 -0.247 (-44.44) (-43.43) Unemployed 0.464 0.171 0.445 0.163 (25.94) (24.98) Work More Than 45 Hours per Week -0.205 -0.068 -0.112 -0.037 (-24.05) (-13.22) Skilled Job -0.165 -0.055 -0.025 -0.008 (-18.29) (-2.85) Professional Job -1.219 -0.263 -1.057 -0.240 (-33.81) (-29.39) Repional Dummies North West 0.493 0.177 0.437 0.155 (35.90) (31.78) North East 0.510 0.186 0.517 0.188 (34.63) (35.26) Cuyo 0.397 0.144 0.394 0.141 (26.75) (26.60) Pampeana 0.113 0.039 0.136 0.046 (8.51) (10.34) Patagonia -0.444 -0.133 -0.408 -0.122 (-26.71) (-24.60) Number of Observations 181,927 181,927 Pseudo R2 0.3423 0.3374 Log-Likelihood for Normal -77,641 -78,228 Sources: INDEC, EncuestaPermanente de Hogares (EPH), May and October, 1998. Notes: 1/ The marginal effects (8F/Sx) of dummy variables are discrete change from 0 to 1. 2/ All the variables are significant at 99% level except for: (i) household head being married (Model 1); and (ii) dummies for individuals aged under 15 and female head in Model 2. 30 Argentina Poverty Assessment Background Paper No. I VI. INCOME DISTRIBUTION One of the important factors in understanding the failure of poverty to decline in the 1990s is the change in the distribution of income. It was generally observed that the poor experienced more steep decline in income during the recession than the non-poor while the non-poor benefited to a far greater extent than the poor in the recovery of the economy, resulting in the worsening of income distribution. A. Trends in Income Distribution During the two decades of 1980s and 1990s, Argentina has experienced a steady worsening of income distribution by any measure. As Table 21 shows, the poor in the Metropolitan Buenos Aires had significantly lost ground compared to their non-poor counterpart, and this trend was markedly pronounced during the hyperinflation of 1989 and 1990. Table 21: Changes in Income Inequality in Argentina, 1990-1998 Earnings from the Primary Jobs Per Capita Household Income Average Gini Top 20% / Bottom Monthly Coefficient Mean Income Gini Coefficient 20% Income Share Earnings Ratio Survey Year Period All Urban All Urban GBA All Urban GBA All Urban GBA All Urban Areas Areas Areas Areas Areas (Sept. 1998) (Sept. 1998 Peso) 1980 October -- -- 479.9 -- 0.391 -- 7.6 -- 1985 October -- -- 380.0 -- 0.409 -- 8.5 -- 1988 May - - 345.8 -- 0.462 -- 11.6 -- October -- -- 306.1 -- 0.449 -- 11.1 -- 1989 May -- -- 364.3 -- 0.466 -- 11.3 -- October -- -- 259.8 -- 0.515 -- 14.5 -- 1990 May 575.5 0.439 258.1 229.4 0.471 0.463 12.1 11.9 October 600.3 0.418 273.0 249.8 0.461 0.451 10.8 10.5 1991 May 598.3 0.426 284.7 248.3 0.462 0.467 11.2 12.6 October 643.4 0.409 319.3 276.9 0.461 0.456 10.4 10.5 1992 May 662.0 0.411 323.9 281.8 0.431 0.442 9.3 10.0 October 714.2 0.410 350.3 299.7 0.441 0.453 9.8 10.8 1993 May 718.7 0.426 374.1 318.6 0.457 0.461 10.9 11.4 October 727.8 0.415 361.9 314.3 0.443 0.453 10.6 11.2 1994 May 734.0 0.425 381.3 330.1 0.447 0.457 10.5 11.1 October 716.9 0.415 356.5 313.3 0.457 0.461 10.9 11.4 1995 May 669.6 0.420 334.2 292.1 0.457 0.461 11.1 11.7 October 657.9 0.435 338.9 288.0 0.484 0.479 12.8 12.6 1996 May 646.0 0.428 327.0 281.6 0.476 0.476 12.3 12.9 October 641.6 0.435 329.4 282.1 0.484 0.483 13.0 13.3 1997 May 640.7 0.443 335.9 288.3 0.485 0.485 13.4 13.3 October 629.0 0.434 345.9 295.7 0.480 0.482 13.3 13.3 1998 May 669.7 0.449 365.2 310.4 0.491 0.491 13.5 13.8 October 660.1 0.455 359.5 304.6 1 0.501 0.498 15.2 14.7 Sources: INDEC, Encuesta Permanente de Hogares (EPH), 1980-1998. 31 Argentina Poverty Assessment Background Paper No. I Following the Convertibility Plan of April 1991, the Argentine economy achieved macroeconomic stability and resumed robust economic growth. During 1989-1993, there had been some narrowing of income disparity between the rich and the poor, and the Gini coefficient went from 0.49 to 0.45. As Table 22 shows, real income gain was felt across the income scale during this period. In particular, population in the bottom quintile saw their income rise by 20% in real terms. Table 22: Changes in Income Distribution, 1990-1998 Per Capita Household Income by Decile 1990 1991 1992 1993 1994 1995 1996 1997 1998 Average Income 1 42.6 49.5 51.3 50.4 52.3 42.1 39.2 40.7 39.2 2 75.0 84.4 90.0 92.4 90.1 79.3 74.4 76.8 74.6 3 98.7 112.1 120.8 125.2 121.8 108.5 103.4 107.5 105.2 4 124.4 141.2 153.8 159.7 155.7 138.4 132.7 139.5 137.4 5 156.0 172.9 185.3 198.1 194.7 169.8 165.5 176.6 173.8 6 190.5 207.4 228.9 243.6 237.9 212.6 207.3 217.3 216.3 7 231.0 256.1 282.4 301.8 294.3 261.7 260.0 273.9 276.0 8 292.1 325.2 359.2 387.4 378.7 337.4 338.4 353.7 364.0 9 398.5 445.2 498.3 528.9 517.0 476.4 476.2 499.0 515.1 10 839.4 974.7 1,027.0 1055.6 1,090.1 1,054.1 1,023.8 1,071.4 1,144.1 Overall 244.8 276.9 299.7 314.3 313.3 288.0 282.1 295.7 304.6 Income Share 1 1.7 1.8 1.7 1.6 1.7 1.5 1.4 1.4 1.3 2 3.1 3.1 3.0 2.9 2.9 2.8 2.6 2.6 2.5 3 4.0 4.1 4.0 4.0 3.9 3.8 3.7 3.6 3.5 4 5.1 5.1 5.1 5.1 5.0 4.8 4.7 4.7 4.5 5 6.4 6.2 6.2 6.3 6.2 5.9 5.9 6.0 5.7 6 7.8 7.6 7.6 7.7 7.6 7.4 7.3 7.3 7.1 7 9.4 9.2 9.4 9.6 9.4 9.1 9.2 9.3 9.1 8 11.9 11.7 12.0 12.3 12.1 11.7 12.0 12.0 12.0 9 16.3 16.1 16.6 16.8 16.5 16.5 16.9 16.9 16.9 10 34.3 35.2 34.3 33.6 34.8 36.6 36.3 36.2 37.6 Gini Coefficient 0.451 0.456 0.453 0.453 0.461 0.479 0.483 0.482 0.498 Share of Income Claimed 50.6 51.3 50.9 50.4 51.3 53.1 53.2 53.1 54.5 by Top 20% (A) Share of Income Claimed 4.8 4.9 4.7 4.5 4.5 4.2 4.0 4.0 3.7 by Bottom 20% (B) To to Bottom 20% Ratio 10.4 10.5 10.8 11.2 11.4 12.6 13.3 13.3 14.7 (C=A/B) Population represented by the samp1e(in 1,000s) 15,005 14,998 16,174 17,051 17,746 19,145 17,435 19,670 20,327 Number of urban agglomerations 24 24 25 25 25 28 27 28 27 Sources: INDEC, Encuesta Permanente de Hogares (EPH), October of 1990, 1992, 1994, 1995, 1996, 1997, and 1998. Note: Household per capita income are from all sources including wages and salaries from primary and secondary jobs, retirement incomes, capital incomes and other transfer incomes. 32 Argentina Poverty Assessment Background Paper No. I However, the rich and the poor saw their fortunes diverge in the aftermath of the Pesos Crisis in December 1994. During this period, the Gini coefficient increased from 0.46 in 1994 to 0.50 in 1998. And the share of income received by the top 20% went up from 51% to 55%. If we subdivide up the post-Tequila period to the Peso Crisis (1995/96) and the Recovery (1997/98), the following story emerges: Income Class (deciles) Pesos Crisis (1995/96) Recovery (1997/98) The Poor (1,2,3) Income collapses (20% drop) Never recovers Middle Class (5,6,7) Income suffers (15% drop) Marginally recovers The Rich (9,10) Slight income loss (5% drop) Fully recovers One of the significant factors that might help explain this phenomenon is the interplay of supply of labor and the changing pattern of labor demand since apertura. And at least in the short run, the nation's education system was unable to turn out workforce capable of meeting the Argentine economy's needs for specific skill categories. Table 23: Earnings from Primary Jobs by Skill Type Year Unskilled Skilled Professional All (1998 Pesos) 1990 366.2 570.2 1176.7 588.1 1991 408.9 610.2 1287.6 621.4 1992 438.9 685.2 1483.6 688.0 1993 420.7 714.9 1685.7 723.1 1994 424.4 725.4 1715.8 725.5 1995 383.0 656.8 1660.6 663.8 1996 362.7 635.9 1613.0 643.8 1997 348.2 626.6 1653.2 635.0 1998 356.2 644.8 1794.1 664.9 (Base=1992) 1990 83.4% 83.2% 79.3% 85.5% 1991 93.2% 89.0% 86.8% 90.3% 1992 100.0% 100.0% 100.0% 100.0% 1993 95.9% 104.3% 113.6% 105.1% 1994 96.7% 105.9% 115.7% 105.5% 1995 87.3% 95.9% 111.9% 96.5% 1996 82.6% 92.8% 108.7% 93.6% 1997 79.3% 91.4% 111.4% 92.3% 1998 81.2% 94.1% 120.9% 96.6% Sources: INDEC, Encuesta Permanente de Hogares (EPH), October of 1990, 1992, 1994, 1995, 1996, 1997, and 1998. When unemployment rate shot up to 18% in May 1995, the poor were more hard hit by a sharp rise in jobless rate. As has been pointed out in Table 11, the poor always suffered from high incidence of unemployment throughout 1990s, hovering above 20%. 33 Argentina Poverty Assessment Background Paper No. 1 This gives strong evidence that the type of jobs mostly likely to be held by the poor have not been created during this period. As Table 23 shows, the earnings of professionals have increased by about 20% during 1992 and 1998, while those of skilled and unskilled workers dropped by about 6% and 20%, respectively. In the midst of sluggish labor market, the professionals fared much better than their less-skilled counterparts. And this trend continued for an extended period, thereby contributing to deteriorating income distribution in general. Table 24: Earnings from Primary Jobs by Education Level Year No Primary Some Secondary Some College All Education Completed Secondary Completed College Completed (1998 Pesos) 1990 348.5 432.9 509.1 658.9 839.1 1033.5 588.1 1991 377.9 472.5 535.2 704.0 723.5 1154.7 621.4 1992 413.6 534.6 605.1 779.9 844.8 1185.6 688.0 1993 430.8 537.2 651.8 801.8 867.9 1283.5 723.1 1994 418.7 537.7 629.4 809.7 875.8 1313.2 725.5 1995 378.4 478.7 548.0 711.8 841.2 1223.5 663.8 1996 352.5 450.7 527.0 700.1 809.2 1168.3 643.8 1997 335.8 443.0 513.0 688.2 784.6 1173.5 635.0 1998 335.0 449.2 529.8 703.4 818.0 1275.9 664.9 (Base= 1992) 1990 84.2% 81.0% 84.1% 84.5% 99.3% 87.2% 85.5% 1991 91.4% 88.4% 88.4% 90.3% 85.6% 97.4% 90.3% 1992 100.0% 100.0% 100.0% 100.00/0 100.0% 100.0% 100.0% 1993 104.2% 100.5% 107.7% 102.8% 102.7% 108.3% 105.1% 1994 101.2% 100.6% 104.0% 103.8% 103.7% 110.8% 105.5% 1995 91.5% 89.5% 90.6% 91.3% 99.6% 103.2% 96.5% 1996 85.2% 84.3% 87.1% 89.8% 95.8% 98.5% 93.6% 1997 81.2% 82.9% 84.8% 88.2% 92.9% 99.0% 92.3% 1998 81.0% 84.0% 87.5% 90.2% 96.8% 107.6% 96.6% Sources: INDEC, Encuesta Permanente de Hogares (EPH), October of 1990, 1992, 1994, 1995, 1996, 1997, and 1998. Table 24 presents the flip-side of skills-earnings profile by looking at time-series data on earnings by education level. This clearly shows that the lowest income scale suffered most from erosion of real income. Interestingly enough, there exists a gradual pattern of better performance by better educated workers going from the least to best educated. It appears that earnings premiums for more education persisted and the supply response has been rather weak at least during 1992-1998. It is beyond the scope of this paper to answer why this has been the case. However, the Mincerian framework of earnings equation provides a partial answer to the question of how individual workers of given characteristics take advantage of opportunities offered by the market. 34 Argentina Poverty Assessment Background Paper No. 1 The resulting parameter estimates of the Mincerian earnings function behaved in a stable manner during the periods. And the estimated parameters were highly significant, including all regional dummy variables.12 Table 25: Returns to Education and Predicted Earnings by Education Level Years of Returns to Education Predicted Earnings Schooling (S) 1992 1998 1992 1998 1 6.6 4.5 378.9 339.9 2 6.8 5.0 405.1 356.5 3 7.0 5.6 434.0 376.0 4 7.2 6.1 465.7 398.6 5 7.3 6.7 500.7 424.9 6 7.5 7.2 539.3 455.4 7 7.7 7.7 582.0 490.7 8 7.9 8.3 629.3 531.6 9 8.1 8.8 681.6 579.1 10 8.3 9.4 739.7 634.1 11 8.5 9.9 804.2 698.1 12 8.6 10.4 876.0 772.8 13 8.8 11.0 956.0 860.0 14 9.0 11.5 1045.2 962.2 15 9.2 12.0 1144.8 1082.4 16 9.4 12.6 1256.3 1224.2 17 9.6 13.1 1381.2 1392.0 18 9.8 13.7 1521.4 1591.4 Sources: INDEC, Encuesta Permanente de Hogares (EPH), October of 1992 and 1998. As Table 25 shows, a simple simulation exercise reveals that the returns-to- education has become more steep with respect to years of schooling during 1992-1998. That is, individual workers with higher education attainment were reaping higher returns for their investment in educating themselves in 1998 than in 1992. For those with only two years of schooling, the returns to education dropped from 6.8% to 5.0%. However, those with 18 years of schooling, the returns increased from 9.8% to 13.7%. 12 The estimated Mincerian earnings functions are given by: October 1992: log Earnings = 5.155 + 0.064 S + 0.00093 S2 + 0.051 EX - 0.00078 EX2 - 0.423 FEM (t-ratios) (167.2) (12.6) (3.9) (43.4) (-31.8) (-47.0) + 0.543 EMP - 0.057 SELF + E (Regional Dummies) (23.4) (-5.4) adj. R2=0.389, n = 20,511 October 1998: log Earnings = 5.016 + 0.040 S + 0.00269 S2 + 0.053 EX - 0.00078 EX2 - 0.477 FEM (t-ratios) (156.5) (7.6) (11.3) (43.3) (-30.5) (-51.1) + 0.565 EMP - 0.223 SELF + , (Regional Dummies) (23.0) (-20.0) adj. R2 = 0.351, n = 24,183 where S = years of schooling, EX = years of experience, FEM = female dummy variable, EMP employer, and SELF = self-employed. 35 Argentina Poverty Assessment Background Paper No. 1 However, we can ask the following question. How much of the change in the income distribution is attributed to the changes in the returns to education, and how much to other factors? During 1992-1998, the Gini coefficient of earnings from the primary job went up by 4 percentage point from 0.406 to 0.446.'1 Using the estimated parameters of the Mincerian earning equations for 1992 and 1998, a series of simulation has been performed in order to quantify the relative magnitude of various determinants of changes in earnings inequality, to wit, returns to education, returns to acquired labor experience or other observable assets, changes in endowments, or simply unobservable factors. Table 26: Decomposition of Changes in Earnings Inequality, 1992-1998 Gini index Absolute %Change of earnings changes in ne ity inequaty 1992 0.406 -- -- Decomposition of changes in earnings inequality I. Returns to education (R) 0.421 0.015 37.5% II. Returns to all observable human assets (A) 0.422 0.016 40.0% III. Variance of error terms (returns to 0.438 0.032 80.0% unobservable assets) (B) IV. C=A+B 0.451 0.045 112.5% 1998 0.446 0.040 100.0% Source: INDEC, Encuesta Permanente de Hogares (EPH), October 1992 and October 1998. The first simulation in Table 26 matches the 1992 distribution of endowments with the education parameters estimated from the 1998 model. Decomposition shows that about 38% (or 1.5 percentage point) increase in the Gini coefficient of earnings inequality can be attributed to changes in returns to education, i.e., increase among the skilled (the more educated) and decrease among the unskilled (the less educated) during this period. Simulation with the simultaneous change in returns to all observable assets is not much different from the results with returns to education, which means that all other observable assets have either neutral effects or cancel each other from the distributive point of view. This leaves a large part of the inequality changes unidentified. Indeed two-thirds (2/3rds) of the total increase in inequality, almost twice the impact of education returns, are attributable to such factors such as changing patterns of leisure-work choice, family size, and others which are also influencing income distribution. Growth, Income Distribution, and Poverty Future efforts to reduce poverty will be clearly related to overall growth in income, as well as changes in income distribution. One can always learn from past experiences how performance of the economy and changes in income distribution interact with each other to produce poverty reduction outcomes. 13 The calculation of Gini coefficients for 1992 and 1998 was carried out only for earnings of the employees, self-employers, and owners of business. 36 Argentina Poverty Assessment Background Paper No. I What would happen to the poverty rate if there had been no deterioration in income distribution? One way of analyzing this question is to decompose the changes in poverty rates into changes in distribution, and other factors including income change. Table 27 provides the results of a pair-wise comparison of poverty rates of two years between 1990 to 1998, where decomposition is made up of distribution-neutral growth effect and the distribution effect.14 The average Gini index during 1990-94 was about 0.454, and it jumped to 0.486 during 1995-1998. During the 1990-94, poverty rate hovered around 24%. However, there seems to have occurred a one-time permanent shift in inequality regime, which was responsible for 5 percentage points increase in poverty rate, i.e., an increase to poverty rate of 29%. Had income distribution not deteriorated between 1992- 98, the poverty rate in 1998 would have stayed at approximately 24% instead of 29%. Another way to decompose the effects of growth and distribution on poverty is to perform regression analysis. Using a pooled time-series and cross-section data from the EPH for 1990-98 by regions, we can estimate the elasticities of poverty rates with respect to changes in per capita income and the Gini coefficient. 15 The estimated elasticity for poverty with respect to growth is 1.38; a 10% increase in per capita income will reduce the poverty rate by 13.8%, provided there is no change in the Gini. Table 28 shows alternate combinations of changes in per capita income and the Gini. For instance, a 20% growth in per capita income would imply a growth rate over 10 years of about 1.8% per annum. This would reduce the poverty rate by 28%, if there is no change in the Gini, but by only 20% if the Gini were to increase by 10%. If we start with a poverty rate of 29.4%, then the forecast poverty rate would vary between 21.3% and 23.5%. 14 For instance, the income effect from the period i to j is given by percentage point change in poverty rate when the poverty line corresponding to the period j were to be applied to the income distribution of period of i with its income adjusted for real income change during the period i and j. 15 The estimated equation regresses the change in the log of the poverty rate (dlnPR) against the change log of per capita income (dinPCY) and change in the log of the Gini (dlnGINI), with dummy variables used to capture the fixed effects of different regions. The estimated equation (without the dummy variables) is: dlnPR = -0.004 -1.3 83dInPCY + 0.768dlnGINI R2=.64 (-0.454) (-9.635)* (3.670)* n=102 * significant 1% level. The equation is estimated using semi-annual data for the period 1990-98 for six urban areas. 37 Argentina Poverty Assessment Background Paper No. 1 Table 27: Decomposition of Poverty, 1990-1998 Changes in Poverty Rate Decomposition of Changes in Poverty Rate 1990 1991 1992 1993 1994 1995 1996 1997 1998 From Component Oct. Oct. Oct. Oct. Oct. Oct. Oct. Oct. Oct. Oct. 1990 -- Income effect 39.4 -11.1 -15.3 -16.8 -16.4 -12.5 -11.1 -13.3 -14.1 Poverty rate w/o dist A 28.3 24.1 22.6 23.0 26.9 28.3 26.1 25.3 --Distribution effect -1.3 -0.8 -0.8 -0.2 1.5 2.4 2.9 4.7 -- Overall effect -12.3 -16.1 -17.7 -16.6 -11.0 -8.7 -10.4 -9.5 Oct. 1991 -- Income effect 27.1 -4.1 -5.9 -5.7 -1.4 -0.0 -2.1 -2.9 Poverty rate w/o dist A 23.0 21.2 21.3 25.7 27.0 24.9 24.2 -- Distribution effect 0.3 0.6 1.5 2.7 3.7 4.0 5.8 -- Overall effect -3.8 -5.3 -4.3 1.3 3.6 1.9 2.9 Oct. 1992 -- Income effect 23.3 -1.8 -1.4 2.4 3.7 1.6 1.1 Poverty rate w/o dist A 21.5 21.9 25.7 27.0 24.9 24.4 -- Distribution effect 0.2 0.9 2.7 3.7 4.1 5.5 -- Overall effect -1.6 -0.5 5.1 7.4 5.7 6.6 Oct. 1993 -- Income effect 21.7 0.2 3.8 5.2 3.1 2.6 Poverty rate w/o dist A 21.9 25.6 26.9 24.8 24.3 -- Distribution effect 0.9 2.8 3.8 4.1 5.6 -- Overall effect 1.1 6.6 9.0 7.2 8.2 Oct. 1994 -- Income effect 22.8 4.2 5.6 3.3 2.6 Poverty rate w/o dist A 27.0 28.4 26.1 25.4 -- Distribution effect 1.3 2.3 2.8 4.6 -- Overall effect 5.6 7.9 6.2 7.2 Oct. 1995 -- Income effect 28.4 1.0 -1.0 -1.7 Poverty rate w/o dist A 29.4 27.3 26.6 --Distribution effect 1.3 1.6 3.3 -- Overall effect 2.3 0.6 1.6 Oct. 1996 -- Income effect 30.7 -1.8 -2.6 Poverty rate w/o dist A 28.9 28.1 -- Distribution effect 0.0 1.8 -- Overall effect -1.7 -0.7 Oct. 1997 -- Income effect 29.0 -0.6 Poverty rate w/o dist A 28.3 -- Distribution effect 1.6 -- Overall effect 1.0 Poverty Rate 39.4 27.1 23.3 21.7 22.8 28.4 30.7 29.0 29.9 Per Capita Household Income 249.8 276.9 299.7 314.3 313.3 288.0 282.1 295.7 304.6 (September 1998 Price) Adult Equivalent Effective Poverty 161.2 151.8 148.6 149.3 149.9 151.6 152.9 152.6 154.1 Lines (September 1998 Price) I I I I I _ I Gini Index of Income Inequality 0.451 0.456 0.453 0.453 0.461 0.479 0.483 0.482 0.498 Sources: INDEC, Encuesta Permanente de Hogares (EPH), October of 1990, 1991, 1992, 1993, 1994, 1995, 1996, 1997 and 1998. 38 Argentina Poverty Assessment Background Paper No. 1 Table 28: Estimated Impact of Changes in Per Capita Income and Distribution on Poverty: A: Percent Change in Poverty Headcount: Total Change in Per Capita Income % Change in Gini +10% +20% +30% 0 -13.8 -27.7 -41.5 +5% -10.0 -23.8 -37.7 +10% -6.2 -20.0 -33.8 B: Estimated Poverty Headcount:_ Total Change in Per Capita Income % Change in Gini +10% +20% +30% 0 25.3 21.3 17.2 +5% 26.5 22.4 18.3 +10% 27.6 23.5 19.5 Sources: INDEC, Encuesta Permanente de Hogares (EPH), 1980-1998. This simulation somewhat mechanical, but illustrates both the importance of growth in income in any plan to reduce poverty, as well as the importance of how gains in national income are distributed among the population. However, it also assumes that the elasticities of the Gini and poverty with respect to growth are constant over the cycle. Recent work by DeJanvry and Sadoulet16 suggests that these elasticities can have important asymmetries: they tend to be higher in a downturn than in a recovery. Poverty is likely to be reduced faster if, for the same average growth rate, there is a greater degree of stability in the rate of growth. 16 Alain de Janvry and Elisabeth Sadoulet, "How Effective Has Aggregate Income Growth Been in Reducing Poverty and Inequality in Latin America?" processed, Nov 9, 1999. 39 Argentina Poverty Assessment Background Paper No. 1 B. Regional Income Disparity Regional distribution of income can be characterized by large disparity between the relatively more affluent places (Metropolitan Buenos Aires and Patagonia) and the northern poor provinces (North East and North West). According to the 1996 EPH data in Table 29, the average per capita household income in the northern provinces were only about half that in MBA or in provinces in the Patagonia region. The per capita household income in the Cuyo region was approximately 67% of that in MBA as well, followed by 73% in the Pampeana region. Therefore, the regions were ranked by income from: (i) MBA and Patagonia, (ii) Pampeana; (iii) Cuyo; and (iv) North West and North East. Table 29: Regional Disparity of Income Distribution, 1996 EPH REGION Per Capita Household Metro North North Cuyo Pampeana Patagonia All Income Decile Buenos West East Regions Aires Avera2e Income 1 47.1 26.9 24.7 34.4 37.1 47.6 38.5 2 86.8 50.5 44.3 63.7 71.5 91.9 73.1 3 118.2 68.6 61.0 87.1 99.3 128.0 101.7 4 151.1 88.8 78.8 108.3 126.4 164.4 130.5 5 190.3 108.7 100.3 133.1 155.4 206.3 162.7 6 233.8 135.1 125.8 163.6 191.8 250.4 203.8 7 296.3 163.9 158.4 206.8 234.2 309.7 255.7 8 384.3 210.1 203.7 268.9 296.4 399.1 332.8 9 548.3 294.2 282.4 374.4 399.2 548.4 468.3 10 1,182.3 587.5 597.4 712.0 742.5 1,131.2 1,006.8 Overall 323.8 173.5 167.7 215.3 235.4 328.8 277.4 Income Share 1 1.5 1.5 1.5 1.6 1.6 1.5 1.4 2 2.7 2.9 2.6 3.0 3.0 2.8 2.6 3 3.6 4.0 3.6 4.0 4.2 3.9 3.7 4 4.7 5.1 4.7 5.0 5.4 5.0 4.7 5 5.9 6.3 6.0 6.2 6.6 6.3 5.9 6 7.2 7.8 7.5 7.6 8.1 7.6 7.3 7 9.1 9.4 9.4 9.6 10.0 9.4 9.2 8 11.9 12.1 12.1 12.5 12.6 12.2 12.0 9 16.9 17.0 16.8 17.4 17.0 16.7 16.9 10 36.5 33.9 35.6 33.1 31.5 34.6 36.3 Gini Coefficient 0.484 0.455 0.477 0.452 0.434 0.462 0.483 Share of Income Claimed 53.5 50.8 52.5 50.5 48.5 51.3 53.2 by Top 20% (A) Share of Income Claimed 4.1 4.5 4.1 4.6 4.6 4.3 4.0 by Bottom 20% (B) Top to Bottom 20% Ratio 13.0 11.3 12.8 11.0 10.5 11.9 13.3 Population represented by the sample (in 1,000s) 94 1 31 748 1,31 3 531 17435 Source: INDEC, Encuesta Permanente de Hogares (EPH), October 1996. 40 Argentina Poverty Assessment Background Paper No. I On the other hand, the intra-regional distribution of income is fairly homogeneous. Income in the rich regions tends to be somewhat more unevenly distributed. One exception is the poorest region, the North East. The North East region features not only the lowest income level, but also one of the worst income distribution of all regions. Tables 30 and 31 confirm the patterns in disparity of regional distribution using both income and consumption distribution data from the 1996-97 National Expenditure Survey. The regional distribution ranking is preserved whether it is done by income or consumption. Again, the poorest North East region has one of the most skewed income distribution. Table 30: Regional Disparity of Income Distribution, 1996-1997 ENGHO REGION Per Capita Household Metro North North Cuyo Pampeana Patagonia All Income Decile Buenos West East Regions Aires Averaize Income 1 53.6 31.5 28.0 35.2 42.6 40.4 41.1 2 99.4 56.6 50.1 67.0 80.4 76.8 78.1 3 138.2 76.5 66.0 93.4 112.8 110.0 109.5 4 177.9 98.1 86.7 117.5 143.1 144.1 141.4 5 217.5 119.3 108.8 144.9 179.5 181.3 179.0 6 268.7 146.5 136.0 176.1 220.5 229.3 221.6 7 336.6 182.4 177.5 221.3 273.2 292.4 279.3 8 426.6 236.7 238.1 285.4 347.2 374.7 359.5 9 601.0 337.2 334.1 390.9 473.2 536.4 499.9 10 1,168.3 691.4 768.4 840.8 928.8 1,118.9 1,016.6 Overall 348.8 197.6 199.4 237.3 280.2 310.5 292.6 Income Share 1.5 1.6 1.4 1.5 1.5 1.3 1.4 1 2.8 2.9 2.5 2.8 2.9 2.5 2.7 2 4.0 3.9 3.3 3.9 4.0 3.6 3.7 3 5.1 5.0 4.3 4.9 5.1 4.6 4.8 4 6.2 6.0 5.5 6.1 6.4 5.9 6.1 5 7.7 7.4 6.8 7.4 7.9 7.4 7.6 6 9.6 9.2 8.9 9.3 9.8 9.4 9.5 7 12.2 12.0 12.0 12.0 12.4 12.1 12.3 8 17.2 17.1 16.8 16.5 16.9 17.3 17.1 9 33.5 35.0 38.5 35.4 33.2 36.0 34.7 10 100.0 100.0 100.0 100.0 100.0 100.0 100.0 348.8 197.6 199.4 237.3 280.2 310.5 292.6 Gini Coefficient 0.455 0.465 0.506 0.467 0.451 0.486 0.470 Share of Income Claimed 50.7 52.1 55.3 52.0 50.1 53.3 51.8 by Top 20% (A) Share of Income Claimed 4.4 4.5 3.9 4.3 4.4 3.8 4.1 by Bottom 20/o (B) Top to Bottom 20% Ratio 11.5 11.6 14.2 12.1 11.4 14.0 12.6 Population represented by , the sample (in 1,000s) 11,807 2,916 2,042 , 1 9,412 1,326 29.369 Source: INDEC, Encuesta Nacional de Gastos de los Hogares (ENGHO), February 1996-February 1997. Table 31: Regional Disparity of Consumption Distribution, 1996-1997 ENGHO 41 Argentina Poverty Assessment Background Paper No. 1 REGION Per Capita Household Metro North North Cuyo Pampeana Patagonia All Consumption Decile Buenos West East Regions Aires Averaie Consumption 1 41.9 28.6 25.2 32.3 36.0 30.0 34.6 2 78.5 51.1 42.3 60.0 68.7 57.1 65.3 3 109.8 70.0 56.7 80.9 93.7 87.7 90.7 4 141.6 87.6 73.6 101.1 120.9 115.0 117.6 5 177.1 107.5 92.3 124.7 149.1 143.8 147.4 6 219.0 132.4 115.3 149.9 180.9 178.7 182.8 7 277.7 165.6 148.5 186.2 223.3 227.5 229.2 8 356.0 211.5 198.4 241.6 281.7 295.5 296.4 9 513.1 286.6 279.9 333.9 382.9 420.4 414.6 10 1,010.3 585.2 592.4 654.9 688.6 805.1 824.7 Overall 292.5 172.7 159.5 196.6 222.6 236.1 240.3 Consumption Share 1 1.4 1.7 1.6 1.6 1.6 1.3 1.4 2 2.7 3.0 2.7 3.1 3.1 2.4 2.7 3 3.8 4.1 3.6 4.1 4.2 3.7 3.8 4 4.8 5.1 4.6 5.1 5.4 4.9 4.9 5 6.1 6.2 5.8 6.3 6.7 6.1 6.1 6 7.5 7.7 7.2 7.6, 8.1 7.6 7.6 7 9.5 9.6 9.3 9.5 10.0 9.6 9.5 8 12.2 12.2 12.4 12.3 12.7 12.5 12.3 9 17.5 16.6 17.5 17.0 17.2 17.8 17.2 10 34.5 33.9 35.3 33.3 30.9 34.1 34.3 Gini Coefficient 0.470 0.452 0.478 0.448 0.429 0.473 0.466 Share of Consumption 52.1 50.5 52.8 50.3 48.1 51.9 51.6 Claimed by Top 20% (A) Share of Consumption 4.1 4.6 4.2 4.7 4.7 3.7 4.2 Claimed by Bottom 20% Top to Bottom 20% Ratio 12.7 11.0 12.6 10.7 10.2 14.0 12.3 Population represented by the sample (in 1,000s) 11,796 2,915 2039 1 9 1 29,4| Source: INDEC, Encuesta Nacional de Gastos de los Hogares (ENGHO), February 1996-February 1997. The results from the 1997 Social Development Survey (EDS) in Table 32 show a similar pattern of regional income disparity as before. Interestingly, the distribution of income is shown to be slightly more uneven when estimated by the EDS data. However, the average per capita household income for each decile are remarkably similar between the EPH and the EDS, in particular up to the 9th decile. Robustness of poverty estimates with respect to data sources probably stems from this regularity of income distribution over a wide range of income scale. 42 Argentina Poverty Assessment Background Paper No. I Table 32: Regional Disparity of Income Distribution, 1997 EDS REGION Per Capita Household Metro North North Cuyo Pampeana Patagonia All Income Decile Buenos West East Regions Aires Average Income 1 46.3 28.6 28.3 41.2 34.1 32.8 36.0 2 92.0 53.1 49.6 67.4 71.6 69.9 70.5 3 129.1 72.1 64.9 90.8 102.5 104.6 99.7 4 172.4 91.5 81.8 113.0 132.1 139.1 130.0 5 219.0 113.3 102.5 140.9 162.6 177.6 164.3 6 279.2 136.2 126.1 167.7 202.4 221.1 207.8 7 354.8 172.5 156.3 204.0 248.7 287.3 263.6 8 455.9 221.2 203.5 268.4 312.3 373.5 345.3 9 678.0 309.5 287.9 365.7 429.0 524.8 489.4 10 1,524.5 693.8 745.3 774.7 905.7 1,104.5 1,145.9 Overall 395.3 189.2 184.6 223.5 260.1 303.6 295.3 Income Share 1 1.2 1.5 1.5 1.8 1.3 1.1 1.4 2 2.3 2.8 2.7 3.0 2.8 2.3 2.6 3 3.3 3.8 3.5 4.1 3.9 3.4 3.7 4 4.4 4.8 4.4 5.1 5.1 4.6 4.7 5 5.5 6.0 5.5 6.3 6.3 5.8 5.9 6 7.1 7.2 6.8 7.5 7.8 7.3 7.3 7 9.0 9.1 8.5 9.1 9.6 9.5 9.2 8 11.5 11.7 11.0 12.0 12.0 12.3 12.0 9 17.2 16.3 15.6 16.3 16.5 17.3 16.9 10 38.6 36.7 40.4 34.7 34.8 36.4 36.3 Gini Coefficient 0.511 0.477 0.506 0.454 0.465 0.496 0.508 Share of Income Claimed 55.8 53.0 56.0 51.1 51.3 53.7 55.4 by Top 20% (A) Share of Income Claimed 3.5 4.3 4.2 4.9 4.1 3.4 3.6 by Bottom 20% (B) Top to Bottom 20% Ratio 15.9 12.3 13.3 10.4 12.5 15.8 15.4 Population represented by the sample (in 1,000s) 8,521 2,592 1,740 1,662 7,851 1,269 23,636 Source: SIEMPRO/ INDEC, Encuesta de Desarrollo Social (EDS), August 1997. 43 Argentina Poverty Assessment Background Paper No. 1 VII. SUMMARY AND CONCLUSIONS This report significantly expands the scope of the statistical information available on poor people in Argentina. Not only did the report construct a time series of micro- data on poverty (the EPH) to increase the coverage from 33% to 60% of the Argentine population (70% or urban population), but also drew on the data sets that have become recently available (1996-97 ENGHO; 1997 EDS) to complement the analysis. Description of poverty picture from this analysis is not far removed from the existing literature on poverty in Argentina in terms of general findings. However, the rich set of statistical information made it possible to formulate alternative hypotheses to identify and test the significance of underlying factors that determine the incidence and changes in poverty during the study period. Argentina enjoyed a stable and buoyant economy following the Convertibility Plan in 1992. Inflationary pressure was in check, economy grew robustly, and poverty rate was sharply reduced until 1994, when the Peso Crisis struck. What followed in the aftermath was revealing, in that it exposed various signs of structural weakness in the Argentine economy in coping with poverty problems in a global economy. Persistent and worsening poverty situation during 1994-1998 has been highly conditioned by the growing income disparity between the rich and the poor. Riding the cycle of recession and recovery, the poor saw their economic position being eroded because of shifting demand away from low skill jobs that does not require high educational attainment. This changing demand patterns for labor has been reflected in chronic discouraged worker effect among the poor vis-A-vis the non-poor, and chronic high unemployment rate among the poor, in particular the indigent poor. A simulation analysis showed that eliminating unemployment would have a large pay-off in reducing poverty. While sustained economic growth over a medium- to long-term would produce substantial poverty reduction, reversing the widening income disparity would require a micro-economic intervention to upgrade the income-generating capacity of the poor. A detailed profile of the poor shows that they are likely to have significantly larger families and have younger families with higher dependency ratio. Because the poor families are larger and younger, children are more likely to be poor. The poor are more likely to be unemployed and have fewer years of schooling. And the main breadwinner of the family are more likely to employed in the unstable jobs and sectors (informal sectors and construction industry) adding to vulnerability of the economic position of the family members. Given that the public sector has limited resources the strategy to reduce poverty in Argentina will have to adopt an integrated approach to maximize its impact. Resumption of economic growth in a stable macroeconomic environment would constitute the first step in the direction toward poverty reduction. While a bulk of job creation should be left to the private sector, public investment should be directed to enhancing the human capital endowments among the poor segments of the population. Analysis shows that a rich country like Argentina have large capacity to mobilize the needed public resources to invest in anti-poverty human capital programs as well as providing the safety nets to the 44 Argentina Poverty Assessment Background Paper No. 1 most vulnerable population. Companion background papers and the main project report provides concrete proposals to address this issues. Finally, it should be reiterated that poverty data is still very deficient in Argentina. Rural poverty still is under-documented due largely to lack of representative data in the rural areas. So expansion of the coverage of household surveys to the rural areas is urgently called for. And many attempts to improve poverty statistics are frustrated due to lack of basic data such as the national price index. Recent recommendations by the Working Group to the Executive Committee on Poverty Study is highly relevant to improve the basic data on poverty int Argentina. 45 Argentina Poverty Assessment Background Paper No. 1 REFERENCE Altimir, Oscar (1999), Revision de las lineas oficiales de pobreza, processed. B.A. de Morales, Elena (1988), Canasta básica de alimentos: Gran Buenos Aires, Documento de Trabajo No. 3, IPA, INDEC, Buenos Aires. Britos, Sergio (1999), Canasta basica de alimentos del Area Metropolitana: actualización 1999, processed. CEPA (1993), Evolucion reciente de la pobreza en el Gran Buenos Aires, 1988-1992, Documento de Trabajo No. 2, Ministerio de Economia, Buenos Aires. de Janvry, Alain and Elisabeth Sadoulet (1999), "How Effective Has Aggregate Income Growth Been in Reducing Poverty and Inequality in Latin America?" processed. Epszteyn, E. and A. Orsatti (1989), Caracteristicas de una linea de pobreza para Argentina, 1985, Documento de Trabajo No. 8, IPA, INDEC, Buenos Aires. Ferreira, Francisco H.G., and Julie A. Litchfield (1998), Calm after the Storms: Income Distribution in Chile, 1987-1994, The World Bank, mimeo. Foster, James (1998), Absolute versus Relative Poverty, American Economic Review, 88 (2): pp. 335-341. INDEC (1999), Encuesta Nacional de Gastos de los Hogares, 1996-199 7, Resultados Definitivos, Buenos Aires. INDEC (1988), Encuesta de Gastos e Ingresos de los Hogares, 1985-1986, Buenos Aires. INDEC, Encuesta Permanente de Hogares, 1980-1998. INDEC, INDEC Informa, 1980-1998. INDEC, Indice de precios al consumidor, contained in Estadistica mensual, 1988-1998. INDEC (1989), Indice de precios al consumidor (revisión base 1988=100), sistesis metodologica, 1980-1989. Lee, Haeduck (1994), A Poverty Profilefor Argentina, mimeo. Ministerio de Economia (1997), Economic Report, Second Quarter, 1997, Number 22, Buenos Aires. Ravallion, Martin (1998), Poverty Lines in Theory and Practice, Living Standards Measurement Study Working Paper No. 133, The World Bank, Washington, D.C. 46 Argentina Poverty Assessment Background Paper No. I SIEMPRO (1999), Estimacion de lineas de pobreza para Argentina, processed. Sen, Amartya K. (1983), Poor, Relatively Speaking, Oxford Economic Papers, 35 (2): pp. 53-69. Sen, Amartya K. (1973), On Economic Inequality, Clarendon Press: London. SIEMPRO (1999), Avances de las estimaciones de linea depobreza, (documento preliminar), Buenos Aires. The World Bank (1995), Argentina: Argentina's Poor: Profile, Report No. 13318-AR, Washington D.C. The World Bank (1994), Argentina: Argentina's Poor: Profile, Volume H- Technical Annexes, mimeo, Washington D.C. 47 Argentina Poverty Assessment Background Paper No. 1 Annex Tables 48 Argentina Poverty Assessment Background Paper No. 1 Annex Table 1: A Profile of the EPH Data Sets, 1990-1998 Available EPH Data Sets by Type of Data Format Region/Province Urban Centers % o %Residing 1990 1990 1991 1991 1992 1992 1993 1993 1994 1994 1995 1995 1996 1996 1997 1997 1998 1998 (Aglomerado) Argentin in Urban May Oct. May Oct. May Oct. May Oct. May Oct. May Oct. May Oct. May Oct. May Oct. Po latio Centers # of data dets 20 24 25 24 24 25 25 25 25 25 25 28 28 27 28 28 28 27 # of missing datasets 5 1 0 1 1 0 0 0 0 0 0 0 0 1 0 0 0 1 % of Argentine population represented by the EPH 56.1 58.4 58.7 58.4 57.9 58.7 58.7 58.7 58.7 58.7 58.7 61.1 61.1 57.5 61.1 61.1 61.1 60.9 Metropolitan Buenos Aires33.5 33.4 I1.lMetro B.A. 32. FederalCapital 9.1 T T T T T T T T T T T T T T T T UA UA 33. 19 Districts 24.3 North West 11.3 5.4 10.Catamarca 22. Gran Catamarca 0.4 U R2 R2 R2 R2 R2 T T R2 T T T T T T T UA UA 38. Jujuy 19. SS de Jujuy 0.7 U R2 T T T T T T T T T T T T T UA UA 46. La Rioja 25. La Rioja 0.3 R2 R2 R2 R2 R2 R2 R2 R2 T T R2 U T T T T UA UA 66. Salta 23. Salta 1.2 U R2 T T T T T T T T T T T T T T UA UA 86.S.delEstero 18. S. del Esterot 0.8 0 R2 R2 R2 R2 T R2 R2 T T T T T T T T UA UA 90.Tucuman 29. SM de Tucuman 2.0 U U R2 T T R2 T T T T T T T T T T UA UA North East 8.7 2.9 18. Corrientes 12. Corrientes 0.8 U R2 R2 R2 0 R2 R2 R2 T U T T T T T T UA UA 22. Chaco 08. Resistencia 0.9 R2 R2 R2 R2 R2 R2 R2 R2 T T T T T T T T UA UA 34. Formosa 15. Formosa 0.4 0 R2 R2 R2 R2 R2 R2 R2 R2 P2 R2 T T T T T UA UA 54. Misiones 07. Posadas 0.8 0 R2 R2 R2 R2 R2 R2 R2 R2 R2 R2 T T T T T UA UA Cuyo 6.8 3.8 50. Mendoza 10. Mendoza 2.4 U R2 T T T R2 T T T T T T T T T T UA UA 70. San Juan 27. San Juan 1.1 U R2 R2 R2 T T R2 R2 T T T T T T T T UA UA 74. San Luis 26. San Luis 0.4 U R2 R2 R2 T T T T T T T T T T T T UA UA Pampeana 35.2 14.3 06. Resto B.A. 02. La Plata 2.0 U R2 R2 T T7 T T T T TT T T T 1T UA UA 03. Bahia Blanca 0.8 U R2 R2 R2 R2 R2 R 2 R2 R 2 T U T T T T UA UA , - -_IIl- U 34. Mar del Plata 1.6 T T T T T UA UA 14. Cordoba 13. Cordoba 3.6 U I T T I I TI T T 0 T T UA UA 36. Rio Cuarto 0.4 iT T T T T UA UA 0.06.Parana 0.6 R2 R2 R2 R2 T T T T T T T T T T UA UA 14. Corcordia 0.4 T T T T T UA UA 42 La Panta Rosa 0.2 U Ri T T T T T T T T T T T T T T UA UA 82. Santa Fe 04. Rosario 3.4 U RI R2 R2 T R2 T T T T T T T T T T UA UA 05. Santa Fe R2 R2 R2 R2 R2 T T T T T T T T T T T UA UA Patagonia 4.2 1.3 26. Chubut 09. Com.Rivadavia 0.4 0 0 R2 0 1 T T T T T T T T UA UA 58. Neuquen 17.Neuquen 0.6 U U T T T T T7T T T T T T T T T UA UA 78. Santa Cruz 20. Rio Gallegos 0.2 U U T T U T T T T T T T IT T T T UA UA 94. T. del Fuego 3T1. delFuego0.2 0 R2 R2 R2 | T T T T T T T T T I UA 0 Note: Nomenclatures for data file formats are as follows: T = base de trabajo, U base usuaria, P2= base usuaria reducida (R2), UA= base usuaria ampliada; and 0 = not available. 49 Argentina Poverty Assessment Background Paper No. 1 Annex Table 2: Adult Equivalent Basic Food Basket for the Current Poverty Line, MBA Component Grams Specification Bread 6060 Salad 420 Sweets 720 Rice 630 Wheat Flour 1020 Other Flour (Corn) 210 Noodle 1290 Potatoes 7050 Sweet Potatoes 690 Sugar 1440 Sweets 240 of Milk (Dulce de Leche) of Sweet Potatoes of Marmalade Dried Legume 240 Lentil Beans Peas Vegetables 3930 Beet Onion Lettuce Tomato Carrot Pumpkin Canned Tomato Fruits 4020 Banana Tangerine Apple Orange Meat 6270 Roasted Carnaza Came Picada Hipbone Falda con Hueso Buttock Shoulder Blade Chicken Eggs 630 Milk 7950 Cheese 270 Fresh Cream Cuatirolo De Rayar Oil 1200 Mixed Soft Drinks 4050 Juice for Dilute Carbonated Drinks Carbonated Soft Drinks 3450 Soda Fine Salt 150 Coarse Salt 90 Vinegar 90 Coffee 60 Tea 60 Yerba 600 Sources: Elena B.A. de Morales, Canasta bdsica de alimentos: Gran Buenos Aires, INDEC Documento de Trabajo No. 3, Buenos Aires, 1988. CEPA, Evolucion reciente de lapobreza en el Gran Buenos Aires, 1988-1992, Documento de Trabajo No. 2, Ministerio de Economia, 1993. 50 Argentina Poverty Assessment Background Paper No. 1 Annex Table 3: A Matrix of Adult Equivalence Coefficients Energy Needs Adult Equivalent Age and Gender Activity Level (K Cal) Coefficient CHILDREN UNDER 10 Children of 6 - 12 months 880 0.33 Children 1 year 1,170 0.43 Children 2 year 1,360 0.50 Children 3 year 1,500 0.56 Children 4-6 year 1,710 0.63 Children 7-9 year 1,950 0.72 MALE ADOLECENT 10 - 12 years 2,230 0.83 13 - 15 years 2,580 0.96 16 - 17 years 2,840 1.05 FEMALE ADOLECENT 10 - 12 years 1,980 0.73 13 - 15 years 2,140 0.79 16 - 17 years 2,140 0.79 MALE ADULT Light 2,490 0.92 18 - 29 years Moderate 2,860 1.06 Intense 3,260 1.21 30 - 59 years Light 2,350 0.87 Moderate 2,700 1.00 Intense 3,070 1.14 60 and more Light 1,930 0.71 Moderate 2,210 0.82 Intense 2,520 0.93 FEMALE ADULT Light 1,950 0.72 18 - 29 years Moderate 2,000 0.74 Intense 2,260 0.84 30 - 59 years Light 1,950 0.72 Moderate 2,000 0.74 Intense 2,260 0.84 60 and more Light 1,690 0.63 Moderate 1,730 0.64 Intense 1,960 0.73 Pregnant Light 2,230 0.83 (over 18 years old) Moderate 2,290 0.85 Intense 2,590 0.94 Lactating Light 2,450 0.91 (over 18 years old) Moderate 2,500 0.93 Intense 2,760 1.02 Sources: Elena B.A. de Morales, Canasta bdsica de alimentos: Gran Buenos Aires, INDEC Documento de Trabajo No. 3, Buenos Aires, 1988. CEPA, Evolucion reciente de lapobreza en el Gran Buenos Aires, 1988-1992, Documento de Trabajo No. 2, Ministerio de Economia, 1993. 51 Argentina Poverty Assessment Background Paper No. 1 Annex Table 4: CPIs for the Metropolitan Buenos Aires (MBA), 1980-1998 General Consumer Price Index General CPI CPI for the Poor (CPI) Year Reference Currency Adult Eq. Overall Food Non-food Overall Basic Food Non-food Period a/ Unit Basic Food Basket Necessities Basket (100.0) (40.1) (59.9) (40.4) (1988=100) (Sept.98=100) (1988=100) 1980 Sept. Pesos 118,927 0.004351 0.004197 0.004454 0.000001327 0.004035 0.003861 Leyes 1985 Sept. Australes 18.01 6.910 6.456 7.214 0.00210805 6.111 6.004 1988 March Australes 149.2 51.41 49.86 52.45 0.0156838 50.64 50.86 April Australes 173.4 60.27 57.83 61.90 0.0183867 58.84 60.72 Sept. Australes 427.6 147.4 144.7 149.1 0.0449524 145.1 151.9 1989 April Australes 978.3 337.9 340.8 335.9 0.103078 331.9 361.2 Sept. Australes 15,045 5,784 5,500 5,974 1.764416 5,105 5,647 1990 April Australes 137,139 57,454 50,050 62,410 17.5276 46,529 57,688 Sept. Australes 245,789 109,926 98,784 117,386 33.5354 83,393 100,890 1991 April Australes 400,212 210,832 172,847 236,261 64.3191 135,786 199,627 Sept. Australes 487,457 236,393 208,479 255,080 72.1170 165,387 208,038 1992 April Pesos 55.5 263,5161 237,633 280,844 80.3915 188,338 224,402 Sept. Pesos 58.0 278,917 250,748 297,774 85.0898 196,616 233,963 1993 April Pesos 60.9 294,236 262,325 315,599 89.7632 206,591 247,494 Sept. Pesos 62.4 303,661 271,419 325,246 92.6386 211,850 257,778 1994 April Pesos 61.6 306,981 266,038 334,390 93.6513 208,966 263,116 Sept. Pesos 62.8 314,876 273,073 342,861 96.0599 213,173 275,012 1995 April Pesos 64.8 321,252 276,805 351,006 98.0049 219,993 281,189 Sept. Pesos 66.1 321,706 278,057 350,927 98.1436 224,336 282,502 1996 April Pesos 65.9 320,571 274,473 351,431 97.7972 223,454 281,504 Sept. Pesos 67.4 322,365 280,111 350,651 98.3445 228,645 280,938 1997 April Pesos 65.4 322,643 271,900 356,613 98.4295 221,825 286,529 July Pesos 65.9 323,827 273,819 357,306 98.7908 223,555 286,015 Sept. Pesos 67.4 324,205 277,354 355,569 98.9059 228,577 285,354 1998 April Pesos 68.3 326,418 278,184 358,708 99.5811 231,664 291,906 Sept. Pesos 69.0 327,791 282,866 357,866 100.00 234,107 290,220 Source: INDEC, INDEC Informa, 1980-1998. Note: a/ Incomes reported in the survey data refer to those earned one month prior to the survey period. Accordingly, poverty lines will be defined for the month which precedes the fieldwork period by one month. Therefore, the reference periods are given by: (i) April and September each year for the Permanent Household Surveys (EPHs), and: (ii) July 1997 for the Social Development Survey (EDS). In the case of the 1996-97 National Expenditure Surveys (ENGHO), it was made to coincide with the second round of the EPH (september 1996). 52 Argentina Poverty Assessment Background Paper No. 1 Annex Table 5: Adult Equivalent Poverty Lines for the Metropolitan Buenos Aires (MBA), 1980-1998 Current Poverty Lines (Adult Male Equivalent) Year Reference Currency Indigent Non-food Poverty Implicit Indigent Poverty Official Period a! Unit Lines (A) Items (B) Lines Engel Lines Lines Poverty (C)=(A+B) Coeff. Rate (Current price) (A)/(C) (Sept.1998 price) 1980 Sept. Pesos Leyes 118,927 121,234 240,161 0.495 89.6 180.9 -- 1985 Sept. Australes 18.01 18.85 36.86 0.489 85.4 174.9 -- 1988 March b/ Australes 149.24 159.69 308.93 0.483 95.2 197.0 -- April Australes 173.41 190.66 364.07 0.476 94.3 198.0 29.9 Sept. Australes 427.61 477.06 904.67 0.473 95.1 201.3 32.4 1989 April Australes 978.28 1,134.18 2,112.46 0.463 94.9 204.9 25.9 Sept. Australes 15,045 17,731 32,776 0.459 85.3 185.8 47.4 1990 April Australes 137,139 181,135 318,273 0.431 78.2 181.6 42.6 Sept. Australes 245,789 316,785 562,574 0.437 73.3 167.8 33.8 1991 April Australes 400,212 626,810 1,027,022 0.390 62.2 159.7 28.8 Sept. Australes 487,457 653,221 1,140,678 0.427 67.6 158.2 21.6 1992 April Pesos 55.5 70.5 126.0 0.441 69.0 156.7 19.3 Sept. Pesos 58.0 73.5 131.4 0.441 68.1 154.4 17.8 1993 April Pesos 60.9 77.7 138.6 0.439 67.8 154.4 17.8 Sept. Pesos 62.4 80.9 143.4 0.435 67.4 154.8 16.9 1994 April Pesos 61.6 82.6 144.2 0.427 65.8 154.0 16.1 Sept. Pesos 62.8 86.4 149.2 0.421 65.4 155.3 19.0 1995 April Pesos 64.8 88.3 153.1 0.423 66.2 156.2 22.2 Sept. Pesos 66.1 88.7 154.8 0.427 67.4 157.8 24.8 1996 April Pesos 65.9 88.4 154.2 0.427 67.3 157.7 26.7 Sept. Pesos 67.4 88.2 155.6 0.433 68.5 158.2 27.9 1997 April Pesos 65.4 90.0 155.3 0.421 66.4 157.8 26.3 July Pesos 65.9 89.8 155.7 0.423 66.7 157.6 -- Sept. Pesos 67.4 89.6 157.0 0.429 68.1 158.7 -- 1998 April Pesos 68.3 91.7 159.9 0.427 68.6 160.6 -- Sept. Pesos 69.0 91.1 160.1 0.431 69.0 160.1 -- Source: Ministerio de Economia, Economic Report, Second Quarter, 1997, Number 22, October 1997, Buenos Aires. Notes: a/ Reference time for those poverty lines which correspond to each survey is the month preceding the survey period by one month: (i) April and September each year for the Permanent Household Surveys (EPHs), and July 1997 for the Social Development Survey (EDS). In the case of the 1996-97 National Income and Expenditure Survey, the reference period is one month before the mid-point of the fieldwork period (August 1996). b/ The Engel Coefficient of 0.483 (inverse=2.07) was used to calculate poverty lines for March 1988. Annex Table 6: Adult Equivalent Indigent Lines by Region, 1980-1998 53 Argentina Poverty Assessment Background Paper No. 1 Indigent Lines by Region (Adult Male Equivalent) Year Reference Currency Metropolitan North West North East Cuyo Pampeana Patagonia Period a! Unit Buenos Aires (Current price) 1980 Sept. Pesos Leyes 118,927 116,474 112,577 107,670 114,453 113,587 1985 Sept. Australes 18.010 17.638 17.048 16.305 17.332 17.201 1988 March b/ Australes 149.2 146.2 141.3 135.1 143.6 142.5 April Australes 173.4 169.8 164.2 157.0 166.9 165.6 Sept. Australes 427.6 418.8 404.8 387.1 411.5 408.4 1989 April Australes 978.3 958.1 926.0 885.7 941.5 934.4 Sept. Australes 15,045 14,735 14,242 13,621 14,479 14,369 1990 April Australes 137,139 134,310 129,816 124,157 131,979 130,981 Sept. Australes 245,789 240,718 232,664 222,522 236,542 234,752 1991 April Australes 400,212 391,955 378,841 362,328 385,155 382,241 Sept. Australes 487,457 477,400 461,428 441,314 469,118 465,569 1992 April Pesos 55.5 54.4 52.5 50.3 53.4 53.0 Sept. Pesos 58.0 56.8 54.9 52.5 55.8 55.3 1993 April Pesos 60.9 59.6 57.6 55.1 58.6 58.2 Sept. Pesos 62.4 61.2 59.1 56.5 60.1 59.6 1994 April Pesos 61.6 60.3 58.3 55.8 59.3 58.8 Sept. Pesos 62.8 61.5 59.5 56.9 60.5 60.0 1995 April Pesos 64.8 63.5 61.4 58.7 62.4 61.9 Sept. Pesos 66.1 64.8 62.6 59.9 63.6 63.2 1996 April Pesos 65.9 64.5 62.3 59.6 63.4 62.9 Sept. Pesos 67.4 66.0 63.8 61.0 64.9 64.4 1997 April Pesos 65.4 64.0 61.9 59.2 62.9 62.4 July Pesos 65.9 64.5 62.4 59.7 63.4 62.9 Sept. Pesos 67.4 66.0 63.8 61.0 64.8 64.3 1998 April Pesos 68.3 66.9 64.6 61.8 65.7 65.2 Sept. Pesos 69.0 67.6 65.3 62.5 66.4 65.9 Source: NDEC, Linea de indigencia para el hombre adulto, mimeo, 1980-1998. SIEMPRO, Avances de las estimaciones de linea de pobreza, (documento preliminar), 1999. Note: a/ Reference time for those poverty lines which correspond to each survey is the month preceding the survey period by one month: (i) April and September each year for the Permanent Household Surveys (EPHs), and July 1997 for the Social Development Survey (EDS). In the case of the 1996-97 National Income and Expenditure Survey, the reference period is one month before the mid-point of the fieldwork Income and Expenditure Survey, the reference period is one month before the mid-point of the fieldwork period (August 1996). 54 Argentina Poverty Assessment Background Paper No. I Annex Table 7: Non-Food Consumption in Adult Equivalent Poverty Lines, Indexed by Non- Food Necessities CPI, 1980-1998 Non-food Component in Poverty Lines by Region (Adult Male Equivalent) Year Reference Currency Unit Metropolitan North West North East Cuyo Pampeana Patagonia Period a/ Buenos Aires (Current price) 1980 Sept. Pesos Leyes 121,234 113,383 103,452 117,390 102,502 119,419 1985 Sept. Australes 18.852 17.631 16.087 18.255 15.939 18.570 1988 March b/ Australes 159.7 149.3 136.3 154.6 135.0 157.3 April Australes 190.7 178.3 162.7 184.6 161.2 187.8 Sept. Australes 477.1 446.2 407.1 461.9 403.4 469.9 1989 April Australes 1,134.2 1,060.7 967.8 1,098.2 958.9 1,117.2 Sept. Australes 17,731 16,582 15,130 17,169 14,991 17,465 1990 April Australes 181,135 169,405 154,568 175,392 153,149 178,423 Sept. Australes 316,785 296,271 270,322 306,741 267,840 312,042 1991 April Australes 626,810 586,219 534,876 606,937 529,966 617,426 Sept. Australes 653,221 610,920 557,413 632,510 552,296 643,442 1992 April Pesos 70.5 65.9 60.1 68.2 59.6 69.4 Sept. Pesos 73.5 68.7 62.7 71.1 62.1 72.4 1993 April Pesos 77.7 72.7 66.3 75.2 65.7 76.5 Sept. Pesos 80.9 75.7 69.1 78.4 68.4 79.7 1994 April Pesos 82.6 77.3 70.5 80.0 69.9 81.4 Sept. Pesos 86.4 80.8 73.7 83.6 73.0 85.1 1995 April Pesos 88.3 82.6 75.3 85.5 74.6 87.0 Sept. Pesos 88.7 83.0 75.7 85.9 75.0 87.4 1996 April Pesos 88.4 82.7 75.4 85.6 74.7 87.1 Sept. Pesos 88.2 82.5 75.3 85.4 74.6 86.9 1997 April Pesos 90.0 84.1 76.8 87.1 76.1 88.6 July Pesos 89.8 84.0 76.6 87.0 75.9 88.5 Sept. Pesos 89.6 83.8 76.5 86.8 75.8 88.3 1998 April Pesos 91.7 85.7 78.2 88.7 77.5 90.3 Sept. Pesos 91.1 85.2 77.8 88.2 77.0 89.8 Source: INDEC, Linea de indigencia para el hombre adulto, mimeo, 1980-1998. SIEMPRO, Avances de las estimaciones de linea de pobreza, (documento preliminar), 1999. Note: a! Reference time for those poverty lines which correspond to each survey is the month preceding the survey period by one month: (i) April and September each year for the Permanent Household Surveys (EPHs), and July 1997 for the Social Development Survey (EDS). In the case of the 1996-97 National Income and Expenditure Survey, the reference period is one month before the mid-point of the fieldwork period (August 1996). 55 Argentina Poverty Assessment Background Paper No. 1 Annex Table 8: Adult Equivalent Poverty Lines by Region, 1980-1998 Poverty Lines by Region (Adult Male Equivalent) Year Reference Currency Metropolitan North West North East Cuyo Pampeana Patagonia Period a! Unit Buenos Aires (Current price) 1980 Sept. Pesos Leyes 240,161 229,856 216,029 225,059 216,955 233,006 1985 Sept. Australes 36.862 35.270 33.135 34.560 33.272 35.771 1988 March b/ Australes 308.93 295.51 277.54 289.74 278.64 299.83 April Australes 364.07 348.15 326.85 341.61 328.09 353.43 Sept. Australes 904.67 864.96 811.87 849.07 814.88 878.33 1989 April Australes 2,112.5 2,018.8 1,893.9 1,983.9 1,900.4 2,051.6 Sept. Australes 32,776 31,317 29,372 30,789 29,470 31,835 1990 April Australes 318,273 303,714 284,383 299,549 285,128 309,404 Sept. Australes 562,574 536,989 502,986 529,263 504,382 546,795 1991 April Australes 1,027,022 978,174 913,717 969,264 915,121 999,667 Sept. Australes 1,140,678 1,088,320 1,018,841 1,073,824 1,021,414 1,109,010 1992 April Pesos 126.0 120.3 112.7 118.5 113.0 122.4 Sept. Pesos 131.4 125.5 117.5 123.6 117.9 127.7 1993 April Pesos 138.6 132.3 124.0 130.4 124.3 134.7 Sept. Pesos 143.4 136.9 128.2 134.9 128.5 139.4 1994 April Pesos 144.2 137.6 128.8 135.8 129.1 140.2 Sept. Pesos 149.2 142.3 133.2 140.5 133.5 145.1 1995 April Pesos 153.1 146.1 136.7 144.2 137.1 148.9 Sept. Pesos 154.8 147.7 138.3 145.8 138.6 150.5 1996 April Pesos 154.2 147.2 137.8 145.2 138.1 150.0 Sept. Pesos 155.6 148.5 139.1 146.4 139.4 151.3 1997 April Pesos 155.3 148.2 138.7 146.3 139.0 151.1 July Pesos 155.7 148.5 139.0 146.6 139.3 151.4 Sept. Pesos 157.0 149.8 140.2 147.8 140.6 152.6 1998 April Pesos 159.9 152.6 142.8 150.6 143.2 155.5 Sept. Pesos 160.1 152.8 143.1 150.7 143.5 155.7 Source: INDEC, Linea de indigenciapara el hombre adulto, mimeo, 1980-1998. SIEMPRO, Avances de las estimaciones de linea de pobreza, (documento preliminar), 1999. Note: a/ Reference time for those poverty lines which correspond to each survey is the month preceding the survey period by one month: (i) April and September each year for the Permanent Household Surveys (EPHs), and July 1997 for the Social Development Survey (EDS). In the case of the 1996-97 National Income and Expenditure Survey, the reference period is one month before the mid-point of the fieldwork period (August 1996). 56 Argentina Poverty Assessment Background Paper No. I Annex Table 9: Implicit Engel Coefficients by Region, 1980-1998 Implicit Engel Coefficients in Poverty Lines by Region Year Reference Metropolitan North West North East Cuyo Pampeana Patagonia Period a/ Buenos Aires 1980 Sept. 0.495 0.507 0.521 0.478 0.528 0.487 1985 Sept. 0.489 0.500 0.515 0.472 0.521 0.481 1988 March 0.483 0.495 0.509 0.466 0.515 0.475 April 0.476 0.488 0.502 0.460 0.509 0.469 Sept. 0.473 0.484 0.499 0.456 0.505 0.465 1989 April 0.463 0.475 0.489 0.446 0.495 0.455 Sept. 0.459 0.470 0.485 0.442 0.491 0.451 1990 April 0.431 0.442 0.456 0.414 0.463 0.423 Sept. 0.437 0.448 0.463 0.420 0.469 0.429 1991 April 0.390 0.401 0.415 0.374 0.421 0.382 Sept. 0.427 0.439 0.453 0.411 0.459 0.420 1992 April 0.441 0.452 0.466 0.424 0.473 0.433 Sept. 0.441 0.452 0.467 0.424 0.473 0.433 1993 April 0.439 0.451 0.465 0.423 0.471 0.432 Sept. 0.435 0.447 0.461 0.419 0.468 0.428 1994 April 0.427 0.438 0.453 0.411 0.459 0.420 Sept. 0.421 0.432 0.447 0.405 0.453 0.414 1995 April 0.423 0.435 0.449 0.407 0.455 0.416 Sept. 0.427 0.438 0.453 0.411 0.459 0.420 1996 April 0.427 0.438 0.453 0.411 0.459 0.419 Sept. 0.433 0.444 0.459 0.417 0.465 0.426 1997 April 0.421 0.432 0.446 0.405 0.453 0.413 July 0.423 0.434 0.449 0.407 0.455 0.416 Sept. 0.429 0.441 0.455 0.413 0.461 0.422 1998 April 0.427 0.438 0.452 0.411 0.459 0.419 Sept. 0.431 0.442 0.457 0.415 0.463 0.423 Source: INDEC, Linea de indigencia para el hombre adulto, mimeo, 1980-1998. SIEMPRO, Avances de las estimaciones de linea depobreza, (documento preliminar), 1999. 57 Argentina Poverty Assessment Background Paper No. 1 Annex Table 10: Earnings Equation Estimation to Prediet Earnings Among the Unemployed, 1990-1998 North North Year Month Intercept S S2 Exp Exp2 Female West East Cuyo Pampeana Patagonia N Adj-R2 (t-statistics in parentheses) 1990 May 4.58 0.082 0.0010 0.053 -0.00083 -0.531 - -- - -- - 2,404 0.1383 (28.82) (2.88) (0.78) (9.27) (-7.61) (-10.33) October 4.83 0.072 0.0007 0.051 -0.00075 -0.494 -- - - -0.117 - 4,037 0.1930 (51.15) (4.35) (0.90) (15.98) (-12.43) (-17.00) (-4.06) 1991 May 4.93 0.050 0.0020 0.052 -0.00075 -0.436 - - -0.391 -0.086 0.137 9,361 0.1904 (72.46) (4.23) (3.65) (22.22) (-16.92) (-20.87) (-12.47) (-3.37) (5.00) October 5.03 0.063 0.0011 0.050 -0.00074 -0.432 -0.395 -- -0.329 -0.149 0.108 13,807 0.2944 (120.19) (8.78) (3.36) (35.67) (-27.59) (-34.65) (-22.46) (-13.75) (-8.55) (5.52) 1992 May 5.21 0.056 0.0013 0.048 -0.00071 -0.434 -0.473 -- -0.392 -0.268 0.198 19,592 0.3467 (163.47) (10.50) (5.17) (45.71) (-36.34) (-46.02) (-29.60) (-24.56) (-18.45) (12.21) October 5.13 0.067 0.0010 0.050 -0.00074 -0.434 -0.532 -- -0.451 -0.278 0.172 21,014 0.3756 (169.35) (13.18) (4.16) (49.52) (-38.21) (-47.88) (-36.13) (-26.49) (-20.42) (11.89) 1993 May 5.19 0.054 0.0015 0.051 -0.00076 -0.438 -0.480 -- -0.290 -0.267 0.193 23,260 0.3594 (174.50) (10.80) (6.67) (53.18) (-41.79) (-50.85) (-35.10) (-17.96) (-21.32) (14.06) October 5.22 0.052 0.0016 0.051 -0.00076 -0.439 -0.504 -- -0.292 -0.249 0.162 23,137 0.3615 (178.32) (10.54) (7.22) (52.79) (-41.52) (-51.29) (-36.81) (-18.24) (-19.84) (11.93) 1994 May 5.16 0.059 0.0015 0.051 -0.00075 -0.437 -0.473 -0.488 -0.350 -0.227 0.167 29,293 0.3724 (195.93) (13.48) (7.10) (58.67) (-44.66) (-56.75) (-37.50) (-29.91) (-24.74) (-18.29) (12.45) October 5.17 0.051 0.0020 0.050 -0.00074 -0.423 -0.444 -0.399 -0.349 -0.240 0.173 27,846 0.3685 (190.57) (11.36) (9.56) (55.41) (-42.05) (-53.53) (-34.52) (-19.27) (-24.24) (-18.82) (12.61) 1995 May 5.12 0.035 0.0027 0.052 -0.00076 -0.410 -0.431 -0.455 -0.316 -0.191 0.254 26,998 0.3695 (186.05) (7.98) (12.98) (55.78) (-41.95) (-49.91) (-31.09) (-26.94) (-21.39) (-14.51) (18.05) October 5.02 0.048 0.0022 0.050 -0.00072 -0.422 -0.411 -0.396 -0.369 -0.190 0.279 31,259 0.3589 (189.86) (11.41) (11.23) (56.07) (-41.54) (-53.44) (-28.70) (-27.13) (-24.37) (-14.54) (18.89) 1996 May 4.97 0.054 0.0019 0.052 -0.00075 -0.406 -0.349 -0.414 -0.353 -0.221 0.261 32,884 0.3476 (188.16) (12.85) (9.94) (58.31) (-43.52) (-52.52) (-25.07) (-28.09) (-23.00) (-16.77) (17.48) October 4.93 0.053 0.0021 0.050 -0.00072 -0.410 -0.358 -0.425 -0.377 -0.190 0.229 29,553 0.3350 (173.03) (11.69) (9.88) (51.84) (-38.15) (-48.78) (-24.62) (-25.47) (-23.76) (-13.43) (14.79) 1997 May 4.87 0.055 0.0021 0.054 -0.00078 -0.441 -0.344 -0.413 -0.335 -0.196 0.253 33,227 0.3523 (180.51) (12.78) (10.87) (59.31) (-44.14) (-55.28) (-24.49) (-27.60) (-21.52) (-14.82) (16.82) October 4.89 0.053 0.0022 0.052 -0.00074 -0.437 -0.361 -0.382 -0.308 -0.193 0.258 33,241 0.3421 (176.58) (11.92) (11.08) (57.61) (-42.89) (-54.14) (-25.36) (-25.05) (-19.55) (-14.33) (17.12) 1998 May 4.90 0.055 0.0023 0.054 -0.00076 -0.464 -0.386 -0.437 -0.346 -0.234 0.187 32,491 0.3541 (177.45) (12.44) (11.30) (59.51) (-44.00) (-57.00) (-27.50) (-28.13) (-22.32) (-17.74) (12.25) October 4.92 0.047 0.0027 0.051 -0.00072 -0.479 -0.379 -0.401 -0.333 -0.207 0.065 28,782 0.3393 (167.64) (9.92) (12.33) (51.92) (-38.30) (-54.92) (-25.81) (-25.31) (-21.33) (-15.36) (3.69) Sources: INDEC, Encuesta Permanente de Hogares (EPH), 1990-1998. 58 Argentina Poverty Assessment Background Paper No. 1 Annex Table 11: Distribution of Adult Equivalent Per Capita Income in Argentina, 1990-1998 Adult Equivalent Per Capita Income Mean Income Gini Coefficient of Income Inequality Survey Period Year GBA All Urban Areas GBA All Urban Areas (Sept. 1998 Peso) 1980 October 604.1 -- 0.383 -- 1985 October 477.0 -- 0.400 -- 1988 May 434.9 -- 0.456 -- October 384.7 -- 0.445 -- 1989 May 456.0 -- 0.458 -- October 324.4 -- 0.506 -- 1990 May 324.5 288.8 0.465 0.458 October 311.0 290.6 0.475 0.453 1991 May 357.0 311.4 0.454 0.460 October 401.5 348.3 0.453 0.449 1992 May 407.2 354.0 0.421 0.434 October 437.9 375.1 0.433 0.446 1993 May 464.6 397.4 0.448 0.453 October 449.5 391.9 0.435 0.445 1994 May 471.7 410.3 0.438 0.448 October 443.0 390.6 0.450 0.455 1995 May 415.7 364.2 0.452 0.456 October 420.0 358.5 0.477 0.473 1996 May 405.7 350.6 0.468 0.470 October 408.9 351.2 0.477 0.476 1997 May 416.9 358.8 0.479 0.479 October 430.0 368.3 0.476 0.477 1998 May 452.6 386.0 0.486 0.485 October 443.9 377.5 0.494 0.491 Sources: INDEC, Encuesta Permanente de Hogares, 1980-1998. 59 Argentina Poverty Assessment Background Paper No. 2 Poor People in a Rich Country A Poverty Report for Argentina Background Paper No. 2 Perceptions and Self-Perceptions of Poverty in Argentina Sandra Cesilini and Enrique Zuleta May 1999 1 Argentina Poverty Assessment Background Paper No. 2 This report attempts to develop a face of poverty rarely explored in Argentina. It is an analysis of how the poor see themselves, how they are viewed by other social sectors and the elements associated with the definition of poverty. This study used both it's own data sources as well as data from a variety of institutions, such as universities and NGOs. The study's own data was based on an exploratory qualitative and quantitative study, , which intended to address the following issues: -Identify the different ways that the social sectors define poverty. -Characterize the interests and the prevailing opinions of the different social sectors in regards to the changes in the poverty situation produced over the last few years. -Identify the areas of institutional responsibility for poverty. -Identify sectors that are in position to resolve the primary problems associated with poverty. -Analyze factors associated with poverty and exclusion of the extreme poor. -Explore the survival strategies of the extreme poor and those of the "new poor." More specifically, the study has focused on identifying: 1. the way that the participants characterize poverty; its defining characteristics (income, exclusion, unemployment, discrimination, etc.) 2. the expectations of the population regarding the experienced socio-economic changes and their impact on individual and family conditions. 3. the significance of social variables (socioeconomic level, gender, occupation, etc.) on the creation of current opinions in the population. 4. the interactions between the -poor (both the structural poor and the income poor) and the participants with the capacity to resolve the problems of poverty. 5. definitions and opinions associated with poverty, and their differences in rural and urban areas. 6. the methods of participation used by the impoverished populations that have led to compromises in the judicial reform process. 7. the amount of information about the reform process available to the individual participants. The quantitative portion of the study was performed during the last two weeks of December 1998, and the qualitative portion was performed during the second two weeks of February and the first two weeks of March 1999. The basic background studies originate from the consultant Sofres-Ibope and from the study "Consultations with the Poor on the theme of Social Capital as part of a World Bank study in twenty countries, for the World Development Report 2000. 2 Argentina Poverty Assessment Background Paper No. 2 PERCEPTIONS, POVERTY'S HISTORY AND NATIONAL PARTICULARITIES In an analysis of social exclusion (The Social Exclusion FIACSO/Guidelines, 1996) Juan Villareal speaks of the history of "poverty" , referring to the historic evolution of the social roles played by poverty, marginalization and the different forms of social exclusion. Villarreal's opinion is that in Argentina and in Latin America, these different concepts have superimposed themselves much like geological layers, thus forming a fractured society. The study, carried out using a questionnaire given to twelve hundred people, focus groups and participation studies (see the appendix, tech); has as a goal this search for the different types of poverty that have accumulated in the collective imagination. This has encompassed the characterization of the interests and prevailing opinions on the subject, the detection of areas of institutional responsibility and competence in the development of specific social policies and the analysis of factors associated with poverty and exclusion in the case of sectors of extreme poverty. The issues analyzed reflect, in general, the new agenda of concerns about poverty in relation to inequality and exclusion in urban areas. In a way, it tries to respond to some of the main conclusions of the recent Forum on Poverty, Inequality and Vulnerability, organized by the World Bank in Buenos Aires on October 19th and 20th, 1998. The abrupt changes which have taken place throughout the last decade in Argentina (market liberalization, privatization, reduction of the state apparatus, etc.), entail a challenge to compare their impact at a subjective level, which implies the need to accomplish a systematic effort to evaluate the impact on poverty as well as the fairness of the diverse reforms and the character of the institutional innovations that are being developed in the region. Traditionally, there has been a tendency to focus social development programs on services such as education, health care or drinking water. Over time, however, new services--such as justice and citizen safety--have begun to be recognized as essential. In the same manner, although in a strict sense they can't be regarded as services, access to housing and credit also constitute critical elements in the establishment of necessary strategies to fight poverty. This emphasis on the provision of services as a strategic aspect in the fight against poverty is consistent, both with an approach focused on basic needs (that emphasizes the consumption condition of these services) as well as an approach that underlines the fact that these services constitute assets of diverse order in a way that improving the access to the services allows for: the accumulation of human capital (for example, in the case of education), infrastructures (for example, housing) and equity (for example, security). At the same time, looking at it from a social exclusion point of view suggests that greater equality in access to basic services is a precondition to promoting greater social cohesion. Much like the rest of the countries in Latin America, Argentina is developing a series of reforms that can be summed up in four main groupings: decentralization, privatization, participation and private-public alliances. Decentralization implies the transfer of responsibility over providing services from a 3 Argentina Poverty Assessment Background Paper No. 2 national level, to a sub-national state level (the province, the municipality).Privatization implies the transfer of responsibility for the provision of services from the state to the private sector (particularly toward that one looking for financial gain).By participation, it is understood that institutional reforms are intending to promote the role of the beneficiaries in the effort made to deal with the services. Finally, he private-public alliances constitute the creation of institutional forms that transcend the traditional separation state-market-civil society, or the dichotomy state-market or benefit-no-benefit, adopting models using joint management. In all these areas, one can find examples of reform and innovation that, although not always motivated by the objective of fighting poverty--have the potential to radically change the ways in which in which services are provided, improving or worsening the access of popular services to them. The New Poverty Studies such as this one specifically attempt to investigate the effect that social reforms have over the social segments that were supposedly, in principle, on the margin of the principal costs of structural reforms, but who in fact suffer difficulties in adapting to the new social conditions. The primary focus of interest falls on situations stereotyped as "new poverty," which is generated by the impact of economic reforms, such as those imposed on Argentina over the last few years. Among the teachings and suggestions that are offered up by the most recent poverty studies, it is important to keep in mind, for the purposes of the design of studies such as this one, some central themes: In the first place, it is necessary to recognize that the standard of living-and as result poverty-is a multi-dimensional reality. This implies a special urgency in the overcoming of traditional approaches inspired by methodological perspectives which are unilaterally economic. In analyzing poverty and inequality situations, it is essential to introduce complementary indicators, pertaining to health, education, vulnerability and risk levels, and relationships between poverty, exclusion and the increase of urban insecurity Secondly, there is a clear demand for approaches capable of developing a micro point of view, intended to empirically reveal the actual self-perceptions and the poverty awareness of those who find themselves in situations of "new poverty." That implies a frank and integral recourse to the knowledge and instruments of the rest of the social sciences; particularly the quantitative and qualitative analysis of the social demands and expectations. In third place an interest in the opinion of the poor on poverty, and their evaluation of the social interventions, public and private, oriented to help them. The self-perception of poverty and new poverty is one of the principal centerpieces of interest for this investigation. 4 Argentina Poverty Assessment Background Paper No. 2 Precisely from that point, the option to carry out a survey over the general population, complemented with a specific study of the sub-sample made up of those whom recognize themselves as subjected to poverty conditions, and oriented towards knowing, in detail, their perceptions about poverty and their evaluation of the compensatory interventions, not only public but also private. The investigation's theoretical frame has generated a body of basic questions, referring to the impact of the institutional reforms, their sustainability and their indirect effect on the relationship between social actors and the necessary process of change. Among these questions, it is worth noting the following: a) What has been the impact of specific institutional reforms--for example, the privatization of public services-- on the access to basic services by different social groups? b) What aspects of those institutional reforms explain such results? c) What degree of sustainability do such reforms have; not only from a financial point of view but also from a political one? d) Are effects noted only short term--as a result of the reforms introduced? Or, on the other hand, do they represent a deep lasting deterioration in the established standards? e) What are the principal effects of the reforms in question? f) How can other initiatives either strengthen or facilitate the fight against poverty and social vulnerability? g) Is it possible to develop compensatory mechanisms that will increase the voice of socially-excluded sectors; and therefore alter the local political dynamics without the need to impose bureaucratic rules that are generally relatively easy to disregard? h) How to guarantee the transparency and efficiency in the process of transferring the responsibility for the provision of services from the government to the private sector? i) What institutional designs are the most conducive for the development of the conditions needed for the development of authorities and capacities, and accountability mechanisms that make these schemes efficient mechanisms to fight poverty and exclusion in a sustainable mode? j) What institutions possess the social legitimacy necessary to face compensatory efforts like those demanded by the new situations? The dimensions of the definition of Poverty The analysis is based on an open question to all of society's sectors, as to how they would define poverty. The following table shows the categorization of the questions, realized by Sofres-Ibope 5. Argentina Poverty Assessment Background Paper No. 2 General definition of the poverty situation TABLE No 1 DEFINITION OF TOTAL SOCIO-ECONOMIC LEVEL High Medium Low UNEMPLOYMENT 20.6 16.0 17.9 22.9 LABOR-RELATED PROBLEMS 2.6 3.3 1.6 3.1 TO BE UNDER-EMPLOYED 0.5 0.0 0.5 0.6 UNABLE TO PROVIDE FOR HIS (HER) FAMILY 7.9 6.1 8.7 7.7 NOT HAVING ENOUGH TO EAT 6.7 4.8 6.3 7.2 NOT HAVING THE MEANS NECESSARY TO PROVIDE 25.6 38.8 27.7 22.0 FOR THE BASIC NEEDS UNABLE TO MAKE USE OF PUBLIC SERVICES 0.1 0.0 0.3 0.0 LACKING IN EVERYTHING 13.6 13.1 13.4 13.9 LACKING IN CASH FLOW 3.9 1.3 4.2 4.1 HAVING A LOW SALARY 2.1 0.5 2.4 2.2 MEMBER OF THE WORKING CLASS 0.5 0.0 0.3 0.7 UNABLE TO EDUCATE/BETIER ONESELF 2.6 8.7 2.2 1.7 LACK OF OPORTUNITIES 2.6 1.3 2.1 3.1 LACK OF VALUES 1.0 0.0 1.9 0.6 BE UNHAPPY, LACK OF FRIENDS 1.1 0.0 2.7 0.4 LACK OF SPIRITUAL VALUES 2.0 3.9 2.0 1.7 BE DISCRIMINATED AGAINST, MARGINALIZED 1.7 0.0 1.6 2.1 NOT HAVING ACCESS TO APPROPIATE HEALTHCARE 0.7 0.0 0.7 0.7 NOT HAVING A PLACE TO LIVE 1.7 1.9 0.6 2.2 LIVING IN A SMALL TOWN 0.9 0.0 0.7 1.2 UNWORTHY RETIREMENT 0.0 0.0 0.0 0.0 OTHER 0.1 0.0 0.0 0.1 DON'T KNOW/NO ANSWER 1.7 0.4 2.0 1.8 To those interviewed, the general definition of what constitutes poverty implies a first step in the process of identifying the different ways in which the distinct social sectors define poverty. 6 Argentina Poverty Assessment Background Paper No. 2 Table I highlights the results to the questions that are meant to convey what is personally considered as "poverty." A descriptive definition is required, referring to economic and social circumstances that are indicative of poverty. The two most important answers are without a doubt those that refer to the notions of "unsatisfied basic needs" (25.6) and "unemployment" (20.6). The rest of the questions fall into a series of susceptible options, even though they are grouped together. The analysis of the data provided by the Socio-Economic Level (SEL) is very interesting. The lack of work adds up to 16.0 in those high SEL sectors, while it rises to 22.9 in low SEL sectors. The unsatisfied basic needs rise to 38.8 in the high SEL sectors, and drops to 22.0 in the low SEL sector. Those sectors with a lower socio-economic are correlated with conditions of structural deficiency that form part of the structural context. What is new for them is the crisis in labor opportunities, which leads to the distribution described. An accurate illustration of what has been said is the reference to the difficulty of access to education and training (2.6). The perception of the importance in the deficiencies and educational shortages rise to 8.7 in the high SELs and falls to 1.7 in the Low. Using this table, it is easy to make a comparison with the Latin-American context, where access to education services, health services and urban infrastructure plays a large role in the definition of poverty, while in Argentina, where the hypothesis is that the access is not restricted to any social sector; and, therefore, defining characteristic of poverty is unemployment. By reviewing the Table, we could establish the following standard of definitions: The answers related to the employment problem comprise 26.3 % of the answers for all the SEL. Following in order of importance; not having the means necessary to cover basic needs 25.6%, which could easily be associated with employment situations. The rest of the answers are very diverse. While analyzing these two SEL, we can see that employment has much more weight for the low SEL, while not being able to provide for the basic needs, is a preeminent answer in the high and medium sectors. This allows for the establishment of a primary work hypothesis. For the poor the basic definition of their condition is unemployment, while for those falling in the high SEL, spiritual values and education rise strongly. Here is where you begin to notice the fragmentation of the different sectors' perceptions. 2.Self-perception of poverty Table 2 registers the responses to the question relating to self-perspective in regards to the poverty situation of those being interviewed. The Table constitutes a second step in the analytical strategy adopted by the investigation, and is based in a simple question: Do you consider yourself as being poor? TABLE No 2 DO YOU CONSIDER YOURSELF POOR? 7 Argentina Poverty Assessment Background Paper No. 2 TOTAL SEX AGE SEL MALE FEMALE 18-29 30-49 50-64 65- HIGH MEDIUM LOW YES 25.9 24.4 27.3 18.1 30.6 25.6 31.4 2.1 17.1 35.3 NO 70.0 70.5 69.6 77.9 65.2 69.8 65.9 96.3 80.1 59.4 DON'T 4.1 5.1 3.1 4.0 4.3 4.6 2.7 1.6 2.8 5.2 KNOW/NO ANSWER Positive responses registered a 25.9, as a national average, A very similar figure as that registered in the metropolitan survey. One fourth of the Argentine population considers itself poor. The percentage is higher amongst women (27.3) than men (24.4), however this difference is not statistically significant, and can result from characteristics of sample. Even so, the most significant analysis is the one referred to the SEL. The national average of 25.9 it raises up to 35.3 in the Lower SEL and it drops to 2.1 in the High SEL and 17.1 en the Middle SEL. This implies that 17.1% of the respondents located in the middle7- sector consider themselves poor. Notably, 59.4 % of the people with low SEL .don't consider themselves poor. This phenomena is related to the new poverty, such that it appears that almost one fifth of the people with high and middle level are considered poor, while those which we characterized as poor because of their SEL do not perceive themselves that way. The "stigma" of poverty has a much larger impact on the poor, and in this sense the "pride to see oneself included in the poor category" does not exist. For the new poor, on the other hand, there does exist some kind of rebellion against their new situation that translates in the decision to classify itself below its SEL. Based on some of the positions presented at the First International Congress for the Poor and Poverty in the Argentinean Society (Analysis of the valued structures in the marginal sectors: new poor and structural poor in the bell party, Duprat and Aguirre, November 1997); "the value structure of the new poor coincides with the values traditionally categorized by the medium stratums", and coincides with family structure (Poverty is related primarily with income). In the case of structural poor, the valuation of needs and the family structure is different (larger quantity of children and over crowding). Follow S. Alvarez (The naturalization of poverty, and the other face of exclusion) it is interesting at this point to analyze "the actual manifestations of poverty linked to some form of curse, reconsidering the involved social groups in these categories," there is also some discussion of the classification provoked by the stereotyping used by specialists. In this interplay the self- perception is resisting a classification as is considered from the classical classifications of SEL. In relation to age and given the insufficient retirement income (around $200 per month), it is logical to find that a third of the population perceives itself as being poor. The 8 Argentina Poverty Assessment Background Paper No. 2 affected sector is that in the 30-49 age group, where family responsibility and unemployment can increase the chance of low self-perception. 3. General characterization of the poverty situation. Table 3 analyzes the identity of perceptions, reasons, and operating values in the process of "social construction" that all self-perception, in the end, implies. The answers belong to the group of interviewees out of the whole sample that perceive themselves as poor (25.9) TABLE No 3 REASONS WHY YOU CONSIDER YOURSELF POOR BASE: THOSE THAT CONSIDER THEMSELVES POOR (25,9% OF THE TOTAL) (THE PERCENTAGES DO NOT ADD UP TO 100%, MORE THAN ONE ANSWER WAS ACCEPTED) TOTAL UNEMPLOYED 31.2 EXPERIENCING LABOR PROBLEMS 6.6 UNDER-EMPLOYED 0.6 UNABLE TO SUPPORT DEPENDANTS 4.4 DON'T HAVE ENOUGH TO EAT 2.4 UNABLE TO PROVIDE FOR THE BASIC NEEDS 14.2 NO ACCESS TO PUBLIC SERVICES 1.2 LACK OF EVERYTHING 3.3 LACK OF MONEY 13.4 LOW SALARY 18.1 BELONG TO THE WORKING CLASS 2.8 UNABLE TO GET EDUCATION/TRAINNING 1.0 LACK OF OPPORTUNITY FOR ADVANCEMENT 3.6 LACK OF VALUES 0.1 UNHAPPY, LACK OF AFFECTION, FRIENDS 0.2 LACK OF SPIRITUAL VALUES 0.8 DISCRIMINATED AGAINST/MARGINALIZED 0.8 NOT HAVING ACCESS TO APPROPRIATE HEALTH CARE 1.3 DON'T HAVE A PLACE TO LIVE 3.4 LIVE IN A SMALL VILLAGE 0.0 UNWORTHY RETIREMENT 7.3 OTHERS 0.4 DON'T KNOWINO ANSWER 0.1 The distribution of answers confirms the results from the previous statements. In the 9 Argentina Poverty Assessment Backgrounid Paper No. 2 sectors that perceive themselves as poor, the perception of lack of labor and labor insecurity, associated with the impact of new social and economic conditions which have occurred in the last years, is more important than the perception that they are suffering structural conditions that prevent the adequate satisfaction of basic needs that exist in other social levels. 38.4 percent of the answers were centered on the labor subject (Bialakawsky, Kurvsky, Bilder and Chan, UNQUI, id.ant.). The new forms of precarious and unstable employment impact the construction of identities and the formation of social ties and, particularly, impact the image of possible social mobility and "a future". This might explain the enormous weight of labor in poverty's perception, without considering the impact provided by education, public services, etc. The tradition of social mobility in Argentina is broken by unemployment; and, thus, a vision of being poor due to a lack of a future arises, this is particularly important for the youth In a Social evaluation carried out for the project PRODYMES III, (WORLD BANK, 1997) this was the principal explanation used to demonstrate why the youth perceive themselves as poor. This affirmation coincides with the results of the participation diagnostic, carried out by the SDS. The adolescent population in the neighborhoods of the periphery from the large cities (march 1998), where the most common answer for the population in this sector it is to not know what their future is going to be in the next 4 to 10 years. A subject that must be evaluated as it was mentioned in "Sin Trabajo", is the characteristics of unemployment and its effects in the Argentinean society (Beccaria and L6pez) (Beccaria y L6pez researchers, UNICEF/LOSADA, 1996.) It mentions that parents' sufferings could stay "inscribed" as a character disturbance in the following generation, and therefore imply as a consequence losses of great magnitude on the social capital, fragmentation, frustration and very difficult problems caused because of the relation between identity and employment. EFFICIENCY IN THE FIGHT AGAINST POVERTY Within the frame of these answers, one can observe the enormous prestige of the church as an institution destined to fighting poverty. The Church and Caritas (the church NGO with the most prestige, trajectory and national coverage that depends individually from each one of the country's parishes) concentrate 29.6%, taking most of the preferences; the national government concentrates 23.1% in second place, and then a great dispersion is noted. Contrary to the decentralization policies carried out throughout the country in the last decade (especially in education and health) people's perception assigns a place of preponderance to the central level. The image of effectiveness in the fight against poverty could be related to two different phenomena: food services and social protection services provided by the church, and the fight against unemployment and the factors that cause poverty by the national government. The NGOs concentrate (if we add Caritas to other NGOs) 22.7%, reaching almost the same percentage as the national government. Another possible interpretation is the tension between the state and the private sector from the efficiency standpoint. Adding all the levels of government, the evaluation of efficient institutions in the first place is 31.4 % against 49.1 % from the total spectrum of the civil society organizations shown in Table 4. This should indicate the need to promote a larger decentralization in fighting poverty towards civil society, building consensus around a new way to make policies. 10 Argentina Poverty Assessment Background Paper No. 2 TABLE No 4 INSTITUTIONS CONSIDERED EFFECTIVE IN FIGHTING POVERTY IST PLACE 2ND PLACE 3RD PLACE TRADE UNIONS 3.5 4.0 5.9 POLITICAL PARTIES 4.1 8.0 6.9 CATHOLIC CHURCH 14.0 14.1 12.2 OTHER CHURCHES 1.5 4.1 5.6 NON-GOVERNMENT 7.1 9.9 8.5 ORGANIZATIONS MUNICIPALITIES 4.3 7.6 20.3 STATE GOVERNMENT 4.0 23.1 9.6 NATIONAL GOVERNMENT 23.1 8.0 8.6 INTERNATIONAL 2.7 5.4 9.9 ORGANIZATIONS CARITAS 15.6 15.8 12.5 OTHERS 0.5 0.0 0.0 NONE 14.7 0.0 0.0 DON'T KNOW/NO ANSWER 4.8 0.0 0.0 Around 15% express that there is no effective institution for the objective under analysis. This fact allows us to observe a resignation to the phenomenon of poverty, as well as a "natural acceptance" of it. 5. EVALUATION OF THE EVOLUTION OF POVERTY IN DURINGTHE LAST FIVE YEARS Around 46% see the situation as worse or much worse, while 35.3% consider it identical. Despite the registered results of growth at the macro-economic level during this period (see Argentine Economy, Economic Report N 27,1998, MEYOSP) Even tough we do not have final data about the this evaluation's reasons (the focus groups investigation is still pending) one of the strongest hypotheses is the relation between the extended family (meaning that the EF includes not only the family members, but also all the relatives, consanguineous or not) and the perception. The hypothesis at this point is that evaluation is not only restricted to family members, and 11 Argentina Poverty Assessment Background Paper No. 2 that, for example, loss of employment or the retirement benefit by one member of the extended family, promotes a solidarity network that directly impacts the worker's salaries, who must totally or partially support EF members. This hypothesis would help to explain the result difference between women and men. Women assess the situation as being worse 52.2 % of the time, while among men such an evaluation falls to a 38.7 (13.5 % difference). In EF, generally, the woman is the responsible for all the solidarity relations. In those homes with a woman as head of the household (where income provisions comes form women without a spouse) an income fall is noted. The most affected were women from the lower stratum which do not have enough opportunity to work as both domestic workers and providers of income for the family. As a consequence of this situation various studies on gender have concluded that in many countries of the region, a gradual "feminization of poverty" has taken place (Source: Latin American women in Ciphers", FLACSO.) TABLE No. 5 RETROSPECTIVE EVALUATION OF THE HOUSEHOLD SITUATION IN THE LAST FIVE YEARS TOTAL SEX SEL MALE FEMALE HIGH MEDIUM LOW MUCH BETTER 2.3 3.6 1.2 7.3 3.1 1.0 BETTER 15.3 18.0 13.0 23.5 17.8 12.5 THE SAME 35.3 38.3 32.5 36.8 34.5 35.4 WORSE 37.5 32.2 42.2 28.4 35.0 40.6 MUCH WORSE 8.4 6.5 10.0 3.3 8.9 8.9 DONT KNOW/NO 1.2 1.4 1.1 0.6 0.8 1.6 ANSWER 12 Argentina Poverty Assessment Background Paper No. 2 TABLE No 6 RETROSPECTIVE EVALUATION OF HOUSE HOLD INCOME LEVELS OVER THE LAST FIVE YEARS TOTAL SEX SEL MALE FEMALE HIGH MEDIUM LOW MUCH BETTER 1.5 1.9 1.1 2.8 2.5 0.6 BETTER 13.9 15.5 12.5 23.9 15.9 10.9 THE SAME 34.3 37.8 31.1 39.6 31.3 35.1 WORSE 39.0 35.1 42.4 29.3 37.8 41.3 MUCH WORSE 9.8 8.1 11.3 3.8 10.2 10.7 DON'T KNOW/NO 1.6 1.6 1.5 0.6 2.3 1.3 ANSWER The general perception is negative. 37.5% considers itself to be in a worse situation than five years ago, this is added to the 8.4 that consider themselves much worse. Only 17.6% (Table #5.) shares a feeling of optimism. This perception does not vary when analyzed by income or by general situation (1-% difference.) This negative perception in respect to the home situation is very similar to the medium and lower levels (48% in medium SEL vs. 52% in low SEL), in regards to the high SEL this percentage is reduced to 33.1%, which, either way is a very negative evaluation. INCOME NECESSARY TO COVER BASIC NEEDS The income structure considered as necessary to satisfy basic needs; set up as a minimum demand, fully coincides with the cost data of "the family basic basket". 35.5% of the population finds itself below the monthly $1000, and only 16% considers itself able to attend to their basic needs under Argentina's current minimum wage. -Table #7. 13 Argentina Poverty Assessment Background Paper No.2 TABLE No 7 MONTHLY INCOME CONSIDERED NECESSARY TO MEET THE BASIC NEEDS TOTAL SEX SEL MALE FEMALE HIGH MEDIUM LOW 600 OR LESS 16.0 14.9 16.9 3.5 4.9 24.7 601-900 PESOS 12.3 11.0 13.5 2.7 7.0 17.1 901-1000 PESOS 17.2 13.7 20.3 8.4 15.3 19.9 1001-1499 PESOS 7.3 9.0 5.8 5.5 7.4 7.5 1500 PESOS 13.0 15.4 10.9 10.1 17.2 11.1 1501-2000 PESOS 15.7 17.6 14.0 24.7 21.9 10.5 MORE THAN 2000 PESOS 13.9 14.2 13.6 40.7 21.0 4.9 DON'T KNOW/NO 4.7 4.2 5.1 4.3 5.3 4.4 ANSWER In this sense, one of the basic findings of the study are those of table 8. Almost a quarter of the population declares itself to be at an income level below that of the minimum wage. It lacks as a consequence the possibility of meeting its minimum necessities, a cipher that coincides, approximately with the percentage of those that perceive themselves to be in a poverty situation. The attached tables allow a profile of the characterization variables of this important segment of Argentine society. 14 Argentina Poverty Assessment Background Paper No. 2 TABLE No 8 TOTAL MONTHLY FAMILY INCOME TOTAL SEL HIGH MEDIUM LOW LESS THAN 200 PESOS 7.8 0.0 1.5 12.8 200-400 PESOS 15.0 0.0 6.1 22.9 401-500 PESOS 8.1 0.0 6.9 10.2 501-600 PESOS 7.9 2.2 5.2 10.5 601-700 PESOS 5.6 0.4 3.7 7.7 701-800 PESOS 5.0 0.9 6.0 5.1 801-900 PESOS 4.4 2.7 6.0 3.7 901-1000 PESOS 7.1 2.4 11.9 5.0 1001-1300 PESOS 6.1 4.3 10.2 3.9 1301-1600 PESOS 6.2 13.3 8.6 3.5 1601-2000 PESOS 4.5 11.9 7.4 1.5 MORE THAN 2000 PESOS 6.1 32.8 7.6 0.5 NO ANSWER 16.4 29.1 18.9 12.6 SOCIAL EXCLUSION AND POVERTY Social exclusion risk factors, other than just poverty in the strictest sense appear particularly intense. This is particularly true for the large cities, according to an analysis performed by SIEMPRO through CERFE. Where diverse social exclusion factors are analyzed (see annex 2), it is concluded that "the confrontation between the index classification of social exclusion ISE, and that of the UBN index (unsatisfied basic needs) provided evidence of. . . some situations where the presence of important urban agglomerations (Gran C6rdoba, Gran Salta, Santiago del Estero - La Banda, Corrientes, Santa Rosa, Rio Gallegos) can worsen the corresponding province's position in the ISE index in respect to the UBN index. 15 Argentina Poverty Assessment Background Paper No. 2 The cases of cities like C6rdoba (and Santa Rosa y Rio Gallegos as well), show that the measurement of social exclusion allows one to gather a series of information relative to various aspects of the difficulties comparable to the individual (analyzed in great part via the disqualification macro-process), and also the unfavorable conditions of the environment where he/she lives and works in. The possibility of gaining access to a good level of income and a good home, as could be in the case of C6rdoba, ends up being (negatively) related and measured by the information provided by other risk factors present primarily in metropolitan areas; factors that the ISE index can measure and synthesize It could then be considered, that the combined risk factors to related metropolitan expansion represents a potentially harmful "cluster of variables" leading towards a future of impoverishment for groups of the population, that will also affect the urban medium segments, which usually are not taken into consideration for the design of social programs. Such consideration becomes more important due to the fact that in Argentina, the urban population represents 87% of the total. INFORMATION PRODUCED FROM THE QUALITATIVE METHODOLOGY APLICATION Given the difficulty of analyzing the multi-dimensionality that emerges from qualitative studies and the large quantity of focus groups conducted conclusions on welfare and poverty's causes and consequences will be summarized; as well as trends in welfare and poverty over the last ten years (note that the question in the survey was to include a period of five years) and institutions' evaluations with conclusions split by rural and urban zones. The focus groups conducted as a result of the poll support the conclusions obtained through the use of the guide for consultations with the poor that is being undertaken simultaneously in twenty countries, and whose results we are waiting to compare with those in Argentina. PERCEPTIONS OF WELFARE AND POVERTY The definitions for Welfare as outlined by the groups, make reference to the access to the means necessary for survival. Some of them were: "have work", "have food", "education for the children", "to have all that is necessary to survive", "live well-off", "to have services--water, light (power), gas." The criteria perceived, both by the urban and rural groups, as those associated with categories (social groups) of Welfare can be gathered in seven groups: Community Infra-structure: lack of pavement (dirt roads in bad shape), lack of potable and running water (community water holes), lack of power in rural areas, lack of public lighting (urban areas), lack of natural gas, lack of public telephones, lack or shortage of public transportation, lack of small businesses. Type of housing: of wood, cardboard, tin; in rural areas of adobe and straw, of cement; without direct access to water (must carry it), with direct access to water. In rural areas, without power (candles or lanterns.) 16 Argentina Poverty Assessment Background Paper No. 2 - Access to social services: the criteria varies depending on the existence of certain institutions such as the different levels of formal education (kindergarten, elementary and high school); in the different health levels, in the different care and nutrition levels of children (Daycare centers and public kitchens.) * Isolation due to Distance: communities removed from the areas which supply social services , appear in two levels, the first pertain so much to the urban as to the rural zones, close to a road but far from the supply center. And the second one pertaining to the rural, not only far away from the supply center but also form the roads that lead to the supply center. * Environmental drainage conditions: floods muddy when it rains. - Work conditions: unemployed, odd jobs, seasonal workers, migrant workers (rural areas), government plans, and steady work. Conditions of land or parcel ownership: not the owners occupy the land, private or government parcels. The majority of these Welfare perceptions, or better said its counter-balance Discomfort; are mentioned as cause and consequence of Poverty. TENDENCIES PERCEIVED OVER THE LAST TEN YEARS. The expansion in the coverage of social services is notable as a general element of the improvement of quality of life on the poorest sectors in urban and rural areas. As soon as the deterioration of the unemployment and security appear , unemployment and instability emerge again within the population of urban areas. This also affects the value of the rural areas' production. The conclusion is one of uncertainty for the consulted, and there is a notable attachment to religious convictions in order to generate optimism in the urban areas, as well as strengthening of the organizations in the rural areas.. Urban Area Improved: The community infrastructure and the access to social services. Population growth. Worsened: the value of currency and the fundamental lack of work and security. Conclusion: "TODAY WE ARE WORSE OFF." UNCERTAIN OR PESSIMISTIC FUTURE because "one can study and get a job, but nothing is guaranteed," and the lack of compromise by the politicians and/or the government. THE OPTIMISM or hope is on God, even though a religious distinction never appeared in the criteria. 17 Argentina Poverty Assessment Background Paper No. 2 kural Area Improved: some social services and community organization. Worsened: The value of agricultural products (like cotton), and that of farm animals, and furs from hunted animals. Conclusion: "TODAY WE ARE WORSE OFF. (The only hope is to strengthen our organizations) PRINCIPAL PROBLEMS PERCEIVED Most of the PROBLEMS prioritized, as well as the welfare categories, are the principal CAUSE AND CONSEQUENCE OF POVERTY, as you will see. In all areas: The lack of work and it's consequences on the shortage of economic resources needed to live with dignity, independent of the state's furnishings as far as social services and infra-structure; appear prioritized by the population, regardless of the methodology used (polls, focus groups, etc.) If we discriminate by the relevant population's residence zone, the perception of problems associated to poverty can be summarized in the following: In the urban areas: Insecurity: robbery, assault, death, violence in general, drugs related delinquency and violence. In rural areas: Isolation of the urban zones and the lack of access to services and infrastructure. CAUSES AND CONSEQUENCES OF POVERTY In order to systematize this subject's information, an ad hoc category was made up of the causes and consequences derived from the groups' relevant information. Nonetheless, find attached, as a table, the expressions used by the different groups' participants; on their perception of the causes and consequences of poverty. 18 Argentina Poverty Assessment Background Paper No. 2 URBAN AREAS In terms of self-perception, free associations of synonyms and poverty definitions were constructed in order to then analyze poverty's causes and consequences, resulting in the following findings in urban areas: Poverty definitions and synonyms: Humble people-Chaotic Situation-More Poverty- Live worse off-Critical situation-Lack of dignity-Malaria- (referring to misery as slang) The perceived CAUSES can be grouped in the following categories: Government Institutions: mentioned in 12 different ways, including the corruption phenomenon, to the assignment of responsibility in the government levels. Unemployment: mentioned in 13 different ways, with variations because of situations of impact in the analyzed areas, associated with privatization as well as modernization and to the lack of investment in employment generation. Different employment barriers (age, appearance, and education) are also considered under other categories were also specified. Education and professional training: mentioned in 9 different ways, specially those associated with the difficulty of going and breaking the inter-generational duplication of poverty. Immigration: mentioned four times, refers to the immigration from bordering countries, as well as that from other states. Within the groups, simultaneously there appeared a lack of control over immigration (especially in the focus groups situated in border areas) as a responsibility of the state. The population increases in new shantytowns and a new wave of rural migration (as in Misiones and Santa Fe.) Housing and infrastructure: appears mentioned in three different ways, and does not appear to constitute a central problem as far as the casual definition of poverty in urban areas, rather it is seen as a consequence. Other: the great existing variety includes a range that goes from solidarity to discrimination, frustration, an excess of children and people's own difficulties in handling their resources. 19 Argentina Poverty Assessment Background Paper No. 2 The CONSEQUENCES The consequences allude to the poverty situation's impact on the every day life of the poor consulted in the various groups. To establish a sequence of the causes/consequences among each of the mentioned causes and each of the mentioned consequences is very difficult. Which is why it has been decided to analyze and categorize the consequences independently. Family Violence Mentioned in thirteen different ways, it appears as the principal consequence derived from poverty. It has a great variety of dimensions as a category, ranging from man's role to women's abandonment of the children. It can be associated to the Depression/frustration category, which includes a scale of consequences from unwillingness, to a tendency of domestic violence and impotence. Both categories could be easily associated and combined. Health Poverty's consequences on health appear very strong (ten mentions), from the deficit in hospital care to malnutrition, shortage of medicine, number of available services at the neighborhood level, new illnesses, etc. Addictions: Alcoholism and drug addictions, according to how they're categorized can be included under health. Insecurity/violence/security forces and repression: these consequences are difficult to categorize, given that the poor tend to perceive themselves equally as the receptors of institutional violence and as victims (four mentions in respect to the increase of insecurity.) Rural Areas In terms of self-perception, construction of free associations of poverty's synonyms and definitions was allowed in order to then analyze the causes and consequences, with the subsequent results in the rural areas as the following: Poverty's definitions/synonyms: Poverty's synonyms that appeared in the groups: Deficient quality of life. Difficult situations. The perceived CAUSES can be grouped in the following categories: Government Institutions: mentioned in six different ways, to include from the corruption phenomenon to the determination of responsibility assigned by government levels. Valuable Dimension: Much more visible than in urban areas, where it was seen as associated with the consequences; here appears as a cause in six different ways, such as dependence, conformism, vagrancy, loss of values, etc. Community Ties, the united 20 Argentina Poverty Assessment Background Paper No. 2 community and the organization could be considered under this VALUABLE DIMENSION. Problems in reference to production: five mentions refer to product prices and the lack of access to resources and credit. Unemployment: mentioned in five different ways in each sector/area. Unlike in the urban areas, the investment on the employment generation does not rest upon the government, rather it is focused towards the assistance of the product-generating sector. Education and professional training: mentioned in two different ways, more likely to be associated to the difficulty of transportation access, to distance and the consequences of the rural life, than to poverty in of itself. Infrastructure only appears mentioned in relation to the lack of transportation. The perceived CONSEQUENCES can be categorized as: Health: Consequences associated to this category are mentioned in nine different ways. This means that there is a very detailed description of the illnesses that are perceived as a consequence of poverty, and in some the limit of association to this category is difficult to define. For example, to perceive pregnancy as a consequence of poverty is not necessarily to associate it with health issues. Violence/delinquency/addictions: With seven different ways of being mentioned, it appears as an important consequence, but without being defined as family violence as in the urban areas. Violence, in all it's forms, although it is mentioned, it is perceived as a phenomenon of urban poverty, not rural. Infrastructure: Mentioned six times, it appears as the most visible consequence of poverty to the population. It also can be associated with other dimensions, such as health and education where the lack of services appear as consequences. Lack of basic inputs: Mentioned four times, from food to clothing and shoes. Education: Mentioned four times, same as the data found in the questionnaire; in the sectors suffering from extreme poverty, the lack of education is neither seen as a cause or a consequence. Access to land: Mentioned three times. It is the lack of land or the means to acquire land. As well as farming tools, it is a dimension of poverty that is perceived as much as a cause or consequence. It is a characteristic more specific to rural areas. 21 Argentina Poverty Assessment Back,round Paper No. 2 INSTITUTION'S IMAGES As much as in urban communities as in rural ones, institutions are valued according to the positive answers they provide to the demands and immediate needs of the population: "if they don't help, they are worthless." That said, since the majority of relevant communities are poor, have historically suffered, and constitute the least favored part of a society that is unable to meet its most basic needs, the strong presence of one or more institutions in the community generates a perception of social restraint. An important difference is observed between urban and rural communities. In the latter, there appears an institution playing an almost fatherly role, around which the community gains cohesion. Urban Zone The heterogeneity of the mentioned institutions, and the degree of importance placed on them, change from one group to the next, sometimes even in groups within the same community. Regardless of this, it can be noted that the institutions mentioned in all the groups that are perceived positively are: Kindergartens/Schools/Colleges: because they play a double role, as providers of knowledge and providers of food: "educate", "have cafeterias", "give out glasses of milk." Emergency Rooms/First AidefHospitals: "freely attend to the poor and provide medications", "save lives." Other institutions mentioned in some of the groups were: Community Cafeterias/Kitchens: because they assist the neighborhood's families with a daily meal (positively valued.) Churches: are only recognized as needed to provided religious rites such as baptisms or masses (neutral value, save in those communities where the priest or father is himself a strong institution.) The institution perceived negatively in all the groups is: Precinct/Police: "security does not exist", "they don't provide security, nor do they do anything", "treat us badly", "discriminate", "you'll never see them here", "you never see a patrol officer", "you will never see them arresting anyone", "they only show up after it's all over", "they treat abused women as if they were to blame." The negative image of this institution is seen on one hand as due to the lack of 22 Argentina Poverty Assessment Background Paper No. 2 control over crime, and on the other hand as the aggression and mistreatment towards the people in the neighborhood. The institutions related with the government Municipality / Local Province Government appear according to the community, as in a border zone, and end up being perceived in either a positive or a negative manner. However, people do assign them responsibility, as in "they have to install running water," "they have to install electric light (power)." Rural Zone As opposed to urban communities, in rural areas one can observe the presence of an institution that establishes a strong link to the community; it provides cohesion and through it, the common objectives of solidarity and improvement are pursued. Valued institutions according to rural communities: Villa Atamisqui (Santiago del Estero): the school, Juan XXIII appears as one of the strongest institutions in the community. Los Juries (Santiago del Estero): the Peasant Co-Operative Union appears as the backbone of this community. Through it, the farmer's rights are organized and defended. As interrelated and fortifying these institutions were perceived as important for the Neighborhood Commission and the Community Center. Isla Talavera (Zlrate-Campana Pcia. Bs.As.): School #25 appears as the base for this community, through it a wide web of provincial programs is instituted to bring work to the local population. In both of the santiguefia communities, the Municipality and the Police are perceived in a negative manner: The Municipality: "only makes appearances when they need our vote," "they don't attend us when we go to them," they don't resolve anything," "they make promises they don't keep." The Police: "are pretty much useless", "are not educated", "doesn't exist", "discriminates, they won't record your complaints." Among other valued institutions, local radio stations play a prominent part because through them the population is able to remain informed. 23 Argentina Poverty Assessment Background Paper No. 2 Interview with a poor, urban man. Manuel Lopez, 46 years old, seven sons. Why don't you tell me about the shantytown's characteristics? Look, some 148 families live here, mostly young people. We do not want it to be some emergency town. It can be a neighborhood, humble or poor, but a neighborhood nonetheless. Here, some thirty families, we're with the Work Plan. We're digging ditches, roads, paths and septic tanks. We're working on the matter of defining the lots and doing the land ownership transactions as per the expropriation laws. The land belongs to two people. Before it was a dump, a real rats' nest. Two years ago this was an area of violence and rape. But from the occupation on, we began to build huts of sheet metal, cardboard, wood. The electricity is precarious; we are in negotiations to resolve the situation with a plan by EDENOR. We have to make a presentation with plans; they are long and bureaucratic transactions. Look, in order to reassign the lots, we have to request the use of bulldozers from the municipality, and in the last three months we only got them to come work on one day, and that was for only four hours. Seventy five percent of the people have no jobs or are under-employed. And this isn't from just now, it comes from factories closing--with privatization there is no investment on new work sources, it's this neo-liberal politics of the last ten years. From politicians we cannot expect anything. They only give us something during the elections; there is a great lack of sensibility towards the people. I came from El Tambo; it's a shantytown on government lands, on September 1, 1986. Now there are small, real houses. We got the titles in September 1994. Is there a way out? No, no there is no way out. Although I'm an optimist there are no social policies to help the poor, those of little recourses. The Work Plan is mostly a subsidy, we have lost our dignity--we have no shoes, food is lacking, we can't send our kids to school, we have no access to credit. How do you see the situation for the young? For the youngest, there is a possibility as long as we have the minimum resources in order for them to be able to study; otherwise they will not have a way out. My parents always told me the same thing, they gave me the tools but I did not want to take advantage. I studied two years of art drawing and publicity. Did you see the auto ads, Peugeot, Ford--I did those, but I was unable to graduate. The institutions have no credibility; the politicians try to corrupt us. It's all a big corruption, before they teach them how to talk they are taught how to make money the easy way, so that each one can make his own. We all fall into that, we don't do anything differently, and I'm talking about generating our own resources. The theme is always--you have so much, you're worth so much--my kids come from school and they say that they want certain shoe brands, certain name brand clothing. It's 24 Argentina Poverty Assessment Background Paper No. 2 a consumer society, it gets to the kids, and they learn it from when they're young. My little seven-year-old girl is a great contradiction. Man alone doesn't learn, he is not born. There are some people who can do more, but they do not. That bothers me--that they'll have one or two pets--why don't they adopt? There are so many kids on the street, why don't they adopt one or two of them? There is a lack of respect for human life. Later, the government, that has all of the resources, doesn't do anything about it, only reformatories and overcrowding places. Rather than giving them a home, an education, a trade--they end up with a gun in hand out to rob people-they have no other option. I'll tell you one, yesterday there came a woman with three kids, she was 25 years old, had no home, had been thrown out. Your heart breaks. We told her there was no room for her here; she should go to Monica Carranza. Without resources, we cannot become responsible. It's crazy; there is more and more inequality. There are poor, there always were, but now there are those who are lacking everything, they have been excluded from all systems, they have no food for their children. To the people in the shanty towns, we tell them: you've been excluded, here we are organized, you'll have a place but you have to work, make the compromise to pay back, take charge, responsibility. It's an election, don't dwell on things coming from above. It takes more for some, less for others; but at least as they see that those next door are bettering themselves, then the man or the woman or even the child will do it, they will come out of their inertia. Inside the shanty town there are people that ended up here due to fear, or due to consequences, due to life's conditions. They take leave of their father or some relative. People here bank on harshness, because there are still people who need it. Just like everywhere else, there are scam artists that take advantage and try and take over a lot. You know who needs it, you also see it in the lack of participation, and the common worries of people don't interest anyone. I'm part of them, I come from El Tambo, and I had some experience. My daughter lives here; she is 21 years old and has a baby. It was a solution as much for her as for me, and we're not on top of each other. What does the commission take care of? The commission takes cares of, from pregnancies to problems in between couples, there are cases of abused women; it gets in contact with social workers so they will come and talk to them. We don't want to politicize, we don't know what kind of ideas they will bring, and we don't want it used as partisan interests. All that is social policies is welcome. Another issue is that of the ditches and paths. Here you have the plans; the side of the traffic circle is more complicated for relocating people. Some don't want this land because it's at the edge. The lots are only 160 square meters. We began with the triangle, followed the three, then the two and then the one. Those on the side are more complicated, so now we have seven or eight leftover families. Tito is our internal surveyor. The other important subject is that of housing. We want to build it ourselves, there is the possibility of credit, and we have to study, read plans. At this time, there is a girl looking for a place. Her cousin lives on the creek's shore, she 25 Argentina Poverty Assessment Background Paper No. 2 has no room. She is desperate, she interrupts us, and they try to calm her. Yes, we have provided a place for a school, in the kitchen's lot. We'll make a community space, where every day at least ten families look for a way out, God where are you? Interview of a poor woman, from rural area. Hipolita del Valle de Paez (POLI) 25 years old married two kids. How would you define good quality of life, or living well? A good quality of life for us would be having a good education. That the children get educated, that they have good health or good services in case the person is in need of medical attention. Around here we don't have it, we can only be treated in the hospital, and that is at 35 kilometers from here in the country, over in Los Juries. And the people out in the country have a need for this hospital, but at the moment of illness there is no good medical service. Another issue would be that the peasants' products have any worth, that they bring a good price. The animals are one of those things that the peasants live off, but today they have no worth. Neither the production nor the animals is worth anything. In this community, do you think there are people that live better off than others? Some can live better than others, but only a little bit better. They might have some things more than others might, I mean, there isn't much change in people that are this far down, with a lot of needs. Around here one of the families has a little house, better made than others. It is made of cement and brick, but in other cases they are little ranches made of mud and sticks, and in an instance of a lot of rain--like the storms we had last month, the houses all leaked, everything got ruined. At those times we needed the town hall (municipality) to give us some nylon (plastic) to cover the roofs, so the water won't leak through. We haven't had any help, they forgot us at that moment, and they forgot what we were going through. As far as the animals, for example, there are people that may have some, but very few because the lots are very small and they aren't enough to have animals to plow and plant. In this area it's happened to all of us, that people have to give away their animals because they were unable to keep them, and it was a disgrace for us, it was so sad to have to give the animals away. Because in order to be able to live, we couldn't have the animals, it was a real disappointment for us. What does it mean to give the animals away? I'm going to give you the animals, because you have a bigger lot. They have room to roam; they have enough to eat enough to drink. And there comes a time when more animals are going to be born, and then it will be time to give them out, if they are born. If ten are born, five are for you, and five are for me. There is also the problem of water. Some lots have very good water, and there are others that don't have any water. In our lots' case, wherever you look the water is salty; it's not good for anything, even for the animals, that is the case in my house. 26 Argentina Poverty Assessment Background Paper No. 2 This situation was it always like this or was it different before? Before it was a little different because we lived in a large field, when we had the problems we were relocated. Before we had a place to plant, a place for the animals, we could raise chickens, goats, sheep, cows, and horses. Besides, they were open fields, the animals were free and safe to roam around, they had plenty to eat, and the water was good. When we were evicted, we had to relocate. We didn't have any other choice. We were evicted from our lot, and so we ended up with this problem. This started in 1984. Who evicted you? Supposedly it was the owner, but now there is a large corporation, from Santa Fe. They're an oil corporation; they are working the land. They have leveled the hills, have burned it. Now they are planting grain that was in lot 40 it's very large; what's happening is that they are dividing it into parcels. How did you get here? The supposed owner, of lot 40, donated one thousand hectares in order to relocate 45 families in 1985, some with 18-hectare parcels, other with 22-hectare parcels and the maximum of 27 hectares. The parcels were given out according to the animals. To the people that didn't have animals--chickens, turkeys; were given 18 hectares. The one's who had some animals were given 22 hectares, and those that had a lot of animals were given 27 hectares. But the amount of hectares didn't matter much. It has been too little for everyone. What is needed for people to be able to get out of this situation and have more opportunities? The important thing would be to have a good government, because that is the biggest problem, by not having a good government we have nothing. There is no justice, justice does exist but it doesn't work. Like in the case of La Simona, it would have to be all solved in the case of a good government, good justice. In this case the governor is in place for all of us. We're the ones who voted him in, we suffer heat, cold, and so many things in order to vote for them so they can be where they are today, and they don't remember us. They take their post and forget all the peasants, the people. The governor, in the case of La Simona, is on the company's side and against the poor peasants. That's why that conflict continues, there is no solution, justice is on the side of the corporation. Do you feel forgotten by society? That is what we feel. I used to work with a program (Strengthening for the Civil Society), that enables us as community activists in our society. It is a very pretty program; it helps you a lot, and depends on the Santiago del Estero province's Secretary of Social Development. The objective of the program is to promote the organization, to make it grow, so it doesn't weaken. To strengthen the co-op created by the peasants that people unite that 27 Argentina Poverty Assessment BackruUnd Paper No. 2 they work in tandem for the community's good. That is the principal objective, so through these programs, the community can grow, it gives it dignity. Because we see that from the government we get nothing, it's very difficult to try and grow through the organization. Move forward, leave the bad times were having, through the organization we achieve .many things, such as fixing problems. We were able to bring the program Prohuerta to the community. It has progressed along well. It consists in that people receive their seeds for their gardens free. They are given it at no charge, then they can work on their land, tend to their garden, do everything they need to grow their own food so they don't have to be shopping; because people before they could eat a salad they would have to go into town. But not now. Now they go to their garden, pull their own vegetables and prepare their meals. It's a great help for the home, the children, and the elderly, for the whole family. I'm working as a promoter for ProHuerta, I'm the one who delivers the seeds, sets up the meetings, tour the gardens to see what people are and are not doing. Some families don't have water for their gardens. They don't have water, good water, so they have to bring it from afar. But it is not easy to bring it, so it is hard to have a good garden due to the lack of water. What is the FOPAR? The FOPAR came to the zone; they did a lot of things, valuable things. And in lot 40 they got involved in a project of five community wells over in the places where there is no water, so that the people that don't have water could draw it from the well and take it to their homes. That is what the project consisted of, but so far it hasn't happened. It was approved, but it hasn't arrived, nothing has been done. What are the most important problems? The water, not being able to raise animals because of the lack of land. You can only have the smaller ones, so they're not worth anything. The roads are a big problem. While one is healthy there is no problem, but in case of say snakebite it's not easy. You have to go out to Los Juries, and you hope to god it isn't storming or raining much. How does the sick person get out from here? By road it's impossible, a vehicle won't get through. You can get out on horseback, but how long does it take to go on horseback? The sick person could die on you. In the case of pregnancy, not all pregnancies are normal. Some don't come out right; they have too many problems. You could wait days, hours, night or day, and when the baby comes, if it comes, it could have problems. On a night like that, you could die. There is a clinic on this lot, but the doctor only comes when he remembers to. He doesn't have a set schedule. But now we're happy because here, at a two-kilometer distance we have a phone booth, it's been there more or less a year. That phone booth helps out in a lot of things. Things haven't changed. It has always been the same. In election years, when people from the municipality come here, they come on the day the candidates are here. They pat our backs, give us a kilo of sugar or mate. They trick you, lie to you, play on people's problems. The promise that "if we win, we'll make the roads, we'll raise the roads," "anything you need, just go to town hall, knock on the door; we're right there for you." These promises are false, they have been coming around here for years with the same 28 Argentina Poverty Assessment Background Paper No. 2 thing, and the roads remain the same. So when people need something, they go to the town hall, knock on the doors but there is no one to attend to them because they are not there, they're out, they have gone to the capital, no one knows anything. And the people around here are not in any condition to be going into town every day. If you go to the hospital, you have to take a number, so you go to the desk and see that the nurses are just chatting and you have to wait until the young ladies finish talking. And what are they talking about that we have to wait until they feel like coming over and giving us a number? But when it doesn't happen like that, they're there for the community's service, for the people. They're paid, I suppose to attend to the people, so they come and give you a number. And we ask, is the doctor in? And they say, no, no the doctor is not in. They lie to you. Something that happened to me once, they told me the doctor wasn't in. So I asked, when is he coming in? We don't know, they told me. So I stand to the side, and there comes an elderly woman who asks for a number and the same doctor I'm looking for- see they treat each other all friendly-and the woman asks if there is a number for the same doctor. Yes there is. So I go over to the counter, and I ask them, how come you told me there wasn't any numbers for that doctor? And she says, no ma'am, I must have missed you. I tell them, you know you're trying to trick me ... why don't you just give me the number, what difference is there between her and me? And they didn't know how to answer me. With the doctors is the same too, you ask them to see you and they don't respond. They tell you impolitely to wait a turn. Do you think they discriminate against you because you're peasants from the country? Yes, yes. The people in town have the possibility that since they live in town, they can go in the afternoon. We in the country wake up at six in the morning so we can take a bus and go see the doctor in the hospital. We get there at eight o'clock in the morning and some times you end up waiting up to thirteen hours sitting there, not being able to eat or drink anything. Sometimes, so as not to lose the time, you go with a friend or relative and you end up hours and hours, hungry. So you have to return without the doctor having seen you, and you miss the bus and have to hitchhike, get back somehow. You go out on the highway and hope that a pick-up or a truck will get you there partway, then get down and hitchhike again until you finally end up walking part of the way home. The hospital here is a very tough situation, we stopped dealing with the hospital. We had contact with a doctor that has a clinic out at Juries. That doctor, no matter what time you go, he'll see you, be it day or night, it doesn't matter what time it is. He'll wake up and see you. It's a clinic, so we go and tell him, look doctor we don't have enough for the appointment today, and he will tell you there's no problem, don't worry, come on in-and he sees us and takes care of us. Now he has a pharmacy (the doctor), and if we only have half of what we need for medications, he says there is no problem, take all of the medications that you need and you can pay me when you have it. We, for example, all the family are thankful for that doctor, that clinic. In the private clinic, how much is it for an appointment? At this moment, it's $20 for the appointment and the medications we have to pay. I have paid for her (the mother), taking her for check ups, she has gotten three medications that cost me $30. My mother has problems with Parkinson's, and her medications can cost up 29 Argentina Poverty Assessment Background Paper No. 2 to $80 because one of costs $40, another one costs $40, plus the $20 for the doctor and in twenty minutes she has spent $100. Do you think that these problems have a solution? I don't even know what to think. I have lived here for many years, and it never changes. And well, what we think is that when there is a change in government, that getting new people, that they are capable which is why they're in; so they'll run things well and do things the way they're supposed to be done and things would change, life would change. That's what I think, but without having good government, we don't know. What institutions are important or positive for the community? The schools, as much the elementary schools as the high schools. With education we could be educated, and the health, there would be hospitals. The co-op gave out credit in time of planting. They give out credit so that people can plow and plant. When the crop is in, you can have it cleaned and then wait for the co-op to buy your cotton. But around there, we're not all equal in the co-op. We need the co-op and take from it, but we need to see that the co-op needs us too because if we take and take from it, then it won't be able to grow. There are people that take their credit, but I go and take only one credit. I prepare everything, work my land, plant, clean and when the time comes to pick the crop I tell the co-op that I'm not going to sell to them. They pay twenty cents less, and I'm going to pay to whoever will pay more. In the end, it won't matter to the co-op whether or not I sell to them. They'll still make it, they'll go and sell it to another middleman, twenty or thirty cents more. But they'll sell it and then will forget that that production they were able to come up with and sell at a high price is because the co-op helped them and not another middleman. They think they're smart. There are people who will do this, and a few of them think that they screwed over the co-op. But they're the ones who are screwed, because after all the effort that went in to create this co-op, we get to see it back on the ground. We have to think since it was our effort, we should maintain it, help it grow, get strong so it won't fail, that is how I see it. Women's situation in relation to the past, is it better? In this case yes. Maybe women's life today is better than last year, because today women have participation. They can belong to women's groups and can take part in neighborhood commissions and the in the co-op's Council. Before, women were at home, nothing else. They were with the kids, and it was the men that went out. For me, with the start of this organization; when we began to organize ourselves and with the start of neighborhood commissions, that is where women's participation began. In many cases, it is the man who gives the woman participation. In my case, I have two little girls; I'm going out a lot. All for the good of the community. As a woman, I have a lot of participation. And in other lots there are many women in neighborhood commissions. But not all women are interested, they're working. Then there are women that I can't tell you they're not interested, or maybe their husband won't let them, or maybe I'll have a lot of desire to work, to participate but my husband won't let me. It is very difficult, one depends on one's husband. Even if we have the desire, if they don't let 30 Argentina Poverty Assessment Background Paper No. 2 us then we can't. Does that happen here, is there a lot of machismo? It doesn't exist much, not too much. Not really. But there are women who I don't understand whether they don't participate because they're not allowed, or because the woman herself doesn't want to take part. Around there we're closed, my husband lets me and I don't want to, or at home the family is very large what with little children, it's a problem. Do you want to add anything else? Yes, this program where I'm working as community activist, we work a lot with the social policies. This is the objective, to work together, for the whole community. We don't use partisan politics, where we're going we don't pay attention to race or color, nor what political party you're affiliated to, nor do we say we won't work with you. We want for the adults to work, to get organized and for them to better themselves. 31 Argentina Poverty Assessment Background Paper No. 3 Poor People in a Rich Country A Poverty Report for Argentina Background Paper No. 3 Employment, Informality and Poverty in the Argentine Labor Market Carlos Arango William F. Maloney The World Bank LCSPR January 31, 2000 Argentina Poverty Assessment Background Paper No. 3 I. Introduction This chapter has two objectives related to our understanding of poverty in Argentina. The bulk of the work is dedicated to understanding the relationship among the various sectors of the labor market and in particular, the role of the informal sector. This is central, first because of what it may reveal about the impact of rigidities in the labor code, and second because the informal sector has traditionally been seen as the most disadvantaged and vulnerable group. The second part of the paper more broadly studies the mobility patterns of workers among different income classes. This offers a dynamic view of poverty-ease of exit from poverty, and who is most likely to fall into it. Informality Roughly 54% of males and 62% of females in the urban labor force in Argentina is unprotected by labor legislation and works in small informal firms.' The origins and dynamics of this sector have attracted renewed interest for at least two reasons. First, increasing labor market efficiency and flexibility is considered an essential complement to the market-based reforms underway.2 To the degree that the sector results from segmentation driven by government or union imposed distortions, its large size stands as a measure of the magnitude of market inefficiencies and required reforms. ' Second, in this view the informal sector represents the disadvantaged sector of a dualistic labor market, and merits intensive focus from a poverty point of view. As one example among many, a recent World Bank document argues that "Protected workers in the 'modern' or 'formal' sector ... enjoy high wages, social security, vacation, pension and employment security as mandated by legislation. By contrast, those unable to find work in such firms resort to the next best alternative, the so-called 'informal sector' in small firms or self-employment, engaged in labor-intensive activities, without job security or benefits."(Ozorio de Almeida et al. 1995, p.1). With wages in the formal See, for example, Harris and Todaro (1970), Sabot (1977), Mazumdar(1983) or Fields (1990). See Fields (1990), Tokman (1992), Portes (1994), Rosenzweig (1988), Thomas (1992) for excellent overviews of the informality literature. See Stiglitz (1974), Esfahani and Salehi-Isfahani (1989) for efficiency wage models of LDC dualism. 2 See, for example, World Bank (1995) World Development Report. 3 A related literature with a very different emphasis sees informality as the result of an ongoing effort by large modern enterprises to evade mandated protections through subcontracting to unprotected workers, a process accelerated by heightened global competition in labor intensive manufactures (See Portes, Castells and Benton, 1989). The existence and behavior ofthe sector are thus directly relevant to the debate over establishing common labor standards throughout free trade agreements: whatever safeguards may be enshrined in labor codes, LDC- based firms could employ an effectively unprotected work force to compete with industrialized country firms. See Portes, Castells and Benton (1989), Tokman (1992).The U.S. Department of Labor has sponsored two studies on the informal sector. See Workers Without Protections: Case Studies of the Informal Sector in Developing Countries (1993) and The Informal Sector in Mexico (1992). 2 Argentina Poverty Assessment Background Paper No. 3 sector pushed above market clearing by government or union intervention, migrants, young workers, and those laid off queue up for good jobs and subsist in the meantime in the inferior informal sector. Maloney (1997, 1999) provides evidence that, for Mexico, the traditional conflation of issues of formality and dualism is probably conceptually inappropriate. The labor market for relatively unskilled workers may be well-integrated with both formal and informal sectors offering combinations of job characteristics from which workers may choose, with little queuing. Formal salaried work may not necessarily be the first choice for two reasons. First, labor protections often levy an implicit tax on workers. In a market with downwardly flexible wages, the cost of non-wage benefits reduces wages by an equivalent amount. For numerous reasons, among them the inefficiencies and rigidities that often accompany such protections, workers may prefer to evade these labor taxes by operating in the informal sector. Second, beyond being unprotected, the various modalities of informal work may offer other characteristics-greater flexibility or possibilities for training- that suit workers better than a salaried formal job. The paper examines two such modalities for Argentina and comes to substantially more guarded conclusions. Self-Employed: owners of informal firms with or without additional employees. The literature has long recognized that some fraction of the self-employed are prosperous and dynamic entrepreneurs and coexist with those rationed out of the formal sector, but the large size of this group in LDCs has often seemed pathological and indicative of distortions. It may be, however, that LDC workers choose to start or work in small firms with the same desire for independence or entrepreneurial ambition as their industrialized country counterparts, but their incentives to do so are greater. In addition to avoiding labor taxes discussed above, the low level of formal sector productivity for the mass of poorly educated workers in developing countries may reduce the opportunity costs of being independently employed. The voluntary upper tier of self-employment may in fact, be the majority of the sector. Informal salaried: Those working for informal firms. The literature has been far more uniform in viewing informal salaried workers as the least advantaged of a segmented urban work force, receiving none of the benefits of being self-employed, yet none of benefits of formal employment either. The evidence is partially supportive of this view, however the mobility patterns do not suggest the standard queuing story. Further, in addition to the logic above, the unique demographics of the sector suggest that rather than being primarily the reserve army of mature workers rationed out of the formal sector, informal salaried work may correspond the entry point and perhaps training area for young workers. For both the analysis of sectoral and income mobility, the paper employs detailed and relatively rare panel data to take a more comprehensive and dynamic approach than has been previously feasible. The panel structure of the Permanent Household Survey permits us to follow individual workers across 2 years and to examine their earnings differentials and mobility patterns as 3 Argentina Poverty Assessment Background Paper No. 3 they transition among formal salaried employment and the two modalities of informal work. This has several advantages over traditional cross sectional work. First, comparisons of sectoral earnings differentials have been the mainstay of segmentation tests to date.4 However, they suffer from the drawback that unobserved individual characteristics may be correlated both with sector choice and earnings, as well as being unable to capture the value of unobserved characteristics of jobs. The earnings differentials generated by individuals transiting between sectors presented in section III largely ameliorate the first problem and constitute more precise and reliable estimates than those previously possible. However, due to unobserved job related characteristics, even these are shown to be fundamentally unable to prove or disprove segmentation. As an alternative approach, the paper examines the patterns of worker mobility. Section IV, generates a dynamic overview of movement through the Argentine labor market, characterizes the interactions among the three classes of work, and compares them to the patterns implicitly predicted by the standard dualistic view. This has not been possible with conventional cross sectional data sets which permit tabulations of the composition of the work force and perhaps changes in those stocks over time. The paper bolsters the results with time series cross sectional evidence on whether informality is, in fact, countercyclical as predicted by standard dualistic theory. Finally, it employs cross country regressions on determinants of the size of the informal sector as well as labor turnover to provide a measure of distortion in the Argentine labor market. While unable to provide conclusive evidence, this approach offers substantial reason to question the dualistic view as the primary explanation for the existence and dynamics of the sector. In fact, in many ways the dynamics of the various sectors are very close to those found in Mexico, a market with extremely weak unions and irrelevant minimum wages. However, there is some evidence that the Argentine labor market is more distorted than some others in the region and hence a larger share of the informal are involuntarily so. It must also be stressed that a well-integrated work force does not preclude distortions impinging in other ways that do not imply segmentation. The poverty results suggest a high degree of mobility both in and out of poverty. Five key results emerge. First, roughly 25% of the lowest quintile of the distribution appears to be well off workers either on vacation or temporarily unemployed who rapidly return to the upper two deciles. Second, in general there appears to be a high degree of mobility even abstracting from these "statistical poor." Third, a spell of unemployment does seem to have a significant impact on falls in income quintile and into poverty although this is not getting worse over time. Fourth, informality there is only ambiguous evidence imply greater probability of falling into poverty or duration in poverty. Fifth, women are more likely to fall into poverty and spend longer in it although duration has decreased over time. 4 See Rosenzweig (1988). 4 Argentina Poverty Assessment Background Paper No. 3 Table la: Mean Age, Experience, School and Real Wage Differentials During Transitions (Male Workers) Period 1993-1997 Transition Inic. Real Wage Ral. Wage Diff. Initial/Final Nobs Age Experience School Med Med Self Employed/Self Employed 516 43.55 27.95 9.72 13.99 -5.66 (0.02) (0.02) (0.01) (0.66) (3.99) Informal Salaried/Informal Salaried 173 32.82 19.02 8.01 8.39 -4.51 (0.07) (0.07) (0.02) (0.27) (4.07) Formal Salaried/ Formal Salaried 705 38.93 22.86 10.03 12.58 3.75 (0.02) (0.02) (0.01) (0.33) (1.67) Self Employed/Informal Salaried 74 35.88 21.10 8.49 11.54 -15.09 (0.16) (0.16) (0.04) (1.66) (8.77) Informal Salaried/Self Employed 66 36.05 20.05 8.85 9.33 1.28 (0.18) (0.19) (0.05) (0.94) (16.74) Self Employed/Formal Salaried 49 39.82 23.68 10.51 10.50 31.77 (0.29) (0.29) (0.09) (1.86) (20.81) Formal Salaried/Self Employed 44 40.68 23.77 10.42 15.67 2.86 (0.25) (0.23) (0.10) (3.41) (20.57) Informal Salaried/Formal Salaried 44 34.91 20.05 8.72 8.99 7.82 (0.29) (0.32) (0.09) (0.95) (15.59) Formal Salaried/Informal Salaried 35 33.69 18.44 8.84 10.07 -16.08 (0.36) (0.39) (0.13) (0.75) (11.41) Notes: Standard errors in (). Real wage differential in % and based on the sample of those workers that have positive income in both the initial and the last period of the transition. "Med" = median from quantile regression using bootstrapped standard H1. Data: The study employs two sources of data on men aged 16-65 with a high school education or less in Greater Buenos Aires. The Permanent Household Survey (EPH) conducts extensive bi-annual syearly interviews and is structured so as to generate panels that allow tracking a quarter of the sample across two years. Though two years does not permit a full description of the life cycle of an individual, it is nonetheless possible to sketch patterns of mobility among sectors and to identify worker characteristics that correspond to them. To generate a sufficiently large sample of roughly 5,700 observations, seven contiguous EPH cohorts were combined beginning in May 1993 and rotating every six months. 5 Table 1 b: Mean Age, Experience, School and Real Wage Differentials During Transitions (Female Workers) Period 1993-1997 Transition Initial Real Wage Ral. Wage Diff. Initial/Final Nobs Age Experience School Med Med Self Employed/Self Employed 203 43.98 28.71 9.65 13.42 -9.40 (10.23) (12.12) (5.07) (1.29) (6.20) Informal Salaried/Informal Salaried 192 36.06 22.21 7.97 10.50 1.14 (12.85) (14.91) (4.17) (0.59) (3.48) Formal Salaried/ Formal Salaried 359 37.40 19.00 12.47 12.58 4.73 (10.82) (12.22) (4.14) (0.56) (2.45) Self Employed/Informal Salaried 58 41.60 28.44 7.19 12.75 -13.29 (9.37) (11.47) (4.14) (0.86) (6.49) Informal Salaried/Self Employed 37 39.89 26.47 7.39 15.67 -9.46 (13.14) (13.14) (4.54) (2.15) (16.73) Self Employed/Formal Salaried 16 34.12 17.44 10.69 16.78 27.16 (13.52) (15.10) (4.27) (3.89) (30.46) Formal Salaried/Self Employed 21 39.43 21.19 12.24 23.49 -25.47 (12.99) (15.56) (4.77) (6.33) (16.73) Informal Salaried/Formal Salaried 24 35.33 18.75 10.58 10.00 -5.66 (13.02) (14.58) (3.96) (2.25) (15.58) Formal Salaried/Informal Salaried 26 41.08 26.83 8.96 9.13 1.42 (13.20) (15.12) (4.12) (2.43) (12.49) Notes: Standard errors in (). Real wage differential in % and based on the sample of those workers that have positive income in both the initial and the last period of the transition. "Med"= median from quantile regression using bootstrapped standard The definition of informality used here focuses specifically on the issue of "protectionism" and comprises owners or workers in firms of fewer than 16 employees who do not have social security or medical benefits. III. Wage Differentials as Evidence of Segmentation Traditional efforts to identify segmentation by comparing conditional means between sectors are unsatisfying for three reasons. First, as is well-documented in the literature, unobserved worker characteristics that affect productivity-- ability to tolerate authority, punctuality, entrepreneurial ability- may also influence in which sector an individual chooses to work and bias estimates of the sector differential.6 But a greater concern is that the specific characteristics of work that pertain to or even define the formal and informal sectors affect the earnings paid in each sector and make it unclear what the 6 Recognizing this, Marcouiller et al. (1997) attempt to mitigate the selection bias but are clear about the limitations of the standard techniques when there is no clear consensus about how workers sort themselves between the informal and formal sector. They, in the end, report higher earnings in the informal sector. 6 Argentina Poverty Assessment Background Paper No. 3 Figure 1: Sectoral Participation and Median Real Wages in Argentina, 1988-1997 (includes -mall firm o-ner, in sf-employed) 70.0% - 14 13 60.0% 12 11 50.0% 10 9 40.0% 8 7 30.0% 6 20 0% 4 10.0% 2 0.0% 0 1988 1991 1993 1995 1997 M IM% form lllil % info i % self ------med(w) form --- med(w) info med(w) self magnitude or sign of the differential should be even in an unsegmented market. In a market without distortions, wages should move to equalize the utility arising from working in each sector. Most generally, informal earnings should exceed formal sector wages by the expected value of unreceived benefits, and fall below them by the taxation that is often evaded. More specifically, earnings both in contract and self-employment may reflect a premium for risk, lifestyle, and in the latter case, the implicit costs of capital invested, and the value of unpaid work by family members. Finally, the fact that informal salaried workers are among the youngest (table 1) and surveys from other countries suggest that roughly 30% are related to their employer. Their reported earnings may therefore incorporate training costs (see Hemmer and Mannel, 1989 and Roberts, 1989) or unobserved payments in kind, which could imply that they are paid less than they would in a formal firm, again, even in the absence of segmentation. The problems these unobservables pose will become apparent in interpreting the differentials below. A third objection arises readily from figure 1. It is obvious that relative wages change dramatically over the business cycle and that the composition of voluntary vs. involuntary moves may change substantially. i. Estimates of earnings differentials Figure 1 shows the relative median hourly earnings over time in each sector and tables la and lb show the percentage change in hourly real earnings generated by movements of individuals among sectors. This holds worker characteristics constant and leaves the variations in the characteristics of the work itself as the residual explanatory factor. The Mexican Micro-enterprise survey interviews in detail the relationship between employees and employer. 8 Heckman (1979) is insistent that the standard two stage methods of correcting for selectivity bias depend on having confidence in the underlying model of how workers choose among sectors so we don=t employ them here. A bad first stage selection specification may induce bias rather than correct for that existing. This is likely to be the case as the premise of this work is that we have little knowledge of the role each sector serves. Further, since a principle argument of this work is that, in the absence of knowledge of the value of unobservable components 7 Argentina Poverty Assessment Background Paper No. 3 The results, are ambiguous, but not entirely consistent with the conventional view of the relation between formal and informal sectors. Though those moving into formal salaried employment tend to gain and those leaving lose, on average, the self-employed earn roughly 10% more than formal salaried workers. This apparent conflict could be resolved if we were to argue that start up costs make it so transitions to self-employment understate the true gain, and that only failed entrepreneurs return to self-employment. But figure 1 also suggests that our sample is taken from a period where the relative gap between the formal and informal wage is shrinking after having expanded greatly from 1988 to 1993. The low incomes of those leaving to move into formal work relative to the sample median suggests that those moving across this period may be entrepreneurs who are failing. In short, the higher overall median in self-employment may still imply that self-employment was more desirable. But here we immediately fall afoul of our inability capture the value of unmeasured characteristics of different kinds of work discussed above: in the absence of information on the value of benefits, compensation for risk, the value of independence, or in-kind payments or implicit training costs, the magnitude of the distortion-free differential cannot be known ex-ante and we have no benchmark against which to compare for segmentation. If, for example, workers fully valued benefits in the formal sector which Pessino argues are roughly 55% of the wage,9 and other unobservables were negligible, the differential in total remuneration going out of formal employment would be negative for all transitions and in the reverse movement positive as would be the case in a segmented market. Unfortunately, we cannot assume that workers fully value benefits, or that the other unobservables are negligible: the value of not having a boss, while difficult to measure, may also be of these magnitudes, but work in the other direction. In sum, neither these, nor any previous reported sectoral differentials are reliable measures of segmentation. For the same reason, we cannot with confidence interpret the 7.8% increase entering formal salaried work from informal salaried, the 16% reduction in the reverse direction, or the substantially lower median income as evidence of the generally assumed inferiority of informal salaried work. It may only suggest that the composition of the premium in the absence of distortions is distinct for this sector: There may be a lower premium for risk since they are not entrepreneurs, the value of in-kind benefits for the informal salaried who are related to the owner may be large, and the implicit training costs may also be substantial. Any of these could account for a 25% differential. of the differential, we cannot use the differentials as evidence of segmentation, the issue becomes of somewhat less importance. It is worth noting that, although it is frequently asserted that Argentina has especially high levels of non wage benefits, this is not obviously the case. For example, The World Bank and Davila-Capalleja(1 996) estimate rates of 40-60% in Mexico 8 Argentina Poverty Assessment Background Paper No. 3 ii. Are the differentials consistent with an integrated labor market? Though we cannot credibly prove or disprove segmentation based on these differentials, we can ask whether they seem plausibly consistent with an absence of segmentation: is the 10% average differential, (or worse, the 12% fall entering self-employment) of reasonable magnitudes to cover the value of benefits, some return to risk, and whatever value is placed on one style of work vs another? Three factors make this plausible and suggest a reason why the attraction of formal sector employment may have been overstated. First, it is often the case that the medical benefits program covers a worker's entire family, the marginal value of benefits to the second formal sector worker in a family is zero. This would seem particularly important for informally employed workers in households whose principal bread winner may be formally employed: There is no reason to pay again the implicit tax for benefits already received. Second, administrative overhead costs are high and the benefits may be of low value given their cost.12 Third, although perhaps less relevant to Argentina, rapid rates of turnover mean that leaving does not necessarily imply the loss of nominally very generous separation benefits and pensions since as Baldn et. al (1973, p. 212) found in their extensive surveys of worker career trajectories in Monterrey, Mexico, Amany change enterprises quite often and thus they cannot benefit from the seniority accumulated in each of them. In each case, the value to workers of formal sector benefits is below their value on paper, and in a market with reasonably flexible wages, workers implicitly pay. In sum, earning differentials do not offer compelling evidence in favor of the segmentation hypothesis in Argentina and, given the difficulty of quantifying the unobservables, they are unlikely ever to be convincing tests. IV. Patterns of Mobility In contrast, the patterns of worker transitions, as well as overall sectoral behavior across the business cycle can offer additional information on the validity of the dualistic view. Ideally, a model of the behavior for each of the four sectors and workers= choices to enter them would be postulated and held up against the evidence. However, this is a vast research agenda in itself and the present paper seeks only to develop a few provocative stylized facts about the dynamics of the market and sketch some characterizations of the nature and the interaction of the sectors. To begin, the dualistic view predicts some general patterns should be evident. If formal sector work is preferred to informal work, we would expect that workers would queue up for formal sector jobs.14 This should imply: 12 In his interviews with Guadalajaran workers, Roberts (1989, p. 50) found that many informants cited the deduction made for welfare as a disadvantage of formal employment, particularly since the services they received were poor. 14 The Constitution conceives of the employment relationship as a lifetime contract and workers may only be 9 Argentina Poverty Assessment Background Paper No. 3 i. Very low rates of formal sector turnover. ii. A largely unidirectional flow of workers who graduate from the informal sector into the formal sector where they stay until retirement. Flows in the other direction should be largely involuntary and in relatively prosperous times, far less. iii. Given a probability of being selected from the queue in each time period, the probability of entry into formal salaried work should be an increasing function of experience. iv. Movements into informality should be countercyclical. This section tests for these patterns in two ways: First, for each definition of informality, Matrices la and lb provide summary data on transitions among sectors by tabulating the conditional probability of finding a worker in sector j at the end of the period (columns) given that the worker began in sector i at the beginning (rows), Pij.15 The row percentages sum to 100% and the totals at the bottom represent the share of the workers to be found in each category at the end of the period Pj. The first three columns and rows represent individuals who are not working: those out of the labor force (OLF), not currently working and not searching; those studying; and those looking for work, the unemployed. The bold rectangle borders five categories of work beginning with unpaid labor and the shaded area comprises the paid jobs described earlier which are our chief focus. fired under extreme circumstances and at great cost. ith and without signed working cards, and hence worker projections, and found little evidence of strong barriers to mobility. 10 Matrix la: Male Worker Transitions Among Sectors of the Labor Market Across Two Years Informal Defined as Unprotected Argentinean Permanent Household Survey, 1993-1997 Probability of moving from initial to final sector, Pij, in percent Initial Sector OLF School Unemp Unpaid SE IS FS Other Total Out of Labor Force 741 I 10 1 6 4 5 0 100 School 4 56 20 0 7 9 0 100 Unemployed 11 4 37 I 15 17 16 0 100 Unpaid 15 0 8 23 23 15 8 8 100 SelfEmployed 4 I 10 11, 68 10 6 2 100 Informal Salaried 4 3 14 01 18 48 12 0 100 Formal Salaried 3 8 01 5 4 80 0 100 Other 0 0 2 01 1 2 41 60 100 Total (P.j) 11 6 14 01 24 13 30 2 100 Pij standardized by final sector size, Pij/P.j Final Sector Initial Sector OLF School Unemp Unpaid SE IS FS Other Out of Labor Force 18 69 200 24 32 17 0 School 38 144 101 14 58 30 0 Unemployed 104 68 144 61 131 52 0 Unpaid 144 0 55 95 121 26 442 SelfEmployed 33 I3 69 71 7 21 I21 Informal Salaried 39 441 100 75 75 41 16 Formal Salaried 24 58 0 21 26 Other 0 0 14 01 132 16 13 Vij = Pij/P.j*(1-Pii)*(1-Pjj) Final Sector Initial Sector OLF School Unemp Unpaid SE IS FS Other Out ofLabor Force 151 412 981 283 232 321 0 School 326 518 297 96 255 333 0 Unemployed 623 244 297 299 402 409 0 Unpaid 706 0 114 382 304 165 1437 SelfEmployed 386 88 337 286 455 33 931 Informal Salaried 283 194 305 1811 449 389 77 Formal Salaried 4561 20 455 0 3141 2991 323 Other i 0 57 01 10171 761 165' Notes: Sample aggregates 7 panels, each panel composed by individuals followed for four semesters beginning with a first panel in May, 1993. This generates roughly 3,000 males 16-65. Informal salaried defined as individuals in firms under 5 workers not covered by medical or social security benefits. Self employed are those who declared so and individuals working in firms with less than 5 workers. Formal salaried are those with medical or social security benefits. Boxed cells represent work, light shading=paid work, darker shading-probability of remaining in same sector. By compensating for final sector size, Pij/P.j gives a measure of fluidity among sectors. By compensating for rates of turnover in initial and final sectors, Vij gives a measure of disposition (logic) to move to a sector. Figure 2b: Female Worker Transitions Among Sectors of the Labor Market Across 2 years Informal Defined as Unprotected Argentinean Permanent Household Survey, 1993-1997 Probability of moving from initial to final sector, Pij, in percent Initial Sector OLF School Unemp Unpaid SE IS FS Other Total Out of Labor Force 8_ ? I - 4 2 F _ ._ 100 School ' 4 21 I 3 79 0 1Oo UnemploNed 341 6 28 I 7 13 10 0 100 Unpaid 25 6 21 18 20 64 - 4 100 SelfEmplo\ed 20 II i 481 14 4 I 100 Informal Salaried 17 ! 9 0 5I1 55 7 0 100 Form al Salaried 7 216 C 4 5 75. 0 100 Other 20 0 0 01 401 01 10 30 100 Total (P.j) 47. 6 11 1 10 11 13 0 100 Pij standardized by final sector size, Pij/P.j*100 Final Sector Initial Sector OLF School Unemp Unpaid SE IS FS Other Out ofLabor Force 10 64 71 39 35 12 39 School 13 194 118 25 59 67 0 Unemployed 73 93 131 72 120 79 0 Unpaid 54 92 181 195 53 15 1333 SelfEmployed 43 15 103 1_ 1 125 29 322 Informal Salaried 36 22 - 00 5 R! 52 0 Formal Salaried 15 26 59 22 43 49 0 Other 42 04 - 3981 0 76 Vij = Pij/P.j*(1-Pii)*(1-Pjj) Final Sector Initial Sector OLF School Unemp Unpaid SE IS FS Other Out ofLabor Force 119 501 486 425 433 269 312 School 161 579 309 106 283 569 0 Unemployed 565 279 219 192 367 429 0 Unpaid 369 241 304 456 143 71 2312 SelfEmployed 4631 62 274 288 532! 218 887 Informal Salaried 4-12 10 25 U, 4460 451 0 FormalSalaried 3421 221 324 103' 3301 428 0 Other 311 0 0 01 10951 01 427 Notes: Sample aggregates 7 panels, each panel composed by individuals followed for four semesters beginning with a first panel in May, 1993. This generates roughly 3,700 females 16-65 Informal salaried defined as individuals in firms under 5 workers not covered by medical or social security benefits. Self- employed are those who declared so and individuals working in firms with less than 5 workers. Formal salaried are those with medical or social security Boxed cells represent work, light shading=paid work, darker shading=probability of remaining in same sector. By compensating for final sector size, Pij/P.j gives a measure of fluidity among sectors. By compensating for rates of turnover in initial and final sectors, Vij gives a measure of disposition (logic) to move to a sector. Argentina Poverty Assessment Background Paper No. 3 matrix standardizes the transition probabilities by terminal sector size, Pj/P.. Though this is a better measure of fluidity among sectors, it is an imperfect measure of ease of entry since a low desire to leave the initial sector will yield a low value as well as a distortion induced low level of turnover in the terminal sector: P, / P.j = (1 - P) V, (1 - Pjj) Only if we compensate both for the likelihood of leaving the initial sector, and the probability of a place opening up in the terminal sector can we get an idea of the "disposition to move" the disposition or economic or institutional logic that compels a worker leaving the initial sector to enter an open position in j. Vj, calculated by dividing the second panel by turnover in the initial and terminal sectors is tabulated in the third panel. As an example, though both third and fourth grade elementary school classes may fully turnover every year, we would expect V to be large for transitions in the ascending direction and zero in the reverse. In the present case, the disposition to enter formal employment from school is nine times that of the reverse transitions as would be expocted if workers generally graduate from school to employment. If the dualism hypothesis that workers graduate from informal to formal employment is correct, we should expect similar asymmetric Vs between the sectors. Appendix I contains the analogous matrix from Mexico for comparison. Further characterization of the patterns of mobility is offered by a multinomial logit model of determinants of transitions among sectors. Tables 2a & 2b present three sets of regressions that correspond to the three initial sectors in which we may find the worker in the first period of the panel. Since these regressions condition on the initial sector which is likely to result from a self-selection process on the part of the workers, they must be interpreted as such.16 The coefficients in the first five columns of each set of results are those from the standard exponential form where the vector 3j measures the degree to which an increase in worker characteristic X, listed across the top of each panel, increases the probability of a worker going to sector j relative to the probability of staying in sector i. The variables in X include experience, experience squared, and schooling. Since these are often the factors included in Mincerian earnings equations, the initial period real wage is included to ensure that it is not simply the wage effect that is reflected. The interpretation 16Arguably, this is exactly what we want. As an example, ifworkers first choose formal salaried work in order to accumulate capital to open their own business later, we are interested in their transition decisions, not individuals randomly selected into the sector, while formal. 13 Argentina Poverty Assessment Background Paper No. 3 of the coefficients of the first three variables is therefore the impact of a rise in education or experience given a certain level of income. By the same logic, the wage term must be interpreted as the wage given the first three variables, that is, earnings above or below what would be predicted by human capital, rather than as absolute income. Though the impact of this measure of unobserved individual characteristics may have interesting interpretations, the variable is included primarily to isolate the school and experience effects. The survey also tabulates if workers who experience a spell of unemployment during the transition between sectors across the five quarters left their previous jobs voluntarily or not. The binary variable "bdlk" takes a value of I in the event of a spell of unemployment in the second and third periods and a 0 otherwise. Two other tools are also brought to bear. First, figure 1 shows the evolution of the sector sizes as a share of the work force and the median hourly wage. The observed patterns provide additional information on the role each sector plays. The traditional dualistic view of the informal sector would postulate that the sector grows when people are either laid off from, or cannot get jobs in the formal sector. During downturns, the informal sector wage is the "flex-wage" that adjusts to clear the labor market while the formal sector wage remains downwardly rigid. In this view, sector size and wage relative to the formal sector wage should move oppositely. On the other hand, if informality is a closer substitute for formal work, a rise in the informal wage relative to the formal should lead to an inflow of workers and an expansion of the informal sector. 14 Table 2a: Male Worker Characteristics Affecting Sectoral Transitions Across Two Years Multinomial Logit, Informal Defined as Unprotected Initial Sector: Self Employed Final: C Exp 1/ Exp2 School Rwage Bdlk Informal Salaried 0.830 -0.060 -2.48E-05 -0.115 -0.103 0.005 (1.334) (1.480) (0.031) (2.954) (1.554) (0.012) Formal Salaried -1.403 -0.098 1.48E-03 0.064 -0.159 0.726 (2.003) (2.336) (1.910) (1.576) (2.131) (1.815) Unemployed -0.418 -0.020 -5.04E-05 -0.126 -0.143 1.816 (0.617) (0.491) (0.070) (3.062) (1.735) (6.218) Number of ohs = 680 chi2(27) = 122.92 [00] Pseudo R2 = 0.102 Log Likelihood = -540.05 Initial Sector: Informal Salaried Final: C Exp 1/ Exp2 School Rwage Bdlk Self Employed -2.577 0.055 -7.75E-04 0.098 0.058 0.385 (3.841) (1.244) (0.853) (1.935) (0.509) (0.825) Formal Salaried -2.172 -0.008 5.OOE-04 0.089 -0.034 0.385 (2.899) (0.177) (0.536) (1.531) (0.231) (0.743) Unemployed 0.628 -0.108 2.16E-03 -0.181 -0.046 1.614 (0.697) (2.098) (2.188) (2.261) (0.220) (3.968) Number of obs = 312 chi2(27) = 47.42 [00] Pseudo R2 = 0.064 Log Likelihood = -357.50 Initial Sector: Formal Salaried Final: C Exp 1/ Exp2 School Rwage Bdlk Self Employed -4.830 0.146 -2.74E-03 0.056 -0.068 1.856 (5.586) (2.396) (2.218) (1.323) (1.014) (4.116) Informal Salaried -0.769 -0.093 1.15E-03 -0.112 -0.063 1.840 (0.921) (1.787) (1.086) (1.907) (0.528) (3.789) Unemployed -0.923 -0.078 1.46E-03 -0.080 -0.102 2.575 (1.464) (2.069) (2.006) (1.875) (1.124) (7.651) Number of obs = 806 chi2(27) = 106.04 [00] Pseudo R2 = 0.101 Log Likelihood = -473.876 Notes: coefficients of logit reflect how experience, experience squared, schooling, and the initial real wage affect the probability of moving from the initial sector to the final sector relative to the probability of staying in the initial sector. The variable "bdlk" is a dummy variable=1 if an individual has suffered unemployment during the two years of transition. "z" statistics beneath coefficient ino; "P-values" in [I. 1/ Sign of compound value of Exp, Exp2 evaluated at mean is that of Exp with few exceptions. 15 Argentina Poverty Assessment Background Paper No. 3 Table 2b: Female Worker Characteristics Affecting Sectoral Transitions Across Two Years Multinomial Logit, Informal Defined as Unprotected Initial Sector: Self Employed Final: C Exp 1/ Exp2 School Rwage Bdlk Informal Salaried 0.797 2.40E-03 -6.76E-04 -0.168 -0.066 1.325 (0.819) (0.044) (0.730) (3.453) (0.601) (1.321) Formal Salaried 2.111 -2.46E-01 3.03E-03 -0.155 0.226 -36.691 (1.589) (3.513) (2.348) (1.823) (1.944) 0.000 Unemployed -0.054 -2.OOE-04 -8.49E-04 -0.076 -0.195 3.279 (0.053) (0.003) (0.746) (1.485) (1.521) (4.256) Number of obs = 304 chi2(27) = 82.97 [00] Pseudo R2 = 0.132 Log Likelihood = -273.38 Initial Sector: Informal Salaried Final: C Exp 1/ Exp2 School Rwage Bdlk SelfEmployed -2.779 0.099 -1.41E-03 0.052 -0.480 0.560 (3.035) (1.901) (1.441) (0.907) (1.821) (0.903) Formal Salaried -3.954 0.027 -2.21E-04 0.186 -0.248 1.107 (4.127) (0.488) (0.188) (2.855) (1.007) (1.841) Unemployed -2.110 -0.003 -1.20E-04 0.043 -0.470 2.718 (2.230) (0.047) (0.100) (0.642) (1.470) (5.803) Number of obs 274 chi2(27) = 63.82 [00] Pseudo R2 = 0.115 Log Likelihood = -244.50 Initial Sector: Formal Salaried Final: C Exp 1/ Exp2 School Rwage Bdlk SelfEmployed -2.907 -0.076 2.12E-03 0.054 -0.177 1.035 (2.632) (1.397) (1.850) (0.813) (1.051) (0.932) Informal Salaried -0.201 -0.052 1.53E-03 -0.119 -0.824 1.221 (0.180) (0.928) (1.411) (1.897) (2.799) (1.414) Unemployed -1.223 -0.131 2.28E-03 -0.024 -0.252 4.051 (1.137) (2.226) (1.690) (0.351) (1.275) (7.055) Number of obs = 424 chi2(27) = 102.05 [00] Pseudo R2 = 0.181 Log Likelihood = -230.83 16 Argentina Poverty Assessment Background Paper No. 3 Finally, we locate the Argentine case in international context (See appendix II). Krebs and Maloney(1998) build a model where the self-employed sector is a desirable sector but where, as the economy develops, the opportunity cost of being self-employed rises with formal productivity. Figure 4 strongly supports this view by showing a strong log-linear relationship between the size of the self- employed sector and formal sector productivity. .461 Peru E 0 Bol E EIS E Col Ven 4L Hon Gua Par Uru _ Arg Gre Pan Chi Meurk O CR Bra Kor Ita 0 t LL - Por o Spa .0 New I Ire A,ei 0A C AUP Den .0576Nbax I II I - T 6.63332 10.5322 Log of Industrial V.A./ Worker Figure 4: Self-Employment vs. Development Level The model also attempts to model turnover and postulates that formal sector employers must pay a premium above the market wage to prevent workers from moving either to the informal sector or to other formal sector jobs. The model postulates a variety of factors that should influence both self-employment size and turnover in a market without labor market rigidities. Maloney(1998) then estimates this relationship using OECD and Latin American data and find support for the model (See appendix II for summary). The residuals from an estimation on the basic economic and demographic variables are plotted in figure AII.5 and offer a measure labor market distortions and rigidities that would lead to above average self-employment and below average turnover rates. Together, these first offer a view of overall labor force dynamics and then of how the three sectors interact. 17 Argentina Poverty Assessment Background Paper No. 3 I. Overview of the Labor Market Levels of turnover and flexibility First, the matrices reveal relatively low turnover rates in the formal sector (and implicitly the length of tenure (calculated as 2/(1 -Pii)) at 10 years in the formal sector, more than in Mexico ( 5.21- 5.7).17 In a result very similar to that observed in Mexico, self-employment and formal salaried work show relatively similar rates of turnover (80 vs. 68% remain) , informal salaried workers turnover much more quickly (48% remain). But again, this implies much lower rates of turnover than in Mexico at self-employed (self-employed 6.25 vs. 3.68 in Mexico, salaried informal 3.8 vs 2.0 in Mexico) It is hard to interpret the relatively low turnover as "high rigidity" for three reasons. First, if, in Mexico the informal sector is a close substitute for the formal, as Balan et al suggest, there will be more movement between sectors that might not be seen in a country like Argentina with a longer tradition of salaried labor. In fact, then, we miss transitions within each category and are really propensity to move to other categories. Second, the panel misses intermediate transitions that might drive the rate of turnover higher. Third, as Krebs and Maloney show (See appendix II), raw turnover rates must be adjusted for country specific economic and demographic variables. In particular, higher education levels imply lower rates of turnover, perhaps due to more firm specific human capital that would lead to lower separation rates. Pages and Marquez (1998) also argue that younger workers should turnover more and Argentina's workforce is somewhat older compared to that of Mexico. Median formal sector worker wage is 38 compared to 34 in Mexico, informal salaried 31 vs. 29 in Mexico. Nonetheless, using consistent mean tenure numbers, figure AII.4 suggests that Argentina's adjusted for these factors turnover rate in the manufacturing sector is far above its predicted value. On balance, the Argentine formal sector seems relatively rigid, and in fact, the entire labor force, seems relatively rigid. Symmetry Also in striking contrast with similar evidence from Mexico, the absence of symmetry of Vs across directions of movement in some sectors of paid work suggests a less than perfectly integrated labor market. For informal salaried work the V's into formal employment are roughly 30% higher than those for the reverse. That is, workers seem more likely to move from informal positions into 17 In the U.S. median tenure for all workers over 16 years old in 1991 was 5.1 years. BLS NEWS, USDL 92- 386. The implicit tenure based on turnover rates on mean July/August 1994 median separation rate of 1.1% /month was 7.6 years. Bureau of National Affairs, Bulletin to Management, Dec. 8, 1994. Mean tenure calculated as 1/(1-Pi). 18 Argentina Poverty Assessment Background Paper No. 3 formal than the reverse. While this might seem at odds with the idea of barriers to entry in the formal sector, as discussed above, it is difficult to distinguish barriers to entry from lack of desire. It may be that informal workers seek permanent status in the formal sector and stay until they retire. As in Mexico, however, movements among self-employment and formal salaried work are symmetrical suggesting that they may be closer substitutes. But strikingly, the logit results show that from neither self-employment nor informal salaried work does the probability of moving into the formal sector relative to staying increase in overall experience as would be expected if there were queuing to enter the sector. In the former sector, the compound coefficient becomes positive only after 33 years of experience. In fact there is no significant relationship between formal and informal salaried work and experience in either direction suggesting movement between them at any point and perhaps more integration than might be suggested by the observed asymmetry. Finally, the logit results suggest that movement out of formal salaried work into informal salaried work decreases with education and the reverse transition is correlated with higher. This is consistent with more educated workers finding better skills matches in the formal sector, and also with unskilled workers facing a lower opportunity cost of becoming informal. The absence of education as an important determinant of movements in either direction between self-employment and formal salaried work suggest that these sectors may be more complementary. ii. Self-Employment as an Alternative to Formal Sector Employment For men, self-employment constitutes the largest source of employment (34%), after formal salaried employment (43%). Although it may serve the traditionally postulated holding pattern or safety net for the latter, the data are consistent with it being a desirable sector in itself. As a first approximation it may be more correct to assume that small scale LDC firms have origins and dynamics similar to their industrial country counterparts, rather than being a distinct phenomenon.20 First, motivational responses from Mexico show that at least two-thirds of those entering self-employment from formal salaried employment report moving voluntarily, with a desire for greater independence or higher pay cited as the principal motives. 21 These results are very close to Gottshalk 20 See Levenson and Maloney (1997) for a discussion of alternative conceptions of LDC small firms, their dynamics, and their decisions to become formal. 21 This percentage remains relatively unchanged even when the sample is restricted to those previously working in firms over 50 employees (available on request).They are also supportive of Balin et al.'s finding that being one's own boss was well-regarded and that movements into self-employment from salaried positions often represented an improvement in job status Of those moves from formal into self-employment they studied, 57% were to upward moves in job quality, 30% horizontal (which the authors argue is considered welfare improving 19 Argentina Poverty Assessment Background Paper No. 3 and Maloney's (1985) finding that roughly 70% of U.S. job changes are voluntary. Put differently, if self-employment is a close substitute for formal salaried work, the implied rates of involuntary entry in Mexico would be normal by U.S. standards. Some evidence from Argentina suggests a similar story. A small survey in Jujuy revealed that 80% of the self-employed had no desire to change jobs and under 18% saw self-employment as a temporary activity before they found a "real" job.22 In Gran Buenos Aires, a SIEMPRO/FONCAP survey found that while 36% would have preferred to work more hours, only 26% were looking for other work. 22 Both are consistent with an "entrepreneurial view" of the sector. The voluntary entrance view is corroborated by the transitional evidence in Argentina as well. As in the U.S., self-employment is not an entry occupation from school (Aronson 1991) and there is little evidence that the sector serves as a holding pattern for young workers. The V values from school are only a third of those for entry into informal salaried work or formal salaried work Further, the V values from unemployment to self-employment are 25% smaller than those into either formal salaried work or informal salaried work which are about identical. Transitions into self-employment from formal employment occurs at the higher age (40) of any transition. This contributes to the median age in the sector being 5 years older than formal salaried workers and 12 years older than the informal self-employed. From formal employment the logit results reveal that the probability of moving into self-employment relative to not moving at all is associated with greater experience. These patterns, and Bal6n et al=s longitudinal interviews, are supportive of the recent industrialized country literature on liquidity constraints, exacerbated in the LDC context, that dictate a threshold level of financial and human capital necessary to start a business that can only be accumulated with time and work as a salaried employee.22 As in the U.S., self-employment for some workers may serve as a form of partial retirement, and there is some evidence for this position. It is unlikely however that the sector serves primarily because of the greater independence), and 11% downward (which also could also be welfare improving eg., a supervisor who buys a grocery store may still consider himself better oft). 22 Micro-enterprise Study for Jujuy, Consultoria Nordeste, Fondo de Capital Social (FONCAP). 22 SIEMPRO/FONCAP (1998) "Perfiles de la Microempresa". 22 See Evans and Jovanovic (1989), Aronson (1991) p 23. Baldn et al. argue that AFirst, the man must accumulate capital. This is no easy matter when he has a manual job and must provide for a large family, so it generally takes years to accumulate enough capital. There must be sufficient fluds not only to set up the business, but also to keep it going during the months or years while it runs at a deficit. ..these kinds ofcapital requirements are modest enough, but the capital is not easy to come by for the working classes of Monterrey or elsewhere in Mexico. A p. 217. As with Evans and Leighton, they find that the percentage ofworkers entering self- employment was roughly constant across age cohorts. See also Aroca and Maloney. 20 Argentina Poverty Assessment Background Paper No. 3 this purpose, or as a safety net for laid off older workers who are unable to find new salaried jobs. First, the decreasing quadratic term on experience implies that the influence of experience on the probability of moving into the self-employed sector increases at a decreasing rate, a result consistent with the findings of Brock, Evans, and Phillips (1986) for the U.S.. In fact, after 26 years, experience becomes negatively correlated with transitions between formality and self-employment. Although, the dummy for some period of unemployment is significant in transitions from formal salaried work to self employment, it is also true for the reverse (marginally) and the V statistics show a greater predilection for movement into unemployment rather than into self-employment. Further, self- employed and informal salaried movements into unemployment are more likely than those of the formal. In fact, as discussed later, the largest contribution to unemployment is from those in the informal sector . When responding to the question of why they no longer are engaged in their previous job, roughly half of the unemployed report that they are cuenta propistas who had insufficient demand for product. The fact that, regardless of destination, the less-experienced are more likely to leave is in line with the mainstream literature on firm dynamics that younger firms (on average also less experienced workers) have higher failure rates.23 This is supported by figure I which shows the evolution of self-employment as a share of total employment and the relative incomes of both sectors. What is striking is that, as the differential with formal employment increases, so does the size of the self-employed sector. This is exactly the opposite of what would be expected if the sector were a residual whose wage would fall relative to the formal sector as the formal sector contracts, dumping workers into self-employment. The evidence is far more consistent with either more workers entering self-employment when its relative return rises, or fewer existing entrepreneurs failing. Why would workers choose self-employment? But what would compel workers to give up the ostensibly large benefits in the formal sector? First, it may be that the decision process of the self-employed is not fundamentally different from that of their counterparts in the industrialized countries who also take on responsibility for medical insurance, or saving for retirement that was previously covered by their employers. Second, since the cost of benefits to employers reduces the wage component of formal sector remuneration, a perceived value below that cost as suggested by section II will lead workers to seek out jobs in the unregulated sector where remuneration is entirely monetary. Third, Balin et. al.'s interviews suggest that the very legislation that is thought to induce rigidities into the labor market in fact stimulates such turnover and encourages workers to leave salaried employment. The paucity of openings for promotion on the rigid job ladders as well as the ceiling on mobility opportunities for manual workers makes self-employment the remaining outlet for further advancement. These last two issues suggest 23 See Jovanovic (1982), Evans and Leighton (1989). 21 Argentina Poverty Assessment Background Paper No. 3 that, in contrast to the usual view, the extant labor protections may make formal sector work less desirable, rather than less attainable for some workers. This logic, which applies to both informal sectors, is most compelling where small scale firms can offer remuneration comparable to that earned in the formal sector- among low education workers unlikely to generate much firm specific capital. As figure 4 shows, the opportunity cost of being one=s own boss rises with formal sector labor productivity, we may see a decline in the share of the labor force self-employed from its present levels. Does this imply there are no distortions? The fact that the majority of the self-employed is voluntary is consistent, however, with a large fraction that would prefer formal sector work and who may make the sector larger than it would be in the absence of labor market distortions. Figure 5 suggests that controlling for demographic variables and level of development, there are perhaps 3 percentage points of the work force. Nonetheless, the remaining 18% may consider the sector desirable. Adjusted Mean Tenure . 1 - Ven - CGre (1) 0 E Ita (0 >. 0 0 Arg E Spa O W Fir Bel UK Swit C0- Ge&Po Par Ausi Bol Mex Neth Jap US Can N n Fra U) Ire (D 0 a)- 00 Bra -2.6 0 2.6 F lexib le ..................................................... R ig id Figure 5: Distortion and Rigidity? 22 Argentina Poverty Assessment Background Paper No. 3 iii. Informal salaried employment and entry into work. Even if the self-employed benefit from being their own bosses, the mainstream view is that those who work for them are the very worst off of the urban workforce: salaried, yet without benefits. In Mexico, this was unlikely to be the case. Rather than being a stagnant group of disadvantaged workers, the sector appeared to serve primarily as the principal, although not exclusive, port of entry for young, poorly educated workers into paid employment. The transition matrices show a cluster of high mobility between school, unpaid work and, to a lesser extent, unemployment that suggests a pool of workers not yet tracked into regular employment. Those leaving school and those unemployed show disproportionate movement into unpaid labor (and, to a lesser extent, informal salaried work). Subsequently unpaid workers move disproportionately into the informal salaried sector. This suggests that while and just after completing school, many help out at the family business, and eventually get paid. They spend on average only 2 years doing this before moving on to other paid work. The brevity of tenure is the same as that found in Brazil by Sedlacek et. al.(1995) and is similar to the U.S. where the median tenure for young workers 16 to 24 years of age is only 1.4 years and 25 to 34, 3.4 years.25 Even if this pattern of graduation from school to unpaid to informal salaried work to other modes of work represents the queuing that the dualistic literature might predict, the wait in informal salaried work is not long. However, if Hemmer and Mannel (1989) are correct that in many countries informal small enterprises train more apprentices and workers than the formal education system and the mostly government job-training schemes together, these years to large degree may constitute continued schooling. Further, the symmetry of the flows back into informal salaried employment from all three of the other sectors suggests that the opportunities there are not considered uniformly worse than those in the other sectors. The logit results suggest that from every sector, entering workers are those with less experience and schooling. Baldn et al (1973 p. 132) provide one possible explanation: "The first years in the labor force are ones for learning skills, 'shopping around,' exploring alternatives... Very few men... held at age 25 the same job they had ten years earlier." The concentration among the very poor and uneducated again suggests the low opportunity cost of leaving formal sector employment. The better educated who push up the mean for schooling and wages in the formal sector (tables 1 a &b), arid who the logit results suggest are more likely to enter formal sector employment, may not consider salaried employment in the informal sector comparable, but those working menial or assembly line jobs at less well paying formal sector firms may. The story in Argentina contains some similar elements, but also some very different ones. The mean age of 31 is 7 years below that of formal sector and 12 years below that of the self- employed. But there is not the same cluster of high mobility between school, unpaid work and, to a lesser extent, unemployment that suggests a pool of workers not yet tracked into regular 25BLS News, USDL 92-386 for 1991. 23 Argentina Poverty Assessment Background Paper No. 3 employment. Those leaving school and those unemployed show far less movement into unpaid labor (and, to a lesser extent, informal salaried work) although the relatively few (under .5% compared to 2% in Mexico) still move disproportionately into the informal sector. In fact, those graduating from school are far more likely to move into unemployment or out of the labor force than in Mexico, by a factor of two, either measured as raw percentages or by the Vs. Entrance into informal salaried work is less likely in Argentina (7 vs 9% of those leaving school), although far more likely from unemployment. Further, the nature of the job seems somewhat different. In particular, informal workers are in the sector for almost twice as long as in Mexico (3.85 years) suggesting that, either the sector is more of a desirable destination, or that the queue to enter formal sector is far greater. Unemployment The matrices can be "inverted" to reveal the origin sector of those found in unemployment. Surprisingly, the breakdown in table 3 suggests that the vast majority of the unemployed during the 1990s came from the informal sector. 30% of the unemployed, 45% of those newly unemployed, and 60% of those entering unemployment from paid employment are coming from the informal sector. This suggests first that the majority was not laid off from large companies due to trade liberalization and that whatever social safety nets are to be developed may need to consider the informal nature of most of the users. Further, as Figure 6 suggests, the young have at least twice the incidence of unemployment as prime level workers. Table 3: Sector of Origin of the Unemployed EPH Panels Previous Sector Percent Informal Self-Employed 18 Salaried Informal 12 Formal Salaried 17 Previously Unemployed 34 Previously Out of Labor Force 6 School Graduates 13 Total 100 24 Argentina Poverty Assessment Background Paper No. 3 Figure 6: Percentage of Males Unemployed Some Time During Two Year Period .3 .2- 01 ....... .. 16 22 27 32 37 42 47 52 57 65 Age Some Observations on Female Labor Market Transitions Figure 2b shows substantial differences in female mobility patterns and highlights the difficulty in separating out barriers to entry from differing preferences. Sector sizes differ radically. 47% of the female sample is out of the labor force compared to 11% for men. Interestingly, the duration of spell in OLF is longer (11.1 vs 7.7 years) but is still of similar orders of magnitude. It is not the case that women spend much longer on average out of the labor force, but rather that they are more likely to move in and out at some period of time. This makes good sense given the higher likelihood of bearing and raising children or becoming primarily in charge of household management. What is striking is that, leaving school, women and men have essential identical patterns of entry into the difference sectors of paid work (Pij). But, transitions from every sector show vastly more leakage into OLF than for men. This suggests that, in fact, for young women, the same options are probably available, but that subsequently, they select out of the work force. 25 Argentina Poverty Assessment Background Paper No. 3 Figure 6a: Male Worker Income Transitions Initial Quintile Final Quintile OLF Unemp 1st Q. 2nd Q. 3th Q. 4th Q. 5th Q. Out of Labor Force 79 9 6 3 0 0 1 100 Unemployed 11 29 21 20 8 7 -1 100 1st Quintile 7 9 29 23 6 12 14 100 2nd Quintile 2 12 18 39 1-4 II 4 101) 3th Quintiile 2 6 10 19 24 33 -4 100 4th Quintiile 2 6 10 11 9 42 19 100 5th Quintiile 2 - 11 3 3 19 59 100 Final Distribution 10 9 15 18 10 20 18 100 Notes:Sample aggregates 7 panels, each panel composed by individuals followed for four semesters beginning with a first panel in May, 1993. This generates roughly 2500 males 16-65 household heads. Shading = probability of remaining in same sector. Boxed Cells Represent Positive Income. Figure 6b Female Worker Income Transitions Initial Quintile Final Quintile OLF Unemp 1stQ. 2nd Q. 3th Q. 4th Q. 5th Q. Out ot'Labor Force 83 7 5 2 2 I I 100 Unemplo%ed 50 22 II 7 - 4 2 100 I st Quintile 30 7 31 8 8 6 10 100 2nd Quintile 16 9 18 29 17 7 -1 100 3th Quintile 12 5 13 17 30 19 100 41h Quinuile 9 6 9 8 17 36 16 100 -th Quinfile 6 4 6 2 -1 16 61 100 Final Distribution 55 7 9 6 7 8 8 100 Notes: Sample aggregates 7 panels, each panel composed by individuals fbllowed for four semesters Beginning with a first panel in May, 1993. This generates roughly 2500 females 16-65 household heads. Shading = probability of remaining in same sector. Boxed Cells Represent Positive Income in the initial period. The logits regressions offer some provocative findings. First, the less educated among the self-employed are more likely to move into formal employment. Second, the dynamics of entering self-employment from formal salaried work is exponential. The older you are, the more likely you enter self-employment no matter what. 26 Argentina Poverty Assessment Background Paper No. 3 VI. Transitional Income Dynamics This section uses the panels to study movements among income groups and, in particularly, those in and out of poverty. Figures 6a & 6b show the mobility patterns of males and female household heads among income quintiles across the two year period of transition. Several caveats are in order. First, we work with heads of households between the age of 16- 70 only and hence are not capturing total household income. Second, as will become very apparent, income data is a noisy measure of either permanent income and poverty studies should probably be done with consumption data which more accurately profile poverty. Unfortunately, in Argentina, as elsewhere in Latin America, panel consumption surveys are not available. Third, the two year period is not long enough to tell life-cycle stories about income mobility. Argentina may consider setting up a longitudinal study that would follow workers over a several decade span. Fourth, as discussed previously, movements between the formal and informal sector may lead to rises in income that do not correspond to rises in utility: moving to self-employment often shows a rise in income, but there may also be a corresponding loss in benefits that would lead to a net fall in welfare. An immediate problem is how those reporting zero income should be handled and the matrix is designed to handle three different possible approaches. It first divides those individuals reporting non-zero incomes into five quintiles. Since preliminary analysis suggested sharp differences in behavior among two types of zero-reporters, out of the labor force and those unemployed, we maintain this division in the matrix. We can now follow individuals beginning in these two, plus the five quintile groups across the columns to see where each ends up after two years. Table 4 tabulates the percentage of the population in poverty, under three different definitions. It then attempts to get some feel for whether the observed status as "poor" reflects a permanent or purely transitory phenomenon. The first column gives the share of the population below the poverty line defined as the top income in the bottom quintile. The second asks "if we take the average of the non-zero reported wages across the four periods as a measure of permanent income, how many would fall below the poverty line." The third takes the maximum as an even more extreme version of the "permanent income" that captures "potential" income. Table 4: Measures of Poverty: Broad to Narrow Raw Average Maximum Non-Zero Wage Lowest 20th + unemp+OLF 30.6 16.9 13.1 Lowest 20'h + unemp 24.5 10.3 6.4 Lowest 20' 16.9 6.8 3.4 27 Argentina Poverty Assessment Background Paper No. 3 Figures 6 and table 4 raise several provocative issues for each of the groups. 1. Poverty including all zero reported. First, most of those out of the labor force (79%) remain there after two years. Closer examination reveals why. Roughly half are over 55 suggesting that they are retired and enjoying other sources of income. And half declare themselves students and, therefore, also probably have alternate sources of income. Therefore, fully half of those reporting zero wage income almost certainly have alternate sources of income and should not be considered poor. Using this definition, however uninteresting, the two "permanent" measures of income is reported in table 3 results in a reduction from 31% to under 17. 2. Poverty including unemployed The other 8.2% of the sample of male heads of households that report zeros are in fact, unemployed. Though they may be unemployed for some time, chances are they will not remain without income for long. In fact, after two years only 29.2% are still unemployed, 38.8 are emphatically out of poverty ( in either the 2-5 income quintiles), 11.5% have moved out of the labor force for unknown reasons. Only 20.6% are found in the bottom quintile of the income distribution. As table 3 shows, the 24.5% defined as poor using this definition, shrinks to 10.3% when average income is considered and then, if the maximum income is considered, to 6.4%. 3. The third definition eliminates all zero entries and defines poverty simply as those in the bottom quintile. Due to many repeated values in the sample, the quintile line was drawn at roughly 17%. Here too, several striking results emerge. First, only 29% of this group remains poor after two years and strikingly, 25% move into the top two quintiles. Simple cross tabs suggest that the vast majority of the latter are well-educated people. Further, dividing the sample by those with and without a college degree shows the result to be constrained largely to the former: Of those with less education 16.7% move to the top two quintiles, while of those with college, 35.1 % move. For some reason, these individuals are not working full time and are showing temporarily lower incomes. The more plausible patterns of mobility are between the l' and 2nd quintiles and the rapid decline into the 3rd suggests that the population making smaller moves is more interesting. And here, 29.3% move up to the next two quintiles, if we drop the "vacationing" rich, 39% are still in poverty, but 39.3% moved up between lV and 2nd quintiles. This is confirmed by table 4 which suggests that, when average income is used, only 6.8% of the income earning population is below the poverty line and, using the maximum wage, only 3.4%. In the aggregate, attempting to control for "temporary" poverty has mixed effects on standard indicators, as shown in table 5. The share of income going to the l' quintile increases by 72%. This, with a 4% reduction in share to the 5th quintile leads to a reduction in the 5th1' quintile ratio from 13.6% to 7.6%, but only a 7% reduction in the Gini. Using income vs consumption based data to calculate ginis will probably not account for a large share of the difference between Latin America and, for instance, Asia. 28 Argentina Poverty Assessment Background Paper No. 3 Table 5: Measures of Income Distribution Rawlst period Average of panel Income 1st .036 .062 5th .49 .47 5th/Ist 13.6 7.6 Gini .394 .366 The lesson must be that reported income is a poor measure of permanent income and income transitions a very noisy measure of income mobility. To some degree any analysis of the determinants of mobility through the distribution or movement in or out of poverty is also an analysis of determinants of temporary fluctuations in income unrelated to poverty. A few issues, however, we may be able to extract some useful information on. First, whether having experienced a completed spell of unemployment leads to downward mobility. Second, whether over time, this impact has grown more intense. Third, whether different types of formal or informal jobs appear more or less related to mobility or incidence of poverty. Fourth, whether female household heads appear to show different behavior. Duration in Poverty We are interested in how long individuals are in poverty. The first column of table 5 presents an ordered probit where the possible outcomes are the semester in which the individual leaves the Vn quintile. We are asking "which factors influence the probability that are found in a later semester, rather than and earlier." This is a measure of duration. Again, the effect of experience is quadratic and negative until the age of 25, when it becomes positive and schooling is negative throughout. A period of unemployment delays leaving unemployment, as does being female. The sector in which the worker is found does have a significant effect. In every case, staying in the same job, whether formal or informal increases the rapidity of exit, although significantly more for formal salaried workers. This may also be a function of the time period examined where clearly the informal sector is doing worse compared to the formal sector, but may not be a generalizable pattern. As figure 1 suggests, doing the same test during prior to 1993 would have shown very different results. The significant trend term suggests that on average, duration is increasing for men and slightly for women. The severity with which a period of unemployment delays exit does not appear to trend. 29 Table 6: Low Income Job Duration and Probability of Ending up With a Low Income Job Duration in Low Income Jobs Falling into Low Income Jobs Ordered Probit I/ Probit Analysis 2/ Experience -0.039 -0.031 (5.269) (5.397) Expe. Squared 8.72E-04 7.14E-04 (5.896) (6.292) School -0.040 -0.028 (4.832) (4.471) Women 1.118 0.588 (7.543) (5.619) Self Employed -0.476 -0.710 (4.714) (10.985) Informal Salaried -0.588 -1.064 (4.113) (11.949) Formal Salaried -1.053 -1.301 (7.525) (18.954) Trend 0.099 0.017 (4.258) (1.067) Badlk 0.590 0.308 (2.998) (1.757) Women*trend -0.075 -0.027 (2.423) (1.177) Badlk*trend 0.044 0.052 (1.150) (1.607) Initial 2nd Qle -0.363 (5.248) Initial 3th Q -0.672 (9.554) Initial 4th Q -0.749 (8.263) Initial 5th Q -0.650 (7.975) Ini 2nd Q. * Badlk 0.470 (2.655) Ini 3nd Q. * Badlk 0.143 (0.814) Ini 4nd Q. * Badlk 0.550 (2.442) Ini 5nd Q. * Badlk 0.436 (1.802) Constant 0.644 (5.440) Predicted Probabilities: Less Than 6 Months 0.301 Less Than a Year 0.126 Year and a Half 0.085 Two Years or More 0.487 1/ Latent variable the number of periods an individual stays in first quintile of the wagedistribution before moving to a higher quintile. 2/ Discrete variable = 1 if the individuals wage falls into the first quintile of the income distribution Argentina Poverty Assessment Background Paper No. 3 for less good jobs we can't know. The interactive terms between initial cohort and unemployment suggests are significant from the 2nd and 4th quintiles. The former is logical- experiencing an adverse shock while just above the poverty line may push you over. It may be that 4th quintile workers are able to afford to be truly unemployed in a way that 3rd quintile workers are not. This, again, has to be interpreted with caution since receiving no income does not imply poverty. In fact the result is probably simply saying that no matter what sector you are from, unemployment means you get no pay check. The trend term itself is not significant as would be expected given that, by our definition, 20% are always in poverty. Female household heads more likely to fall into poverty although they show no relative trend over time. VII. Conclusions and Policy Recommendations The Informal probably do not represent an especially vulnerable group from a poverty perspective 1. The majority of the informal self -employed sector is probably not especially disadvantaged and does not merit special attention from a poverty point of view. On average, they have higher incomes than formal salaried workers, although the fact that they lack formal sector benefits may offset this differential. Further, the bulk of the sector appears voluntary and behaves procyclically. 2. Understanding the nature of the informal salaried sector will require more work. It does not seem to serve the same role as port of entry to working world that it does in Mexico. Those emerging from school without employment would seem to prefer unemployment. However, its constituents are still among the youngest and do show a graduating pattern from informal salaried to formal work although at a rate much slower than in Mexico. On the other hand, there is no obvious cyclical pattern that would suggest that the sector purely absorbs excess labor as more traditional views might suggest. They do earn ( unconditionally) less than the formal sector, but may represent younger workers in apprenticeship roles or family members paid in kind. It is unclear whether this group merits particular focus from a poverty perspective. There is some evidence that labor codes are distortionary and rigid by LAC standards 3. The existence of a large informal sector is not, in itself, evidence of large distortions in the labor market. However, on several counts, the various indicators suggest that Argentina's market is both more distorted and rigid than most. Rates of turnover in the formal sector are low both in absolute terms and after adjusting for numerous factors that theory suggests should influence them. However, this rigidity seems a characteristic of the market as a whole and it is difficult to 31 Argentina Poverty Assessment Background Paper No. 3 know if this is caused by formal sector legislation or if there is some other factor affecting both the formal and informal sectors. 4. The size of the self-employed sector is larger than would be expected for a country of its level of income and demographics. This suggests that a larger fraction may be involuntarily informal. 5. Finally, even in the absence of strong evidence on segmentation or barriers to entering the formal sector, this says nothing about the impact that labor legislation or unions may have on formal sector productivity. It is probably the case that too much attention has focused on relations between formal and informal sectors as a measure of efficiency and insufficient on indicators within the formal sector. These distortions also reduce the desirability offormal work and impede safety net coverage. 6. It may also imply, however, that the disincentives to being formal are higher. If the implicit tax on wages is far higher than the actual benefits received, workers have the incentive to avoid formality. The current high levels of non-wage costs and the inefficient provision of services make this a likely scenario. Improving the quality and cost-effectiveness of labor benefits is a critical element in any program to extend them to a broader range of the population. However, it is also the case that the vast majority of the self-employed in Chile, when given a chance to contribute to personal retirement accounts (AFP), chose not to. Until this logic is understood, extending safety nets more broadly will be difficult. Fomenting Micro-firm growth. 7. The large share of employment in the micro-enterprise sector suggests that measures to promote growth may yield dividends from a poverty point of view. As yet, there is limited evidence as to what are the barriers to growth. Work by FONCAP suggests that there is a demand for credit not served by formal financial institutions, but it may also be the case that weak institutional structures-contracting mechanisms, judicial access- make expanding risky and hence credit is not a binding constraint. FONCAP is contemplating a longitudinal study of micro-firm dynamics which would be extremely useful in approaching this question. Unemployment 8. The finding that the majority of unemployment derives from those who were previously informal may only suggest the standard finding that small firms have high mortality rates everywhere. To the degree that other work establishes that those unemployed are especially poor, this may suggest policies to extend safety nets to the informal sector. The results suggest that those fired will spend a longer time in unemployment and have a greater chance of falling into low income 32 Argentina Poverty Assessment Background Paper No. 3 jobs. However, if the state of being unemployment itself suggests that the worker can afford to be unemployed, this may not necessarily follow. 9. The bulk of unemployment is among the young, particularly those between 16 and 20. To the degree that these individuals are family members with other means of support, the poverty impact may be less severe than for those in prime age. Income transitions as indicators ofpoverty 10. Individual reported income is frequently used to calculate poverty and distribution statistics. The transition matrices here suggest rapid movement out of "poverty" but also cast doubt on the correspondence between reported and permanent income. At least 25% of those who were initially found in the bottom quintile were found in the upper 2 quintiles at the end of two years. It seems more probable that this represents "vacationing wealthy" rather than astounding income mobility. All of the conclusions below must be interpreted with this in mind. 11. Distribution measures based on permanent ( average non-zero) income reduces the ratio of 5th to l' by 50% and the gini by 7%. 12. Women appear to spend longer in low income (1V quintile ) jobs and have a higher incidence of falling into them from higher quintiles. 13. There is a trend toward longer duration in low income jobs over time for men although probably not for women. However, whatever is driving this does not imply a secular increase in the probability of becoming poor. 14. Informality does not appear to preclude upward movement out of povety. As with those who remain formal, those who are informal in the first and fifth periods spend less time in the bottom quintile and have a lower probability of falling into the bottom quintile. Since the reference group almost by definition consists of those moving among sectors, it may be that this simply captures successful job matches. 33 Argentina Poverty Assessment Background Paper No. 3 References Aroca, P. and W. Maloney (1999) "Logit Analysis in a Rotating Panel Contex and an Application to Self-Employment Decisions, IBRD Working Paper 2069. Aronson, R.L. (1991), Self- Employment, A Labor Market Perspective, ILR Press, Ithaca Baldn, J. and H.L. Browning, E. Jelin (1973), Men in a Developing Society, Institute of Latin American Studies, U.T. Austin. Beneria, L.(1989) Subcontracting and Employment Dynamics in Mexico City, in Portes et al., The Informal Economy, Johns Hopkins. Brock, W.A., D.S. Evans, and B.D. Phillips, (1986) The Economics of Small Businesses:Their Role and Regulation in the U.S. Economy, New York: Holmes and Meier. Esfahani, H. and D. Salehi-Isfahani (1989), "Effort Observability and Worker Productivity: Towards and Explanation of Economic Dualism," Economic Journal, 99, 818-836. Evans, D. S. and L. S. Leighton (1989), ASome Empirical Aspects of Entrepreneurship, American Economic Review, 79:3, 519-535. Evans, D.S. and B. Jovanovic (1989), AEstimates of a Model of entrepreneurial choice under Liquidity Constraints, Journal of Political Economy. Fairlie, R.W. and B.D. Meyer (1996), ATrends in Self-Employment Among While and Black Men:1910-1990, mimeo, U.C. Santa Cruz. Fields G. S. (1990), "Labor Market Modelling and the Urban Informal Sector: Theory and Evidence," in OECD, The Informal Sector Revisited, Paris. Gottschalk, P. and T. Maloney (1985) , "Involuntary Terminations, Unemployment, and Job Matching: A Test of Job Search Theory, Journal ofLabor Economics, 3: 2.109-123. Gregory, P. (1986), The Myth of Market Failure: Employment and the Labor Market in Mexico, World Bank & Johns Hopkins, Baltimore. *Harris, J. R. and M. P. Todaro (1970), AMigration, Unemployment, and Development: A Two- Sector Analysis, American Economic Review, 60:1, 126-142. 34 Argentina Poverty Assessment Background Paper No. 3 Heckman, J. (1979), ASample Selection Bias as a Specification Error, Econometrica, 47:153-161. Hemmer, H. and C. Mannel (1989), "On the Economic Analysis of the Urban Informal Sector," World Development, 17:10. Jovanovic, Boyan (1982), ASelection and Evolution of Industry, Econometrica, 649-670. Krebs, T. and W. Maloney, "Quitting and Labor Turnover: Micro-Evidence and Macro Consequences", IBRD Working Paper #2068. Levenson, A. and W. Maloney (1997) AThe Informal Sector, Firm Dynamics and Institutional Participation, Working Paper, University of Illinois and IBRD Policy Research Working Paper 1988. Maloney, W. F. (1999) "Does Informality Imply Segmentation in Urban labor markets? Evidence from Sectoral Transitions in Mexico, World Bank Economic Review 13:2 p 275-302. Maloney, W. F. (1997), "Labor Market Structure in LDCs: Time Series Evidence on Competing Views." IBRD Working Paper #1940. Maloney, W.F. (1999), "Self-Employment and Labor Turnover: Cross Sectional Evidence," IBRD Working Paper #2102. Marcouiller, D. V., Ruiz de Castilla and C. Woodruff (1995), " Formal Measures of the Informal Sector Wage Gap in Mexico, El Salvador, and Peru" mimeo. Marquez, C. and J. Ros (1990), "Segmentacion del Mercado de Trabajo y Desarrollo Economico en Mexico," El Trimestre Economico. Abril-Junio. Mazumdar, D. (1983). "Segmented Labor Markets in LDC's." American Economic Review, 73, 2 254-259. Ozorio de Almeida, A.L., and S.E.M Graham and L. F. Alves (1994) "Poverty, Deregulation and Informal Employment in Mexico," ESP Working Paper, World Bank, July. Pessino, Carola AArgentina: The Labor market during the Economic Transition, Portes, A. (1994), AThe Informal Economy and Its Paradoxes, in Neil J. Smelser and Richard Swedberg, eds., The Handbook ofEconomic Sociology, Princeton, N.J.: Princeton University Press. Portes, R. and R. Schauffler (1993), "Competing Perspective on the Latin American Informal Sector," Population and Development Review, 19:1, 33-59 35 Argentina Poverty Assessment Background Paper No. 3 Portes, A., M. Castells and L.A. Benton, (1989)., The Informal Economy, Johns Hopkins, Baltimore. Roberts, B. (1989) "Employment Structure, Life Cycle and Life Chances: Formal and Informal Sectors in Guadalajara," in Portes et al., The Informal Economy, Johns Hopkins. Rosenzweig, M.R.( 1988) ALabor Markets in Low-Income Countries, Handbook of Development Economics, H. Chenery and T.N. Srinivasan, eds. Vol 1, North Holland, New York, Sabot, R.H. (1977) AThe Meaning and Measurement of Urban Surplus Labor Oxford Economic Papers 29:389-411. Secretaria del Trabajo y Prevision Social (1993), El Sector Informal en Mexico, no 2. Sedlacek, G.L. and R. Paes de Barros and S. Varandas (1995), "Segmentacao e Mobilidade no Mercado de Trabalho Brasilerio," Unpublished Manuscript. Stiglitz, J.E. AAlternative Theories of Wage Determination and Unemployment in LDC=s: The Labor Turnover Model, Quarterly Journal of Economics, 88:194-227 Thomas, J.J. (1992), Informal Economic Activity, Ann Arbor, University of Michigan Press. Tokman, V.E. (1992) Beyond Regulation, The Informal Economy in Latin America, Lynne Rienner, Boulder. U.S. Dept of Labor (1993) Work Without Protections: Case Studies of the Informal Sector in Developing Countries. U.S. Dept of Labor (1992) The Informal Sector in Mexico.World Bank (1995), "Labor and Economic Reforms in Latin America and the Caribbean," Regional Perspectives on World Development Report 36 Appendix 1: Worker Transitions Among Sectors of the Labor Market Across Five Quarters Informal Defined as Unprotected Mexico National Urban Employment Survey, 1991-1992 Probability of moving from initial to final sector, Pij, in percent Final Sector Initial Sector OLF School Unemp Unpaid SE IS FS Contract Other Total Out of Labor Force T' I 71 9 (10; 1 - School 2 53 5 6 Q 17 3 1 100 Unemplo%ed 14 -4 20 1 II I 615 -I 1 0o0 Unpaid 3 1 6 27 15 I j ~~--- --- Iu--I---- - Self Emplo,ed ISE 2 0 2 I 66 8 9 5 6 100 Liormal Salaried (IS) 1 2 - .' 16 38 23 7 5ool Formal Salaried tFSI 2 I 3 S 15 6 76 4- 100! Contract 0 2 01 4' 12 20 4.5 5 10u O t h e r 3 2 5 ! 2 l I I 3 4 4 1 6 1 0 0 Total (P.j) 6 4 4 2 21 111 40 8 5 100 Pij standardized by final sector size, Pij/P.j Final Sector Initial Sector OLF School Unemp Unpaid SE IS FS Contract Other Out ofLabor Force `3 172 48 421 2 1 1 41 School 26 128 373 Iii 88 42 71 Unemployed 231 95 97 io ls 66 iL1 148 Unpaid 299 168 1 69 27 82 SelfEmployed(SE) 37 8 53 67' 781 221 70 14 Informal Salaried (IS) 251 1081 00 57, 95 98 Formal Salaried (FS) 291 141 68 25; 25 54 71 Contract 241 11 49 IT 64 I 15 s50 101I Other 46 421 134 63 99 107 84 91 Vij = Pij/P.j*(1-Pii)*(1-Pjj) Final Sector Initial Sector OLF School Unemp Unpaid SE IS FS Contract Other Out ofLabor Force 171 7671 2311 438 126 198 97 175 School 104 341 10831 6-2 314 5! 177~i Unemployed 1029' 25-161 18-7 218S 347 135 2 Unpaid 261 867i -871 -__ 277- 145 .9' 14 ___ SelfEmployed (SE) 3831 53, 196 272 370: 274 375 4on Informal Salaried (IS) 1-1 180 218 -440 3531 385 J; 278 9 n Formal Salaried (FS) -424 12-1 355 143 304 357' 411U 354 Contract 1551 -3 Ill -4l 345, 338 381 219 Other 194 106 201 1031 3501 2061 197 Notes: Sample aggregates three panels 1990:3-91:3, 1991:1-1992:1, 1991:2-92:2 generating roughly 15,000 observations. Includes male workers, 16-65 with high school education or less in 16 metropolitian areas. Informal defined as workers in firms under 16 workers not covered by medical or social security benefits. SE, IS, Contract are informal. FS is Formal. Boxed cells represent work, light shading=paid work, darker shading=probability of remaining in same sector. By compensating for final sector size, Pij/P.j gives a measure of fluidity among sectors. By compensating for rates of turnover in initial and final sectors, Vij gives a measure of disposition (logic) to move to a sector. Argentina Poverty Assessment Background Paper No. 3 Appendix II. Argentina in International Context; Cross Sectional Evidence' Krebs and Maloney (1999) Table AII.1: Determinants of Self -Employment and suggest that both the size of the Turnover informal sector and rate of turnover may be strongly affected by Self-Employment Mean Tenure demographic and economic factors including the normal process of growth. Sample Complet Reduced Complete Reduced The question then arises, how distorted e or inflexible is Argentina once we adjust C 0.36 0.50 74.35 43.30 for these factors? (2.46) (3.29) (4.23) (6.03) Table AII.1 presents some Indust. V.A. -0.03 -0.05 -16.21 -3.19 suggestive tests using cross-country (2.24) (3.19) (3.99) (4.88) data for determinants of the share of the workforce in self-employment and the I.V.A. sq 0.92 mean tenure in the manufacturing (3.99) sector. Mean tenure bears an inverse proportional relationship to turnover SS Worker 8.68 and was more consistently reported in (3.62) the data so it is used to proxy for turnover.2 The results must be SS Emp 0.16 0.15 6.23 11.65 interpreted with caution. First, data on (2.81) (1.63) (1.79) (5.86) at most 40 countries are available. This in some cases can make the results Protection 4.OOE-03 0.09 sensitive to the countries included. (2.93) (2.99) Second, the data are not uniform. Most of the OECD variables were gleaned Real Interest 0.23 0.21 -7.10 from presumably consistent publications (4.71) (3.21) (1.89) of that organization while the Latin variables were individually extracted Secondary -1.00E- 0.04 from not necessarily consistent survey 03 data. A Latin American dummy that (2.37) (2.23) might have picked up these data discrepancies was never significant. Youth 4.14 Despite these potential pit-falls, the (2.8) regressions prove surprisingly robust NOBS 40 20 26 17 and consistent with most of the R2 0.90 0.90 0.68 0.81 Note: t-statistics below coefficients. 1 Based on Maloney(1998) Self-Employment and Labor Turnover in LDCs: Cross Country Evidence. The data sources and more detailed descriptions are listed therein. 2 If half of all workers leave their jobs in a year, then the mean tenure is two years. 38 Argentina Poverty Assessment Background Paper No. 3 predictions of the graphical analysis. The explanatory variables corresponding to the graphical exercises and simulations are the following: 1. Formal Sector Labor Productivity/Profitability (Z) Industrial Value Added (Indust. V.A.): The log of industrial value added per industrial worker proxies for formal sector labor productivity specifically and more generally for level of development. Social Security Tax-Employers (SSEmp): Social security (broadly defined) taxes as a share of the wage by worker. The model suggests that there may be different effects. Employment Protection(Protection): An index of employment protection constructed by Gustavo Marquez (1998b) that captures both the difficulty of laying off workers and the cost in terms of severance pay. Unfortunately, this leads to roughly a halving of the available observations and hence a separate set of regressions are run using this reduced sample. 2. The rise in the benefits to being self employed, or a reduction in worker taxes (8) Social Security Tax-Workers (SSWorker): Social security (broadly defined) taxes as a share of the wage by worker. 3. Hiring Costs: Share of appropriate age group with secondary education (Education) Real Interest Rate (Real Interest) 4. Other Variables: Share of the working population found between the ages of 16 and 20 (Youth.) 4 Chief Economist Office of the IDB. 4 The model implicitly assumes homogeneous work forces across countries. This is clearly not true as the share of young workers is much higher in Latin America than in the OECD countries. This variable is most relevant to the turnover regressions where traditionally young workers have higher rates of turnover as they shop around for careers. But it may also have a similar interpretation in the self-employment regressions. 39 Argentina Poverty Assessment Background Paper No. 3 Empirical Determinants of Self-Employment The employers' social security tax, the real interest rate, the level of education, the labor productivity and educational variables and the share of young people, enter significantly and of predicted sign. The first four are all consistent with both the graphical analysis and the simulations. The fact that in no case was the worker's share of social security remotely significant may well be due to the large measurement error in this variable.5 Most important in terms of magnitude is formal sector labor productivity. Taking the extreme values of this variable would account for 9.6 points of the variance in the share of self-employment. Figure 4 suggests the same, but important conclusion: a large self-employed sector is not obviously evidence of distortions, but may rather reflect that the opportunity cost ofself-employment is lower in poorer countries. Table AII.2 presents the raw self-employment level for each country and its deviation from that predicted.given the country's level of development as measured by formal sector productivity. As might be expected from figure 4, there is a complete re-ordering. Costa Rica, Honduras and Panama, with among the largest self-employed sectors appear grouped with relatively undistorted economies such as the U.S. and Canada once level of development is taken into account. Mexico appears roughly at the predicted value given its level of development. The relative youth of Latin America's population also explains much of the variance with the spread across the sample accounting for 8.8% of the higher share in self-employment. The education variable, measuring the degree to which firms must bear the burden of financing general education, would account for roughly 7.5 percentage point difference. Real interest rates are also surprisingly important.6 The difference between 5% as in the OECD vs often above 30% in Latin America is worth over 6% in the share of self-employment. Once again, the importance of ensuring macro- economic stability, and reducing risk to bring down interest rates seems clear. 5 In roughly 25% of the cases, the social security law dictates a progressive tax that varies greatly across the range of incomes. We chose the midpoint of this range, but we can have no way of knowing if, in fact, this represents that average tax on labor. 6 Peru's very high self-employed sector (46%) and high real interest rates (67%) are clearly dominating the relationship although even dropping this outlier yields a significant coefficient. 40 Argentina Poverty Assessment Background Paper No. 3 In general, these effects dwarf the impact of any of the three labor market Table AII.2: Self-Employment variables. Across the range observed, social Shares security taxes on employers explain Country % Self- Dev. From relatively little of the size of the sector, Employed Predicted roughly 3.2 percentage points. Costa Rica 24.4 -7.6 Honduras 31.8 -6.8 The same exercise was repeated with Austria 6.6 -5.9 the smaller sample arising from using the Panama 26.2 -5.1 Marquez protection index, in column 2. US 7.3 -4.8 With fewer observations, the apparent Canada 8.9 -4.6 collinearity of the productivity variable and Guatemala 32.3 -4.1 education variables makes identifying the France 8.6 -4.1 "correct" parsimonious form difficult. Denmark 6.9 -3.6 However, in all cases, the employment Ireland 13.5 -3.3 protection index has the impact of increasing Chile 25.2 -3.2 the size of the self-employed sector. But, Paraguay 31.7 -3.2 again, the difference between the highly Germany 8.5 -2.8 protective Mexico, or Bolivia, Honduras, Netherlands 9.6 -2.7 Peru or Spain compared to the unprotective Norway 5.9 -2.7 US or UK is worth only about 1.35 Luxembourg 5.8 -2.3 percentage points in the share of self- Sweden 9.3 -1.9 employment, a fairly small impact. Brazil 23.2 -1.8 Finland 9.7 -1.8 In sum, the three labor distortion Australia 12.3 -1.4 variables, the tax on salaries, on payroll, and Bolivia 39.2 -1.2 restrictions on hiring and firing have UK 12.6 -0.7 relatively small impacts compared to those Portugal 19.5 -0.7 of the productivity, real interest rate, New Zeal 16.2 -0.5 education variables and, in the larger sample, Mexico 26.5 0.2 the relative youth measure. Thus, it is hard El Salvador 35.2 0.2 to argue that these distortions are Turkey 26.0 0.7 responsible for the size of the sector. Belgium 13.3 0.8 Japan 9.9 1.7 Colombia 34.2 2.3 Spain 18.6 2.4 Switzerland 10.6 2.5 Iceland 15.5 4.0 Korea 23.6 4.3 Argentina 28.5 5.9 Uruguay 32.8 6.8 Venezuela 34.4 8.4 Greece 27.7 8.4 Italy 22.8 9.4 Peru 46.1 18.6 Argentina Poverty Assessment Background Paper No. 3 Empirical Determinants of Turnover The turnover results are more difficult to interpret first, because of the small number of observations available, and second because as the previous section shows, it is far less clear theoretically what the signs should be. Column 3 presents regressions of average tenure in manufacturing employment on the same set of independent variables. In the complete regressions, labor productivity enters both in levels and with its square suggesting a non-linear relationship. Taken at the mean, labor productivity appears to have a negative impact on tenure (a positive impact on turnover), consistent with theory, and in conflict with the simulation results. This can be reversed with the exclusion of all other variables, but the inclusion of the share of the population with secondary schooling reverses its sign. Thus, it appears that the OECD countries have more stable work forces, it appears that it is the fact that they are educated, rather than rich that drives the result. Both taxes on social security appear to increase tenure and real interest rates decrease it. Including the employment protection variable reduces the sample to 17 observations and leaves only the productivity, social security tax on employers and protection variables significant. The latter enters with predicted sign, suggesting that it does negatively affect turnover and importantly. The difference across the range from the U.S.(1) to Venezuela (37) accounts for 3.24 years on a mean of about 9. Distortion and Rigidity Reconsidered The above findings suggest that many economic and demographic factors not directly related to labor market legislation affect both the size of the self-employed sector and rates of turnover. Only by compensating for these variables can more meaningful measures of distortion and rigidity be constructed. Figure 5 attempts, in a rough fashion, to do this. It plots the residuals from the regressions of the self-employment share and tenure on labor productivity, education, real interest rate, and youth to see if the combination of the two can reveal anything about the functioning of a given labor market. In theory, these residuals capture the impact of all variables not explicitly included in the regression, including other labor market distortions or rigidities.8 Though speculative, we will interpret these See MArquez and Pag6s (1998) for a graphical treatment of these issues. The econometric results here are broadly consistent with their findings. s It is absolutely correct to argue that they also include any country specific variables and any error in measuring sector size or turnover, both of which cast doubt on using the residuals for this purpose. However, this critique applies to the use of the raw measures of these variables as well and since they are thought to contain information, then the adjusted values obtained from the residuals are probably more appropriate. 42 Argentina Poverty Assessment Background Paper No. 3 residuals as measuring labor market distortion (self-employment above the predicted value) and rigidity (average tenure above the predicted value). Despite the crudeness of the technique, the results are broadly plausible. In the North East quadrant of figure 5, more rigid and distorted economies, we find Greece, Italy and Spain, all European countries renowned for repressive labor codes. It is, however, interesting to note that the only Latin country appearing in this quadrant, perhaps unsurprisingly is Argentina. Paraguay, in the South-East quadrant also appears quite rigid although less distorted than average. Venezuela, in the North-West quadrant, on the other hand, has high rates of turnover but a very high self-employment sector suggesting perhaps substantial rigidity. Among the most flexible and undistorted, the South West quadrant, we find, unsurprisingly, the U.S. and Canada, but also Bolivia, Panama, and Venezuela in the Western quadrants and Brazil and Panama just over the line. These results require very strong caveats both on empirical and conceptual grounds. The available data is thin and hence the regressions and their residuals must be treated with caution. It also must be remembered that the level of development (formal sector productivity) is not exogenously given, but may itself be affected by labor legislation. 9 Since this variable has great explanatory power in both the self-employment and turnover regressions, the measures of distortion above may be understated. Further, many other social programs and labor market institutions also have historically emerged over the course of development that will be correlated with productivity. Since it is not obvious whether these would be more or less distortionary, the direction of bias induced in the residuals cannot be known a priori. 9 The graphs were redone replacing industrial value added per worker with purchasing power parity adjusted per capita GNP. The placement of countries changed little suggesting that the results are not very sensitive to the exact measure of labor productivity used. Panama, Brazil and Honduras do waver across the line separating the south east and south-west quadrants depending on specification. 43 Argentina Poverty Assessment Background Paper No. 4 Poor People in a Rich Country A Poverty Report for Argentina Background Paper No. 4 Protecting the Poor and Improving Investment in Their Human Capital Vicente Paqueo and Kihoon Lee April 13, 1998 A contribution to the Argentina Poverty Assessment Argentina Poverty Assessment Background Paper No. 4 1. Introduction In poverty assessments, it is important to have a perspective regarding the size of social expenditures and the extent to which the poor shares in their benefits. It is also important to examine how investments in the human capital of the poor and the protection of their well being could be enhanced through public and private efforts. This paper examines certain aspects of these themes. It should be mentioned at the outset that this study is intended to complement two other reports on health and education; hence, it is limited to certain topics and does not go into extensive discussion of health and education issues 2. The first section of the study presents a brief overview of the size and composition of social spending in Argentina. This section draws from the recent Ministry of Economy's report, Caracterizaci6n y Evoluci6n del Gasto Pfiblico (DNGPG 1998). The next sections then discuss sequentially selected issues in social welfare, health and education. Key topics include: (a) the incidence of health, education and social welfare expenditures and the impact of public and private transfers to households on the rate of poverty; (b) the economic feasibility of financing health insurance coverage for the poor within existing budgetary constraint; and (c) determinants of children's school participation and their implications for the poor. The 1997 Encuesta de Desarrollo Social (EDS) of the Secretaria del Desarrallo Social provides the data base for the analysis of these topics. The setting: an overview of social expenditures 3. Argen1tina has the highest social spending in Latin America (DNGPG 1998). In 1997 consolidated public social spending (GPS) stood at 1,585 pesos, as can be seen from Table 1. This constitute some 17.5 percent of its GDP (PBI) and 64.6 percent of consolidated public expenditures (GPC). Table 1: Consolidated Public Social Spending, 1980 - 1997 (average) Period GPS GPS per capita GPS/PBI GPSIGPC (in millions 19Q7 tin thous3nds 1997 (%1 (%) Pes~~~~i i pes,os) ________ 1980-1983 33.19 1.16 14.7 45.1 1984-1988 37.94 1.23 16.8 51.6 1989-1990 36.03 1.12 17.5 56.8 1991-1995 51.58 1.52 18.6 64.0 1996 54.53 1.55 18.3 65.5 1997 56.54 1.59 17.5 64.6 Notes: (i) GPS: Public Social Expenditures (ii) PBI: Gross Domestic Product (GDP) (iii) GPC: Consolidated Public Spending Source: Direcci6n Nacional de programacidn del Gasto Social - Secretaria de Programacidn Econdmicay Regional Caracterizaci6n y Evolucidn del Gasto Ptiblico Social en el Periodo 1980-199 7, Octubre 1998, processed. 4. There are several noteworthy trends regarding social spending in Argentina. First, one finds a higher per capita GPS in the 90s than in the previous decade. This is due to rising national income and a slight increase in the share of GPS in GDP. Second, this trend has 1 Argentina Poverty Assessment Background Paper No. 4 been accompanied by an increase in the share of GPS in public expenditures from 45.1 percent at the beginning of the 80s to 56.8 percent at the beginning of the 90s to 64.6 percent in 1997. These facts can be interpreted as an increase in government commitment to social issues, considering that there has been a downsizing of public spending relative to GDP (Table 2) as the role of government was being re-defined in favor of market solutions to economic problems. Table 2: Consolidated Public Spending, 1980-1997 Year millions of per capita % of PBI (**) 1997 pesos (N) in thousands 1997 pesos 1980 77.79 2.77 32.21 1981 82.01 2.88 36.22 1982 66.46 2.29 32.29 1983 66.15 2.25 29.23 1984 68.34 2.29 28.87 1985 69.57 2.30 33.56 1986 74.14 2.41 32.74 1987 82.04 2.63 35.02 1988 72.66 2.30 31.75 1989 65.25 2.03 33.35 1990 61.95 1.91 28.67 1991 68.81 2.09 28.40 1992 77.30 2.31 29.11 1993 82.73 2.44 29.09 1994 88.29 2.57 29.09 1995 85.48 2.46 29.93 1996 83.29 2.37 27.92 1997 87.47 2.45 27.05 Note: the amounts in 1997 are preliminary *) see methodologic appendix (**) expenditures in pqsos/PBI in pesos Source: Direcci6n Nacional de Programaci6n del Gasto Social - Secretaria de Programacidn Econdmicay Regional Caracterizaci6n y Evoluci6n del Gasio P4blico Social en el Periodo 1980-1997, Octubre 1998, processed 5. Social expenditures in Argentina may be classified into two broad categories (Table 3). One is social insurance spending; the other is social sector (or non-insurance) expenditures. The former consists of expenditures that are financed mainly by beneficiary contributions; the latter are paid for by fiscal resources (see Table 4). These are taxes levied from the general public. 62.2 percent of public social spending is financed by these taxes. The rest are funded by mandatory beneficiary contributions and fees. Public spending on education is paid for 100 percent by fiscal resources; in contrast, 57 percent of public health spending is financed by mandatory fees and contributions. It is interesting to note that fiscal resources finances close to half of social security expenditures. 2 Argentina Poverty Assessment Background Paper No. 4 Table 3: Consolidated Public Social Spending: Social Insurance vs. Non-Insurance (Sector) Expenditures, 1980-1997 Share of GDP Non-Insurance (Sector) Social Insurance Total 1980 6.96 9.13 16.09 1990 6.78 10.75 17.53 1997 7.55 9.93 17.48 % Distribution 1980 43.2 56.8 100 1990 38.7 61.3 100 1997 43.2 56.8 100 Source: DNPGS: Secretaria de Programaci6n Econ6mica y Regional, Caracterizaci6n y Evolucid6n del Gasto P4blico Social en el Periodo 1980-1997, Processed, October, 1998 Table 4: Consolidated Fiscal Social Spending, 1997 (Pe nt) Function Fiscal Fees and Total Expenditures Contributions Public Social Spending 62.20 37.80 100.00 Education, Culture/Science 100.00 100.00 Health 42.10 57.90 100.00 Water and Sanitation 100.00 100.00 Social Security 48.00 52.00 100.00 Employment 22.10 77.90 100.00 Housing/Urban Development 100.00 100.00 Social Promotion/Assistance 79.00 21.00 100.00 Other Urban Services 100.00 100.00 Source: Direcci6n Nacionalde Programaci6n del Gasto Social - Secretaria de Programacidn Econ6mica y Regional Caracterizaci6ny Evolucidn del Gasto Pblico Social en elperiodo 1980-1997, Octubre 1998, processed 6. The pattern of distribution between these two groups of spending (Table 5) shows that two items account for 87 percent of the social insurance expenditures: social security payments (62.9 percent) and (b) expenditures for social health insurance. With respect to social sector spending, the bulk of that goes to education, culture and science (50.9 percent) and health expenditures (21.4 percent). Social promotion and assistance (or social welfare) is third with 9.5 percent. In 1997 social insurance constitute about 56.8 percent of total consolidated public social expenditures (Table 5 ). This is almost 10 percent of GDP, compared to 7.55 percent for social sector spending. 3 Argentina Poverty Assessment Background Paper No. 4 Table 5: Distribution of Consolidated Public Social Spending: Social Insurance vs. Sectors (Non-Insurance) Function Social Sectors Social Insurance Education, Culture and Science 50.9 Health 21.4 24.5 Water and Sanitation 1.9 Housing/Urban Development 5.0 Social Promotion/Assistance 9.5 3.2 Employment 1.9 9.4 Social Security Services 62.9 Other Urban Services 9.3 Total 100.0 100.0 Source: Direcci6n Nacional de Programaci6n del Gasto Social - Secretaria de Programaci6n Econ6mica y Regional Caracterizacidn y Evoluci6n del Gasto Pblico Social en elperiodo 1980-1997, Octubre 1998, processed 7. Public social sector spending can be further sub-divided into two types of expenditures: universal and targeted (focalizado) programs (Table 6). Universal programs like support for education are intended to benefit the general public. In contrast, targeted expenditures are intended to benefit the poor and other disadvantaged groups. In 1997 the share of targeted social spending stood at 16.4 percent of GPS. This is slightly less than the figures in the 80s. 8. It is clear that the total amount of targeted public spending grew between 1992 and 1997. In fact, it grew by 27.2 percent during said period. By ordinary standards, this is a reasonable growth. The question is whether the growth in targeted social spending is proportional to the growth of the number of poor people. The preliminary answer is that the finance of targeted public spending has fallen behind during the nineties (Chart 1). Chart 1: Trends in Poverty Rate and Targeted Social Spending Indices 200 [---Poverty Rate 150 Index 100 - 0 - -Targeted Social 0 ----- --------- Spending a r-e c to Index 4 M 0) M) M) a) a) M0) 0 4 Argentina Poverty Assessment Background Paper No. 4 Table 6: Consolidated Social Sector (Non-insurance) Public Expenditures (1997) Universal vs. Tarneted A. %of GDP Yer Universal I ) Taraeted ") Total 1980 5.54 1.42 6.96 1981 5.79 1.14 6.93 1982 4.22 0.91 5.13 1983 4.57 1.17 5.74 1984 5.36 1.21 6.57 1985 5.78 1.45 7.23 1986 6.27 1.51 7.78 1987 6.69 1.67 8.36 1988 5.98 1.34 7.32 1989 5.56 1.43 6.99 1990 5.54 1.24 6.78 1991 5.72 1.17 6.89 1992 6.14 1.09 7.23 1993 6.57 1.15 7.72 1994 6.61 1.20 7.81 1995 6.69 1.13 7.82 1996 6.32 1.17 7.49 1997 6.31 1.24 7.55 B. Percent Distribution 1980 79.64 20.36 100.00 1981 83.59 16.41 100.00 1982 82.28 17.72 100.00 1983 79.58 20.42 100.00 1984 81.60 18.40 100.00 1985 79.92 20.08 100.00 1986 80.58 19.42 100.00 1987 80.03 19.97 100.00 1988 81.75 18.25 100.00 1989 79.55 20.45 100.00 1990 81.71 18.29 100.00 1991 82.98 17.02 100.00 1992 84.87 15.13 100.00 1993 85.15 14.85 100.00 1994 84.60 15.40 100.00 1995 85.60 14.40 100.00 1996 84.42 15.58 100.00 1997 83.58 16.42 100.00 Note: the amounts in 1997 are preliminary. (*) covers education, culture and science and technical, health, water and sanitation, and other urban development. (") covers housinglurban develoment, social promotion, social assistance, and employment. Source: Direcci6n Nacional de Programacidn del Gasto Social - Secretaria de Programacidn Econ6mica y Regional Caracterizacidn y Evoluci6n del Gasto POblico Social en el Periodo 1980-1997, Octubre 1998, processed. 5 Argentina Poverty Assessment Background Paper No. 4 Table 6 (cont'd): Consolidated Social Sector (Non-insurance) Public Expenditures (1997) Universal vs. Targeted C. In millions of Pesos Yea r Universal ( j araeted (*1 Tota 1980 13.385 3.422 16.807 1981 13.121 2.575 15.696 1982 8.683 1.870 10.553 1983 10.353 2.656 13.009 1984 12.681 2.860 15.541 1985 11.986 3.012 14.998 1986 14.204 3.424 17.628 1987 15.662 3.909 19.571 1988 13.692 3.058 16.750 1989 10.889 2.800 13.689 1990 11.968 2.679 14.647 1991 13.854 2.841 16.695 1992 16.293 2.905 19.198 1993 18.685 3.260 21.945 1994 20.060 3.657 23.717 1995 19.102 3.216 22.318 1996 18.862 3.480 22.342 1997 20.416 4.010 24.426 D. Index base 1980=100 Year -Universal (M Taraee Ig Total 1980 100.00 100.00 100.00 1981 . 98.03 75.25 93.39 1982 64.87 54.64 62.79 1983 77.35 77.64 77.41 1984 94.75 83.58 92.47 1985 89.55 88.02 89.24 1986 106.12 100.07 104.89 1987 117.01 114.25 116.45 1988 102.30 89.36 99.66 1989 81.35 81.82 81.45 1990 89.42 78.29 87.15 1991 103.51 83.02 99.34 1992 121.73 84.91 114.23 1993 139.60 95.27 130.57 1994 149.87 106.88 141.12 1995 142.72 93.97 132.79 1996 140.92 101.71 132.94 1997 152.53 117.19 145.34 Note the amounts in 1997 are preliminary. (*) covers education, culture and science and technical, health, water and sanitation, and other urban development. (**) covers housing/urban develoment, social promotion, social assistance, and employment. Source: Direccidn Nacional de Programacidn del Gasto Social - Secretaria de Programacidn Econdmica y Regional Caracterizacidn y Evolucidn del Gasto P0blico Social en el Periodo 1980-1997, Octubre 1998, processed. 6 Argentina Poverty Assessment Background Paper No. 4 9.The consolidated public social expenditures above came from different levels of government. Over half (54.1 percent) of those expenditures in 1997 were spending by the national government, which is less than its value of 62.7 percent in 1980 (Table 7). The provincial government, on the other hand, accounts for 39.2 percent, up from 31.7 percent. With respect to the municipal government, it share remains small, although slightly increased. Table 7: Percentage Distribution of Public Social Spending by Level of Government, 1980-1997 Year Nation Province and C8A Municipalities Total 1980 62.7 31.7 5.6 100.0 1990 59.7 34.3 6.0 100.0 1997 54.1 39.2 6.6 100.0 Source: DNPGS - Secretaria de Programaci6n Econ6mica y Regional, Caracterizaci6n y Evoluci6n del Gasto POblico social en el Periodo 1980-1997, Processed, October 1998. 10. As a percent of their own total spending, social expenditures form a large part of the budget of various levels of government (Table 8). But the provincial government shows the highest percentage of 55.2 percent, compared to 47.6 and 51.4 percent of the national and municipal governments, respectively. Table 8: Percentage of Consolidated Public Social Spending in Total Public Spending by Level of Government, 1997 Yr' Naton Provnc ag BA Muipatg Tol 1980 47.6 55.2 51.4 50.0 1990 58.4 68.0 55.3 61.1 1997 66.0 64.9 54.1 64.6 Source: DNPGS - Secretaria de Programaci6n Econ6mica y Regional, Caracterizaci6n y Evoluci6n del Gasto POblico social en el Periodo 1980-1997, Processed, October 1998. 7 Argentina Poverty Assessment Background Paper No. 4 Social Assistance Programs 11. Argentina has numerous targeted social assistance programs that seek to improve the quality of life of the poor, vulnerable, and disadvantaged population. Many of them are also meant to strengthen the ability of communities to address social and cultural issues. Annex 1 provides a list of these programs by thematic areas and describes their objectives and services. This section examines some of these interventions and lays out key questions regarding social assistance spending. 12. Benefit incidence of social welfare programs. Provincial government spending on social welfare (social promotion and assistance) programs include expenditures on food, clothing, shelter and other assistance to the poor and vulnerable population. In 1997 the Statistical Yearbook reports that provincial social welfare expenditures amounted to 3.4 billion pesos. 13. Because these are targeted programs, the distribution is expected to be strongly pro-poor. The interesting question here is: to what extent is there leakage? That is, how much of the benefits from these expenditures go to the non-poor? Although it is neither feasible nor optimal to have zero leakage economically and politically, it is important to know its size. An equally relevant question is: to what extent are the poor and vulnerable getting social assistance? 14. How does one measure benefits from this spending, given data availability and the variety of programs financed under the rubric of social welfare? In the 1997 Encuesta Desarrollo Social (EDS), there is a series of six questions asking whether during the last 12 months anybody in the family has received the following type of assistance: (a) money to pay for housing rent and services; (b) money to pay for other services (e.g. health, food, etc.); (c) in-kind food items; (d) food in community feeding centers or non-school infant feeding programs; (e) clothing; and (f) mattresses and bedspread 15. For each of these items, respondents were asked whether they received assistance regularly or occasionally and whether they received any of it from public agencies, NGOs, Obra Social/trade union, or relatives not living in the household. An index was constructed from the above items to measure (albeit, roughly) the extent to which a family received social assistance from the public sector. A family gets zero, one or two points for each of the above questions if it got no, regular or occasional assistance from government agencies, respectively. 16. The distribution of these benefit points among households by income class was calculated for each province and then aggregated to get the national distribution. Families are classified into income quintiles. The results are shown in Tables 9 below . As expected, 8 Argentina Poverty Assessment Background Paper No. 4 social welfare assistance appears to be strongly pro-poor on the whole. 47.2 percent of the social welfare benefits goes to the poorest quintile. In contrast, the highest quintile gets only 2.4 percent of the benefits. Table 9: Benefit Incidence of Government Social Assistance Income Distribution Population % households quintiles of benefits distribution receiving assistance (%) (%) Public All sources First 47.2 20 29.5 45.0 Second 29.0 20 18.8 31.4 Third 15.0 20 9.8 22.0 Fourth 6.4 20 4.9 16.1 Fifth 2.4 20 1.4 9.8 Total 100 100 12.8 24.8 Poor 64.6 31.0 25.5 39.8 Non-poor 35.4 69.0 7.2 18.1 Total 100 100 100 24.8 (a) Weighted by provincial government spending on social welfare Source: 1997 EDS 17. These results are laudable. Still, they indicate room for improvement. Significant amounts of the social welfare benefits are going to the non-poor ( 35 percent). In fact, about a quarter have been enjoyed by the third, fourth and fifth quintiles. Moreover, Table 9 shows that about 70 percent of the poorest quintile are not getting public assistance and over half got neither public nor private help. Several questions arise. First, what difference does it make in the lives of the poor if they do or do not receive assistance from public programs? Second, how many more poor people could be reached by a more efficient use of resources? Third, how much of the urgent needs of the unassisted poor could be addressed by changing government priorities in the allocation of scarce social development resources? 18. It would appear that these questions have yet to be analyzed comprehensively, systematically and rigorously. A review of available "evaluation studies" reveals that they have not really addressed the impact question. And to the extent that they sought to analyze benefit incidence among socioeconomic groups and issues of efficiency, the analysis has been limited to a few provinces and programs. 19. To answer the above questions comprehensively and rigorously would require much more time and resources than are available for this study. Nevertheless, the 1997 EDS and some of the above-mentioned studies could be used to further frame the issues that need to be addressed regarding social welfare spending and poverty. 20. Food and nutrition. One of the thematic areas in the inventory of social assistance programs (Annex 1) relate to feeding and nutrition of children and senior citizens. Programs in this area include POSOCO (Politicas Sociales Comunitarias), PRANI 9 Argentina Poverty Assessment Background Paper No. 4 (Programa Alimentario Nutricional Infantil), PROHUERTA (Program Huertas) and PROSONU (Progama Social Institutional). Feeding and nutrition related assistance are also included in other thematic areas. In health, programs like PMI (Programa Materno Infantil) and PROMIN (Programa Materno Infantil y Nutricion) have feeding and nutrition components. In Tercer Edad, assistance to help senior citizens meet their basic needs is provided through ASOMA (Apoyo Solidario a Los Mayores). Taken together, the food and nutrition assistance programs are targeted to pregnant mothers, children 0- 14 years old and senior citizens (60 years old and over). In addition to the above interventions, government is funding the PROHUERTA program which seeks to improve the nutrition of the urban and rural poor (NBI families) through the promotion of family and community gardens and farms. 10 Argentina Poverty Assessment Background Paper No. 4 Table 10: Benefit Incidence of Targeted Food and Nutrition Assistance Programs Total Household per capita income quintile Structural Poverty Programs for Quintile Quintile Quintile Quintile Quintile NBI Not NBI 1 2 3 4 5 1. Children 3-4 years old Coverage rate (a) 19.7 20.4 13.8 19.5 21.6 20.2 20.2 19.5 Distribution of benefits Nurseries, preschool & others 100 38.5 26.7 10.9 17.0 6.9 20.4 79.6 Infant feeding centers 100 56.4 25.1 14.8 3.7 0.0 38.6 61.4 2. Primary school students (b) Coverage rate (c) 50.1 69.7 49.8 33.0 30.5 13.2 78.0 41.1 Distribution of benefits 100 56.6 22.9 10.7 8.0 1.8 38.2 61.8 3. 65 year olds and over Coverage rate 6.0 21.2 7.0 5.3 1.7 0.1 32.8 4.7 Distribution of benefits 100 48.4 25.8 19.1 6.3 0.4 25.6 74.4 4. Pregnant mothers and children 0-2 yrs Coverage rate (d) 38.1 44.4 34.8 29.9 34.7 32.5 46.7 34.5 Distribution of benefits (e) 100 69.9 21.2 6.6 2.1 0.2 49.8 50.2 5. PROHUERTA (f) I_I_I_I (a) Includes beneficiaries from preschool, infant, and other feeding centers; (b) Comodores Escolares; (c) Over children aged 6-13 years; (d) Includes beneficiaries of milk delivered by Obra Social; (e) Excludes Obra Social milk beneficiaries; (f) Program for the promotion of community and family gardens/farms. Source: 1997 EDS. Table 11: Distribution of Population By Demographic and Income Groups Income uintiles First Second Third Fourth Fifth Total Children 3-4 years old 32.2 23.6 19.6 15.1 9.5 100 Primary school age children (6-13) 29.4 23.7 18.7 16.2 12.0 100 65 year olds and over 9.4 16.5 23.6 24.6 25.9 100 Pregnant mothers and children 0-2 yrs 30.1 24.1 19.7 14.4 11.7 100 All ages 20.0 20.0 20.0 20.0 20.0 100 Source: 1997 EDS 11 Argentina Poverty Assessment Background Paper No. 4 21. Consistent with Table 9, the figures in Tables 10 and 11 again show the strongly pro- poor character of the food and nutrition programs. The poorest quintile is getting disproportionately more of the benefits than the upper income quintiles. Besides, a high 70 percent of the poorest children appears to have been assisted by these programs. As mentioned, there has been no evaluation of the value added of these programs. The evaluation of ASOMA, however, indicates that senior citizens that have benefited from the program felt better off after receiving assistance, despite the national economic crisis. This finding should, of course, be treated with caution since it does not control for other variables. 22. It is interesting to note that despite the progressiveness in the distribution of benefits, the number of very poor preschool children that have not been reached remains substantial. The percent of children aged 0-2 in the lowest quintile, which are the most vulnerable group to long-term negative effects of malnutrition, is still about 56 percent. The corresponding rate for preschool children aged 3-4 years in the same first quintile is even higher at about 80 percent. Of the poorest 65 year olds and over, only a fifth received food and nutrition assistance. 23. It could be argued that government might have to spend more on those programs to increase their coverage of the neediest children. Given, however, budgetary constraint, it is necessary to see if increased coverage could be increased without significantly raising total spending on these social programs. At present, it is not clear what the scope is for increasing coverage through improved efficiency. An evaluation of the nutrition and feeding programs in a few provinces indicates that, given the decentralized structure of the programs, the number of target beneficiaries reached has been in conformity with what has been expected on the basis of allocated resources. On the other hand, the study points to the possibility of substantially reducing unit cost of food packages through centralized purchasing. In practice, however, this hypothesis is of doubtful validity in light of international experience. Centralization of purchases has its own inefficiencies that could make the program less effective. These include distribution delays, wastage, and failure to delivery to targeted beneficiaries. It is for these reasons that increasingly countries are decentralizing the management of social programs. 24. The study of ASOMA also points to substantial variability in unit cost of assistance, with low and high cost outliers. While this could indicate inefficiency due to lack of cost control, such a conclusion is again not warranted because regional price differences and variations in the components of the food packages have not been taken into account. 25. Depending on the availability of additional resources and the scope for efficiency improvements, hard choices might have to be made to further concentrate feeding program benefits on the poorest and most vulnerable children. There appears some scope for this, as shown by the fact that about 61 to 80 percent of the benefits from nutrition and feeding programs directed at 3-4 year olds come from non-NBI households. Moreover, three fourths of food assistance to senior citizens go to non-NBI households. 12 Argentina Poverty Assessment Background Paper No. 4 26. Finally, the Government might want to consider giving greater budgetary priority to certain demographic groups; that is, reallocate some resources devoted to primary school age children to the neediest of pregnant mothers and children 0-4 years old. It can be argued that the unborn and these very young children are more vulnerable to the ravages of malnutrition. This, of course, is a tough choice that involves value judgments on top of economic cost-benefit calculations. In any case, what is indicated here is the need to establish informed, clear priorities and develop a more efficient system of targeting. In this regard, SIEMPRO's proposed system of identifying and registering poor and socially vulnerable families and their socioeconomic characteristics (SISFAM) would be a step in the right direction. 27. Employment and productivity development. This thematic area consists of various training, employment, and labor productivity promotion initiatives and include workfare, employment subsidy to private firms, and economic assistance to unemployed workers (Annex 1). As with other interventions, little rigorous assessment of the value added and incidence of these programs have been systematically undertaken. It is worth noting here that while an evaluation of Programa Social Agropecuario (PSA), a program designed to help small rural producers through financial, training and technical assistance, had been undertaken, the study focused on organizational and administrative issues, and has little to say about impact, incidence and cost-effectiveness. The exception to the above state of knowledge is TRABAJAR, for which noteworthy evaluation studies have included impact assessment. 28. TRABAJAR is a social safety net program based on work-fare principles. The program provides temporary financial assistance to poor workers, who are suffering from high levels of unemployment, and assists in improving social and economic infrastructure in poor communities. Municipal and provincial governments can propose from a menu of public projects for financing by the program. Such projects include minor construction, repair, expansion or remodeling of schools, health facilities, basic sanitation facilities, small roads and bridges, small dams and canals, community kitchens and centers, and tourist centers. The national government pays for one-third of the cost of the proposed projects; the rest is financed by local government. 29. TRABAJAR II cost about $1.2 billion, of which $192 million are for workers' wages and $8 million for administrative expenses. Management information system data reveal a coverage rate of about 20 to 30 percent of the poor unemployed persons, depending on whether repeaters are excluded or included in the calculation. Studies also reveal that TRABAJAR II, which uses low wage rate to attract only those who are poor and with few prospects of employment, is well targeted. Experience thus far shows that self- targeting is effective: evaluation of the program indicates that on average 40 percent of its resources have reached those in the bottom 5 percent of the income distribution, while 75 percent have benefited the bottom 20 percent. Moreover, targeting of poor communities has also been effective. 30. In terms of impact, an econometric study using a sophisticated version of the matching technique reveals that the net income gain of workers employed by the program's 13 Argentina Poverty Assessment Background Paper No. 4 projects is about half of the workers' wages. The other half is equivalent to the amount that the workers would have earned anyway without the project. Another study indicates that the quality of the projects and their valuation by beneficiaries are on par with those of Social Fund projects on the whole. 31. Community development and social integration. There are several interventions in this thematic area (see Annex 1 ). One of them is a new program called FOPAR (Fondo Participativo de Inversion Social). FOPAR, which now has a budget of about $ 118 million, finances projects prepared and proposed by poor communities. Unlike TRABAJAR, it seeks to build institutional capacity among the poor, and simultaneously to provide them with means to meet specific needs. This means that the projects under this program are defined not only to satisfy specific needs of the poor, but also to be a training vehicle to build and strengthen their ability to identify, define and prioritize their needs, mobilize and manage resources, as well as design and implement their own projects. 32. The program started in 1995, when the Government requested the Bank to support the establishment of a pilot Social Investment Fund (FOPAR), which would operate in the northern provinces with future expansion in mind. Evaluation of this pilot program has been conducted, reviewing 119 projects and interviewing a stratified sample of beneficiaries. These various assessments indicate that 75 percent of beneficiaries come from NBI families. In addition, four-fifths of the remaining 25 percent were from households with monthly income of $300 or less. This effective targeting of the poor comes from a combination of careful geographic targeting and the establishment of a menu of projects eligible for financing that are small and of such a nature as to attract only the poor. 33. The ex-post technical assessment of the civil works considers the quality of the projects as good or better, while interviews with beneficiaries indicate that they were satisfied by the projects. They rated community and beneficiary participation as excellent. A high 96 percent of the beneficiaries felt that the projects effectively satisfied their needs or solved their problems as designed. Furthermore, 97 percent thought that FOPAR was indispensable in implementing their projects. Finally, a large majority of the beneficiaries (83 to 97 percent, depending on the nature of the project being evaluated) emphasized that the financial administration of FOPAR-provided resources were carried out in an honest and transparent manner. 34. The incidence of public and private income transfers on poverty rate. In most countries, low income households are often helped not only by the government but also by relatives, friends and other members of society. This assistance can be in a form of in- kind or monetary transfers such as pensions, remittances from relatives and friends, and social assistance from the governments. In this section, we ask the question: What difference does it make to the poverty rate if these transfers were included or excluded from the calculation of the poverty rate? Who would be most affected by it? 14 Argentina Poverty Assessment Background Paper No. 4 35. We examine this question, again using the 1997 EDS. In this survey, there are two kinds of income: earnings from work and non-work income. Included in the latter are incomes from: (a) becas de estudio (scholarships) (b) seguro de desempleo (unemployment insurance) (c) jubilaci6n contributiva del pais o del extranjero (retirement income) (d) pensi6n ordinaria (income from regular pension program) (e) pensi6n graciable o asistencial (pension to assist the poor) (f) subsidio del gobierno provincial (provincial government subsidy) (g) envios de familiares que no viven en el hogar (remittances from relatives) (h) ayudas de amigos y vecinos (assistance from friends and neighbors) (i) cuota de alimentos (food assistance money) (j) otros ingresos (other income) 36. We interpret "otros ingresos" as income from interest and rents, since the questionnaire does not have any other items asking for income from property. To measure income transfers from all sources, therefore, all the items from (a) to (i) were added. Item (j) was excluded, but added to work income to form an estimate of the family's total income from work and property, which we refer to in the discussion as its "core" income. 37. In Table 12, it is clear that the effect of transfers on the incidence of poverty varies, depending on demographic groups and the definition of poverty. On the whole, the effect is quite modest in regard to indigent households. The difference in poverty rate with and without transfers is only 2.6 percentage points. It rises to 3.9 percent when moderate poverty is included. 38. The pattern of the effects of these transfers is interesting. The most affected group is the age group 60 and over, particularly those above 65 years old. Without transfers, the 65 and over age groups would have experienced an increase of poverty rate from 21 percent to 38.9 percent. The percentage of indigent persons would have risen by 12.2 percentage points. In contrast the prime age group of 25-59 years would have experienced an increase of less than 1.5 percentage points. In terms of gender, poverty rate of female- headed households (widows and single mothers) would have increased from 33.2 percent to 40.7 percent. 39. This pattern apparently makes sense: senior citizens, widows and single mothers are either benefiting from targeted social protection programs or they are simply enjoying the benefits of social security and the life insurance of their spouses. The latter, of course, could also be the recipient of generous gifts and assistance from grown up children, nieces and nephews. 40. Table 13, however, reveals that the better off in terms of core income (labor earnings and asset income) are heavily favored. While the richest quintile get 125 pesos of income transfers per capita, the poorest quintile get only 8 pesos. This pattern does not necessarily imply inequity. It can be argued that these transfers are delayed payments for 15 Argentina Poverty Assessment Background Paper No. 4 past beneficiary contributions to the social security system and assistance from relatives. Two comments are in order here. First, transfers from relatives are comparatively small. Second, as seen in the previous discussion regarding the sources of finance of the various expenditures, close to half of the financing of pension payments are from general fiscal revenues. The issue which is worth examining further is whether re-targeting of this public subsidy to senior citizens with low core income could even further reduce poverty or indigency rates. Table 12: Household Income Transfers and its Effect on Poverty Rate Percent Poor (%) Difference Percent Indigent (%) Difference With w/o % With w/o % transfers transfers transfers Transfers All 31.0 34.9 3.9 10.5 13.1 2.6 Male Household Head 30.4 33.9 3.5 9.9 12.0 2.1 Female Household Head 33.2 40.7 7.5 13.0 18.9 5.9 Age Group of Household Head 15-24 32.7 35.7 3.0 13.5 14.6 1.1 25-59 33.3 34.7 1.4 11.4 12.5 1.1 60-64 26.1 30.5 4.4 8.5 12.1 3.6 65 and More 21.0 38.9 17.9 6.0 18.2 12.2 Source: 1997 EDS Table 13: Average Monthly Income Transfers Per Capita by Source and Core Income Status Core income class All sources Public and private Relatives/friends (quintile) organizations First 10.2 8.2 2.0 Second 21.1 18.9 2.2 Third 36.7 32.1 4.6 Fourth 56.8 49.0 7.8 Fifth 125.1 108.8 16.3 Total 50.0 43.4 6.6. Source: 1997 EDS 41. Multiple regression analysis of income transfers confirm the above results (Table 14). It is interesting to note, however, that holding for age constant, the amounts received by households in the second to fifth core income quintiles are largely the same. In contrast, the difference between what they and the poorest quintile are getting is substantial at about 50 pesos per month. 42. It is also notable that there is no significant correlation between transfers received and the level of provincial government social welfare spending. There are several possible explanations. One is that social welfare assistance are mostly in-kind; hence, its receipt 16 Argentina Poverty Assessment Background Paper No. 4 will not be reflected in the households' reported income, since EDS does not measure the value of in-kind transfers. Another is that few benefits are actually received by the people. Yet, another is that public social welfare spending is simply substituting for non- governmental and private assistance. Table 14: Determinants of Income Transfers: 1997 EDS Variable Parameter Estimate T-statistic Constant 79.02 17.38 Dummy for the Quintile 2 (Q2)') -52.17 -12.36 Dummy for the Quintile 3 (Q3) -46.45 -10.95 Dummy for the Quintile 4 (Q4) -52.71 -12.53 Dummy for the Quintile 5 (Q5) -52.64 -12.59 Male Household Head (SEXOHH) -39.67 -11.83 Age Group 25-59 (E2559) 30.51 10.94 Age Group 60-64 (E6064) 156.35 22.23 Age Group Older than 65 (E6500) 298.34 63.27 Per Capita Social Welfare Expenditure (PSOCWEL) 0.01 0.32 Adjusted R-Square 0.1214 Sample 43,944 Memo: 1) Dummy for the Quintile 1 is omitted 2) Dummy for Age Group 15-24 is omitted Source: 1997 EDS 17 Argentina Poverty Assessment Background Paper No. 4 Health Sector. 43. Relative to its GDP, Argentina's health expenditure is among the highest in the world (Table 15). In 1994, for example, it spent about 10.6 percent of its GDP on health, including public and private expenditures; while the Latin American and Caribbean average was only about 7.2 percent. 44. Consumption of health services in Argentina is financed through various means. These basically include: (a) social health insurance, financed through obligatory contributions by formally employed public and private workers who are members of Obras Sociales; (b) voluntary health insurance, which families buy from the market; (c) provincial government subsidy of publicly provided health services; (d) national government public health programs; and (e) out-of-pocket fees. Table 15 shows that about 41 percent of total health spending in Argentina is financed by the public sector. The rest are paid for by social health insurance and out-of-pocket expenses, a pattern that is in line with Latin American and Caribbean practice. Table 15 Health Expenditures: International Comparison (Circa 1990-95) Per capita health Per capita public Percent of GDP Public expenditure health expenditures Sector (%of total) US$ official PPP$ US $ official PPP$ Total Public Private exchange rate Exchange Argentina (1994) 877 929 356 377 10.6 4.3 6.3 41 World 463 526 278 316 9.1 5.5 3.7 60 Latin America and 234 412 96 169 7.2 3.0 4.2 41 Caribbean 2329 2243 1421 1368 9.1 6.1 3.9 61 High - income 138 293 70 149 6.1 3.2 3.0 51 Middle - income 16 85 7.7 41 5.6 2.8 2.9 48 Low - income I I_I_I_I_I_I_I __ Source: Bank staff estimates 45. In 1997, 64.7 percent of the population is covered by health insurance. Eighty-one percent of the insured population is covered by obligatory Obra Social alone. Another 6.5 percent is insured by both Obra Social and prepaid care, bringing Obra Social coverage to a total of 87.4 percent. The Obra Social coverage accounts for a high percentage of the insured population for all income classes. 46. A sizable percentage of the population (35.3 percent) without health insurance coverage remains (Table 17). Two things stand out. One is the strong negative correlation between insurance coverage rate and income class. The percentage of the population who are uninsured falls drastically from 65.8 percent in the poorest quintile to 9.3 percent in the richest quintile. Not surprising, therefore, 60 percent of the uninsured are poor (Table 19). The other striking feature is that in fact there are still relatively many well-off 18 Argentina Poverty Assessment Background Paper No. 4 families that are uninsured. Of the uninsured non-poor, it turns out that a quarter do not have health insurance coverage. Table 17: Distribution of Population by Health Insurance Coverage Percent coverage Without coverage Income Obra Mutual or Obra Social Emergency quintile Social Prepaid & Prepaid only First 28.3 3.4 1.3 1.2 65.8 Second 51.7 6.2 2.5 1.2 38.4 Third 59.7 7.8 4.6 1.6 26.3 Fourth 66.8 10.2 5.2 1.5 16.3 Fifth 62.3 16.1 11.2 1.1 9.3 Total 51.0 8.1 4.3 1.3 35.3 NBI 33.1 2.5 1.0 0.5 62.9 No NBI 54.4 9.2 4.9 1.5 30.0 Total 51.0 8.1 4.3 1.3 35.3 Poor 35.8 4.4 1.8 1.3 62.3 Non-poor 61.8 8.4 6.4 1.1 28.1 Total 52.3 7.0 4.2 1.2 35.3 Source: 1997 EDS 47. Incidence of provincial health spending. The responsibility for providing basic health services lies principally with the provincial governments, which spend about 11 percent of their total expenditures on health in 1996. This provincial health spending constitute about 17.2 percent of consolidated public social expenditures. The question which this section seeks to address is how the benefits from this provincial health spending are distributed among income classes. 48. To measure the distribution, we constructed an indicator of the number of times household members visited public health clinics and hospitals, using the 1997 EDS. One issue that has to be dealt with relates to the weights that should be put on inpatient versus outpatient care. The usual practice is to weigh them equally. In fact, inpatient care should be given greater weight because it is more resource intensive. In this study, inpatient care is given 12 points against one point for outpatient consultation. This is largely an arbitrary choice, based on an educated guess. In a micro-study of a hospital in Tucuman, the ratio of doctors' time input into inpatient care relative to outpatient care is about 18. If inefficiency from overstaying inpatients is factored out, the ratio is reduced to 12. 49. The above measure still understates health service utilization. The reason is that some preventive care visits might have been omitted because of the structure of the questionnaire. It instructs the interviewer to skip the question on consultation when the individual was not sick or did not feel well during the previous month. To partially address the issue, an alternative indicator was constructed, adding to the above measure prenatal consultations and birth delivery in public health facilities. 19 Argentina Poverty Assessment Background Paper No. 4 50. Keeping these caveats in mind, we measured the distribution of these visits by income class for each province and applied it to the province's public spending on health. The provincial estimates were then aggregated to get the national distribution of benefits from provincial government health expenditures. Implicitly, we assume subsidy per visit to vary by province, but not within it. It should also be noted that the estimates did not take into account the fees paid by users of public health facilities. Considering, however, that fees constitute a tiny fraction of provincial government health spending, that might not be a serious problem. 51. Various estimates of the benefit distribution are presented in Table 18 below. All four measures show that the distribution of the benefits from public health favors the poorest quintile. The extent to which they have benefited, however, depends on whether differences in unit subsidy are taken into account or not. The share of the first quintile is about 30 percent, as measured by the physical utilization indicator. Using measures that take into account inter-provincial differences in government health spending, the share of the first quintile falls to about 26 percent. With the latter measure, we also note that the share of the poor falls even more (from 44.4 percent to 37.9 percent), when prenatal care and birth delivery are included. One wonders about the extent to which progressivity will be further reduced when differences in unit subsidy within provinces are factored in. 52. Despite the progressivity of the distribution of provincial government health spending, it is clear that there is substantial leakage of benefits away from the poor. Over half of the benefits from the provincial health expenditures goes to the non-poor. Table 18: Distribution of public health facility visits and provincial government health spending benefits (percent) Income Curative visits Prov'1 health Curative & Prov'1 health quintiles (a) expenditures prenatal/birth expenditures (b) delivery visits First 30.2 26.4 30.5 26.6 Second 24.3 19.9 24.3 19.8 Third 20.0 22.0 20.0 22.0 Fourth 15.7 19.5 15.6 18.5 Fifth 9.8 13.2 9.6 13.1 Total 100 100 100 100 Poor 44.2 40.8 44.4 37.9 Non-poor 55.8 59.2 55.6 62.1 Total 100 100 100 100 (a) Distribution of visits to public health facilities from not feeling well. Inpatient care is multipledby 12. (b) The visits are weighed by provincial government health expenditures. Source: 1997 EDS 53. Improving health protection for the poor. The government has embarked on several major national level reforms of the health insurance system to improve its efficiency and equity. Essentially, the reforms have sought to rationalize the process and rules of the game governing the obligatory national Obras Sociales. One of the noteworthy reforms is 20 Argentina Poverty Assessment Background Paper No. 4 providing members of these Obras some choice and establish some competition among them. Furthermore, it sought to address funding for catastrophic illnesses and raising the minimum health package for members of low income national Obras Sociales. 54. The next big push in Argentina's health system reform should be to improve health care services for the poor without insurance. The public health system by providing virtually free services acts as a safety net for the poor. The problem, however, is that because of inefficiency and lack of focus it is unable to ensure that the poor get good quality services and needed medicines. This problem is indicated by the extent to which poor families are disadvantaged in terms of the length of time they have to wait to be attended, the amount of unfilled medical prescriptions, the unrealized recommended diagnostic analyses and the lower prevalence of prenatal care (Alex's paper might contain information that we could cite. If not, we could get this information from 1997 EDS). 55. The experience of Tucuman, Salta, San Juan and Rio Negro highlights the above issue. Recent study of these provinces reveals that the public health facilities are underutilized. They spend too much on salaries and too little on medicines, equipment and other materials. They not only are overstaffed; their staff mix also lacks balance (they have too many administrative and general support staff). The problem is that incentives to focus on the poor, perform efficiently and deliver good quality services are not embedded in the way public health facilities are financed. Moreover, their frontline managers do not have the necessary authority to manage effectively and efficiently. A study of a major hospital in Tucuman illustrates the above problem (Box 1). 56. To address the above issue, Rio Negro, Tucuman, Salta and San Juan decided to adopt a health sector reform strategy that would concentrate provincial government health expenditures on the uninsured poor and change the payment mechanism (See Box 2). The hope is to both improve health care protection of the poor and raise the quality of services rendered to them by creating "choice and voice" mechanisms.' 57. Below, the study explores the extent to which provinces are in a position, at least financially, to adopt the above-mentioned poverty-oriented health reform without significantly increasing public spending on health. At the core of this issue is the extent to which current provincial government spending would be sufficient to finance purchase of health insurance coverage for the uninsured, especially the poor, within current budget constraints. 58. To answer this question, we estimate the number of uninsured for each province, using projected population and health insurance coverage rates drawn from the 1997 EDS data. These rates are probably underestimates as the survey does not cover isolated rural areas. The total provincial government health expenditure is then divided by the number of uninsured and the poor population to get an estimate of the monthly health insurance premium the government could afford to buy coverage for the uninsured and the poor, if that is so desired. 1 For a description of the design of the PRL II strategy and its rationale, see the World Bank (1997), Argentina: Improving Health and Education (Salta, Tucuman and San Juan). 21 Argentina Poverty Assessment Background Paper No. 4 Box 1. The Inefficiency of Public Sector Health Services: A Case Study Using a combination of direct recording of time use, interviews and data from hospital records, an analysis of a major hospital in Tucuman shows gross inefficiency in resource use. It appears that on the whole at most 70 percent of total hospital personnel is needed to produce current outputs. Of the available staff time of outpatient facilities at least half is idle. Moreover, only half of doctors' time is devoted to patient care. Inefficiency in a public health system usually translates into poor quality of care. When resources are misused, the ability of health facilities to provide timely and credible health services suffers. Data shows that in the above hospital delays in scheduled surgical operations are quite high at about 16 percent. Three-quarters of these are avoidable and over half is due to lack of materials and timely diagnostic analyses. The scope for possible improvement in the health care of the poor is further highlighted by field visits of two private and other public hospitals. Comparative observations reveal that the private providers have: > Better (state of the art) equipment and infrastructure - compared to the dilapidated state of public hospitals > Reception and waiting rooms of the private facilities that are much more comfortable and relaxing (with TV sets), reflecting client-orientation and sensitivity to patients' dignity > Much shorter waiting time in terms of appointment and return of diagnostic findings Knowledgeable local observers in Tucuman expressed the opinion that private health care providers could produce better quality services at a cost that is probably less than that of public hospitals. . For example, birth delivery at the Hospital de Maternidad is about $400, compared to $250-300 in private hospitals. The clients of the above-mentioned private hospitals are from all economic classes. They include poor patients whose Obra Social has entered into a service contract with these providers. The point here is that for the same amount of resources being spent now by the government on public health facilities, the poor should be able to get better quality services and more timely attention, which in turn should lower patient cost of waiting and repeat visits. Source: Argentina: Improving Health Education (Tucuman), World Bank, 1997 22 Argentina Poverty Assessment Background Paper No. 4 Box 2. Reforming the Provincial Health System There are three main ideas that underpin health sector reform under the Provincial Reform Loan (PRL II) to improve poor families' access to quality health service: (a) adjustment in provincial government infrastructure and staffing for the purpose of saving and shifting resources to more productive uses in the health sector; (b) targeting of provincial government health spending on financing access to a basic health service coverage plan for those who cannot afford health insurance; and (c) gradual separation of financing of health services from their delivery in order to change the institutional incentive structure underlying the public health system, including greater transparency and accountability as well as more discretionary authority for front-line managers and health authorities. The key assumption of the reform is that aside from ensuring that health sector "public goods" are financed adequately, the fundamental role of provincial government is to ensure that all citizens have access to good quality "basic health services". Given Argentina's social health insurance system, this largely means making sure that the uninsured poor have access to those services. A corollary to this principle is that the government health subsidy should be used to pay for services actually rendered to target beneficiaries. Under the current system, government spending is used to maintain public hospitals and their staff regardless of quantity and quality of services rendered and the ability to pay of the clients being served. The consequence is that there is no incentive from the financing system to encourage provision of good quality service, to be productive in terms of the number of clients served and to improve subsidized services to the poor. The hospitals received their budget on the basis of past allocation and some allowance for inflation. In PRL II, it is envisioned that in the long-run government subsidy would eventually no longer be given to public hospitals directly and automatically. Instead, it would go to some purchasing agency or fund holder (FH), which would be responsible for negotiating with public and private service providers for delivery of a pre-specified package of health services to a pre- determined group of eligible population ( the uninsured poor) on the basis of some agreed price. This FH could be an autonomous body under the supervision of the Ministry of Health or could be an agency that would handle health service contracts for both the poor and the government workers (members of Obra Social Provincial). Under this vision, public hospitals, which would be given management autonomy, would have to earn income to pay for their staff and their operations. They can do this by charging the above-mentioned FH and other insurance agencies capitation fees, reflecting the full cost of providing good quality services to their members. The idea here is to give to FH the financial means to motivate and enable public hospitals to provide poor patients improved health service. The practical question is how to make the transition from the current to the new system of financing hospitals. Clearly, the envisioned system cannot and should not be done overnight. From the political economy standpoint, the new system could not come to fruition unless transitional arrangements are put in place that would allow hospital staff to make necessary adjustments smoothly and create incentives for stakeholders to support the implementation of the proposed reform. The purpose of PRL II is precisely to help the provincial government make the adjustments, create appropriate incentives for reform and make the transition to the new system. Argentina: Improving Health Education (Tucuman), World Bank, 1997 23 Argentina Poverty Assessment Background Paper No. 4 59. The results of the analysis are presented in Table 19. In general, it appears that all the provinces are in a financial position to adopt the above-mentioned strategy of Rio Negro, San Juan, Tucuman and Salta. Monthly provincial health spending (PHS) per poor person per month ranges from a low of 36 pesos in Corrientes to as high as 1,504 pesos in the Federal Capital. The data also show that with the exception of Corrientes (15 pesos), PHS per uninsured, including both poor and non-poor, ranges from about 18 to 147 pesos. What seems surprising is the low figure for the Province of Buenos Aires. The reason for this is that unlike other provinces municipalities in Buenos Aires are responsible for a large fraction of public health expenditures. With contributions from the municipalities, the Province of Buenos Aires could certainly afford to pay for a health insurance coverage for the uninsured. 60. The above policy is especially feasible financially, when limited to the uninsured poor. Rough estimates have shown that a good basic health care insurance can be bought for about 17 pesos per person per month. In comparison, the lowest provincial health expenditure per poor person is about 55 pesos. 61. In conclusion, it is clear from the figures presented in Table 19 that there is scope for improving access by the poor to better quality health services within the budgetary constraint. There is enough money to finance health insurance coverage for the poor or its equivalent that is of the same quality that families with prepaid care costing $17 per person per month are receiving from private providers. What is needed is a change in the way the current provincial health system works. 24 Argentina Poverty Assessment Background Paper No. 4 Table 19: Health insurance Coverage Provincial Government Health Expenditure, and the Poor Expanded Population Coverage Uninsured population Uninsured poor Uninsured non-poor Provincial PHS per PHS per Sample Size government uninsured uninsured health spending Monthly poor (PHS), Million Monthly Persons % of Persons % of poor Persons % of un- population insured Capital 1,602,320 3,036,891 52.8% 658,893 21.7% 64,845 42.9% 257,346 20.3% 1,170 147.98 1,503.59 Buenos Aires 8,259,497 13,711,746 60.2% 5,223,662 38.1% 1,157,803 61.2% 1,423,333 27.2% 1,158 18.47 83.35 Catamarca 142,118 300,636 47.3% 95,940 31.9% 23,300 51.2% 19,356 21.1% 58 50.38 207.44 Cordoba 1,694,978 2,994,848 56.6% 1,005,340 33.6% 153,148 39.8% 47,224 17.5% 233 19.31 126.78 Corrientes 622,275 883,567 70.4% 415,614 47.0% 170,029 63.7% 70,710 24.8% 74 14.84 36.27 Chaco 529,829 918,711 57.7% 445,715 48.5% 153,087 71.4% 74,380 28.2% 107 20.01 58.25 Chubut 325,027 418,579 77.7% 148,735 35.5% 56,339 73.9% 47,295 20.6% 90 50.43 133.12 Entre Rios 749,876 1,087,274 69.0% 423,734 39.0% 150,635 60.0% 117,392 25.6% 114 22.42 63.07 Formosa 223,821 469,573 47.7% 215,818 46.0% 56,144 60.4% 23,705 23.2% 62 23.94 92.03 Jujuy 386,099 574,468 67.2% 256,671 44.7% 133,262 64.5% 31,413 18.7% 62 20.13 38.77 La Pampa 159,142 291,979 54.5% 123,302 42.2% 35,567 75.9% 25,953 26.7% 60 40.55 140.58 La Rioja 143,513 260,321 55.1% 71,024 27.3% 13,322 32.8% 8,609 20.2% 48 56.32 300.26 Mendoza 991,705 1,548,729 64.0% 522,897 33.8% 170,611 63.2% 153,274 22.6% 174 27.73 84.99 Misiones 480,036 928,076 51.7% 439,840 47.4% 123,688 70.3% 73,034 28.0% 93 17.62 62.66 Neuquen 311,570 500,886 62.2% 224,769 44.9% 70,518 79.2% 60,474 29.2% 114 42.27 134.72 Rio Negro 335,766 583,079 57.6% 228,589 39.2% 53,172 72.2% 73,556 29.0% 94 34.27 147.32 Salta 624,708 1,000,983 62.4% 473,358 47.3% 200,583 64.3% 85,711 28.9% 134 23.59 55.67 San Juan 336,576 564,898 59.6% 236,383 41.8% 71,916 63.1% 53,020 27.0% 78 27.50 90.38 San Luis 208,957 338,269 61.8% 135,446 40.0% 36,905 64.9% 35,614 26.0% 46 28.30 103.87 Santa Cruz 137,347 191,292 71.8% 52,528 27.5% 10,398 49.6% 23,734 22.3% 51 80.91 408.73 Santa Fe 1,957,239 3,008,956 65.0% 897,771 29.8% 299,279 60.2% 188,114 15.3% 197 18.29 54.85 S.Estero 349,265 710,718 49.1% 298,668 42.0% 91,325 62.2% 48,578 25.5% 71 19.81 64.79 Tucuman 761,289 1,247,551 61.0% 353,675 28.3% 132,113 53.3% 72,054 15.5% 102 24.03 64.34 T.Fuego 69,539 99,862 69.6% 27,595 27.6% 4,619 74.0% 13,074 21.6% 42 126.83 757.69 National 21,402,4901 35,671,894 60.0% 13,008,6881 36.5% 3,432,606f 60.4%] 3,026,954 24.0% 4,432 28.39 107.60 Source: 1997 EDS, 1998 Statistical Yearbook and 1998 Informe Economico Regional 25 Argentina Poverty Assessment Background Paper No. 4 Education 62. Over the years, Argentina has achieved educational progress that compares favorably with other developing countries. This advance is reflected in Table 20 , which shows the difference in educational achievement of children aged 20-29 years old and their parents by income status. It is remarkable that all income groups appear to have achieved significant improvements in educational attainment and that the poor has gained as much as the average person (3.4 years). Incidentally, the biggest gainer is the middle class (4.0 years). Table 20: Years of Schooling of Children Aged 20-29 and Their Parents Per capita income quintile Structural Poverty Total 1 2 3 4 5 Total NBI No NBI Parents 8.3 7.2 6.7 7.6 9.0 11.6 8.3 5.2 8.6 Children 11.7 10.6 10.4 11.6 12.5 13.9 11.7 8.6 12.1 Difference 3.4 3.4 3.7 4.0 3.5 2.3 3.4 3.4 3.5 Source: 1997 EDS; SIEMPRO calculation 63. Table 21 also shows that on the whole there is a high enrollment rate among children in the primary education age group 6-14, corresponding to the first and second cycles of basic education (EGB1 and EGB2). Despite this achievement, however, there is still much to be done in raising educational attainment, especially among the low income groups. Among the poorest 12-14 year olds, there continues to exist a significant percentage of the population not going to school (about 14 percent). And among children aged 15-17 years old, corresponding to polimodal age group, about 31 percent did not go to school, on average. Non-enrollment rate among the poorest quintile reaches 43 percent for this age group. This premature school leaving together with repetition, has led to a low percentage of children aged 15-19 who had completed secondary education (Chart 1a). 64. This section explores the benefit incidence of public education expenditures among income groups and examines selected hypotheses regarding determinants of school participation that might be impeding further advances in educational attainment, effectiveness and equity. 26 Argentina Poverty Assessment Background Paper No. 4 Table 21: Public and Private School Enrollment By Age Group and Income Class, 1997 Part A Age Group 6-14 years (primary) 15-19 years (secondary) 20-24 years (tertiary) % enrollment % public % enrollment % public % enrollment % public rate rate rate Income Class Quintile 1 91.9 91.3 49.5 85.27 19.1 83.3 Quintile 2 94.6 87.0 53.4 86.87 20.0 87.1 Quintile 3 96.8 77.5 64.6 82.85 23.4 87.1 Quintile 4 98.0 65.6 68.6 75.28 40.0 80.7 Quintile 5 97.9 46.8 80.7 59.87 50.1 79.5 Overall 95.8 73.6 63.3 78.0 30.5 83.5 Part B Age Group 12-14 years (3er ciclo 15-17 years (polimodal) 13-17 years EGB) % enrollment % public % enrollment % public % enrollment % public rate rate rate Income Class Quintile 1 86.2 86.5 56.7 82.2 66.9 86.2 Quintile 2 89.4 86.3 61.9 87.0 70.7 86.8 Quintile 3 93.8 83.8 76.5 80.2 82.9 80.7 Quintile 4 96.1 76.0 81.6 72.2 86.4 69.8 Quintile 5 92.8 65.5 93.1 53.9 94.4 51.7 Overall 91.51 78.4 70.71 76.01 67.81 76.7 Source: 1997 ED Chart la: Education Attainment Profile, Age 15- 19,1997 e 120.0% 100.0% -a on 80.0% - + - Quintile 1 cM 60.0% - - - - Quintile 5 c 40.0% - - Mean 20.0% 0.0% 0 1 2 3 4 5 6 7 8 9 10111213 Highest Grade Attained 27 Argentina Poverty Assessment Background Paper No. 4 65. Benefit incidence of public education spending To what extent is the poor benefiting from public education spending? How much room is there for improving the incidence of education spending? These questions are discussed in the following paragraphs. 66. Methodologically, we again follow standard incidence analysis. However, instead of the usual practice of assuming that unit cost or subsidy is the same for all children within each education level, we allow for inter-provincial differences in unit cost for each level. Public education expenditure per student varies substantially among provinces. Consequently, the incidence estimates were calculated province by province before aggregating them to arrive at the national distribution of benefits. 67. Table 22 presents the results of the analysis. Not surprisingly, it shows expected patterns. With universal enrollment, the distribution of primary education benefits appears to be somewhat favorable to the poorest group, as higher income families tend to send their children to private schools. The first quintile, which accounts for 29.3 percent of primary school age children, takes 32.6 percent of primary education benefits. In contrast, the share of the richest quintile, which has a population share of 12.3 percent, accounts for only 8.0 percent. Table 22: Distribution of Public Education Expenditure Benefits by Level of Education and Income Status (Percent), 1997 Quintiles Total Primary Secondary Tertiary Bene Pop Bene Pop Bene Pop Bene Pop fits 6-24 fits 6-12 fits 13-17 fits 18-24 First 25.5 24.7 32.6 29.3 20.9 26.3 10.4 18.4 Second 23.4 22.7 25.1 23.5 24.0 24.7 12.1 20.3 Third 20.8 19.8 19.1 18.8 23.4 19.5 17.4 21.2 Fourth 18.0 18.0 15.1 16.1 19.3 16.7 26.0 21.1 Fifth 12.3 14.8 8.0 12.3 12.3 12.9 34.1 19.0 Total 100 100 100 100 100 100 100 100 Poor 40.4 38.5 47.2 42.7 37.9 42.9 17.6 30.4 Non-poor 59.6 61.5 52.8 57.3 62.1 57.1 82.4 69.6 Source: 1997 EDS 68. The distribution of public secondary education benefits in contrast appears to be pro- middle class and anti-poor. The share of the middle income (third and fourth) quintiles in secondary education benefits is greater than its population share. The favorable distribution to the middle class is clearly at the expense of the poorest income group. While the richest group breaks even, the first quintile, which has 26.3 percent of the relevant population, gets only 20.7 percent of the benefits. 69. The extent to which the distribution of secondary education benefits is biased in favor of middle income classes is probably understated. The reason for this is that private education subsidy and differences in per student public spending within each province, which have not been taken into account in the above analysis, is likely to be favorable to these families. In addition, it is possible that inclusion of rural children, which are 28 Argentina Poverty Assessment Background Paper No. 4 excluded in the 1997 EDS survey could further worsen the distribution of. secondary education benefits. 70. As in most other countries, tertiary education clearly favors the upper income groups. It is only the two richest quintiles whose share of benefits exceeds their population share. The rest of the income groups get a disproportionately lower percentage of the benefit distribution. Only 17.6 percent of the tertiary education benefits go to the non-poor, which accounts for 30.4 percent of the relevant population. The above analysis probably underestimates the degree of inequality in the distribution at this level for the same reasons as in secondary education. 71. To summarize, we conclude with two observations. First, the distribution of public education spending benefits on the whole appears more or less proportional to population (age 6-24). In reality, however, given that the estimates are probably biased against low income groups as explained above, the actual distribution is likely to favor the better off. Second, it would appear that distribution of education benefits becomes increasingly unfavorable to the poor, particularly the poorest quintile, as one moves to higher levels of education, where the cost of education becomes more burdensome and schooling participation rate drops to lower levels. 72. Differential returns to education. There are two motivating factors for examining this topic here. First, economic theory suggests that the rate of return to education is a key determinant of demand for education. In this regard, the hypothesis has often been expressed that early school leaving by the poor is due to low rates of return to further education. Second, analysis of differential education rates of return between poor and well off children could improve our understanding of income inequality and help us make a more realistic assessment of the extent to which the poor are truly benefiting from public education spending relative to the well off. 73. The rate of returns to education, it could be argued, is greater the more well off the parents of the children are for several reasons. One is that well off children might have better public schools than the poor. Another reason is that the ability of poor children to profit from schooling could have been reduced by underdevelopment and lack of stimulation of their mental faculty due to past malnutrition and other disadvantages. More generally, there are probably many significant complementarities between the productivity of child's schooling and parental socioeconomic status. For example, more educated parents are able to help children learn more in school and guide them more intelligently in their choice of courses and jobs. Besides, in a economy with relatively inflexible formal labor markets where non-price rationing of protected high wage jobs could be a common practice, high socioeconomic families are probably able to exert more influence in getting their children jobs with better pay or training opportunities. 74. In looking at the rate of income growth from education, it would be useful to also explore the effect of the learning achievement of children in school. The few empirical studies on the relationship between individual earnings and learning achievement support the 29 Argentina Poverty Assessment Background Paper No. 4 hypothesis that individuals with high learning achievement score tends to have higher education rate of return. 75. In this study, it is not possible to control for individual learning achievement in assessing the effect of schooling on income growth. The required data are not available. What is available is the average provincial learning achievement test scores for math and language. These test scores differ substantially among provinces, which probably reflect persistent differences in the learning environment of children across provinces. In this regard, it is not unreasonable to assume that the provincial test score measure (taken as the simple average between math and language) correlates well with a host of interacting school and peer factors that are usually associated with high learning achievement, such as high academic expectations, better teachers and school governance, and availability of school materials and textbooks. In a sense, differences in this measure could be seen as indicating variations in the quality of provincial education systems and its student population. 76. The above hypotheses were tested by estimating earnings functions in which the rate of increase in income with respect to years of child education depends on their parents' schooling and the provincial learning achievement test score. These functions were estimated on a sample of unmarried children 20-29 years old from the 1997 EDS. We chose parent's education (measured as years of schooling of the head of the household) to characterize the children's socioeconomic status. 77. Several specifications were tried. But the basic specification is one that relates the natural logarithm of hourly wage rate to child's age, gender, and years of schooling, which is interacted with his parents' education and the average test score of the province in which the child is residing. 78. A few of these regression equations are presented in Table 23. The regression results indicate that parents education tends to raise the rate of income gain from a year of child education. The results, including those not presented here, show that the interaction term between provincial test score (SCORE) and parents' education (SF) is statistically significant. 30 Argentina Poverty Assessment Background Paper No. 4 Table 23: The Average Rate of Income Gain from Education and Learning Environment: Unmarried Children Ag e 20-29 1 2 3 Dependent: Ln WAGE Independent variables: Intercept -0.7457 -0.9764 -0.9417 (-3.10) (-8.14) (-7.86) Child's age 0.0431 0.0483 0.4892 (4.54) (10.11) (10.26) Child's gender (male=l) 0.0436 0.0469 .0398 (0.86) (1.78) (1.52) Child education completed (S) 0.0531 -0.0776 -0.0819 (7.58) (-7.10) (-7.49) S*SCORE 0.0022 0.0021 (13.20) (12.50) S*SF (SF= education of household head) 0.0135 0.0010 (2.44) (4.75) Adjusted R-square 0.119 0.166 0.172 Sample size 797 3032 3032 Ordinary Least Squares I I Data base: 1997 EDS. Figures in parenthesis are t-values of the regression coefficients. Table 24: Rate of Income Gain (IG) from Education at Low and High Levels of SF and SCORE Income growth (%) Elasticity of IG wrt National Lowest SF SCORE SCORE = 61 SCORE = 47 High SF = 12 5.82 2.88 0.06 3.42 Low SF =3 4.92 1.98 Notes: sample mean of SF is 3.7 and 6.5 for poor and non-poor children, respectively. The average rate of income growth (IG) from a year of child education for the age group 20-29 can be derived from equation 3 in Table 23 as: IG = -0.0819 + 0.0021* SCORE + 0.0010*SF 79. At the sample mean value of SCORE and SF, which is about 61 percent and 6 family members respectively, the IG is about 5.21 percent. It is remarkable that the elasticity of the rate of income gain from education is quite high at 3.42 (Table 24); in contrast, that of SF is quite low. This suggests the possibility that the negative effect of poor socioeconomic background of children could be overcome by improving the quality of the education system. 80. The above equation implies that at some low level of SCORE and SF the rate of income gain from a year of education could become negative. Although counter-intuitive, it actually makes sense. An income loss could result, when children are not learning in 31 Argentina Poverty Assessment Background Paper No. 4 school. As is well recognized in labor economics, the time spent by children in school could mean loss of work experience and, hence, reduced wage, particularly during the first few years of his entrance into the work force. If the loss from experience and earnings foregone while the child is in school is not compensated sufficiently by increased productivity over time because of lack of learning, the rate of returns to education can indeed become negative. In Argentina, it appears that that no province has an average test score that is so low that a year of schooling would mean an income loss, even among those whose parents have very low education. 81. The effects of education on income gain, which appear to be small, need to be carefully interpreted. They should not be strictly interpreted as an education rate of return. The sample is limited to children aged 20-29 and, hence, the estimated wage effects might not fully reflect yet the impact of education on income. In addition, because the sample consists of unmarried children, the results may not be generalizable to other groups. 82. To get a better sense of the rate of return to education, we estimate an earnings function for the age group 20-65. The results, presented in Table 25, reveal an average Mincerian rate of return (RR) to an additional year of education of about 7.17 percent. For a province with a low SCORE of 47, RR falls to about 5 percent. This implies that at about the mean value of SCORE, the elasticity of RR with respect to SCORE is about 1.36. It rises to 1.68 as the SCORE reaches the lowest observed value of 47, implying that RR becomes more responsive in provinces with very poor learning environment. Incidentally, Table 25 also shows the importance of the health effect on wage. A chronic illness reduces wage rate by 7 percent. Finally, gender appears to be a significant factor as well. Male workers earn close to 9 percent more than females. 32 Argentina Poverty Assessment Background Paper No. 4 Table 25: Wage Equation for Workers Age 20-65: Education Rate of Return Dependent: hourly wage rate Coefficient t-value Intercept -0.1582 -5.47 Age 0.0152 27.56 Gender (male=1) 0.0881 7.36 Years education completed (S) -0.0259 -4.83 SCORE*S (interaction) 0.0016 18.92 Health (ILLNESS) -0.0728 -5.74 Adjusted R-square 0.184 Sample size 17,649 Assumptions about SCORE Rate of return (RR) 7.17% 61 (mean) 4.93 47 (lowest) RR elasticity wrt SCORE 1.36 From 61 to 67 1.68 From lowest to mean score Source: 1997 EDS 83. Correlates of early school leaving. We now come to the question: why do some children dropout of school prematurely? Why is there a huge gap in the educational attainment of poor and well off children? The economic view is that educational attainment is primarily determined by expectations regarding its rate of return, household capability to finance schooling, and ease of access to schools. 84. It is not possible in this study to quantify the extent to which education demand changes in response to differences in income gains from schooling. It is, however, possible to increase our understanding of why children of the poor leave school early by examining various factors that are arguably related to education rate of return. 85. One such factor that naturally comes to mind is the provincial learning environment. This variable, as measured by SCORE, has just been shown to have a strong positive correlation with income growth. A positive correlation between SCORE and enrollment decision should, therefore, be expected . On the other hand, a high provincial SCORE might reflect a more highly selective educational system that discourages average and disadvantaged students from enrolling in post-basic education. In principle, therefore, it is possible to have a positive or negative correlation, depending on the relative strengths of these two opposing forces associated with SCORE. A negative effect would mean that the education system is generating high student performance, at least partly, by being highly selective of student intake. 33 Argentina Poverty Assessment Background Paper No. 4 86. This process usually goes against low income children. As often observed in educational systems where good schools is highly limited and admissions to them are rationed through price and non-price mechanisms (such as examination and other student performance measures), well off parents have the resources and incentives to prepare their children for examinations and use their influence to get their children admitted to and retained in those schools. Hence, the discouragement effect of a highly selective process would probably tend to be lower among well off children. To test this hypothesis, we specify a school enrollment regression equation with an interaction term between SCORE and the income of the head of the household. The expectation is that this interaction term is positive. 87. In testing these hypotheses, the following factors were explored and controlled for at the same time: (a) physical distance to public secondary education (DISTSE); (b) being able to enroll in a pubic school that the child wanted (SAV); (c) secondary education provincial spending per student (PEXS); (d) the gender of the child (SEXO = male) and of the household head (SEXHHO = male); (e) family size (FAMSIZE); and (f) child health (ILLNESS), as measured by number of chronic disabilities and diseases. A logit model was used to estimate the probability of school enrollment among age groups 15-19 and 15-24. 88. The results, which are presented in Table 26, are consistent with expectations. First, there is a positive interaction between SCORE and the income of the head of the family. Second, evaluated at mean income, the effect of SCORE appears negative. These findings suggest that on average high performing provincial education systems are associated with social selectivity that is sufficiently strong to discourage some poor children from pursuing further study. 34 Argentina Poverty Assessment Background Paper No. 4 Table 26: L git model of enrollment Dependent: logit of enrollment probability No Chil Wage Effect of Expected Child Wage Age 15-19 Age 15-24 A e 15-19 Independent variables Coefficient Coefficient Coefficient Coefficient Intercept 11.4009* 6.7571* 12.6599* 14.1103* Age (EDAD) -0.7908* -0.5385* -0.8528* -09309* Age of household head (EDAD-F) 0.0206* 0.0229* 0.0166* 0.0216* Child's education (S) 0.3494* 0.3153* 0.2849* 0.2847* Education of household head (SF) 0.0566* 0.0616* 0.0497* 0.0542* Education of spouse (SM) 0.0545 0.0369* 0.0428* 0.0598* Per capita income of household head (PY-F) 0.0009* 0.0005* 0.0341 -0.00027 Pub sec educ spending per student (PEXS) -0.0032 -0.1691** 0.0727 Consumer Durable Assets (ASC) 0.1779* Child's gender (SEXO) -0.1556* -0.0534* -0.1687* -0.0658 Gender of household head (SEXOHH) -0.3619* -0.0382* -0.4828* -0.2775** Admission to school of one's choice (SAV) 0.3499* 0.1670* 0.3959* 0.3892* Child's health (ILLNESS) 0..2952* 0.1950* 0.4203* 0.5521* Distance to secondary pub school (DISTSE) -.4158* -0.2933* -.03736* -0.4400* Family size (FAMSIZE) -0.1028* -0.0824* -0.0832* -0.0740* Provincial average test score (SCORE) -0.0239* -0.0152* -0.0444* -0.0511* Single-headed family (NOSPOUSE) -0.0184 0.2419** 0.0517 0.1166 Interaction: SCORE*YFF (head's income) 7.362E-6** 8.154E-*6 0.000808 Child predicted wage (WGEHAT) 0.6060* 1.2291* -2 LOG L 6227 10727 4727 6352 Sample size 6725 11329 6725 6852 * and ** mean significant at 1.0 percent and 5.0 percent level. Source: 1997 EDS 89. Third, distance to public secondary school remains a significant cost and disutility factor that is discouraging some children from schooling participation. In this regard, we note that children in the first quintile live further than those in the fifth quintile. Forty-one percent of children in the poorest group live at least a kilometer away from the nearest public secondary school, while the corresponding figure for the richest quintile is only 35 percent. Fourth, children with more chronic illnesses tend to enroll in school. Although counter-intuitive, this result points to the fact that on average the opportunity cost of going to school is lower for children with chronic illness and disabilities. Analysis of the EDS data reveals a negative correlation between ILLNESS, their chances of working in wage employment and their predicted wage rate. Fifth, holding per capita income constant, family size has a strong negative impact on demand for education. Since the poor has a larger family size than the well off, this clearly is a significant factor that reduces educational capital in poor children. Family size among children in the first quintile is about seven compared to only four among the richest group. 90. Wage expectations, opportunity cost and school enrollment. It is conventional wisdom to argue that a principal reason why poor children leave school early is the high opportunity cost of going to school. In this section, we explore this view to complement the preceding discussion. 35 Argentina Poverty Assessment Background Paper No. 4 91. To this effect, a wage equation was calculated from a population of 15-24 year olds with wage earnings, following Heckman's two stage methodology. In the first stage, a probit model was estimated, relating the person's background characteristics to the probability that he is going to work in a wage earning activity. In the second stage, the wage equation was then estimated by OLS, using the inverse of Mill's ratio from the probit model as an independent variable to correct for selection bias due to unobserved heterogeneity. The wage equation corrected for selection bias is presented below in Table 27. Table 27: Hourly Wage Rate Equation Corrected for Selectivity: Age 15-24 Ln Wage Coefficient t-value Intercept -2.6982 -2.53 Age (EDAD) 0.0871 2.59 Age of household head (EDAD-F) -00016 -0.97 Child's education (S) 0.0421 1.378 Education of household head (SF) -0.0008 -0.11 Education of spouse (SM) -0.0077 -1.48 Income of household head (YYF) 0.0564 3.39 Consumer durable assets (ASC) 0.0372 5.59 Child's gender (SEXO) -0.0342 -0.36 Gender of household head (SEXOHH) -0,0192 -0.50 Child's health (ILLNESS) -.1305 2.46 Family size (FAMSIZE) -0.0271 -3.94 Provincial average test score (SCORE) 0.0108 1.91 Provincial average prime age wage rate (WGEP) 0.0884 7.40 Interaction: S*SCORE -0.0001 -0.25 LAMDA (inverse of Mill's ratio) 0.1179 0.41 Adjusted R-square 0.227 Sample size 2673 Source: 1997 EDS 92. The results indicate that sample selection bias is not significant. As expected, provincial test score, household wealth and family head's income are significant determinants of child's wage. Most interestingly, the negative effect of child's chronic illness on his wage rate is considerable. 93. The above equation was used to get predicted wage rates (WAGEHAT) for all children 15-19. This variable was then introduced into the enrollment logit model as an independent variable, as a measure of opportunity cost of school participation. Contrary to expectation, WAGEHAT turned out to have a significant positive effect. 94. An possible explanation is that children with high WAGEHAT might have higher expectations regarding returns to their education. This is plausible since much of the wage variations is due to factors that arguably could enhance returns to education. This hypothesis needs further exploration due to its implications. Perhaps, the problem with early schooling leaving among the poor is not due to high opportunity cost, but rather to 36 Argentina Poverty Assessment Background Paper No. 4 low returns to their time and financial investment. As observed above, wealthier children tend to get higher predicted wage rates, yet more of them go to higher levels of education. 95. Incidentally, the fact that WAGEHAT is positively correlated with schooling should not be taken to mean that it would be ineffective to provide cash grants to poor families (conditional on children going to school) to compensate them for short-term foregone income from child work. Rather, what the data seem to be saying is that the negative effect of high opportunity cost could be overcome by giving children and their families higher expected returns to schooling. This can be done not only by reducing the cost of education but also by increasing income gain from schooling, such as improving the quality of the learning environment. 96. Education and social assistance policy. The study on the whole points to the following conclusions and implications: 97. First, while the overall distribution of public education spending benefits appears to be proportional to the share of the population by income class, in reality it is probably biased against the poorest children. There are several reasons for this. One reason is that the incidence analysis is biased towards greater equality because of the exclusion of the rural population from the data base and the assumption of uniform unit subsidy across income groups within each province. 98. Second, reform of the provincial education system that improves learning standards can lead to substantial increases in the rate of return to education. The provinces, however, need to pay care attention to how this is being achieved. If improving standards is done through greater student selectivity, it would discourage school participation, particularly among the lower income groups. 99. Third, social selectivity can be attenuated by a well targeted compensatory program to increase school participation and learning achievement. The national government through its Plan Social Educativo (PSE) are currently assisting schools in disadvantaged areas through financing of school infrastructure improvements and provision of teaching/learning supplies and materials. The adequacy and effectiveness of PSE, however, have not yet been rigorously examined, although feed backs received have been positive. Indeed, it appears that PSE is strongly pro-poor. The 1997 EDS reveals that 60 percent of the PSE beneficiaries come from the first quintile and 40 percent from NBI families (Table 28). Moreover, the two lowest income quintiles account for over 80 percent of the beneficiaries. 100. Still, it can be seen that about half of the children in the poorest quintile have not received assistance from PSE. Whether such a figure is high or low depends on what it means to children who have not received PSE assistance. This issue should be included in a rigorous and comprehensive evaluation of PSE. 37 Argentina Poverty Assessment Background Paper No. 4 Table 28: Coverage Rate and Distribution of Beneficiaries by Income Class Total Per capita income quintile Poverty 1 2 3 4 5 NBI No NBI Coverage rate (a) 42.0 51.8 36.6 33.9 23.8 25.3 55.7 36.2 Distribution of beneficiaries 100 59.9 21.3 11.3 5.5 2.0 39.5 60.5 (a) percent of children enrolled in public school. Source: 1997 EDS. SIEMPRO Tabulation 101. Fourth, limited household resources, cost/disutility associated with distance to school and needs of a large family impact negatively on demand for education, while labor market incentives for poor children appears weak for many of them to persevere and overcome educational costs and difficulties. An integrated well targeted compensatory program is needed to deal not only with the supply side aspects of the problem such as what PSE has done, but also address the incentives or demand-side issues. In this regard, the proposal of the Secretaria de Desarrollo Social (SDS) deserves support. 102. SDS is proposing to try out a program2 that would provide cash grants to families with eligible children, provided they enroll and keep them in school (as part of the government's Programa de Garantia de Renta Minima). This proposed experiment, which draws from the experience of Brasilia's Bolsa Escola and Mexico's PROGRESA, is worth pursuing. It is a promising way to systemically strengthen incentives for the poor to send and keep their children in secondary school longer, while at the same time providing social welfare assistance to poor families. This would have, of course, to be accompanied with the establishment of SISFAM, an system for continuous identification and monitoring of eligible households. 103. How effective and practical the program will be depends critically on the size of the cash to be given, which in turn will determine long term feasibility. One of the objectives of the experiment is precisely to determine the size of the grant that is effective and at the same time affordable. SDS proposes to try out grants initially amounting to about $80 -$100 per child for families. This amount does not appear unreasonable to begin with. We estimate roughly that a child age 15-19 who is not currently enrolled in school would forego about $80 per month, if he had to go to school (Table 29). 2 "Programa bolsa familiar para educacion y su posible implementacion en Argentina: poblacion objetivo, focalizacion y presupuesto", SIEMPRO, Subsecretaria de Politicas, Secretaria de Desarrollo Social, version preliminar, Octubre de 1998. 38 Argentina Poverty Assessment Background Paper No. 4 Table 29: Estimates of Foregone Earnings if Un-enrolled Child Goes to School Age group 15-19 Student Not student Working Not working Working Not working Predicted hourly wage rate (a) 1.31 1.19 1.15 1.09 Predicted hourly wage rate 1.20 1.11 Hours work per week 2.70 17.22 Hours per month (a) 74 Foregone income if child goes to 82 school ($) (a) calculated by using 4.3 weeks in a month (b) predicted wage derived from equation in Table 27. 104. In thinking about implementation of this program, the issue of how it should be financed is critical, given the budget constraint. Rough estimates by SDS indicate annual cost ranging from half a billion to over 2 billion pesos, depending on the schedule of cash grants to be followed. An apparently sensible financing approach is the consolidation and conversion of existing programs (as appropriate) into cash grants that can then be used to finance the Renta Minima Program. To advance this idea, an independent commission of eminent persons should be appointed to review and evaluate existing social programs and subsidies with the aim of identifying which of them can be folded together into the Renta Minima Program and of determining the amount of money that can be made available to it from this source. 39 Argentina Poverty Assessment Background Paper No. 4 Annex 1 Inventory of Social Programs for the Poor and Vulnerable, 1998 Thematic Area Obj$etives Services Provided Food and Nutrition * Community Social Policies * Improve the quality of life of the NBI population * (i) Provision of free meals; and (ii) Other social (Prov. Gov./POSOCO) services * Infant Nutrition and Feeding * (i) Improve living conditions and access to Program sufficient food for nutritionally vulnerable * (i) Daily dietary rations in infant feeding centers, (Prov. Gov./PRANI) children from 2 to 14 years old who live in infant care centers and school cafeterias; (ii) unfavorable socioeconomic conditions, through Strengthening PRANI (monthly box containing 9 free meals and support for basic education; (ii) meals - 1200 calories/day); (iii) Technical Evaluate and restructure the system for infant Assistance and training for Provincial, Municipal feeding centers and school cafeterias; and (iii) and NGO officials and technicians; and (iv) Gradually transform infant feeding centers into Equipment and infrastructure support for feeding * Community Farm Program infant care centers centers/cafeterias (Prov. Gov./PROHUERTA) * Improve nutrition of the NBI population with insufficient nutrition * (i) Community, family and school-attended farms * Social Nutrition Program supported by key investments, technical assistance (Prov. Gov./PROSONU) and training; and (ii) Training sessions * Improve the nutritional status of the target population * Free meals Training and Strengthening * National Center for * (i) Articulate community actions in a framework * (i) Coordination, training and various services for Community Organizations complimentary to public and private fields; community organizations; (ii) Distance training; (iii) (CENOC) (ii) Facilitate development of community Workshops for the National Volunteer Program; (iv) organizations through human resource training, Training for Indigenous People (Program for Training formation of networks and alliances at the Indigenous Peoples) regional, local and national levels; and (iii) Argentina Poverty Assessment Background Paper No. 4 Thematic Area a bjectives SerSices Provided Recognition and diffusion of the sector's activities through the community organizations considered as active players in the development Information, Monitoring and of social policies * (i) Training ;(ii) Technical Assistance to social Evaluation System for Social programs; (iii) Tables, CD and documents with Programs (SIEMPRO) * Providing Government organizations with the information on the social needs of the population; (iv) necessary resources to operate a social Social survey; (v) Studies on qualitative aspects of the information system, including monitoring, poor social sectors; (vi) Evaluations of social evaluation and training in social administration programs; (vii) Information on the provision of social programs (services on demand); (viii) Reports on the status and analysis of the social situation; (ix) Maps or documents with indicators; (x) Technical plans for dissemination and equipment; (xi) Publications/bulletin; and (xii) Integrated monitoring system. Community Development and Social Integration * National Institute for * (i) Promote integrated development of * (i) Financial assistance; (ii) Technical assistance; Indigenous Issues (INAl) indigenous communities, with a focus on and (iii) Training preserving and valuing their cultural heritage; (ii) Plan, design and supervise the designated resources for technical assistance and financing for the communities and public and private organizations in relation to their area of competence; and (iii) Organize a national registry of indigenous communities (RENACI), * Attention to Priority Groups in accordance with Decree 155/89 * (i) Technical assistance and financing for institutions specializing in care for the handicapped; (ii) Training * (i) Improve living conditions of the populations for providers of home care; and (iii) "Tourist days" with social risks (handicapped, adolescents, (semi-complimentary and complimentary) children, adults over 60, the chronically and Argentina Poverty Assessment Background Paper No. 4 Thematic Area Objective Services Provided terminally ill, etc.), by facilitating the satisfaction of basic needs; and (ii) Promote Attention to Vulnerable strategies of social support, favoring integration * (i) Financing of projects with participatory planning Groups (PAGV) and participatory solidarity processes (PPP); (ii) Financing projects for direct assistance; (iv) Financing projects for institutional * (i) Contribute to decreasing exposure to social strengthening risks and improving the quality of life for the most vulnerable groups in urban centers and indigenous populations through implementation of a model of participatory administration at the local level and articulation with community initiatives based on the strengthening of local public organizations and civil society; (ii) Implement a method of administration which supports the reform process and social administration at the local level, promoting coordination of community initiatives to decrease Participatory Fund for Social risks; (iii) Support the development and integration Investment (FOPAR) of vulnerable groups through actions which * (i) Financial and organizational assistance for facilitate their access to social services; and (iv) community projects; (ii) 4-hour training modules for Maximize the use of public resources at the local project design; and (iii) Training for institutional level strengthening * (i) Financing projects for social investment in locations and municipalities with extreme poverty which respond to the most pertinent needs of the * Plan for Strengthening Civil community; and Society (ii) Develop local administrative capacity through participatory experiences which are concrete in * (i) Financial assistance for projects supporting terms of structure, management and execution of community initiatives, and those focusing on parents; projects designed to improve the socioeconomic (ii) Financial assistance for projects for training conditions of groups and communities in poverty (equipment for NGOs, initial products, training, situations preparation of training material); (iii) Technical Argentina Poverty Assessment Background Paper No. 4 ThematicAra Objectives Services Provided assistance for projects; (iv) Training for social (i) Strengthening the operational and administration in municipalities; (v) Training administrative capacity of the community community leaders (initial, specific and more organizations through the training and technical advanced); .(vi) Basic and complimentary training;; support for the initiatives they develop; and (ii) (vii) Training for promoters; (viii) Training * Plan for Strengthening Guide the links between the State and Civil Society internships; (ix) Financing projects for social Youth Development (PFDJ) for the development of social programs administration in municipalities; (x) Administration of projects (workshops); and (xi) Financing projects * (i) Technical assistance, training and economic support for the design and execution of solidarity projects; and (ii) Technical assistance, training and economic support for the design and execution of innovative projects * (i) Strengthen the capacity of youths in the design and execution of projects through the guidance of solidarity groups of youths which work for * Action Program for Integral improvements in the quality of life for the poor, or Support for the Socialization socially vulnerable; (ii) Support for governmental of Marginalized Minors in the and non-governmental organizations which provide GBA initiatives and innovations linked to youth issues; (PRO,AMBA) (iii) Strengthen the autonomy and consolidation of * (i) Financial support for the design and execution of youth groups through agreements on the small construction works and teaching equipment; and * Program for Attention to construction of a just and integrated society with (ii) Technical assistance and training for NGO Minors in Especially Difficult solidarity; and (iv) Strengthen the democratic forms personnel Circumstances (PROAME) of participation, favoring the model of decentralization from the Federal Government * Improve the social conditions and amplify the * Financing of projects for shelters, meal rations, possibilities for social and labor integration of medical attention, production initiatives, promotion of children and adolescents at risk through technical rights, school support, recreation, and labor market support and financing for projects in execution, or training Argentina Poverty Assessment Background Paper No. 4 Tbcmtic rea bjecivesServices Providtd to be executed, by NGOs Advisory Programs for * (i) Contribute to improving the living conditions Youths and their Families of children and adolescents in especially difficult circumstances, providing opportunities for them to realize their physical, psychological and social potential, oriented toward the exercising of their rights through projects led by NGOs and through * (i) Legal assistance to minors; (ii) integral attention strengthening provincial governments and/or to mothers in situations of risk or abandonment; (iii) municipalities with obligations to youth-related temporary housing for children and the handicapped at issues; and (ii) Promote adequate methodologies for risk or abandoned; (iv) equipment, repair and * Integrated Project for intervention which promote agreement on, and expansion of institutes; (v) training of specialized Aboriginal Communities in the development of, community capacity and initiatives personnel; (vi) treatment and assistance for Ram6n Lista Dept. (Formosa) handicapped children and elderly people; (vii) (DIRLI) * (i) Protect youths and families at great social risk research; and (viii) re-establishing families for as consequence of abandonment, abuse, exploited street children mistreatment, addictions, law-breaking, etc.; and (ii) Coordinate the participation of institutions, * Institutional Subsidies organizations and/or publicity entities for the programming, execution and diffusion of local and * (i) Expansion of schools; (ii) Grants for secondary regional activities to orient and promote family students; (iii) Teaching materials in Wichi; (iv) Works integration, using technical assessment, specific for water provision; (v) Provision of ambulances; (vi) material, studies, investigations, congresses and Provision of medicine and milk; (vii) Improved and/or training new wells; and (viii) Living quarters, possible with latrines and baths * Improve the living conditions of the Wichi population, using their cultural principles as a basis, * Subsidies to institutions which present proposals for valuing, and spreading these principles. The interventions in the areas of extreme need and/or Project would promote education, nutrition, health, poverty capacity for self-administration, production and organization. Argentina Poverty Assessment Background Paper No. 4 Tbi6utfit Areaitetie: Ser~vices Poie Improve the quality of the NBI population at social risk, through the participation of governmental and non-governmental organizations, generating a process of self-sustaining development Education * Social Education Plan - * (i) Improve the quality of education through * (i) Specific equipment for special education schools; Program 1: Better Education for pedagogical support to poor schools; (ii) Promote (ii) Equip schools (laboratory, sciences, etc.); (iii) All (MCYE/PSE-PI) participation of the education community; and (iii) Cabinet for information in the school; (iv) Textbooks Focus available resources toward the poorest and manuals (primary and secondary); (v) Institutional schools of the country projects; and (vi) Services * Social Education Plan - Program 2: Improvements in * Make possible a Federal Law for Education, * (i) Administrative areas; (ii) Teaching areas; (iii) School Infrastructure improving and extending the physical infrastructure Construction of service areas; (iv) Classroom (MCYE/PSE-P2) of poor schools in the whole country construction; (v) Constructions of polymodal rooms; (vi) School establishments; and (vii) School * National Program of Grants Reconstruction for School Retention * (i) Increase years of school attended by poor (MC YE/Becas) youth; (ii) Strengthen the school institutions to * Grants improve quality and equity of education; and (iii) Contribute to improved employability of poor youth Emergencies and Others * Actions from the Office of * Plan, coordinate and execute actions for the * (i) Food, coats, mattresses, etc.; and Social Emergencies (DIES) prediction, prevention and resolution of states of (ii) Training municipal and/or provincial agents in social need, caused by emergency situations which dealing with emergencies and catastrophes compromise the survival of communities Employment and Productive Development * Labor Training for Sectors * Increase the competence and productivity of the * (i) Grants for living expenses and maintenance for or Types of Activity (MTSS) workforce in a specific sector or type of activity at unemployed people who receive no unemployment the regional, provincial or local level insurance or other social security services; (ii) Free labor training; and (iii) Civil responsibility insurance Argentina Poverty Assessment Background Paper No. 4 Thematic Area Objectives Services Provided Training for Employee * Re-establish the workforce in sectors and/or * (i) Free training; (ii) Civil responsibility insurance; Support regions in crisis, or in the process of restructuring, and (iii) Grants for living expenses and maintenance improving the employability of workers for unemployed people who receive no unemployment insurance or other social security services * (i) Establish a program of transitory employment * Non-remunerative economic aid, civil responsibility * Local Employee to carry out economic and social infrastructure insurance and health coverage Development (II) projects, or to provide services which contribute to community development and improvement in the employability of workers; and (ii) Present the legal framework with which executing organizations can provide non-remunerative economic aid * (i) Credits for investment and operation; and (ii) * Facilitate the productive capacity of human and training for producers * Rural Development for natural resources of small producers and indigenous Northeast Argentina peoples from NEA through sustainable increase in (SAPYA/PRODERNEA) investment and training in self-administration * Civil risk insurance ($5/month) and variable grants . Increase employment and decrease training costs * EMPRENDER of firms (MTSS/EMPRENDER) * Monthly non-remunerative economic aid ($200), civil responsibility insurance and medical assistance * FORESTAR * Private employee promotion to stimulate coverage (MTSS/FORESTAR) transitory occupation for unemployed workers in rural areas to carry out work linked to forest activities * (i) Subsidies to private employees with base salaries * PROEMPLEO not more than $1000 (people between 38 and 44: (MTSS/PROEMPLEO) * Generation of private employment for $100/18 months; persons 45 and over: $150/18 undetermined periods of time, targeting the months); (ii) Training to support the employees, and unemployed with difficulties integrating in the grants for living expenses and maintenance; and (iii) labor market Coverage of moving expenses Argentina Poverty Assessment Background Paper No. 4 Theiatic Area Objectives Services Provided Program for Support of * (i) Courses on the search for employment (5/6 days) Productive Reconversion ("Proyecto Imagen"); (ii) Training for developing (MTSS/PARP) * (i) Increase the probability of maintaining or occupational skills (including living expenses, accident finding productive employment; (ii) Improve and medical insurance, childcare, etc.) ("Proyecto conditions of the beneficiaries through an Joven"); and (iii) Courses on business administration orientation on how to look for employment and technical assistance ("Proyecto Microempresas") ("Proyecto Imagen"); (iii) Facilitate insertion in the labor market, through specific training ("Proyecto Joven"); and (iv) Collaborate in the development of * Program for Support for micro-enterprises and other forms of self- * (i) Subsidies for protected production workshops; Protected Production employment ("Proyecto Microempresas") and (ii) Subsidies for acquisition of goods for training Workshops (MTSS) and other expenses (i) Promote labor market adaptation for people with handicaps (without discrimination based on type and grade) of working age, for integration into the protected and non-protected labor market; and (ii) Promote new workshops and support existing ones, in order to improve the achievement of the * Program to Assist economic and social goals in order to generate in * Monthly subsidy ($100) for converted workers Reconversion of Shearers each area a plan for business development (MTSS) * Promote labor contracting in the private sector, as well as the reconversion of the workforce in firms * (i) Productive projects with technical and financial * Local Development which contract shearers assistance; (ii) Training; (iii) Monitoring and Promotion Program evaluation; (iv) Institutional development; and (v) * Strengthen the institutional capacity of the Strategic planning municipalities to induce and manage economically sustainable development, oriented to social investments through the creation and * Agricultural Social Program consolidation of productive entities which * (i) Strengthening self-consumption projects; (ii) (MEYOSP/PS) generate employment Technicians and promoters trained; (iii) Innovative productive activities; (iv) Traditional productive Argentina Poverty Assessment Background Paper No. 4 Thematic Area Objectiv Services Provided * Improve the productive capacity of small rural activities; (v) Courses for specific demand; (vi) producers by increasing regular income Training of rural agents; (vii) Financial assistance to productive projects; (viii) Technical assistance for small producers; and (ix) Training for producers Special Employment Programs * Monthly aid (between $100 and $200) and civil (MTSS) responsibility insurance * Implement, through agreements, actions to promote private employment, responding to regional demands or productive sectors in emergency or crisis which contribute to a Special Training Programs improving the situation of the unemployed and * (i) Grants from $80 - $160/month and civil (MTSS) most vulnerable responsibility insurance; and (ii) Payment for training * (i) Labor training to increase the employability of workers linked to certain sectors or branches of activity; and (ii) Stimulate the association among * Non-remunerative economic aid ($160/month), civil * Community Services actors who need to develop training responsibility insurance and medical assistance coverage (i) Establish transitory employment through social service projects for the community, oriented to improving the quality of life of the * Integrated System of population; and (ii) Reduce the impact of the * Unemployment insurance ($150-$300/month), Services for the Unemployed loss of earnings due to loss of employment medical services and regular corresponding payments (SEGURO DES) *Establish income and medical coverage among * Occupational Workshops those who lost their jobs and are newly * (i) Subsidies (up to $300,000) for training (MTSS) employed workshops for civil society; and (ii) Technical assistance and training for workers in micro-, small and medium- enterprises * Develop quality labor training through the creation or adaptation of institutions designed to Argentina Poverty Assessment Background Paper No. 4 Thematic Area Obetives Services Provided * TRABAJAR II (TRABAJAR meet the needs of occupational training for 2) human resources directly related to the demands * Monthly non-remunerative economic aid (up to of the local productive markets, and those in the $200), accident insurance and health coverage zone of influence Develop transitory employment for unemployed workers in conditions of poverty or social vulnerability, reducing the impact of the loss of earnings in poor areas due to loss of employment, and improving the employability of these workers through works considered socially relevant Health * Immunization Program * Prevention of immuno-preventable diseases * Doses distributed (AS; BCG; D.AD; DPT; SABIN; (MS YA S/PA l) TT) * Maternal and Child Health * (i) Reduce infant mortality from mother-child Program (MSYAS/PMI) malnutrition; (ii) Health promotion for children and * (i) Technical assistance for provincial maternal and adolescents in the entire country within the child programs; (ii) Elaboration of norms for framework of health policies already established; prevention, health protections and attention to mothers, and (iii) Promote and develop the strategy for children and adolescents; (iii) Powdered milk; (iv) primary care for maternal and child health Medicine; (v) Training workshops for health teams * Maternal and Child Health and Nutrition Program * (i) Reduce infant morbidity/mortality, improving * (i) Extension, revision and equipping hospitals, (MSYAS/PROMIN) the focus, design, extension and coordination of the health centers, infant feeding centers, nurseries and services and programs for health, nutrition, free daycare centers; (ii) Training; (iii) Common and feeding and infant development; (ii) Promote the modified milk; (iv) Provision of modified milk and psycho-social development of children from 2 to 5 boxes of free meals years of age; and (iii) Reduce the prevalence of * National Program for maternal-child malnutrition Chagas Control (MSYAS/CHA GAS) * In five years, eliminate the occurrence of new * (i) Insecticides; (ii) Training workshops; (iii) Argentina Poverty Assessment Background Paper No. 4 Themitic Area Objectihes Services Pro%ided cases of Chagas, de%eloping actions for control of Control ol blood sampling: (i Studies of those the parasite in the entire country infected; (v) household surveillance; (vi) Serology in those from 0 to 14 years of age * National Program for the Fight Against Human * (i) Prevent sexual, pre-natal, blood, hemophiliac * (i) Reagents for studies on the viral "load" of Retroviruses (AIDS and STDs) and transplant, and drug use-related transmission; HIV/AIDS patients; (ii) Anti-retroviral medications (MSYAS) (ii) Reduce the individual, family, and socio- specifically designed for HIV/AIDS and other economic impacts; and (iii) stimulate the opportunistic infections; (iii) Reagents for blood banks production of information which allows evaluation to test for HIV and STDs; and (iv) Medication for of the tendencies and projections of the epidemic STDs * Project for the Fight Against AIDS and Sexually * (i) Reduce incidence of infection of HIV and Transmitted Diseases STDs; (ii) Inform the population and stimulate the * (i) Financing for projects for prevention, selected (MSYAS/LUSIDA) adoption of behavior to prevent HIV and STDs; and from those presented by OCSs; (ii) Training of health (iii) Improve the administrative capacity of the care personnel; (iii) Prevention campaigns for programs for the national and provincial programs HIV/AIDS, and the production of informational against AIDS and STDs, as well as giving integral materials; (iv) Formation of a network of education attention to the population living with HIV/AIDS disseminators; (v) Teacher training for AIDS and/or STDs at the hospital and ambulatory care prevention in schools; (vi) Institutional strengthening levels for hospitals and health centers Income Subsidies * Attention to Non- * Administration of the system for non- * (i) Others; (ii) Pensions for assistance to invalids, Contributing Pensioners contributing pensions, and medical coverage with soldiers in the Malvinas War, families of the the limits determined by Decrees 292/95 and Disappeared, mothers of more than 7 children, and the 492/95 aged population. * Social Aid Attention to emergency situations for the * (i) Housing or food provided by treatment centers in extremely poor, without medical coverage, and at Buenos Aires; and (ii) Subsidies for health care high risk Care for the AFed . Solidarity Support for * (i) Improve the living conditions of seniors (60+ * (i) Free medication; (ii) Food ration; and (iii) Seniors years of age) in situations of social risk, facilitating Tours/visits Argentina Poverty Assessment Background Paper No. 4 ThemOtic, Are# becie Semices Provided (ASOMA) the satisfaction of basic needs; and (ii) Promote the integration of seniors with other age groups to promote solidarity and share experiences and values * PAMI-Geriatrics (PAMI) * Provide boarding in Geriatric Institutes for the * Boarding rooms in Geriatric Institutes elderly poor * PAMI- Economic Subsidies * (i) Rent assistance; (ii) Complimentary clothing; and (PAMI) * Improve living conditions of retirees and (iii) Medication pensioners in states of emergency * PROBIENESTAR * Food packages (PROBIENESTAR) * Improve conditions of the poor, and reduce social isolation of retirees and pensioners in PAMI, and in situations of social risk Housing and/or Social Infrastructure * (i) Legalize the properties of land settlements; * (i) Construction and improvements in housing; and * ARRAIGO and (ii) Construction and/or reconstruction of (ii) Legalize properties of land settlements (CTFN/ARRAIGO) habitats * (i) Credit support to productive sectors; (ii) Credits * (i) Support to communities and municipalities for equipment; (iii) Credits for institutional * Financing to Municipalities with actions for improving the quality of life of the development; and (iv) Credits for works (PRODISM) population; (ii) Improve financial administration at the provincial and municipal levels; and (iii) Reinforce the institutional capacity of local governments * (i) Housing solutions; and (ii) Housing constructed (constructed through contracted firms, intermediate * National Fund for Housing * Reduce the housing deficit and improve housing entities, municipalities or communities) (Prov. Gov./FONA VI) and sanitary conditions, as well as the quality of life of the target population * (i) Provision of basic equipment for communities; (ii) Infrastructure works (rooms for multiple uses, * Improvements in Housing * Contribute to the development and improvement rooms for first-aid, etc.); (iii) Regularization and urban and Basic Social Infrastructure of conditions of rooms, housing, basic planning in irregular settings and poverty-stricken Argentina Poverty Assessment Background Paper No. 4 Thematic Area Objectives Services Provided (PROGl 7 SSV) infrastructure and access to land in NBI homes and areas;(iv) Housing solutions (improvement, for vulnerable groups in situations of risk, complimentary housing, credit funds); and (v) Houses emergency and marginalization, as well as those affected by environmental emergencies * Construction and/or expansion and improvement of small works for drinking water and basic health * Program of Social Assistance for Provision of * Improve the sanitary conditions of small Healthy Drinking Water, and localities Good Health * (i) Financing of project for improved drinking water; (MEYOSP/PASPAYS) and (ii) Financing for project for basic health * Program for Social Development in Border Areas * Contribute to improving the quality of life of the * (i) Public infrastructure with inter-sectoral of NOA and NEA with NBI NBI population in border areas in the La Plata connections, included the health unit; (ii) social populations (PROSOFA) Valley support, technical assistance and training; (iii) guarantees of land entitlements to neighborhood * Neighborhood Improvement populations; and (iv) technical assistance and training Program * (i) Improve the living conditions of the NBI for institutional strengthening urban population, situated in neighborhoods with infrastructure needs, enviromnental problems and/or legalization of land settlements through the execution of a group of works and services, as well as the strengthening of community organization; and (ii) Facilitate access to basic services, with a social focus to guarantee the sustainability and replicability of the program Poor People in a Rich Country A Poverty Report for Argentina Background Paper No. 5 Poverty and Health Alexandre V Abrantes, Human Development Country Sector Leader for Argentina Chile and Uruguay and Dariush Akhavan, Consultant April 16 1999 Argentina Poverty Assessment Background Paper No. 5 Demography of Poverty The poorest households in Argentina tend to be younger, larger and have more children. Reproductive health services need remain a priority in a poverty alleviation strategy. The average age of the lower income quintile household is just 25 years, while the mean age of the upper income quintile households is 41 years. The average size of the lower income quintile household is 5.1 people and only four percent are single person households. Sixty percent of these households have children aged 0-19 years and only 1 percent of these households have no children. By contrast, while the average size of upper income quintile households is just 3.1 persons and about 29 percent are single person households. Also, only 36 percent have children aged 0-19 years and 10 percent have no children at all. Table 1: Average family age, size and composition by household income quintile Per capita IHousehold income uintiles 1 2 3 4 5 Average age 25 31 35 37 41 Averagefamily size 5.1 4.2 3.8 3.5 3.1 Percent of households with no children 1 1 3 3 10 Percent of households w/children aged 0-19 66 59 53 48 36 EDS. Social Development Household Survey Total fertility rate was 2.7 per 1,000 women aged 15-45 in 1995, which is about the average for the Region, but are high when compared with other middle income countries such as Chile and Uruguay and higher than one would expect from the country economic and educational situation. This average parameter hides significant variation in the country and a negative association between birth rates and income level. Crude birth rates in the 14 percent poorest jurisdictions is relatively high, at 21.1 per thousand, contrasting with only 15.2 per thousand in the 14 percent upper income jurisdictions. Health Status and Poverty. Are the poor at a disadvantage in terms of death, disease and disability? As elsewhere, the poor in Argentina carry a disproportionate part of the total burden of disease and are disproportionately affected by a number diseases. Diseases which affect disproportionately the poor are mostly avoidable communicable disease, maternal, perinatal and nutritional conditions, which could be controlled with universal access to a comprehensive package of quality public health and personal health services costing no more than US$30 per capita, about eight percent of 1997 public health expenditure per capita. The overall health indicators of Argentina are good when compared with those of other countries in the Region, and have improved markedly over the past ten years. In 1 Argentina Poverty Assessment Background Paper No. 5 1995, life expectancy stood at 76 years, 12 more than in 1960, the crude death rate at 8 per 1,000 population, infant mortality at 22 per 1000 births in 1995, down 16 percent from 1990, and maternal mortality ratio at 14 per 10,000 live births in 1995, down by 14 percent from 1990.1 However, these rates are worse than those of other middle income countries, and lower than one would expect for a from the economic and education indicators. For example neighboring Chile, with lower income than Argentina, had in 1995 an infant mortality rate of 12 per 1,000 live births and maternal mortality ratio of 6.5 per 10,000 live births. ILLUSTRATE with health gap. Table 2: Association between Average Per Capita Household Income and Years of Life Lost per 1,000 Population Di%ea%e whi.h affeci L inur Disease% which afiect Linear dispriportionatel, the poor Correlation disproportionatel the rich Currelatient Tuberculo;is -(1.361 Colon and rectum cancer 1.1.347 Syphilis -0.415 Pancreas cancer 0.326 Diarrhea -0.455 Breast cancer 0.387 Tetanus -0.455 Ovary cancer 0.361 Bacterial meningitis -0.334 Endocrine disorders 0.347 Chagas' Disease -0.383 Rheumatic heart disease 0.316 Hypertensive disorder of -0.370 Inflamatory heart disease 0.472 pregnancy Abortion -0.435 Protein-energy malnutrition -0.450 Congenital heart diseases -0.319 Cervix Uteri Cancer -0.318 Infant mortality ratio -0.62 Neonatal mortality rate -0.56 Maternal mortality rate -0.48 Proportion of low birth weight -0.58 babies (<2,500 g) The relatively good national health indicators hide significant variations between different income groups, the poor having much worse health status than the rich and having a different pattern of death, disease and disability. Subjectively, about 23 percent of the poor and rich report having been ill in the past 30 days; and about 16 percent of those from the lowest household income quintile report some form of chronic ailment, while as many as 23.5 of those in the upper quintile report chronic suffering. Objectively, when one looks at the burden of disease in terms of years of life lost 2due to premature death, evidence shows a quite different picture: the poor die earlier 'Bos, E, Hon, V., Maeda, A. et al : Health Nutrition and Population Indicators, a statistical handbook. (Washington, DC: The World Bank, 1998) 2 The calculation of Years of Life Lost in this study are very approximate. There are no corrections for underreporting, incorrect coding and other factors. Since the study is comparative between provinces, the 2 Argentina Poverty Assessment Background Paper No. 5 than they should, and are disproportionately stricken by largely preventable Group A diseases, i.e. communicable, maternal, perinatal and nutritional conditions. In fact there is a strong negative association between per capita household income and the years of life lost per 100,000 population due to syphilis, diarrhea, tetanus, abortion, protein-energy malnutrition, and with infant, neonatal and maternal mortality. Other negative associations exist, albeit weaker, between poverty and tuberculosis, bacterial meningitis, Chagas' disease, hypertensive disorder of pregnancy, and congenital heart diseases, strong associations between average per capita income (Table 2.). Group A diseases account for 19 percent of total years of life lost in lower income and for only for 5 percent in higher income provinces On the other hand, the rich seem more vulnerable Group B diseases, i.e. non- communicable diseases. In general per capita household income is positively associated with non communicable diseases and the association is strong for several cancers (i.e. colon and rectum, pancreas, breast, and ovary), endocrine disorders, and rheumatic and inflamatory heart disease. Rich and poor seem to be equally affected by injuries. Fig 1. Linear Correlation of Burden of Disease due to Premature Mortality (YLL's) With Income Per Capita 0.60 I -- - - -- ' Group I Diseases t Group II Diseases p. 0.40 . 0.20 -0.20 JjIS I -0.40 -0.60 Differences in health status are consistent with differences observed in the major determinants of health: environmental conditions (i.e. access to adequate water and sanitation), health related behaviors (i.e. diet, exercise, smoking habits, alcohol and other substance abuse, and sexual behavior), and health care utilization. lack of correction for the mentioned factors is less important because there are no reasons to believe that their would be a bias by province. 3 Argentina Poverty Assessment Background Paper No. 5 The relatively high incidence of communicable diseases and of infant mortality among the poor is likely to be determined by the direct environment conditions in which the poor live, i.e. sanitary conditions are likely to be associated with the high incidence of diarrhea among the poor, while housing conditions, crowding and malnutrition are likely to be associated with Chagas' disease, tuberculosis, leprosy, meningitis. A high proportion of the poorest live in dwellings of very low quality and with inadequate sanitation: 10 percent of those in the lowest per capita family income group quintile live in "precarious" dwellings (i.e. casilla or rancho), 42 percent have inadequate water supply and 47 percent have inadequate sewerage systems, two or three times higher the national average. In addition, for poorest family income quintile, the average number of people per dwelling is 4.8 living, almost twice as many as for the upper income families. Figure 2: Infant Mortality and Per Capita Family Income ( 401 40 1 30 S 10- .E 0 0 100 200 300 400 500 600 700 Per Capita Family Income The relatively high incidence of abortions, hypertensive disorder of pregnancy, syphilis and cervix uteri cancer, low birth weight babies, perinatal and maternal mortality are likely to be associated with inadequate coverage of poor women with quality reproductive health services, which we will describe later under the relevant section. The health of indigenous of people is a concern, given that they are usually among the poorest and because in most countries they tend to discriminated against by mainstream social services. In Argentina, health statistics are not broken down by ethnic group and therefore the necessary data are not available to accurately diagnose the health situation of the indigenous population. Experience of local health programs, points to generally poor sanitary and health conditions, with high incidence of cholera, Chaga's disease, and poor maternal and child health. 3 Pan American Heatlh Organization: Health Condition of the Americas (Washington, DC: PAHO, Edition, Sci Pub No,, 1998) 4 Argentina Poverty Assessment Background Paper No. 5 Health Care Coverage and Utilization. Are the poor appropriately covered by essential public health services and do they have access to adequate personal health care? Available health sector resources are more than adequate for providing a comprehensive package of public health and personal health services for the entire population. The fact that avoidable, Group A disease incidence and mortality are relatively high, in spite of plentiful health sector resources suggest that there may be problems with the existing organization and quality of health care. Personal health care in Argentina is delivered by both the private and the public sector. The private sector responds mostly to the needs of those with mandatory or private health insurance and the public sector is meant to cater for the needs of the poor and the uninsured. The system has been heavily biased towards specialized and hospital care, but at present there is a move towards strengthening general and ambulatory health care network to improve the efficiency and quality of the system. About 54 percent of hospital beds are in the public sector, but the proportion can be as high as 70 percent in provinces where the population is sparse. Most doctors work part time in both the private and public sectors. As one would expect there are many overlaps: insured patients who seek service in the public sector, mostly for complex or expensive interventions for which the insurance co-payment would be very high, and uninsured patients who choose to pay out of pocket for convenience, to avoid waiting lists or to get what they may perceive as more personal care. The delivery of public health services is largely the responsibility of provincial public health services, with support from some Federal programs. Health Care Supply Health care resources in Argentina are generally plentiful and relatively accessible in every part of the country. Although the poorest in Argentina have more difficulties in accessing health facilities such as hospitals and pharmacies, more than 70 percent live within walking distance from such facilities, and even the poorest provinces the available number of doctors and hospital beds should be enough to provide a comprehensive package of public health and personal health services to the entire population. The health system in Argentina is well provided and distributed: about the 70 to 77 percent of the population lives within 10 blocks from a health center or a hospital, respectively, and 87 percent of the population does not have to walk more than 15 minutes (1 km) to the closest pharmacy. However there is a very strong negative association between per capita household income and access to public hospitals (R= - 0.99) or pharmacies (R= -1.00). Although access of the poorest to hospitals and pharmacies is more difficult, more than 70 percent of the poorest income group quintile have easy access to either a hospital or a pharmacy. 5 Argentina Poverty Assessment Background Paper No. 5 Figure 3: Household Income and Access to Public Hospitals S 3-- - - --. 880 15 0OA 0 0 200 400 600 800 1000 1200 1400 Per Capita Household Income With an average of 367 inhabitants per doctor and 209 inhabitants per hospital bed, Argentina has enough, if not too many, doctors and hospital beds. The average number of doctors per 1000 population hides significant variation by province, from a low of 1 per 1000 population in poor Formosa to almost 9 per 1000 population in the wealthier City of Buenos Aires, as well as a strong association between the number of doctors available and the average per capita family income (R= 0.50). This situation is not as favorable for the number of specialized nurses and nursing auxiliaries: there are over 1800 inhabitants per specialized nurse and a ratio of 5 doctors per nurse. Even if we add the number of nurse auxiliaries to the number of nurses, supply is still very low, with 473 inhabitants per nurse or nursing auxiliary and more doctors than nurses and auxiliaries together, suggesting that many doctors must be carrying out functions which would normally be performed by nurses or auxiliaries. As for the number of hospital beds, the range of variation is much smaller, from 3.3 beds per 1000 population in San Juan and 7.8 in the City of Buenos Aires, but there is no consistent association between the availability of public hospital beds per capita family income. About 54 percent of hospital beds are in the public sector, but the proportion can be as high as 70 percent. As public hospitals are supposed to meet the needs of the uninsured poor, one would expect that in poorer provinces there would be relatively more public hospital beds, but such association does not exist (R= 0.06). Coverage with essential public health programs Overall coverage of the population with essential public health programs is high in Argentina. If we take prenatal care as an example, only about three percent of women do not benefit from some form of prenatal care. However, the poor are less likely to 6 Argentina Poverty Assessment Background Paper No. 5 benefit from such essential public health programs, and when they do the quality of service is lower. As in other countries, the "inverse care law"4 applies -- those who need the most get the least. If we use prenatal care as a tracer for the quality essential public health programs, there is an almost perfect negative association between average per capita income and the probability of having had no prenatal exam (R= -0.92), or having had less than five prenatal exams during pregnancy, which would be the relevant international quality standard (R= -0.99). The same applies for immunization of children, for whom there is a strong association between average per capita income and the probability of having been correctly immunized (0.39<R<0.46). The problem is serious, given that the relative health risks of poor women and children are higher and that for them the potential benefits of prenatal and child care are much higher than those accrued by those coming from upper income groups. Figure 4: Per Capita Household Income and Quality of Prenatal Care 30 V .r 25'---- c -O 15 20 Xju a. I. 0 5 - - ..- 0 200 400 600 800 1000 1200 1400 Per Capita Household Income Low coverage of the poor with essential public health services such as maternal and child health may be in part associated with the relatively low supply of nurses and nursing auxiliaries who typically provide the bulk of such services, i.e. health education, counseling, vaccination, etc. In short supply, these professionals are likely to be absorbed by the private hospital sector, leaving the public ambulatory sector unattended. As for the indigenous people, the Government has set up a health program for Indigenous Peoples in the wake of a cholera epidemic which hit them particularly hard. The program was launched in the Northern Provinces in 1994, and has thus far trained and outfitted over 250 community health workers who provide basic primary health care in their communities, covering about 40,000 people. Gradually these agents are being incorporated in local health teams, but international experience is not very encouraging in 4 Hart, JT: Inverse Care law, Lancet, xxx 7 Argentina Poverty Assessment Background Paper No. 5 relation to the prospects of mainstream health services catering for the special needs of indigenous people. Those who are presently poorly covered by essential public health services and indigenous populations are particularly difficult to reach by traditional health services which assume women will take the initiative to seek preventive health interventions, such as prenatal care or immunizations. That is not the case in most countries and cultures, where the lower social class or income group is relatively alienated from general societal values and social services, live day by day and have little propensity to invest in future gains, such as those which could be accrued from preventive care. For these groups health services have to be proactive and seek the women and children in the household, at school or at work, and cannot wait that they walk in through the health center door. For some years, the Argentina Maternal and Child Health and Nutrition Project, has been trying to change the way these services are delivered, introducing a new proactive and community-based intervention health care model.5 Utilization of personal health services In Argentina the population consults the doctor on average 2.2 times a year, a figure which is what is to be expected in a middle income country. Utilization of ambulatory care varies widely by province, from a low 1.1 consultation per person per year in poor Jujuy, to over 4 consultations in relatively rich La Pampa and Tierra del Fuego, and the variation is strongly correlated with per capita household income (R= 0.49) This association is likely to be determined by the number of doctors available in rich provinces, by the relatively low out of pocket cost of medical consultations, and by the fact that consultation are often initiated by the patients themselves. The utilization of non doctors, i.e. pharmacists, nurses or non professionals is almost non existent for both poor and rich. 5 Maternal and Child Health and Nutrition Project -- PROMIN. Ln. 4164 AR 8 Argentina Poverty Assessment Background Paper No. 5 Figure 5: Per capita Household Income and Consultations per person, per year .5 0 4 3 (40 0 10 20 30 40 50 60 70 Per capita Household Income Although the public system is meant to cover the uninsured poor free of charge, 30 percent of its consultations are provided to patients who are covered by mandatory or voluntary insurance, from which until recently no costs are recovered. The fact that insurers frequently do not pay public health services for services provided to their beneficiaries constitutes a significant cross-subsidy to the middle- and upper-income groups. If such subsidy were eliminated, there would be significantly more resources to improve coverage and quality of service for the poor and uninsured population, without increasing the fiscal burden. The average hospital discharge rate is 8.2 per 1000 population, with relatively little variation by province, and shows no association with family income per capita. This is consistent with what would be expected from a middle income country. The lack of association between hospital utilization and family per capita income is likely to be explained by the fact that hospital beds are better distributed than the doctors, that the decision to be admitted to a hospital is less likely to be initiated by the patient himself, and to the lack of elasticity to price, as public hospitals cover all catastrophic health care expenditures. Quality of care There seems to be no discrimination against the poor in terms of admitting them to a hospital when they need. However, once in the hospital the poor do not get the same quality of care as those in the upper income groups: a) over 25 percent of women coming from the lowest household income quintile are delivered by non doctors, while virtually no woman from the upper quintile has such experience; and b) 32 percent of women coming from the lowest household income quintile have no post delivery follow up, while only two percent of women from the upper income group miss such follow up. In ambulatory care the same pattern appear to be true: 74 percent of patients coming from the lowest household income quintile had to wait more than 24 hours for a consultation, while only 45 percent of those in the upper quintile had such delay. 9 Argentina Poverty Assessment Background Paper No. 5 Medical practice in the public system differs significantly from that of the private system, in spite of the fact that the same doctors work in both systems. The rate of cesarean sections is just one example of different protocols and practices: while 20 percent of public hospital deliveries are cesarean sections, the proportion is 40 or 44 percent in Obra Social or private hospitals, respectively. The different payment system, i.e. historical budget allocation for public hospitals and fee for service in the private sector, is likely to explain most of the observed difference. Table 3: Quality of maternal care by household income per capita. ______________________________ 'Per capha Hlousehold Income indles, 1 2 3 4 5 Prenatal care delivered by non doctor 8.3 3.3 2.1 0.4 0.1 Delivery by non doctor 28.5 17.9 10.2 8.2 0.1 Cesarean section 21.8 33.7 37.7 41.5 42.1 No Post partumfolow up 31.6 17.2 6.5 2.5 2.3 Social Development Household Survey Other health related social programs In addition to providing for a network of public health services, the GoA finances a series of health related social assistance programs such as milk and food distribution programs, to support the poor pregnant women and children. These programs are targeted to the poorest strata, but there are significant leaks to the middle classes, which can amount to 10 to 30 percent: only about 70 percent of social assistance programs for poor mothers and children (milk + maternal and child care) go the poorest per capita family income group quintile respectively, while about 8 percent of mother and 9 percent of children from the middle class mothers and children (i.e. the three upper household income quintiles) receive the benefit. Social protection programs run by the mandatory social insurance schemes, i.e. Obras Sociales, are particularly poorly targeted: 62 percent the milk distribution program benefits go to the 40 percent higher per capita income families. Better targeting would allow improving the relatively low coverage of the lower income quintile Health insurance coverage. 10 Argentina Poverty Assessment Background Paper No. 5 Are the poor adequately protected against health expenditures which could throw them into the vicious circle of poverty and disease. The entire population is covered against catastrophic financial loss caused by disease, either by mandatory or voluntary private insurance, or by public health services. The issue is one of equity, as those with mandatory or voluntary insurance benefit from choice of provider, and better quality of service. Recent hospital autonomy reforms, if not accompanied by some form of health insurance for the poor and uninsured could put the poor at a disadvantage even in public health services. There may be a problem with health insurance for those who become unemployed or are in transition between two different jobs. Thirty seven percent of the general population and 67 percent of the population in the lowest income group quintile is not covered by health insurance, their health care needs being met by a network of public hospitals, health centers and health posts, where they have very variable quality of care. The other 63 percent of the population is covered by mandatory (Obras Sociales) or voluntary private insurance (Prepagas or Mutuales), their health care needs being mostly met by the private sector. Health insurance covers more of those in the higher income groups, but also covers a significant proportion of lower income people: 28 percent of those in the lower income quintile and 33 percent of those with unmet basic needs are covered by an Obra Social. Those without health insurance use mostly the public health network, which is generally well provided and distributed, but which typically offers few amenities, little choice of provider, and has significant waiting times for consultations and tests, important waiting lists for non-urgent surgical interventions, and where the quality of service is very variable, from excellent to less than adequate, as we described before. Traditionally, public hospitals have had no reason to segment or discriminate their clientele, because no costs were recovered from patients or their insurers, when there was one. Recent hospital self management reforms may put the poor at a disadvantage. Because autonomous hospitals are now allowed to collect fees from insured patients, can retain the respective income, can distribute part of the additional revenue to staff as performance bonuses, and have kept their regular historically-set budgets, they have an obvious incentive to increase their share of paying patients by offering them preferred services, i.e. better scheduling, less waiting times, jumping over waiting lists, and better amenities. Sixteen percent of those without health insurance also use the private sector, mostly for ambulatory care, paying out-of-pocket to avoid waiting lists and for convenience. There are no data regarding out of pocket expenditures by income group. Coverage against high pharmaceutical expenditures is a problem for poor families, as drugs tend to be expensive in relation to their income and the public system offers very inadequate coverage: 43 percent of the poor in the lowest family income group quintile had to pay pharmaceuticals out-of-pocket and eight percent could not afford the prescriptions written by their doctors. When looking at the population aged 15 or more, who are not covered by any health insurance plan, we confirm what one would expect: the majority of the unemployed, domestic workers, and informal sector workers are not covered. But it is 11 Argentina Poverty Assessment Background Paper No. 5 surprising to read that 62 percent of the unemployed who were formerly employed, also lack any health insurance coverage, meaning that when a worker becomes unemployed, he or she must loose coverage immediately, or after a very short period of time. It is also interesting to note that 10 percent of those in the highest income quintile choose not to insure against health care expenditures, bearing their relatively low health risk and assuming that they will use the public hospital in case they incur in catastrophic health care expenditures. Figure 6: Household Income Group and Coverage with Health Insurance 100%~ - 80% 60% a Without coverage S 40% With coverage 20% 0% Quintile 1 Quintile 2 Quintile 3 Quintile 4 Quintile 5 The 51 percent of the population which is covered by Obras Sociales' health insurance is entitled to a standard benefits package regulated by the Government, and other benefits which vary by insurance carrier. Beneficiaries can choose from a preferred provider network. The 8 percent covered by Prepagas or Mutuales are typically entitled to a more generous benefits package, better choice of providers, at a premium. Four percent of the population have double coverage. When the health insurance beneficiaries need a particular intervention which is not included in the respective benefits package, or when they do not want to pay the respective insurance co-payment, they seek care at the public hospital where they could until recently get free health care: i.e. 17 percent and 14 percent of the beneficiaries of Obras Sociales and Prepagas, respectively seek, care in the public system, and use about 30 percent of all public health service consultations. This has constituted a significant cross-subsidy from the public to the private sectors, from the poor to the higher income strata.. Starting last year, public hospitals begun identifying health insurance beneficiaries and charging the respective health insurance carriers, which in time should put an end to the cross-subsidy. As of now, invoicing and collection are still very incomplete and inefficient and revenues from cost-recovery typically range from 3 to 10 percent of total hospital budget, while the proportion of insured patients is typically around 20-30 percent, and their cost per case is disproportionately higher because insured patients tend to use the public hospital for complex or expensive interventions only. 12 Argentina Poverty Assessment Background Paper No. 5 The Poor and Health Care Financing6 Total public health expenditures are sufficient to finance a very generous package of public health and personal health benefits for the entire population. Public funds available to finance personal health care for the uninsured population are about one half of those available to finance care for the insured patients. The share of the public health expenditures which is managed at the Federal level maybe too low for promoting national health policies in the Provinces and municipalities. In 1997, Argentina invested about US$12,900 million public monies in the health sector, which correspond to about, 15 percent of total public expenditures and 22.3 percent of total social public expenditures. Although health expenditures have increased by more than 50 percent from 1990 through 1997, they decreased as a proportion of the GDP (from 4.58 percent in 1995 to 3.99 in 1997) and as a proportion of total social expenditures (from 23.58 percent in 1995 to 22.82 percent in 1997). Table 4: Source of health sector finance. 1997. Federl Province Mnc Work-related contribWtns Percent 8% 27% 7% 58% GoA: National Directorate for Social Expenditure Programming Over the nineties, the relative responsibility of different government levels for public health sector finance has changed: the Federal government has been declined, that of the Provinces has remained constant and that of municipalities has grown in importance. Over the same period and in the area of work related contributions, the relative weight of the national union-based health insurance, the Obras Sociales, has declined, while expenditures related with the insurance for the elderly and pensioners (PAMI) and those related to provincial health insurance schemes have increased their share of the total public health expenditures. In 1996, Provinces invested on average about 11 percent of their total public expenditures in the health sector. However the relative weight of the health sector in the total public expenditures varies widely, ranging from less than 5 percent total public expenditures in Formosa, to a high of 27 percent in the city of Buenos Aires. The investment each province makes in its public sector, both in pesos per capita and as a proportion of their geographical product is highly correlated with the respective income per capita (R= 0.7 and R= 0.6, respectively). Thus public health services behave as luxury goods: the richer the Province, the more it puts into its public health system. Public resources available to care for the poor are about half what one would have in an equitable distribution of public health monies. Care for the poor is financed from provincial budgets which constitute about twenty seven percent of total public health 6 Government of Argentina: Caracterizacion y evolucion del gasto publico social en el periodo 1980 - 1997. (Buenos Aires, AR: Secretaria de Programacion Economica y Regional, mimeography, October 1998) 13 Argentina Poverty Assessment Background Paper No. 5 sector financing. Provincial budgets have to cover: a) most public health services for the entire population, b) subsidize the care provided to the insured population at public hospitals, given that such care goes mostly unpaid by the respective insurance carriers, and c) personal health services for 37 percent of the uninsured population. Data suggests that about 8 to 10 percent of provincial health budgets goes to finance general public health services, and that 30 percent of public health services are provided free of charge to insured patients, which means that only 60 percent of provincial public health expenditures, or 18 percent of total public health expenditures, are available to pay for personal health care for the 37 percent of the population who is uninsured, about on half of what one would expect in a equitable situation. Figure 6. Provincial geographical product and health expenditures per capita, Arg. Pesos 1997 700 S 600 S500----- - Q 400 4--- - - - - - 3 200 g200- - - -i 0 0 100 200 300 400 500 600 700 Hith exp per capita The federal share of health sector resources may be insufficient to promote a national health policy. Only eight percent of total public health expenditures are managed federally and 73 percent of the already small federal health budget are directly administrated by the Ministry of Health. The remaining 27 percent correspond to special health programs in the Ministries of Defense and Interior (health services for the armed forces and the police), Education (medical schools), Office of the Presidency (Drug control program), and Legislature (health services for the Congress staff). Of the budget managed by the MoH, the largest expenditures go for the maternal and child health program, to subsidy the decentralized Garrahan hospital, for prevention and control of health risks and diseases, AIDS and sexually transmitted diseases control, to support higher education. Some of these functions, such as financing the Garrahan hospital and higher education, are a drain on resources and not close to what one could foresee as the core functions of a small and strong Ministry of Health, i.e. regulation of the insurance and health care provider market, sanitary and health and disease surveillance, monitoring, control and evaluation, health promotion, pubic information, etc. What is left to use as matching grants to promote national health policy may be too little to have leverage over provinces or municipalities. 14 Argentina Poverty Assessment Background Paper No. 5 Conclusion The poorest households in Argentina tend to be younger, larger and have more children. Reproductive health services need remain a priority in a poverty alleviation strategy. As elsewhere, the poor in Argentina carry a disproportionate part of the total burden of disease and are disproportionately affected by a number diseases. Diseases which affect disproportionately the poor are mostly avoidable communicable disease, maternal, perinatal and nutritional conditions, which could be controlled with universal access to a comprehensive package of quality public health and personal health services costing no more than US$30 per capita, about eight percent of 1997 public health expenditure per capita. Available health sector resources are more than adequate for providing a comprehensive package of public health and personal health services for the entire population. The fact that avoidable, Group A disease incidence and mortality are relatively high, in spite of plentiful health sector resources suggest that there may be problems with the existing organization and quality of health care. The entire population is covered against catastrophic financial loss caused by disease, either by mandatory or voluntary private insurance, or by public health services. The issue is one of equity, as those with mandatory or voluntary insurance benefit from choice of provider, and better quality of service. Recent hospital autonomy reforms, if not accompanied by some form of health insurance for the poor and uninsured could put the poor at further disadvantage, even in public health services. There may be a problem with health insurance for those who become unemployed or are in transition between two different jobs. Total public health expenditures are sufficient to finance a very generous package of public health and personal health benefits for the entire population. Public funds available to finance personal health care for the uninsured population are about one half of those available to finance care for the insured patients. The share of the public health expenditures which is managed at the Federal level maybe too low for promoting national health policies in the Provinces and municipalities. 15 計 的 g 目 p 勿 ，〕 蠶i 辟旦 州 U》．鬥。 賺〉O -.0 ！、！鄴 i一朧 鬥 崗口 目口 莽暑 邑 口‘ 州 弓 需 之 O 以 騙 Argentina Poverty Assessment Background Paper No. 6 INDEX OF TABLE AND FIGURES TABLE 1. N UMBEROFM INIFUNDIOS, 1988................................................................................................ 7 TABLE 2. LAND TENANCY, ARGENTINA, 1988........................................................................................... TABLE 3. LAND TENURE AND POVERTY STATUS, 1988............................................................................... 9 TABLE 4. CONSUMPTION AND NBI-BASED POVERTY, MISIONES AND SALTA.............................................. TABLE 5. OCCUPATIONS OF NON-FARM FAMILIES, MISIONES AND SALTA...................................................16 TABLE 6. INCOMES OF NON-FARM RURAL FAMILIES, MISIONES AND SATA...............................................16 TABLE 7. OCCUPATIONS OF HEADS OF FARM HOUSEHOLDS, MISIONES AND SALTA.....................................17 TABLE 8. FARM HOUSEHOLD INCOME, MISIONES AND SALTA....................................................................18 TABLE 9. GENDER, INCOME AND POVERTY...............................................................................................17 TABLE 10. EDUCATIONAL ACHIEVEMENT BY GENERATION, MISIONES AND SALTA..................................... TABLE 11. INCOME BY QUINTILE AND FARM SIZE, MISIONES AND SALTA...................................................37 TABLE 12. FARM AREA BY TENURE STATUS AND QUINTILE.......................................................................38 TABLE 13. RESERVATION WAGE, MISIONES AND SALTA ...........................................................................31 TABLE 14. NON-FARM SELF-EMPLOYMENT..............................................................................................33 TABLE 15. SOCIAL INFRASTRUCTURE: MISIONES AND SALTA....................................................................35 TABLE 16. CROP AREA BY CROP TYPE AND QUINTILE...............................................................................40 TABLE 17. CROP AREA BY CROP TYPE AND PRODUCTIVITY RANGE ...........................................................40 TABLE 18. MANAGEMENT CAPABILITY AND TA RECEIPT..........................................................................44 1. FIGURE 1. POVERTY HEADCOUNT INDEX (NBI) BY ADMINISTRATIVE UNIT, 1991 (INDEC)........................ 4 FIGURE 2. PER CAPITA HOUSEHOLD EXPENDITURES: MISIONES AND SALTA...............................................14 FIGURE 3. PER CAPITA HOUSEHOLD INCOME: MISIONES AND SALTA..........................................................15 FIGURE 4. SCHOOL ATTENDANCE BY AGE COHORT, MISIONES AND SALTA................................................22 FIGURE 5. ACTIVITIES OF YOUTH BY AGE.................................................................................................29 FIGURE 6. INDIVIDUAL ACTIVITIES BY AGE ..............................................................................................29 Box 1. ANALYSIS OF THE RESERVATION W AGE ........................................................................................28 1 Argentina Poverty Assessment Background Paper No. 6 2. RURAL POVERTY IN ARGENTINA 1. Because 87 percent of the population is urban overall and urban poverty problems are so visible, poverty in Argentina has mainly been viewed as an urban phenomenon. For a country at its level of development and sophistication, surprisingly little is known about the extent of rural poverty in Argentina, as poverty assessments have not covered the rural population, or give little attention to the rural poor in their analysis. However, some im- portant facts are known: * rural poverty is largely a regional phenomenon. While poverty afflicts only around 15 percent of the population, 25-35 percent of the population in the Northeast (NE) and Northwest (NW) regions is poor; * about twice the proportion of the population are poor in rural areas as in urban ar- eas. Thus the majority of the rural populations of the NE and NW are poor; * a minority of the rural poor are farmer operators, although this reaches nearly one- half in the NE and NW, and almost all rural dwellers are poor farmers in Patagonia (where the rural population is very small). Thus rural poverty is by no means strictly an agricultural problem; and * the severity of poverty, however measured, is greatest among the rural poor. 2. This study, based on a desk review of existing data sources and analysis of an LSMS-based rural household survey undertaken by the Secretariat of Agriculture, Live- stock, Fisheries and Food (SAPyA') in the provinces of Misiones (representing the Northeast) and Salta (the Northwest), will provide the analysis of rural poverty missing from a previous Bank-sponsored study of Argentine poverty2. It will describe the charac- teristics of poor, rural households; identify key constraints to improving their standard of living; and suggest appropriate government actions to assist the rural poor to increase their standard of living through improvements in the enabling environment and direct as- sistance. While the analysis will cover all rural households, identification of constraints and opportunities will focus on the economically active rural poor--small farmers, agri- cultural laborers, and rural non-farm wage workers and entrepreneurs. The focus thus will be on means of raising productivity and income among the poor, rather than provision of social safety net assistance. At the time, called the Secretariat of Agriculture, Animal Husbandry, and Fisheries (SAGyP). 2World Bank, 1995 2 Argentina Poverty Assessment Background Paper No. 6 3. The main objective is to contribute to the formation of a better strategy for con- fronting rural poverty in Argentina. The development of such a strategy is required to help avoid social costs due to excessive rates of urban conglomeration, realize the poten- tial productivity advantages of the small farm, improve social equity and social stability, and increase the efficiency of use of public assistance resources. 4. The study draws on a number of new sources of information developed through the assistance of a grant from the Government of Japan, including papers by local consult- ants (see Bibliography), a Social Assessment involving stakeholder analysis, diagnostic workshops and review of case studies carried out by local NGOs and Government officials in several provinces where rural poverty is severe, and, most important, a survey of 597 randomly-sampled rural households in the two provinces of Misiones and Salta in the Northeast and Northwest, respectively. The survey, which collected data on all compo- nents of income, expenditure, labor and other inputs in farm and non-farm activities, pro- ductive and social infrastructure, as well as socio-cultural characteristics of households, was carried out by SAGyP in May-July 1996 and covered the previous 12-month period. 5. The two provinces surveyed--Misiones and Salta--are located in the northeast and northwest corners of Argentina, respectively. Misiones, inserted between Brazil and Paraguay, is a recently-developed rolling frontier area, formerly entirely wooded, now set- tled by a very mixed population which includes Germans, Finns, Japanese, Poles, Ukraini- ans, Brazilians, Swiss, Swedish and Norwegians, as well as a remnant indigenous Guarani population.? The area is covered by red soils, which are only slightly acidic, well-drained, and with some plant nutrients. A high proportion (25 percent) of these are cultivated, and much of the crop area (36.5 percent) is in perennials like yerba mate (an herbal tea), tea, tung oil, and citrus; which are followed by annuals such as tobacco, sugar cane, and aro- matics. Salta also is a geologically mixed and partly rugged frontier area, bounded by Chile on the west and Bolivia and Brazil to the north. A rather small part of the area is cultivated (8.5 percent), this overwhelmingly (73 percent) in annual crops. Overall, its agricultural production is based on sugarcane, tobacco, grain, livestock mainly on natural pasture, wine grapes, and timber. Although for Argentina as a whole, the percentage of cultivated area (17 percent) and the percentage of the latter in annual crops (45 percent) lies about halfway between the corresponding figures for Misiones and Salta, the latter provinces differ considerably in the characteristics of their agricultural economies from the agricultural heartland of Argentina. However, the rural sample indicates only 13 percent of adults in Misiones were foreign-born, and most of these were from Brazil or Paraguay. Salta, despite its border location, has only 3.4 percent foreign-born adults in the rural areas, these being Bolivian in extraction. 3 Argentina Poverty Assessment Background Paper No. 6 6. The provinces were chosen to represent RmsoddGmEkm'Ae the two regions where &P- rural poverty was consid- ered to represent a seri- r ous problem, because 19Paid-"erK-dke1 Sam LmE they have a high percent- age of rural dwellers and I a high proportion of the A. latter are poor (Figure 1). As noted above, 13 per- cent of the entire popula- tion is rural, but 25-30 percent in the NE and NW. While 15 percent of Per1rre 20 the population is poor, a5 the proportions reach 25- 35 percent in the NE and Figure 1. Poverty Headcount Index (NBI) by Administra- NW; and in general about tive Unit, 1991 (INDEC) twice the proportion are poor in rural areas as in urban areas. About 15 percent of the poor are rural in the NE and NW, though less than 10 percent elsewhere. Although the survey sample was non-random from a national or regional point of view, it was a geographically-stratified random sample of the entire rural population of the two provinces, where "rural' refers to administrative units with under 2,000 population. Thus generalizations found below about "the rural population" or "the rural poor" based on the LSMS survey should all be taken as refer- ences to the corresponding populations in Misiones and Salta, and have no statistical validity for populations elsewhere. A larger rural survey is planned as part of the future work program of SAPyA. 7. Due to an earlier lack of statistical attention to the rural poor, a mixture of data sources, and differing objectives of analysis, variations exist in the definition or scope of the rural poor population. The poverty group in Argentina is customarily defined statisti- cally in terms of a head count index based mainly on housing and educational characteris- tics of the poor, which will be compared below with a definition based on a consumption measure of per capita welfare. Discussions in the next section based mainly on data drawn from an agricultural census also refer to small farmers or a subset thereof termed minifun- distas, which obviously excludes poor rural wage workers and the self-employed. 8. The paper is organized as follows: I. A brief headcount enumeration of rural poverty in Argentina, including location and prevalence, based on existing population and agricultural census data, to es- tablish a country-wide background. 4 Argentina Poverty Assessment Background Paper No. 6 II. Evaluation of existing poverty indicators (unsatisfied basic needs approach) through comparison with consumption-based poverty lines developed from the two-province rural household survey. III. Description and simple analysis of characteristics which differentiate the rural poor from the non-poor and typologies of poor, rural households based on the ru- ral household survey in Misiones and Salta. IV. Analysis of the influences on agricultural crop productivity, which circumscribe the options for poor households to improve their situations via agricultural pro- duction alone.. V. Discussion of implications for appropriate actions to be taken by Government to improve the effectiveness of its rural poverty policies and programs. 9. The objectives of the paper are to contribute to a (i) better understanding by Ar- gentine policy makers and program managers of the accuracy of existing poverty measures for targeting assistance programs to the rural poor and additional information needs; (ii) better understanding of the limitations of existing programs relative to the needs of the rural poor; and (iii) better understanding of a comprehensive approach to rural poverty which recognizes rural-urban linkages, farm and off-farm activities and productive and non-productive assistance needs. As a result, it is hoped that a more comprehensive ap- proach to rural poverty reduction will be adopted, with stronger coordination among sec- toral ministries. 5 Argentina Poverty Assessment Background Paper No. 6 National Context 10. In Argentina, as in many other countries where regular surveys of income or expenditure are not carried out, other indicators of poverty are commonly used both to monitor trends in pov- erty and target geographical areas. As there existed two major sources of data, the National Population Census (the last Census took place in 1991) and the National Agricultural Census (which last took place in 1988), collecting different information for different purposes, the poverty indicators developed from each source are distinct and not mutually comparable. Although none of these indicators bears a close relation to income or expenditure, there is rough consistency in as- sessments of the scope of the rural poverty problem and its geographical and occupational concen- trations. The following highlights what is known in this respect, and provides a national context for the two-province analysis which follows. 11. The rural population, defined for Census purposes as persons living in communities with population under 2,000, represented about 13 percent of the total population in 199 .4 Over the last three decades, there has been a progressive decline in this proportion, decelerating in recent years, from 28 percent in 1960, to 21 percent in 1970, 17 percent in 1980, and 13 percent in 1991. More than 40 percent of the rural population is concen- trated in the Pampeana Region (the central region, which is 92 percent urban overall, en- compassing both Great Buenos Aires and the Pampas). About the same absolute number of rural inhabitants (1.8 million) are found in the Northeast (NE) and Northwest (NW) regions combined, where 28 percent of the total population is rural. 12. Rural Poverty: NBI Statistics. The official Argentine poverty indicators, aggre- gated into an NBI index (Necesidades Basicas Insatisfechas, or unsatisfied basic needs), are based on questions contained in the Population Census, and are thus reliably enumer- ated for the entire country at periodic intervals. There are four or (in some reports) five main indicators of households lacking basic necessities, namely those with: (NBI-1) more than three persons per room; (NBI-2) housing materials judged inadequate; (NBI-3) lack of an indoor toilet; (NBI-4) children between 6-12 years of age not attending school; and (NBI-5) more than four times as many family members as workers and a head of house- hold with less than third grade education. Other indicators are available or conceivable, but not commonly used. The overall poverty head count (NBI index) is defined as the number of families which are characterized by one or more of these indicators (see Figure 1 above). 13. On this definition, 18.3% of Argentine households were poor in 1980, declining to 14.5% in 1991 (these figures disguise a substantial rise in poverty up to 1988, and a fall thereafter). At the latter date, 31.7 percent of the rural population were NBI-poor, as op- posed to 13.8 percent of the urban population. However, in the NE and NW, nearly one- half of the rural population was poor. Classed by occupation, of a national economically- There are numerous settlements with more than 2,000 inhabitants that are essentially rural considering socio- economic structure and linkages to the agricultural sector. If the cutoff were settlements of less than 5,000 per- sons, the proportion of the rural population would rise to almost 17 percent, with much of the increase repre- senting agricultural wage laborers or sharecroppers who live in towns but work in the country. 6 Argentina Poverty Assessment Background Paper No. 6 active rural population of 1.6 million, some 20 percent reported their main occupation to be agricultural, and of these, 36 percent were NBI-poor. Again, some two-thirds of the last were found in the NE and NW regions. 14. Rural Poverty: Minifundios. Aside from the poverty headcount and occupational data summarized above, the Census data has not been subject to analysis which would shed more light on the characteristics of the rural poor, particularly those relating to the mandate of SAPyA. However, the National Agricultural Census of 1988, which took the farm enterprise as the basic survey unit, often has been analyzed to roughly identify numbers of poor farmers, although the definitions and consequently numbers vary from source to source. The definitions are based on characteristics of the farm enterprise (implicitly ignoring off-farm income sources, as well as agri- cultural wage laborers). One of the most relevant characteristics is farm size, but unfortunately (a) a definite farm size could not be enumerated for some 10 percent of farms surveyed in 1988, many being poor households without land title; and (b) a farm size adequate to fully support, or occupy, a household varies considerably by agro-ecological zone, type of farming, and size of household. Thus the most frequently-used definition distinguishes small producers,' where farm size is not part of the definition, and minfundios6, a subcategory of small producers whose farms are also relatively small (but vary by region). By these definitions, out of 421,221 agricultural enterprises in 1988, 230,000 (55%) were classified as small producers, within which minifundios came to about 160,000, concentrated for the most part in two regions: Northwest Northeast Argentina Salta Total Misio- Total nes No. of minfundios 6,024 50,913 14,456 51,10 159,712 .7 As percent of national total 4% 32% 10% 32% 100% As percent of all farms in 66% 65% 54% 60% 38% area As percent of farms in region 12% 100% 30% 100% -- Table 1. Number of Minifundios, 1988 5 "Pequefios productores" as defined by PSA to include: "producers living at the production site, their main income derived from own-farm agricultural activity, with net income less than twice the average rural wage of farm workers and with working capital--including animals and equipment--less than $20,000 and no long-term hired labor. No reference is made to size of farm. 6,"Pequenos productores minifundistas." A residual after deduction of an estimate of the number of small producers with an "undetermined or imprecise farm size." This presumably includes only small farms, although without any clear cutoff points. 7 Argentina Poverty Assessment Background Paper No. 6 Comparing the regional NBI data, minifundios accounted for 40-50 percent of the NBI rural poor in the NE and NW, 20 percent in the Pampeana and Cuyo, and almost all of the (small rural) population in Patagonia. 15. At another extreme, an alternative reprocessing of the 1988 Census which ex- cluded the ten percent of farms of undetermined size and defined "smallholders" more pre- cisely as all farms under 50 hectares counted 186,940 smallholders (49 percent of all farms) but only two percent of farm area. On this more inclusive definition, Cuyo, a re- gion of relatively intensive farming, has a larger proportion of smallholders than the NE or NW (83 percent, versus 64 and 73 percent respectively). 16. A difficulty with an equation of smallholders or minifundios with rural poverty arises from the unclear relationship between farm size, agricultural income, and the eco- nomic viability or welfare level of the household in view of the crucial role of off-farm in- come sources for such households. For example, the percentages of farm producers en- gaged part-time in off-farm work were 29 percent in the Pampas, 49 percent in Cuyo, 28 percent in the NE, 36 percent in the NW, and 37 percent in Patagonia. In some provinces in the NE and NW, the proportion employed off-farm in non-agricultural work was twice the proportion employed off-farm in agriculture. A second problem with the definition is the omission of farm laborers: 340,646 permanent laborers living on farms (non-family members) were counted in the Agricultural Census, these being concentrated in Pampeana (53 percent), followed by NW (14 percent) and NE (12 percent). Additional numbers of farm laborers live apart from their employers, either in the country or the towns. 17. The equation of rural poverty with minifundios does facilitate analysis of the rela- tionship between land tenure and rural poverty. As this will be considered below for the two-province sample, some national tenure statistics will provide the context for subse- quent generalizations. Land is held under a variety of tenure arrangements in Argentina, which have reasonably clear counterparts in other countries. Individual property (propiedad exclusiva, unchallenged ownership by an individual with or without registered title), is the most common form of land tenure in all regions, representing about 76% of all farms of known size. Joint Property (propiedad combinada, where ownership is jointly held, or held by parents) and Agricultural Tenancies (contratos agrarios, with verbal or written contracts and defined rents) are 10 percent or less of farms of known size in all regions. Informal Occupancy (ocupaciones, with or without owner's permission but without any documentation) is important in the NE (27 percent), in Patagonia (16 per- cent), and in NW (10 percent), with a national average of 14 percent. 8 Argentina Poverty Assessment Background Paper No. 6 Percent of Farms of Known Size in 1988 Percent of Individual Joint Agricultural Informal Total Farms of Property Property Tenancies Occupancy Unknown Size Pampeana 80.6% 7.3% 8.3% 3.9% 100% 2.6% NE 64.6% 3.3% 5.0% 27.1% 100% 13.4% NW 79.7% 5.4% 4.5% 10.4% 100% 45.5% Cuyo 91.0% 2.4% 3.0% 3.6% 100% 17.9% Patagonia 77.9% 3.4% 2.5% 16.3% 100% 52.3% ARGENTINA 75.9% 4.8% 5.6% 13.7% 100% 22.2% Table 2. Land Tenancy, Argentina, 1988 18. In terms of land area (and including estimates for the farms of unknown size), the Pampas has the highest percentage of area under Property (90 percent), the highest in Ag- ricultural Tenancies (9 percent) and the lowest in Informal Occupancy (1 percent). In contrast, the northern regions, as frontier areas more recently settled, have the highest percentage of area under Informal Occupancy (around 38.5 percent). The national totals are about 62 percent in Property, 8 percent in Agricultural Tenancies, and 30 percent in Informal Occupancy. 19. The high levels of poverty in the northern regions may be seen to coincide with the highest incidence of Informal Occupancy as a tenure form. That the poor farms are more likely to be under Informal Occupancy is true nationwide, but the fact remains that this particularily hazardous form of tenure is a problem for a small minority of the poor. Other characteristics of farms under Informal Occupancy will be considered below in the context of analysis of the two-province survey. Percent of Farms by Land Tenure and Poverty Status Individual Joint Prop- Agricultural Informal With Undef. Total Property erty ITenancies Occupancy ISize Poor AU (A) 59.1 3.7 4.4 10.7 22.2 100.0 Total AU (B) 64.9 12.5 6.2 6.2 10.2 100.0 Table 3. Land Tenure and Poverty Status, 1988 20. The most important "problems" in land tenure are reportedly non-divided succes- sion, referring to farms of clouded title passed on undivided through more than one gen- eration (13.4 percent of farms), informal occupancy of public lands (8.5 percent), infor- mal occupancy of private lands (5.4 percent), communal lands (1.5 percent), farms on land claimed by indigenous communities (0.7 percent), and those in protected areas (0.1 percent), for a total of 29.6 percent of farms with tenure problems. 9 Argentina Poverty Assessment Background Paper No. 6 3. INDICATORS OF RURAL POVERTY 21. Unsatisfied Basic Necessities Approach. It has been pointed out that the NBI indicators "capture various symptoms or manifestations of poverty", rather than being di- rect measures of poverty; and that they relate to physical or human capital development, a process which has a long gestation period and measures of which tend to be stable during economic cycles when income or expenditure measures fluctuate widely (World Bank, 1995). As the NBI analysis was developed and studied mainly on the basis of urban data, it is not clear how well NBI indicators differentiate between poor and non-poor in a rural context (defined as localities with less than 2,000 inhabitants), where the level of human and physical capital is certainly lower and the distribution possibly flatter. At the same time, the NBI indicators are rather easy to survey and less subject to statistical error than surveyed income or expenditure. Even though the consumption-based poverty indicators may be preferred, if the NBI indicators collectively are closely correlated with household consumption, then they might be sufficient for geographical targeting. To evaluate this, the rural household survey included the corresponding NBI questions. For the provinces of Misiones and Salta, the NBIs prove to be 75 percent and 88 percent respectively of the rural population, as estimated from the LSMS survey. These are higher than the 1991 es- timates for the corresponding regions (55 percent in the NE and 54 percent in the NW). 22. Consumption-Based Poverty Line Estimation. The two-province samples, each drawn to provide representative data for a particular province, are not sufficient to enable us to draw conclusions about the entire rural population of Argentina. Nor do they permit us to estimate a poverty line for the rural population based strictly on data from the sam- ple. However, in order to provide some points of comparison with the urban population, which have been subjected to earlier poverty analysis, it is instructive to translate the pov- erty line previously drawn for urban residents into its rural equivalent in the surveyed ar- eas, based on prices in these two provinces. This allows us to estimate the proportions of the rural population in these provinces which are "poor" in terms permitting comparison to the main urban centers. This poverty line can also be used to divide the samples into the "poor" and "non-poor" for purpose of characterizing the nature of poverty. Con- sumption rather than income is conventionally used for such welfare rankings as the for- mer is more reliably estimated from household survey data and less sensitive to temporary fluctuations (bad harvests, transitory unemployment, etc.), with savings or dissavings and errors absorbing the difference between the two. 23. In these two provinces, the poverty line in mid-1996 came to about US$ 950 equivalent per capita per annum, and the line of indigency or severe poverty about US$ 600.7 Because up-to-date urban poverty line data are unavailable, these absolute numbers The urban poverty line, as calculated in World Bank, Argentina's Poor: a Profile (1995), is based on a basic food basket of 27 items, determined from the 1985-86 Income and Expenditure Survey, reflecting a minimum daily dietary standard of 2,700 Kcal per adult-male equivalent. The cost of this food basket, adjusted to the average age/sex composition ofthe population, is defined as the "indigent line" (a consumption level in value terms which would be sufficient to provide a minimum dietary standard only if no money were spent on non-food commodities). The poverty line, on the other hand, was defined as 1.63 times (in Misiones; 1.59 in Salta) the 10 Argentina Poverty Assessment Background Paper No. 6 cannot be compared contemporaneously with urban poverty statistics. However, they permit us to say, in terms roughly comparable to the 1993 Buenos Aires poverty profile, that about 77 percent of rural inhabitants in Misiones and 73 percent in Salta had expen- diture levels in 1996 which fell below the poverty line, as compared to 18 percent of the population in Buenos Aires in 1993. Moreover, some 31 percent in Misiones and 38 per- cent in Salta fell below the indigent line, as compared to 3 percent in Buenos Aires in 1993. Thus the incidence and severity of poverty in the rural areas of these two provinces were, not surprisingly, substantially higher than in Buenos Aires. It was also higher than was reported for the corresponding provincial cities in 1992 (WB, 1995, p. 15, Table 2.5). Location Ppre ne ofPnnulntian Helow ino nf* Poverty Iniec B sml) NIRI (urbqn)21 MicdanPe 77 11 75 1? N19971 Salta 73 38 88 270(992) Buenos Aires' 18 3 190(992) 1/ Poor and Indivrent- for 1993 as estimated in World Bank- 1995 2/ INDEC. Encuesta Permanente de Hogares (base reducida). October 1992. Table 4. Consumption and NBI-Based Poverty, Misiones and Salta 24. Reliability of NBI-based Targeting. Targeting may be directed at the household level, or at small geographic units (the department, a sub-provincial administrative unit of which 35 were covered by the survey, is the unit considered here). Furthermore, the ob- jective of targeting may be to determine eligibility for a program of assistance and/or to establish the amount of assistance in relation to level of poverty. If eligibility is in ques- tion, the obvious criterion is whether the household falls below the poverty line and, for geographic units, whether the proportion of the poor population, thus defined, exceeds some (arbitrary) percentage (e.g., 50 percent, 60 percent, etc.). If we assume that "true" eligibility or amount of assistance should be based on the expenditure-based poverty line, then how well can the NBI indicators "estimate" where households or departments fall on the scale of eligibility?' 25. To start with the simplest example, suppose the department-level criterion for eli- gibility for poverty assistance is those where at least 60 percent of rural households are poor on the consumption criterion (this would include two-thirds of the sampled depart- ments): if the overall NBI index (proportion of households where at least one NBI indi- cator is positive) is used in place of the consumption poverty line, the targeting criterion will result in 14 percent Type A errors (excluding departments which should be eligible) and 17 percent Type B errors (including departments which should not be eligible). That cost of this food basket, where the multiplier reflects the average budgetary proportion (61 percent in Missio- nes, 63 percent in Salta) devoted to food of sample families for whom (a) total expenditure was in the vicinity of the indigent line; or (b)food expenditure was in the vicinity of the indigent line (the average of (a) and (b) for both provinces was used). Cautionary note: the rural sample is biased through selection of two provinces where rural poverty levels are known to be high. Statistical results might differ if all rural households in Argentina had been sampled. 11 Argentina Poverty Assessment Background Paper No. 6 is, this NBI-based criterion leads to correct identification of majority-poor rural depart- ments in about 69 percent of the cases (out of 35 in our sample). 26. However, such a naive approach to prediction can be improved upon by estimating the relationship between either household expenditure levels or "true" proportions of the poor on the one hand and the NBI component indicators on the other. Regardless of how this relationship is estimated statistically, it emerges that two NBI components -- adequate toilets and school attendance -- have no consistent relationship with measured rural pov- erty levels, and so should not be considered relevant. Moreover, although NBI-1 and NBI-2 can be treated as separate indicators, a combined indicator (positive if either of these housing-related predictors is positive) is equally effective. Finally, NBI-5, an indi- cator which combines an excessive household dependency ratio (household size > four times number of workers) with a low educational level for the head of household, can be divided into its two components (NBI-5A and NBI-5B), from which it appears that the educational component (NBI-5B) is by far the most important. 27. At the household level, the overall multiple correlation between yearly household consumption and the NBI indicators in any combination is weak (R2 <0.10 in multiple re- gressions), although those individual indicators described in the previous paragraph are statistically significant. Using logit to classify individual households as poor or non-poor based on NBI indicators results in no more than 70 percent success (with roughly equal proportions of Type A and B errors). 28. At the departmental level, when household consumption is summed over house- holds within each geographic unit and regressed on the aggregate NBI index (percent of households with at least one positive NBI indicator), the relationship is statistically signifi- cant and somewhat stronger (R = 0.31). The number of households per department classed as poor on the NBI criterion is even better at predicting the number of households classed as poor on the consumption criterion (R2 0.65), and might be the best basis for targeting if the amount of funds were to be proportionate to the poor population. To cor- rectly classify eligibility (on the over 60 percent criterion), logit regression indicates that a combination of a composite NBI(1-or-2) indicator with disaggregated NBI-5A (depend- ency ratio>4), NBI-5B (low educational level), and NB15 (interaction term) provides the best predictions (80 percent correct classification of departments). 29. Using logistic regression on the sample data, it is possible to improve slightly (to 88 percent) on the ability to accurately classify departments as "poor" or "non-poor" on any given criteria (such as 60 percent of the rural population below the consumption pov- erty line). This involves bringing in additional explanatory variables such as mean family size, mean number of adult equivalents, and the ratio between these two (the raw materials from which NBI-5A was constructed), and allowing for non-linearities or interactions in the predictive equation. However, as both the number of departments sampled and the number of households sampled within each department were relatively few, and the selec- tion of provinces surveyed far from random, there are no grounds for arguing that such an approach could be used to identify target departments beyond those for which consump- tion-level poverty can already be estimated. It must be concluded that, until household 12 Argentina Poverty Assessment Background Paper No. 6 consumption surveys are carried out in other regions, NBI-based indicators are all that is available, and the high degree of targeting error noted above is inescapable. 30. For the purposes of further analysis below, drawing a dichotomy between "poor" or "non-poor", on whatever criteria, is less interesting than exploring the differences in household characteristics and behavior over the range of welfare levels, from poorest to richest. Dichotomizing the population also entails a risk of overlooking differences within the poor population (which constitutes the majority of this sample), or of missing non- linear relationships over the range of family income with other variables. Consequently, for some comparisons, the rural population is divided into three groups, the indigent, the non-indigent poor, and the non-poor, based on price-adjusted consumption per capita. For other comparisons, households have been classed into consumption quintiles, a meas- ure of general well-being based on rankings of household consumption per capita adjusted for departmental differences in prices. The quintile classifications also differ from per capita household expenditures in that certain expenditures which do not increase welfare are excluded (taxes and cost of transport to work); items which increase consumption without cost are included (subsidized school lunches, pension and other transfers); and durables are valued at annual depreciation value regardless of when purchased. As the term quintile implies, the rankings of per capita consumption are expressed as five catego- ries (1-5, from lowest to highest), each accounting for an equal number of households.9 4. CHARACTERISTICS OF THE RURAL POOR IN MISIONES AND SALTA 31. Household Expenditure in Misiones and Salta. Sample data on all net com- modity and financial inflows to the household (i.e., total net income including transfer) over the last twelve months and the corresponding outflows in the form of real consump- tion, cash expenditure or accumulation of assets were assembled and are shown averaged by quintile in Annex Table 1. As the sample is self-weighting, these averages may be taken to represent all rural households in Misiones and Salta. Inflows and outflows theo- retically should be equal in the absence of errors in the data, and the discrepancy recorded overall was only 2.7 percent. However, the discrepancies for each quintile were substan- tially larger, reflecting reported inflows greater than reported outflows for the lowest An alternative approach is to distinguish quintiles in consumption per adult-equivalent as a measure of relative welfare adjusted for differences in consumption "needs" by age and sex. However, the correlation between the two measures is 0.95 and the maximum difference between the two measures in mean household consumption for any quintile is 5 percent, so any relationship found for one measure will be valid for the other. 13 Argentina Poverty Assessment Background Paper No. 6 quintiles, and the reverse for the rich. Underreporting of income by the wealthy would be expected, but it is unclear what the source of error is for the poor. 32. The calculated Gini index of inequality based on per capita expenditures in these two provinces is 37 (it is 41 for expenditures by household). Given that the provinces were selected for their predominance of poor families, it is not surprising that this is low relative to inequality in rural areas as a whole elsewhere in Latin America (in 1989 these ranged over 41, 44, 45, 50, 57, and 70 for Venezuela, Peru, Mexico, Honduras, Brazil, and Chile respectively"o). Indeed, the main reason for citing the Gini index is as evidence that there was sufficient income variation in the sample for these two provinces to distin- guish the characteristics of the poverty group. 33. The varia- tions in total expen- Per Capita Expenditure by Poverty Line ditures per capita, including savings, across poverty groups are shown in Figure 2, and totals $2,000 per household are tabulated by quintile aSa\Angs along with inflows $1,500 E Durables and savings in An- O Educ & Health nex Table 1. The $1,000 E Daily use graphs and tabula- M Food tions show that, --- while per capita ex- $500 penditure on food nearly doubled be- $_ tween indigents and Indigent Poor Non- non-poor, the pro- poor portion of food ex- penditure in the budget declined Figure 2. Per Capita Household Expenditures: Misiones and from 73 percent for Salta the lowest quintile to 48 percent for the fourth quintile, and then, more sharply, to 36 percent for the highest quintile--the consumption behavior of which looks distinctly "non-poor". The differences in per capita expenditure between poor and non-poor groups were accentuated by a de- cline in the average family size from 7.88 to 2.96 from Q1 to Q5 (see discussion of food security below). Reported savings--defined as accumulation or decrease of assets (mainly 10 Compilation by Haeduk Lee from studies cited in Technical Department, Latin America and the Caribbean, Pov- erty and Income Distribution in Latin America: The Story of the 1980s (World Bank, 1993), Annex 1. 14 Argentina Poverty Assessment Background Paper No. 6 land, herds, and commodity stocks) and depreciation of fixed capital--are negative for only the lowest quintile, rising to eight percent of total outflows (10 percent of inflows) for the highest quintile." 34. Income Sources and Household Activities. Rural households in these two provinces divide evenly (50:50) between those having some income from own-account agriculture (with a few ex- ceptions, those having the Per Capita Income by Poverty Line use of some farmland) and those who should be called non-farm households. The $1,800 - distinction is not a clean one, $1,600 as members of both types of $1,400 households pursue different a Self-employment activities and have multiple $1,200 n Transfers income sources. Farm $1,000 i/ Non-farm Wages households (those with agri- $800 0/Farm Wages cultural income2) represent a Subproducts about 40 percent of indigent $600 zAgriculture households, half of poor but $400 not indigent households, and $200 60 percent of non-poor households. Indigent Poor Non- 35. The first observation poor is that the rural population, including the poorest, are Figure 3. Per Capita Household Income: Misiones and very much a part of the cash Salta economy. Valuing net in- flows of commodities (mainly retentions from agriculture for self-consumption) at market sales prices, they represent only 3.3 percent of total inflows, and the proportion peaks at 10 percent for the median quintile. That is because own-account agriculture as a whole, including crops, animals, and subproducts, contributes only 11 percent of inflows to the poorest quintile (versus 26 percent to the highest) (Figure 3). Wage labor is a crucial source of income to all groups, although the importance of non-agricultural wages is higher for the non-poor. Transfers--pensions and remittances--also are important to the poor, but are replaced in importance by self-employment income among the non-poor. However, these observations treat the rural population as a single group, whereas it is If the difference between inflows and outflows were treated as accumulation rather than a statistical discrepancy, the signs would be the reverse of expectations. 12 Although statistics based on occupational surveys may distinguish farmers from non-farmers based on the "princi- pal" occupation, we prefer the more inclusive definition of farm households because of the focus of this analysis on farm productivity and the fact that some owners of even very large farms may invest most of their time and receive most of their income from non-agricultural activities. 15 Argentina Poverty Assessment Background Paper No. 6 more revealing to differentiate between those who are at least partly farmers and those who are not. Household Head's Main Occupation (percent of non-farm families) Perma- Seasonal Sector Self-Em- nent Wage La- House- Unem- Total ployed Wage La- bor work ployed bor Agriculture 0.3 35.3 12.5 0.7 48.8 Agri- 1.4 0.3 1.7 processing Industrial 1.4 3.1 2.4 6.9 Commerce 2.8 1.0 0.3 4.2 Services 2.8 2.1 3.8 8.7 Public Sector 7.3 0.7 8.0 Other 1.0 1.7 3.5 6.6 9.0 21.8 Total 8.3 51.9 23.5 7.3 9.0 100.0 Table 5. Occupations of Non-farm Families, Misiones and Salta Per Capita Income of Non-farm Rural Families by Source Non- Indi- Poor Poor Total gent Agric. Wages 363 425 1,054 587 Non-Agric. Wages 219 289 449 310 Self-employment 43 57 211 97 Income Transfers 96 206 -60 86 Other 1 0 0 0 Total 722 978 1,654 1,081 Household size 6.61 4.77 3.20 5.00 Pet. of non-farm 38 33 30 100 hh Table 6. Per Capita Income of Non-farm Rural Households, Misiones and Salta 36. The heads of non-farm rural households are divided by type of employment into the self-employed, permanent wage laborers, seasonal laborers, unemployed, and house- hold workers, as shown in Table 5. Relating household income sources to these occupa- tional categories does not make much sense, however, as other members of the family may contribute income from quite different sources (see section on labor market for analysis of wages and employment by worker rather than household). Arraying income of non-farm 16 Argentina Poverty Assessment Background Paper No. 6 households by poverty group (Table 6), note how little change occurs in the composition by source over the three groups. Per capita incomes are also strongly inversely propor- tional to household size: income per household of the non-poor is only about one-third larger than the average for indigents, but the much reduced family size provides a much higher standard of living. Obviously it is possible for a rural household dependent mainly on farm wage labor and with few or no dependents to earn more than the poverty level. On the other hand, a disproportionate part of rural non-farm households falls in the target group of the rural poor. Household Head's Main Occupation (percent of farm families) Self- Perma- Seasonal Land- House- Unem- Sector Employed nent Wage lord work ployed Total Wage La- Labor bor Agriculture 72.3 2.8 3.1 1.0 79.2 Agri- 0.3 0.3 processing Industrial 1.4 0.3 1.7 Commerce 0.7 0.7 Services 1.0 0.7 1.7 Public Sector 5.2 1.4 6.6 Other 0.3 0.7 0.7 2.8 5.2 9.7 Total 74.7 10.4 5.9 1.0 2.8 5.2 100.0 Table 7. Occupations of Heads of Farm Households, Misiones and Salta 37. The corresponding occupational distribution for heads offarm households (Table 7) indicates that less than three-quarters consider themselves farmers by primary occupa- tion. However, this gives a deceptive impression of concentration of income sources, as other family members are more likely to declare non-farm primary occupations. 38. In fact, on average less than half of the income of "farm families" is derived from the farm, counting low-level processing of farm products (Table 8). The proportion is least for indigent households, which owe their survival to more liberal amounts of non- farm wage income. Crop cultivation is the source of the bulk of farm income: many households appear to have seen losses on livestock, and to have reduced their herds dur- ing 1995-96.13 Subsidiary production was generally little more than extraction of edible 13 For households raising livestock, attribution of net income to crop cultivation, livestock raising, and subsidiary production was not infallible, as (a) transfer prices used in attributing income to home-grown feed crops and corresponding feed costs for livestock do not necessarily reflect real market valuations; and (b) the production of milk, meat, hides, and wool is valued in subsidiary production rather than livestock raising. Livestock also supply subsistence demands, and a small operation with an accounting loss (at producer prices) may meet these demands at lower cost than retail purchase. 17 Argentina Poverty Assessment Background Paper No. 6 Per Capita Income of Farm Families Indi- Poor Non- Total gent Poor Farm Income 168 471 980 629 Crops 135 396 807 520 Livestock 12 3 49 26 Subsid. Prod- 21 72 124 84 ucts Agric. Wages 119 97 106 106 Non-Agric. 149 137 273 201 Wages Self-employment Income (Non- 43 47 168 101 Agric) Transfers 56 90 96 85 Rent and Other 1 4 66 31 Total 536 845 1,689 1,153 Household size 7.39 5.51 3.67 5.13 Pet. of farm hh 24 32 44 100 Table 8. Per Capita Income of Farm Households, Misiones and Salta oils or essences from crops and production of milk, meat or sausage, hides, and wool from livestock. Farm wage work is an important income supplement for the indigent, though its importance declines for better-off households. Non-agricultural wage labor plays a crucial role, accounting for 20-30 percent of net income for the poorest households. Non-farm self-employment is relatively unimportant for poor and indigent households, but for a few non-poor households entrepreneurship may represent a path to accumulation. The signifi- cant role of transfers to households at all quintile levels--mainly pensions to retirees-- should also not be overlooked. 39. Again, it is notable that total, as opposed to per capita, household incomes are rather flat across the three groups. There is only a 12 percent difference in total house- hold income between the indigent and the non-poor, but household size of the latter is cut in half, resulting in a tripling of per capita income. The crucial role of household size is further evidenced by the fact that the correlation between the log of expenditure per capita and that of household size is -0.45 and the variance of the log of household size (0.62) is larger than that of total expenditure (0.40). Finally, farming households are slightly under- represented among the poorest quintiles, relative to non-farming households. 40. To better understand differences among rural households, it would be desirable to cluster them by types less abstract than farm and non-farm households. Unfortunately, this has proven virtually impossible, because of a shortage of differentiating data from the survey and because those groups which could be distinguish from the "typical" rural 18 Argentina Poverty Assessment Background Paper No. 6 household usually represented too small a proportion of the sample to permit statistically- meaningful generalizations. For example, households of foreign origin are too few in number, and households which had migrated from other areas of Argentina showed no interesting differences from households which had not migrated. One exception may be the cluster of households residing in administrative or market centers within rural depart- ments, as distinguished from other households. 41. If it is assumed that residences with water supplied by public networks and with electric lighting distinguish the former,14 then it appears that around one-fifth of the sam- ple resided in such community centers. These households were disproportionately those where the household head had a non-agricultural occupation, e.g., as a non-agricultural wage worker, self-employed in commerce, or especially on the public payroll; and where the household head was literate and possessed relatively higher levels of education. Inter- estingly, both overt unemployment rates and the proportion of the unemployed who had given up seeking work, though not the poverty rates, were also higher compared to households in the hinterland outside these community centers. Thus it should be under- stood that even within areas defined as "rural", something persists of the rural-urban dis- tinctions, arising from higher density of settlement and the nature of the the economic functions served by any community center. 42. Other Socio-Cultural Characteristics. Family size was clearly the single most important familial characteristic correlated with per capita income and expenditure, although a number of other factors show some empirical relationship with household economic status. In an analysis of variance (ANOVA) model accounting for expenditure per capita in terms of multiple household characteristics, such as age, gender, literacy, education, and migration status of household head, grandparents' occupations, and number of household members, only the number of household members and completion of primary education by the head of household proved statistically sig- nificant, with the negative correlation between per capita expenditure and household size ac- counting for almost all of the "explained" variance. Household size also is significantly positively correlated with total household income and expenditure. 43. Indigenous households, if broadly defined by the fact of one or more parent's ethnicity, make up only 3 percent of the sample, most being from Salta; and the small number in the sample does not permit very detailed analysis. As only 0.7 percent of heads of household consider themselves part of an indigenous group, and 2.2 percent speak or understand an indigenous language, ethnic consciousness doesn't appear very high. In- digenous origin is strongly associated with poverty--55 percent of indigenous households, on the broadest definition, fall in the lowest consumption quintile. Given the small num- bers of acknowledged indigenes in the sample, it was impossible to demonstrate any inde- pendent relation of ethnic status to poverty status, when the level of education of house- hold head, as well grandparents' education and occupations, were controlled for. That is, 14 No information establishing the location of residence--town, village, or countryside--was collected. 19 Argentina Poverty Assessment Background Paper No. 6 the very few indigenes who had obtained better education did share better jobs and higher economic status. 44. Gender differ- Non-farm HH* Farm HH ences can be examined Gender of Head Gender of Head by looking at the differ- Means of Groups Male Female Male Female ential characteristics of Income per capita 1,079 981 1,177 887 female-headed house- Of which: transfers 51 352 65 276 holds, or by looking at Poverty rate (%) 72% 64% 54% 65% the access of all females Numbers of HH 268 45 257 17 within the household to * including farms with less than 0.5 hectares education, labor mar- ket, etc. The second Table 9. Gender, Income and Poverty type of analysis is more conveniently integrated in other sections below. Female-headed households made up 10.6 percent of the sample. The distribution of such households among per capita expenditure quintiles is roughly equal and the poverty incidence overall is nearly the same as that of male-headed households. This is due to a high proportion of widows, especially in the more numerous non-farm cohort. Households headed by widows are comparable to those headed by elderly men, e.g., they usually have expenditures in excess of income and are dissaving; but poverty levels are not so high if pensions are received. However, female- headed farm households, less likely to receive pensions, are notably worse off (Table 9). Farms with female household heads average only 60% of the size of farms run by males, and net income per hectare which is 66 percent of that of males. Pensions and other off- farm income tend to narrow this "gender gap", but not erase it. 45. The educational and literacy levels of the rural population in the two provinces show surprisingly little relationship to poverty status or gender (Table 10). Among adults, School Attainment Indizent Poor Non-Poor Male Male Fe- Maleemale Illiterate2 1.6 7.0 1.3 2.8 3.3 0.0 Less Primary Incomplete 56.6 50.7 58.6 49.0 18.9 47.7 than Primary Complete 27.9 31.2 23.0 30.1 25.6 31.4 18 Secondary Incomplete 12.8 11.2 16.4 17.5 15.6 20.9 years Secondary Complete 0.4 0.0 0.7 0.7 4.4 0.0 old Tertiary or University 0.0 0.0 0.0 0.0 1.1 0.0 Other 0.8 0.0 0.0 0.0 1.1 0.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 Illiterate2 13.5 20.4 11.2 21.0 5.3 10.6 Primary Incomplete 40.4 38.0 40.5 39.5 42.9 31.0 18 Primary Complete 34.8 27.6 37.1 26.3 32.7 32.9 years Secondary Incomplete 7.4 9.2 5.4 6.2 8.6 8.8 or Secondary Complete 2.8 3.6 3.5 4.9 5.3 6.9 above Tertiary or University 0.4 0.8 1.9 2.1 4.9 9.7 Other 0.7 0.4 0.4 0.0 0.4 0.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 Includes children still in school. 2 Including some with 1-2 years of primary school. 20 Table 10. Educational Achievement of All Household Members Argentina Poverty Assessment Background Paper No. 6 the non-poor are about half as likely to be illiterate and greater percentages have entered or completed secondary or higher educational levels. The same patterns are less pro- nounced among the younger generation, among whom illiteracy is relatively rare. In fact, the educational patterns are clearer if we disregard poverty status and focus on other vari- ables. For example, it is apparent that levels of education, while still low for rural dwell- ers, are improving across generations (Table 11). Among the parents of current heads of households and their spouses, 26 percent of the men and 42 percent of the women were illiterate or unschooled, but this has been reduced to 7 percent and 16 percent for their children, and under 9 and 13 percent respectively for the grandchildren (including those not yet of school age). Similarily, among the previous generation, only 12 percent of the men and 10 percent of the women completed primary school or better, and only less than one percent of either went on to receive some secondary or higher level educations. This has improved significantly among the current household heads and their spouses (7-8 per- cent of whom have received post-primary educations), and current school attendance among their children, as a proportion of age cohorts, shows additional progress, with over 95 percent attending primary school and over 45 percent secondary school (of children still living at home at age 17). Put conversely, the main educational problem is that the majority of rural children fail to enter secondary school (Figure 4; the statistics do not cover children who have left home). Although the household survey did not ask why, in- formants for the locality survey generally cited the need to work as the reason for limited attendance of secondary schools. About 60% of non-attendees of secondary school are from indigent families (who represent 50% of the corresponding age cohorts). 46. A close relationship between educational levels of a household and its economic status is commonly found in cross-sectional surveys. However, the direction of causation can easily be confused, as the economic status of the family affects the ability to educate the children. To keep the direction of causation clear, we can look at relationships which precede each other in time, i.e., the influence of the educational levels of each generation on those of the following generation. For example, adults were about twice as likely to be School Attainment Last Generation Head & Spouse Children' I Male Female Male Female Male Female Illiterate2 24.0 39.9 7.3 16.4 8.8 12.8 None 2.0 1.6 Primary Incomplete 53.2 44.3 61.8 55.2 23.4 28.9 Primary Complete 11.3 9.0 22.9 18.9 51.7 39.0 Secondary 0.9 0.8 6.4 7.6 13.8 13.4 Tertiary or University 0.1 0.1 0.6 0.4 2.0 5.8 Other 0.1 0.4 0.4 0.5 Missing 8.0 4.1 Total 100.0 100.0 100.0 100.0 100.0 100.0 Includes children still in school. 2 Including some with 1-2 years of primary Table 11. Educational Achievement by Generation 21 Argentina Poverty Assessment Background Paper No. 6 School Attendance by Age Cohort (percent in age cohort by type of school) School Attendance 9I0 RNone (Poor) - None (Non-Poor) 0 0 60 jo University Terciary 40 MTech. Second. o_30 20M Ordinary Sec. 10 E Prirrary 01-EMiT Pre-Schol 0 2 4 6 8 10 12 14 16 18 20 22 24 Age in Years Figure 4. School Attendance by Age Cohort, Misiones and Salta illiterate themselves if their parents were illiterate. One out of four children of illiterate or non-schooled grandparents received no education (compared to 12 percent overall), whereas 40 percent of children of grandparents with secondary education received tertiary or university education (compared to 1.7 percent overall). Thus it can be said that educa- tional status is partially transmitted intergenerationally. 47. It is particularly interesting to assess how the achieved educational levels of senior members of a household influence the current economic status of the household. This may depend on the sector of economic activity and occupation of the senior members. For ex- ample, some other studies have found that education, or higher levels of education, have less impact on income from agriculture than on off-farm earning opportunities, and that the marginal return to each year of education is less for the poor than for the non-poor." To test such possibilities, regression analysis was performed on the survey data based on a model which posited that total annual household expenditure (as a proxy for long-run in- come, and including savings) was a function of the summation over individual adult mem- bers' years of education and combination of sectors of occupation (agriculture or non- agriculture) with type of employment (salaried manager, self-employed, uncompensated 15 Lopez and Vald6s 1997; Wiens 1995. A difference in marginal returns to years of education might result from differences in the quality of education available to households of different income levels. 22 Argentina Poverty Assessment Background Paper No. 6 family labor, full-time wage labor, and seasonal wage labor).16 Land owned and land cropped were included as covariates to partially control for effects of differences in house- hold productive assets, and provincial dummies to control for the influence of geography. Interactions between years of education, occupational sector x type, and poverty classifi- cation of the household were included to test whether the hypothesized differential mar- ginal impacts of education occur in Argentine rural areas, and the extent of diminishing returns was tested by inclusion of a squared term for years of education. 48. The results (see Table 12 for a multiple regression including 515 households") suggest that: (a) the level of education has a positive marginal impact on household income level regardless of sector or occupation (about $103 per year for each additional year in the base case of indigent households working in all combinations of sectors x job types other than non-farm wage labor); but (b) the marginal impact of years of education is considerably higher for the non- poor generally ($90 more than for the indigent) and particularily in non-farm wage labor (an additional $222); (c) in addition, and independently of years of education, jobs as farm managers (salaried), non-farm self-employed, and (with 10 percent statistical significance) non-farm wage workers add more to annual household income ($8,754, $1,173, and $1,024 respectively) than other jobs (mainly agricultural sector, whether self- employed, household uncompensated, or wage labor); and (d) there appears to be serious diminishing marginal returns to education, but only in non-agricultural wage labor and most severely for the poor, such that marginal returns fall to zero after 6.6, 8.4, and 13.5 years of education respectively for the indigent, poor, and non-poor." 16 The contribution of each individual adult is assumed a function of education and type ofjob, and total income per household is the sum of functions (assumed identical in form and coefficients) for all individuals, so in the ag- gregate function, the numbers of individuals with each sector/type ofjob and the sum of the years of education for all adult household members (and the sum of squared years of education, to test for diminishing return), be- come the relevant variables. 17 Occupational and educational data was counted for both employed and unemployed (including housework) indi- viduals, in order to avoid selection or specification bias due to exclusion of the unemployed. Only the primary occupation of each individual was counted. Note that poverty status, itself an endogenous variable, is not used as an explanatory variable other than to differentiate coefficients on other variables (even here, the ideal would be poverty status ex ante). Caviat: diminishing returns does not appear if estimated income is used in place of estimated expenditures (in- cluding net savings) as the dependent variable. Points (a) - (c) remain valid. Recall that estimated income is believed more subject to errors than its theoretical equivalent calculated mainly from expenditure data. 23 Argentina Poverty Assessment Background Paper No. 6 49. The strik- Standard ingly low reward Variable Coefficient Error overall to addi- SECTOR X TYPE (count per household) tional education-- Farm Manager 8,753 1,536 * Non-agric. Self-employed 1,173 322 * compared, say, to Non-agric.Wage Labor 1,024 631 ** the $200 per Unemployed or Housework 174 234 month earnings of YEARS OF EDUCATION (sum per hh) a seasonal farm Years 103 51 * laborer"--and the Years squared -2 5 finding in Years x Non-farm wage labor -21 8 ' Years x Poor 38 23 off-farm wage Years x Non-poor 90 25 * labor, the poor Years x Poor x Non-farm wage labor 41 41 apparently receive Years x Non-poor x Non-farm wage labor 222 48 * zero payoff to Years squared x Non-farm wage labor 200 96 * post-primary edu- OTHER VARIABLES: c Crop Area (hectares) 156 45 * cain--hepst e Land Owned (hectares) 0.9 0.5 * plain higher post- primary dropout Province (Salta=-1) -1,148 276* porae aongthe* significant at 5 percent or better ** significant at 10 percent or better poor. Jobs paying I Expenditure including savings proxy for income. a premium most Table 12. Influence of Education and Employment on Annual Income frequently go to (515 households, R = 0.33) primary graduates, and this may keep children in school up to completion of primary school.20 That the marginal returns are lower to the poor than to the non-poor suggests differences in the quality of education accessed by each group, not captured by number of years of schooling. That diminishing returns to education are severe in non-farm wage work, but not in the agriculture sector, suggests a lack of local jobs requiring skill. Over- all it seems that currently the main motive for pursuing education beyond primary school would have to be consumption demand rather than impact on earnings. 50. Although the intergenerational progress in education does give reason for some optimism, for adults the main external agents of change must be training, technical assis- tance, and organizational participation. The rural areas of Argentina prove to be sorely lacking in these means of human capital formation, though it is unclear how Argen- tina compares with other countries or the rural areas with urban. Some 3.8 percent of adults (age > 15) received some form for non-formal training over the previous year, of which 1.2 percent was in labor skills, 1.3 percent technical assistance, and the remainder divided among home arts, health, and other. Members of non-poor families were two to three times more likely to have received some form of training than indigent or poor family 19At a discount rate of 10 percent, a marginal return to education of $400 would be required to offset opportunity cost, if the latter is evaluated at the annual earnings of a long-term farm laborer. 20 Tabulation does not suggest that any minimum educational requirement is associated with these jobs, but modal education ofjobholders is at least full primary. 24 Argentina Poverty Assessment Background Paper No. 6 members. Training was evenly divided between the sexes, though with men receiving most technical assistance and women most of the last three categories. Non-farmers had twice the training incidence of farmers; and Salta had a three-to-one edge over Misiones in training (the bulk apparently being in non-agricultural subjects; see below). Much of this non-formal training was not free, and wealthier groups tended to pay more. 51. Among self-employed farmers, 20 percent received some form of individual technical assistance (TA) over the year, of which the greatest assistance was received from (processing?) industry technicians (6.8 percent ), 5.7 percent government techni- cians, and 13 percent leaders or technicians of cooperatives. No major technical assis- tance was attributed to farmer-neighbors or parents, and little to unattached technicians or sellers of agricultural inputs. The bulk of farm technical assistance was concentrated in Misiones (nearly four-to-one advantage over Salta). Receipt of assistance was also posi- tively associated with quintile status, notably assistance by the government, 59 percent of which went to the top quintile and only 5.9 percent to the lowest two quintiles; by the co- operatives, 64 percent to the top two quintiles; and self-employed agents, 75 percent to the top two quintiles. Most of the agricultural technical assistance was delivered through farm visits, often 6-8 times per year in Misiones, less frequently in Salta. Farm visits were made, in decreasing order of frequency, by agents of commerce/industry, the cooperatives, federal government programs, private technicians, and NGOs. None were made by agents of the province, municipality, or the church. 52. Technical assistance delivered through group meetings is a potential supplement, or substitute, for individual services. About 17 percent of the farmers participated in group meetings, usually 1-3 times per year, disseminating agricultural technique, spon- sored by all sources. If the numbers who received technical assistance in either personal or group form are counted, the total receiving assistance rose to 26 percent of all farmers. These meetings, though a little more frequent in Salta than personal delivery, still were most common in Misiones and attendance underrepresented the poorest quintiles. On the other hand, participation in such meetings, unlike individual technical assistance, was not found to have an impact on farming productivity, in analysis below. Therefore the efficacy of this delivery mode is can be questioned. 53. Farmer organizations have a variety of potential roles, including direct training, obtaining access to technical assistance or credit, processing or marketing products, and providing inputs or consumer goods. Information on two types of organizations, coop- eratives and farmer associations, was collected, and it is possible to assess the extent of their roles in practice. Cooperatives existed in the vicinity of only 21 percent of sample farms, mainly those in Misiones (where 35 percent of the farmers had access to a coop- erative). However, only 15 percent of indigent farmers could access a cooperative, as compared with 25 percent of the non-poor. Farmers' associations were found in the lo- calities of only 18 percent of farms, also more often in Misiones (where 22 percent of farmers had access, compared to 12 percent in Salta); the same degree of bias with respect to poverty status obtained. Where cooperatives exist, about 60 percent of farmers belong to them, whether in Misiones or Salta; and where farmers' associations exist, slightly over 25 Argentina Poverty Assessment Background Paper No. 6 half belong in Misiones, but one-third in Salta. Overall, one out of five farmers belongs to one or both of the two types of organization. However, the effect of the presence of a co- operative or farmers' association on participation is relatively weak among indigent farm- ers, and only one out of 12 indigent households belongs to either type of organization. 54. A rather basic form of organization is represented by exchange of labor among farmers, a form of mutual assistance which could be very informal and was practiced on some 23 percent of farms (46 percent in Misiones, 13 percent in Salta). This is of par- ticular interest as it was the least correlated with quintile status, that is, is almost as com- mon among the poorest (26 percent) as among other farmers. This makes it a potential nucleus for organizing efforts, although the practice itself doesn't seem to have much im- pact on economic status. 55. How much does organizational participation matter? Deferring its impact on pro- duction to a subsequent section, we look here at its role in facilitating access to services. Logistic regression was used to estimate how much membership in cooperatives and farm- ers' associations was associated with a higher probability of (a) receiving personal or group technical assistance, or (b) obtaining credit.21 In the first case, the model posited that membership in a cooperative, a farmers' association, or a work exchange group (with an interaction allowed between the first two) would each positively influence the probabil- ity of receiving either or both individual or group TA. These hypotheses were jointly confirmed, with individual Wald statistics that were signficant at less than 1 percent level and 82 percent correct identification. Cooperative and farmers' association membership were quantitatively about equally important, and more so than work exchange group membership. Although about half the cases of TA actually received were not correctly identified (noting that some farmers lack access to groups, yet receive TA), it seems clear that either group membership substantially helps to gain access to TA or TA recipients have characteristics which make them likely joiners of groups. Further, neither quintile status nor farm size are statistically significant in accounting for TA received if group membership is taken into account--suggesting that the observed bias of TA in favor of the better off is a product of the lack of organizations accessible or attractive to the poorest, rather than discrimination by the technical assistance system. 56. Surprisingly, organizational membership does not seem to relate as well to credit access. Here the basic finding of logistic regression is that past solicitation of loans is more closely associated with current receipt of technical assistance in personal (primarily) or group (less significant) form, to the extent that group membership variables add nothing directly to the equation (indirectly, of course, they facilitate receipt of technical assis- tance). Although common sense suggests that agricultural technicians can be the gateway to credit, neither group membership nor receipt of technical assistance bear any relation- ship with (a) solicitation of loans in the past year, or (b) actual receipt of loans. 21 In the first case, the direction of causation is ambiguous. 26 Argentina Poverty Assessment Background Paper No. 6 57. Rural Savings. Some 15.5 percent of all rural households reported having some form of savings (10 percent among the indigent, 14 percent among the poor, and 21 per- cent among the non-poor). For those with savings, the mean total amount (stock) was $5,705 (ranging from $1,768 for the indigent to $8,118 for the non-poor). Asked to re- port different forms of savings, including bank accounts, financial instruments, buildings other than principal residence, land rented out, money, stocks of food or forage, and loans to non-family members, most frequently families had only one type of savings, invested in buildings (56 percent) or stocks of food or forage (38 percent, mainly in Misiones). These two categories accounted for 85 percent of total value of savings. Only six families had bank accounts, another six lent out money, and two held cash. The current year's net savings, for those with one or more of the above assets, averaged $32 (there were almost as many dissavers as savers). The absence of formal financial institutions was confirmed by the fact that only one out of 20 local central places had a bank branch. Although it is possible that households under-reported their assets in the survey, the statistical evidence collected does not suggest that a program to create small savings institutions would be find it easy going, due to the fact that the few families which have liquid savings are widely scattered geographically, and appear to prefer physical assets over financial instru- ments or cash, perhaps a legacy of past high rates of inflation.22 58. Rural Finance. If only a small minority of rural households save, then even less (11 percent) have ever sought to obtain a cash loan (two-thirds of the latter in Misiones, and ranging from 8 percent for the indigent to 14 percent for the non-poor). The majority who have not tended to be poorer and less educated than those who have, or households with female heads, or members of indigenous groups (none of whom have sought or re- ceived loans). As for why such households haven't tried to borrow, 40 percent have been discouraged by belief that they wouldn't be approved, the excessive documentation re- quired, or fear of the process. Another 36 percent have avoided borrowing as too risky, and 15 percent did not feel a need for credit (both frequently in the highest quintiles). 59. Of the 11 percent who ever tried to borrow, 32 percent also did so in the last twelve months, and 43 percent of these were refused a loan, on one grounds or another (no signficant pattern is apparent). Thus only 12 out of 587 households (2 percent) actu- ally received cash loans in the previous year (and on average, almost two loans per house- hold). The largest proportion of these loans (39 percent) were for "Other Household Ex- penses"--probably consumer durables--and not for purchases of land, housing, food, agri- cultural inputs, machinery, animals, or payments of debt. The average size ranged from $446 for indigent families to $3,640 for non-poor families. By source, the Government accounted for 26 percent of the number of loans, cooperatives 4 percent, state banks 9 percent, family or friends 35 percent, and "other sources" 26 percent. However, the typi- cal scale of loan also differed greatly by source: in the tens or hundreds from family or friends; $1,000-2,400 from a Government program or cooperative; and $10,000-18,000 from a bank. Loans from family and (usually) friends were interest free, those from the 22 This conclusion is subject to revision by a follow-up mission on the rural financial sector, which interviewed some of the same households. 27 Argentina Poverty Assessment Background Paper No. 6 cooperatives or banks charged annual interest rates of 16-18 percent, and those from Government programs a rate averaging only 3.6 percent. Although some refused borrow- ers cited absence of guarantees or titles as the reason they were turned down, in fact only four of 23 loans extended required a guarantee (land title, house, or other goods). 60. Purchases of goods on credit was a more common type of financial transaction, as 13 percent of households reported such purchases in the last year. The credit was mostly extended by stores for the purchase of consumer goods--about 6 percent of cases were for machinery or production inputs, and 2 percent for land or housing. As far as the respon- dants were aware, guarantees were usually not required (72 percent), though in other cases the substance of the good purchased served as the guarantee. The sums involved in these transactions were not notably less than for cash loans, with average transactions of- ten over $1,000. 61. A final type of finance--advances against crops or livestock--proved virtually non- existent, with only one such transaction reported. Thus the survey data suggest that rural producers, agricultural or non-agricultural, in these provinces scarcely benefit from the financial system, and the outreach of Government credit programs, with their substantial subsidies, is extremely limited. Within this context, the poorest in income are also poorest in liquid assets, and the least served by the financial system. 28 Argentina Poverty Assessment Background Paper No. 6 62. Labor Market Participation and Employment. Figure 5 provides a succinct representation of the main activities of some 2,197 household members age 10 or above, Activity Distribution by Age Statu so SednWck Work Unavailable *Ck 30n 30~ Invalid/Adx 20Rete 1o0E Housewovk 10 14 18 22 26 30 34 38 42 46 50 54 58 62 66 70 74 78 Ag. in Years Cass wei&nedby AGEWr Mnor categxies omitted Figure 6. Individual Activities by Age Activities of Youth by Age Group 10 60, Actiity b 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 AgeoGroup Cases wxeitedby A(EWI 29 Argentina Poverty Assessment Background Paper No. 6 expressed as percentages of each age cohort. Figure 6 concentrates on the normal ages of labor force entry, between 10-25, expressing the same data as actual numbers of persons in each age cohort. Together the two figures are suggestive of the life cycle of a family: beginning when primary school is completed, an increasing proportion of age cohorts make major decisions, whether leaving the family to migrate or start new families (hence the rapid decline in size of age cohorts remaining in the family between 15 and 25); seek- ing a job; or (for females only) concentrating on housework. Among the youth who have dropped out of school (age 10-20), a disproportionate percentage (relative to other age groups) are either still seeking jobs or have stopped seeking them, being convinced that no jobs are available. By their 30s, the proportion who have found adequate jobs rises to above the demographic proportion of the age group. Towards the older end of the life cycle, an increasing proportion drops out of the labor force due to retirement, invalidism, or feebleness, while a majority in the rural areas continue to work into their 70s. 63. The size of the potential labor force may be defined in either of two ways: to in- clude only those working or seeking work, or to further include those who are not work- ing but have given up actively seeking work due to their conviction that no work is avail- able. (Neither method counts those women specializing in housework, many of whom may have done so after failing to find jobs.) If the first definition is used, then some 19 percent of the potential labor force was unemployed (actively seeking work); on the sec- ond definition, the unemployment rate (including those who have given up seeking work) was 31 percent; and in both cases, the unemployed numbers split 3:1 between males and females (most of the latter having become housework specialists). As there seem to be far less jobs opening up than active job seekers (see below), those who have given up seem realists rather than slackers, and the second definition seems to best reflect the potentially- available labor supply. 64. Unemployment is especially a problem among the indigent and poor, youth in gen- eral and young women in particular. Depending on definition, unemployment rates are two or three times as high among the indigent as among the non-poor (Table 13). Youth Poverty Line Classification Poverty Group Indigent Poor Non-Poor Total Age Group' 10 20 30 50 10 20 30 50 10 20 30 50 10 20 30 50 All TOTAL (%) 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 WORKING 12.1 39.9 36.8 31.5 13.6 39.3 49.5 39.3 15.1 50.5 68.2 41.6 13.1 42.6 50.7 38.0 33.3 NOT WORKING2 87.9 60.1 63.2 68.5 86.4 60.7 50.5 60.7 84.9 49.5 31.8 58.4 86.9 57.4 49.3 62.0 66.7 Searching 7.3 15.2 13.9 7.1 4.7 17.0 8.5 4.1 1.4 7.4 7.5 1.1 5.4 13.6 10.1 3.8 7.7 Not Searching2 80.6 44.9 49.3 61.4 81.7 43.8 42.0 56.6 83.5 42.1 24.3 57.3 81.5 43.8 39.1 58.2 59.0 Student 57.02.9 0.0 0.0 63.8 4.5 0.0 0.0 70.5 10.5 0.0 0.0 61.75.5 0.0 0.0 22.8 Housework 12.9 25.4 32.8 22.0 10.6 25.9 30.3 24.1 10.1 24.2 20.8 29.2 11.7 25.2 28.3 25.6 21.3 Retired 0.0 0.0 0.0 17.3 0.0 0.0 0.5 166 0.0 0.0 0.0 17.4 0.0 0.0 0.2 17.1 3. 7 Old/Sick 0.5 2.2 2.5 9.4 0.9 0.0 2.7 7.6 0.0 3.2 0.6 6.2 0.5 1.7 2.0 7.6 2.6 Doesn't Exist .9 13.8 11.9 11.04.3 12.58.0 5.5 2.2 3.2 2.9 3.9 62 10.4 7.8 64 7.4 Other 1.3 0.7 2.0 1.6 2.1 0.9 0.5 2.8 0.7 1.1 0.0 0.6 1.5 0.9 0.9 1.6 1.2 nemployment (1)3 7.6 27.6 27.4 18.4 25.7 30.2 14.7 9.4 8.5 12.8 9.9 2.6 29.2 24.2 16.6 9.1 18.8 nemployment (2)4 7.2 51.0 41.2 36.5 9.8 42.9 25.0 19.6 19.3 17.3 13.2 10.7 47.0 36.0 26.1 21.2 31.2 Age groups 10=10-20 years; 20=20-30 years; 30: 30-40 years; 50= 50 and older. 2 Sum of rows in box immediately below. Excluding individuals who claim work does not exist 4 Including individuals who claim work does not exist Table 13. Employment, Job Search, and Reasons by Poverty Class and Age Group (%) 30 Argentina Poverty Assessment Background Paper No. 6 tend to fall disproportionately into the categories of those seeking work or those who be- lieve no work exists. The unemployment rate, on the broadest definition, is 47 percent (males 41 percent, females 56 percent) in the 10-20 age group; it falls to around 36 per- cent in the 20-30 age group, and 26 percent in the 40-50 age group. Women in the labor force in the age range 10-30 were less likely to report sufficient work than their population proportions (56 percent of males, 43 percent of females aged 10-20; 66 percent of males, 56 percent of females aged 20-30); conversely, a higher percentage of women were seek- ing work or had concluded that there was no work available. Ominously, 86 percent of new jobholders (over the previous year) were young males. Contrary to supposition, women are less, rather than more, likely to seek seasonal work, as opposed to permanent jobs. In fact, those looking either for agricultural jobs or seasonal work tended to be ei- ther the youngest or the oldest, whereas those in their prime tended to look for non- agricultural work (though the great majority would take anything they were offered). 65. The survey included information on other labor market variables, including the supply of labor, going market wage rates, and the reservation wage of each potential worker (RW; minimum wage required by the worker if he/she is to accept new work, for all individuals actively seeking new jobs). One would like to be able to use this data to judge the extent of imperfection in the labor market, as a combination of high unemploy- ment rate and going wage rates (Table 14) substantially exceeding the RW of job-seekers might be indicative of a labor market which cleared through rationing rather than price, and would raise the question of what prevents the wage rate from being bid down and clearing the market. However, there are several obstacles to such analysis which made it impractical. 66. First, the usual type of analysis23 assumes that labor market participation (seeking or possessing a job) depends for each individual on the relationship between a market wage and the individual's reservation wage. However, the relevant market wage (which depends on individual human capital as well Average Monthiy Agricultural Non- Total as local labor market Wage Rate agricultural characteristics which Permanent 311 395 350 determine the reward Seasonal 213 209 212 to human capital) is Table 14. Actual Wages, Misiones and Salta known only for em- ployed wage laborers (Table 14) and a reservation wage is known only for active job seekers (the reservation wage for sample adults engaged in active job search averaged about $240 per month, about $28 above wages of seasonal labor and $110 below those of full-time workers). Econometric analysis based on these data would face sample selectiv- ity bias, which is usually overcome through a multi-stage estimation procedure.24 23 E.g., as described in Killingsworth and Heckman (1986) and practiced from comparable Chilean survey data in . Gill (Psacharopoulos and Tzannatos, 199?). 24 This procedure in fact makes no use of the reservation wage data, since the relationship between market and reser- vation wage may systematically differ as between participants and non-participants in the labor market. 31 Argentina Poverty Assessment Background Paper No. 6 67. It is possible to perform the first stage of this procedure by using probit or logit analysis to estimate a participation function, where participation is defined as being em- ployed for wages and where influences hypothetically determining both the market and reservation wage for each individual replace the latter as explanatory variables. For rural areas, however, the concept of participation in the labor market is especially ambiguous: should those actively engaged in own-account farming or self-employment be considered "non-participants" on the same terms as the housewife or the unemployed? Shouldn't those seeking work but not currently employed also be considered "participants? These issues suggest that the traditional binary choice model for participation is inappropriate in the rural setting, or must be redefined to accomodate a richer set of choices. Finally, in the second stage of the estimation procedure, where the market wage and wage labor sup- ply are the dependent variables, the subsample of rural "participants" for which the neces- sary data are available is a small fraction of the total sample--in this case, only forty obser- vations for females, too few to obtain good statistical estimates. 68. Employment Opportunities. There are three options for those who would not serve as part of the unpaid farm or household labor force: migration, local wage labor, or non-farm self-employment. Each of these will be examined separately below, but the sa- liant facts are that migration has been absorbing about 35 persons per year from among the sampled households, representing only 3.3 percent of the potential labor force or 22 percent of the current number of active job seekers. Local jobs for new entrants obtained in the previous year, 23 in number, absorbed 2 percent of the potential labor force or 14.7 percent of current active job seekers. Net increases in self-employment absorbed another 11 persons. Thus less than half of active job seekers found jobs in the previous year. 69. Thirty percent of households have seen an average of two members, almost all sons or daughters, leave for other parts over the last five years. This migration was, in the main, to find work (54 percent of men, 44 percent of women, the difference repre- senting the more frequent departure of women for matrimonial reasons). The frequency of migration was inversely related to poverty status (77, 62, and 45 ex-members per hundred households migrated among indigent, poor, and non-poor households respectively). However, the determining factor was more likely family size: the correlation between number of migrants and number of adults (including past migrants) is 0.85, whether or not per capita consumption levels are controlled for, whereas controlling for number of adults reverses the apparent the poverty effect. Of those looking for work, 77 percent moved to urban areas, the majority of these to towns within the same province, secondarily to the federal capitol and surrounding area, and least to other provinces. Although movement to elsewhere in the rural areas accounted for one-third of all migrations, this was most fre- quently for marital or familial reasons, and disproportionately by females. Migration was more frequent in Salta than in Misiones, where daughters departed much less frequently than sons. 70. Controlling the rate of outflow of migrants to the cities may be viewed as a pri- mary rationale for bettering the situation of the rural poor. From another perspective, 32 Argentina Poverty Assessment Background Paper No. 6 outmigration from poor families or poor areas, by reducing the burden on the natural re- source base, may be viewed as part of the solution to rural poverty. Either way, there are reasons to further milk the survey data for insights into the push and pull factors which cause migration. Unfortunately there is no information on the individual characteristics of the migrants, the pull factors which attracted them to other areas, the wages they receive, or their success in finding jobs after migration, so most observations relate to the supply side, where the salient variable is clearly the number of adults in the family. 71. However, one benefit to the household from migration--remittances--is known from the survey. By relating households with migrants to remittances received25, it was observed that average remittances received from households with migrants were inversely proportional to poverty status ($22, 63, and 107 respectively for indigent, poor, and non- poor households). Furthermore, (a) no remittances were reported from the majority of migrants in all categories; (b) those who migrated in order to seek work accounted for a disproportionate share of remittances; (c) the mean remittance received from male mi- grants ($109 per year) was higher than that from females ($44); and (d) classified by des- tination of migration, mean remittances were highest from those who sought work in greater Buenos Aires ($138), followed by local towns ($120) and towns in other prov- inces ($51), with almost nothing coming from migrants to other rural areas (which were mainly for non-work reasons). If remittances were proportionate to the migrant's savings over cost of living, then the small monetary advantage of Buenos Aires relative to local towns surely didn't compensate for extra travel cost, and this may help explain the limited number of migrants who chose the capitol area as their destination. 72. Non-farm self-employment is a relatively rare occupational category in the sam- ple, but its frequency among members of poor and indigent households is only slightly less than the population proportion. This employment category is often considered an expres- sion of entrepreneurship, an opportunity for individuals or families to substantially better th e ir liv e lih o o d , a n d a p o te n tia l N o . o f e n te r r is e s 1 0 3 2 stimulus to local economic de- velopment. Although the high Avg. no. of employees 0 1 15 lvelsofent slmoymet hi- Avg. monthly net in- $176 $400 $3000 levels of self-employment mn- come found among the highest come quintile appear consistent with Table 15. Non-Farm Self-Employment these notions, the sample sta- tistics (Table 15) support a less optimistic interpretation: for most individuals and for all of the poor, self-employment is an option far inferior in prospects to working as a seasonal farm laborer. Of the non-farm self-employed, over 97 percent (all poor) are individuals with no employees (divided fairly evenly among product production, services, and mer- cantile activities). The average earnings to the principals fall below those of seasonal farm laborers, indeed, below the average reservation wage. The few enterprises--all from non- 25 With the caviat that survey informants were asked about remittances as the last question in a long questionnaire. Remittances are also received from relatives who were not migrants. 33 Argentina Poverty Assessment Background Paper No. 6 poor households--which are successful enough to hire one or more employees better fit the notions of entrepreneurship as a generator of higher income. 73. Food Security. The great majority of rural households in all quintiles (91 percent for the lowest, 80 percent for the highest) spend more money on food than the income they receive from agriculture. This is obvious for the half of rural households that do not engage in agricultural activities as own-account farmers, but it holds true for farmers as well: net food sellers only represent 40 percent of farmers, and are in the majority only among farmers with over one hectare of land (45 percent of all farmers). It also cuts across household income groups: even two-thirds of households with incomes exceeding $5,000 per year (the top 29 percent) are net buyers of food. The conclusion is thus straightforward: uniformly lower food prices bring a net gain to rural inhabitants, whatever the quintile or income group; and only a majority of larger farmers (a small minority of the rural population) benefit from higher food prices (assuming retail and farmgate prices move together). 74. The average annual household expenditure on food is almost constant across the quintiles of per capita expenditure at about $2,600, whereas the average family size de- clines from 7.88 for the lowest quintile to 2.96 for the highest. As the Buenos Aires food basket supplying minimum daily requirements costs on average about $600 per capita in these provinces, only the top two quintiles meet that requirement. More precisely, actual food expenditure falls below the local cost of the Buenos Aires minimum standard in the case of 63 percent of the rural population (57 percent in Misiones, 68 percent in Salta). Even considering that substitution of relatively cheaper local foods presumably reduces the percentage who do not obtain minimum caloric requirements (the basis for estimating ca- loric values of local food consumption is lacking), this is evidence of a serious household food security problem for the rural poor in a country where food is relatively abundant and cheap. 75. Social Infrastructure. The survey captured two sorts of information on rural infrastruc- ture: the presence or absence of (mainly public) infrastructure at the community level (only in the urban core of each departamento), and measures of services delivered by social infrastructure to each household. On the former, it might be supposed that the absence of adequate social infra- structure would correlate with poverty levels, as such correlation is marked in most countries. However, insofar as can be judged from variations in infrastructure within Misiones and Salta (Table 16)26, the geographic distribution of social infrastructure (at the level of rural towns and in quantitative terms) is rather equitable, and, in the case of Government programs, even favorable to departments with a high concentration of rural poor. 26 Comparisons with other areas of Argentina are unavailable, and might lead to quite different conclusions. 34 Argentina Poverty Assessment Background Paper No. 6 76. However, such corn- Presence of Various Types of Social Infrastructure in parisons have no direct Poor and Non-Poor Departments bearing on the extent of ac- cess by the rural poor to the Percent Present services or infrastructure in Infrastructure Tye Poor Non-Poor the communities in which Paved Streets 37 27 the poor reside.27 Extensive All-Weather Roads 90 91 data were collected on Primary School 100 100 household access to services Secondary School 79 80 in health, education, water Bank Office 11 9 and sanitation. Cross- Telephone Lines 84 91 tabulations of these data by Public Telephone 79 91 quintile status reveal the Public Transport 95 100 following differences in Electric Power System 84 82 service access (in which Q1, Potable Water System 19 11 referred to as the "poorest", Sewage System 47 36 is contrasted with Q5, as the Hospital 32 30 "richest" households): in Pharmacy 26 40 health, the poorest are Health Center 67 80 about as healthy as the rich, Agro-technical Assistance Office 21 18 in terms of illness incidence Cooperative(s) 32 64 per capita, but the poorest Government Credit Programs 16 18 households have some three Government Tractor Service 26 0 times as many medical con- Government Seed Supply 42 27 sultations as the richest, due Government Transport Service 26 0 to larger average household 1 Poor department defined as having size. The poorest more fre- more than 60% rural poor on con- quently consult with agentes sumption criteria sanitarios and less often Table 16. Social Infrastructure: Misiones and Salta with doctors, compared to the rich. The poorest also tend to seek out public hospitals, health centers, or "other" facilities, whereas the richest go to private hospitals, clinics, or doctors' offices. The poorest rarely pay for consulta- tions or hospital stays, while the richest pay part or all of the costs. Most of the poorest have no medical insurance, whereas a slight majority of the richest do. Forty to fifty per- cent of all households pay all costs of medication; of the remainder, the poorest tend to pay nothing, and the richest part of costs. 27 Unfortunately all questions in the household surveys relating to the household's access to social infrastructure were moved to a separate questionaire, which collected information on town or village characteristics within some, but not all, of the survey departments. No information was collected which specifies which town or vil- lage is closest to each household, so the impact if any of varying completeness of social infrastructure on house- hold-level poverty cannot be assessed from the survey. The most that can be done is to relate a measure of the degree of concentration of rural poverty in each department to the state of social infrastructure in that depart- ment, as shown in the table. 35 Argentina Poverty Assessment Background Paper No. 6 77. In educational facilities, the recorded distinctions between rural poor and rich are not very sharp: the poorest are more likely to have attended schools with multi-grade classes and never went to private schools (as compared with about 10 percent in the top two quintiles). Overall most students of all classes walked to local schools, though a sub- stantial minority of the richest used public or private vehicular transport to reach school. 78. In sanitory facilities, nearly half of the richest but only 8.5 percent of the poorest had indoor plumbing, and the rest had to bring water from outside. Similarly, 39 percent of the richest but only four percent of the poorest had flush toilets, as opposed to out- houses, and 25 percent of the richest but five percent of the poorest had sewage systems or septic tanks. Similar comparisons held for fuel and lighting: 27 percent of the richest but seven percent of the poorest had'gas cooking fuel, the remainder depending mainly on wood fuel. And 73 percent of the richest but 33 percent of the poorest had electric light- ing, the remainder mainly using kerosene lamps. 5. POTENTIAL FOR AGRICULTURAL PRODUCTIVITY INCREASE 79. The rural poor in Argentine derive only a fraction of their income from agricultural production, relying on off-farm activities for the remainder. Rural poverty can be ad- dressed at the household level either by increasing off-farm earnings or raising farm pro- ductivity. Given limitations of external demand for labor, on which the survey shed little light, education seems the most obvious route to higher individual off-farm earnings; how- ever, the marginal returns to post-primary education appear low. As for agricultural pro- ductivity, the survey indicates considerable variance among farms. If the causes of this variability could be isolated, it might help focus a strategy for addressing rural poverty. 80. Land and Tenure. Because land is the fundamental productive resource of farm families, it is useful--and most revealling--to show incomes cross-tabulated between farm size and quintile status (Table 17). Note that the first four size ranges and the last two taken together contain roughly equal numbers of farms; and while there is some concen- tration of numbers of farms along the diagonal, at least a few percent are found in every cell.28 Among the surpises is how small a proportion of farm area is actually cultivated (i.e., in crops) and how small the variation in total income per household is among the cells, especially those representing farms of radically different scale; and that the poorest and smallest farms still match or exceed household incomes on farms ten or hundreds of 28 Thus there are even four farms in Q1 with more than 500 hectares each of farm area--and family sizes averaging 9.5! 36 Argentina Poverty Assessment Background Paper No. 6 Annual Household Income Farm Crop income in- Farm Area (hec- Ql Q2 Q3 Q4 Q5 Total /ha come tares) /crop ha 0 - 3 3,258 3,57 3,32 2,796 2,029 3,150 358 2,724 2 0 3 -15 3,167 3,46 3,89 6,276 3,592 4,254 255 1,989 5 1 15-25 3,068 4,02 2,61 3,963 7,578 4,743 126 1,934 8 4 25-80 3,245 5,73 6,08 2,785 9,163 6,557 87 2,828 5 1 80-500 2,614 8,84 5,73 4,099 12,158 7,212 17 2,001 5 3 > 500 2,249 -200 4,55 3,278 4,862 3,712 1 512 7 Total 3,077 4,16 4,14 4,187 7,343 4,739 177 2,285 8 1 Crop Ha's/capita 0.2 0.6 0.8 1.3 3.3 1.4 Land 24.5 23.1 75.7 70.7 195.5 87.9 Ha's/capita I Table 17. Income by Quintile and Farm Size, Misiones and Salta times larger. This is so for several reasons: (a) the poorest families typically are very large, with excess family labor to pour into both intensified farming and non-farm activi- ties which supplement farm income; (b) larger farms either lack sufficient labor and/or capital to specialize in crop cultivation or possess land suitable mainly to pasture, hence concentrate on extensive forms of livestock raising. The results are seen in the last two columns: an extraordinary decline in productivity per hectare as farm size increases, due to a lesser decline in crop yields combined with a switch at larger farm sizes to livestock raising (which may have been particularly unprofitable in the year preceding the survey) or simply a failure to cultivate much of their land. These findings alone should negate any argument that small farms are uneconomic and large farms superior, at least in the sense of obtaining the most return from land as a natural resource. That is not to say that the smallest farms are viable in the sense of providing a level of living above the poverty line without extreme reliance on non-farm sources of income. However, some 20 percent of farms surveyed, with areas in the 3-25 hectare range and smallish household sizes, are surely viable, in that these households fall in the Q4-Q5 range, and are still more efficient than larger farms. 37 Argentina Poverty Assessment Background Paper No. 6 81. In Argentina, Mean Area of Farms (< 50% Livestock Gross Product) (hec- most farms have un- clouded proprietary tares) rights, though not Qitl necetssai titlghi n Tenure Status Q1 Q2 Q3 Q4 Q5 To- necessarily title. Thista is true of "farmers" in Misiones and Salta-- Owned w/ title 1.8 5.8 5.8 8.1 55.8 19.6 defined as those for Owned w/o title 2.1 12.9 6.1 7.9 11.1 8.5 whom livestock ac- Rented 0.8 0.6 3.2 0.1 1.4 1.2 count for less than 50 Occupied (gov't) 0.1 3.1 1.4 1.1 1.0 1.3 percent of gross prod- Occupied (pri- 0.1 0.9 0.1 12.8 0.0 3.5 uct--insofar as most vate) farmers have owner- Other 0.5 1.1 1.4 1.0 0.1 0.8 ship rights to their Total 5.4 24.4 18.0 31.0 69.4 34.9 land, there is little ten- Mean Area of Ranches (> 50% Livestock Gross Product) (hec- ancy, and "squatting" tares) on government- or Quintile private-owned land is Tenure Status Q1 Q2 Q3 Q4 Q5 To- relatively infrequent tal (Table 18). However, a (ablnghe 8. oo r, Owned w/ title 0.8 0.3 0.0 2.0 192. 45.9 among the three poor-5 est quintiles, a major- Owned w/o title 0.0 0.0 0.0 0.0 0.0 0.0 itY Of land owned is Owew/til u u 0. 00 00 00 utit ld oed is Rented 0.0 0.0 0.0 0.0 5.0 1.2 ufante he stalion Occupied (gov't) 0.0 37.5 50.0 0.0 0.0 14.7 Povnche, al quineSaita Occupied (pri- 75.0 100.3 0.0 0.0 25.0 47.1 Province, is quite dif- ferent: except for the vate) richest (highest qui- Other 0.0 0.0 0.0 166.7 0.0 29.4 tile), ranch land is al- Total 75.8 138.1 50.0 168.7 222. 138. most all "occupied" 5 3 (squatted on, without Table 18. Farm Area by Tenure Status and Quintile contractual rights), whether it is government or privately-owned. As will be shown below, land titling is asso- ciated with higher investment in fixed capital per hectare, and ranching with extensive cul- tivation and low yields on cultivated land. In addition, squatters, who are all below the poverty line, do not grow trees or perennial fruit and put much less area into other peren- nials on average. They are, however, more diversified and more labor-intensive in crop cultivation, and just as likely to use fertilizer and other current inputs, obtaining as a result levels of crop productivity not inferior to other farmers. 82. Crop Diversification. The choice of crops to grow is a key decision for the farm unit, influenced by land characteristics, climate and environment, farm resources (finance, labor, equipment), and marketing prospects, including expected prices. Do these influ- ences leave enough freedom of choice so that a farmer can, by choice of crop mix, im- prove net income or reduce risk? Are changes in cropping patterns any part of a house- 38 Argentina Poverty Assessment Background Paper No. 6 hold strategy for alleviation of or escape from poverty? Do "diversification" and "spe- cialization" have any roles in such a process? (Refer to the implications of the discussion of the opening of Argentina to international trade and of the thrusts of existing programs for the small farmer.) 83. One way of looking at these questions is to compare the crop choices of the poor and non-poor, on the implicit assumption that the extent of poverty is the dependent vari- able. For example, Table 19 shows a cross-tabulation of the average percentage area in each major crop type against the quintile status of (over 200) households, separated by province to partially control for major differences in land, climate, and market. Amidst the "noise" represented by small proportions of farm area planted to a wide variety of crops which betray no systematic relationship with income status, there are several general pat- terns evident. One is that the poorest quintile stands out in terms of an especially signifi- cant proportion of crop area planted to corn (or corn + beans in Misiones, corn + other cereals in Salta), presumably for subsistence purposes. Second, households at all income levels have two or at most three cash crops which occupy the bulk (over 90 percent) of their acreage. These are basically tobacco and tea or mate in Misiones, and a more diverse mixture in Salta--two or three crops for each quintile group, but differing considerably by quintile, with beans, squash, and tobacco being most important overall. The proportion of area in cash crops increases with household income level, although it is high even for the poorest quintile (80 percent or more in Misiones, nearly 50 percent in Salta). Finally, in Misiones there is a strong tendency for farms heavily focussed on tobacco--and entirely dependent on the tobacco companies for credit, inputs, and TA--to be poorer than those focussed on tea or mate (though in Salta tobacco is the choice of middle-income farmers). 39 Argentina Poverty Assessment ZBackground Paper No. 6 Table 19. Crop Area by Crop Type and Quintile Missiones Salta Crop Quantile of Household Quintile of Household Area Percent 1 2 3 4 5 Total 1 2 3 4 5 Total Corn 4-. 222 24 )3 16 190 5 3 24~ 2 3 C 5 Cereals 24 1 7 0 5 05 ' 0)3 4 18.F 81 116 E ' 4 7 Fodders 0 0 163 11 16 2 6 F ~ 4 Potatoes 1 7 1 1 C7 07 1 1 2~ E 79 1 1A 4 Cassava 10.9 9.8 8.7 7.6 5.1 7 Oils/oils-seeds I' r 0 17 1 3 1 2 C' Beans 1 14 0 2 11 r6 04 13 _j 16 Squash C' 1 2 1 4 C. 1 04 06 1 9 122 30 0, I 7,- 1,34 Vegetables, leaf 0 1 0.3 1.1 1.1 0.4 0.6 0.5 06 10.5 1.7 Vegetables,.root 02 0 7 0'7 0 4 r, 0 2 03 02 r3 4 06 Vegetables. other 056 C 03 ' C E s 1'0i7 14 46 '36 .' 0 23 Spicesicondiments 0,4 0 1 101 1 13 128 27? 4 .s Tobacco 124 17 5 7 6 11 4 6 32318 ' 23 Tea &Mate 11I 350 420C 4E66 666 526 00 Fruit, trees 08 0 3 0. C'- C 1S 0 - 04 Other fruit 0.1 23 0.5 1.9 1.3 0.1 7.0 1.3 Other crops 25 4.5 50 A15 2.4 2.4 5.4 2.8 0.1 0.9 Codes missing 0.5 0.4 11 0.5 0.6 0.3 28 0.7 Total lOu 0c 00IO0 10300 1000 1000 1001 1000 I100: O0 1000 100 00 C00 01 Net Productivity (WhectaFe) 3,152 3.99 2.349 1.984 2.220 2,5611 1,222 1,617 2,807 2,361 681 1,8 78 Note: Shaded areas highest producivity witnin block Boldfacedl=nighi value-added. italicized=low value added crops Table 19. Mean Percentage Distribution of crop Area by Crop Type and Quintile, Missiones and Salta Table 20. Crop Area by Crop Type and Productivity Range Missiones Salta Crop Quantile of Household .Quintile of Household Area Percent <$250 $250-1000 $1000-1750 $1750-3500 >$3600 Total <$250 $250-1000 $1000-1760 $1750-35GO >$3600 Total Corn 7 7 123 220 25 2 29') 190 662 58,: 39 2,69c 20)9 24 1 Cereals 01 011 01: 11 0 F 10 ~ 3E 4 6 Z2 8 47 Fodders 0 0 16r 0 1 E 2e8 A. - Potatoes ri IC 1 0.4 34 11 43 7 9 i8 F.. I 6 24 Cassava 1.0 3.8 10 7 9.4 10.4 7.0 Oils/olls-secas 0 7 1 2 9 1 4 1 1 0)2 Beans 1 4 22 3' 20 , 6 26 03 S ir Squash 03a 0 0 2 4 09, Z'4 A a 36 E. 4 CO C 184 Vegetables leaf 0.9 1.3 0 2 1.1 0.6 0.9 1.7 Vegetables.ro F01 04 1 3~ 0 0 2 t Vegetables ot-er 02, rjl 1 3 G 41 w 1 Spicesicond,ments 0,1 0 3 ') 1 1 7 0 1 142 Z4 ~ E. 42 Tobacco -,7 11 3 80 14 6 1-.6 93 1 2 Tea &Mate 6 7 641ri 471 .12 C 2') - 13 0 0 Fruit, trees 1. 0'0 02 02 4 47 C) Othler fruit 0.1 0.5 11.8 18a 0 2 1.3 Other cropts 0.6 0.4 1.0 8 6 2.4 4.1 3.4 2.3 0.9 Codes missing 0.6 0.9 0.1 16 0.6 90a 0.5 0.7 Total 1100 100O 1000 10lci 0 10')0 100 100 C' 100 C. 1000 1)))100 1)0'ci ' Note: Shaded areas: highest productivity within block. Boldfaced=high value-added; italicized=low value added crops. Table 20. Mean Percentage Distribution of crop Area by Crop Type and Productivity Range ($/hectare), Missiones and Salts 40 Argentina Poverty Assessment Background Paper No. 6 84. Taking the number of crops grown per household as a crude index of diversifica- tion, and the percentage area per household of the most important crop as an equally crude index of specialization, similar cross-tabulations by quintile indicate that there is a direct U-shaped relationship between specialization and quintile status, or inverted U- shaped relationship with diversification. That is, both the richest and the poorest house- holds tend to be somewhat (20-25 percent) more specialized or less diversified than mid- dle households. There is a similar tendency of both the largest and smallest farms to be more specialized (less diversified) than farms in between (e.g., 65 percent of farms over 100 ha, 44 percent of farms under three hectares, but only 17 percent of other farms planted only 1-2 crops). 85. What differences do these differences in cropping patterns make to farm produc- tivity? In fact, gross productivity (crop yields, in value terms) does not vary systematically among quintiles. Nor does net productivity (net value of production per crop hectare) show any consistent relationship with quintile status: the highest net productivity in Misiones is among the two poorest quintiles, while it is found among the middle quintile in Salta (see shaded areas in Table 19). Both gross and net productivity are, however, closely related to diversity and specialization: the log of gross crop productivity is strongly positively correlated with the diversity index (r = +0.58), and the 20 percent of farms with the highest net productivity are, in Misiones, more than twice as diverse in their cropping as the bottom 20 percent (the ratio is 1.7 in Salta). A negative relationship with the de- gree of specialization is also observed. The problem is that the relationship between household income per capita and net farm productivity is obscured in cross-sectional data by the correlations of farm size and off-farm incomes with quintile status, and a strong negative relationship between farm size and crop productivity. That is, the rich tend to have larger farms (more often livestock-oriented) and receive more off-farm income than the poor, leading presumably to decreased managerial interest in crop production and lower gross and net crop income per hectare. 86. Does the data suggest what cropping patterns have the highest net productivity, hence contribute most to income? Regression analysis relating net productivity per hec- tare to the proportion of crop area in each category of crop suggests that some crops in these provinces offer generally higher value-added per hectare than others, though crop choice per se accounts for at most 19 percent of variation in productivity. Relative to corn, certain crops--cassava, leaf vegetables, condiments, other fruit, and miscellaneous crops classed as "other" or "codes missing"--appear to offer higher value added per hec- tare (boldfaced below), whereas root vegetables, oilseeds, fodder, tea/mate, beans, pota- toes and squash are on the low side (italicized below). As shown in Table 20, in Misiones, farmers with the lowest net productivity are heavily specialized (2.5 different crops per farm on average) in tea/mate (87 percent of crop area), accompanied by corn (8 percent), beans, cassava, and tobacco (1-3 percent each); whereas those with the highest produc- tivity are much more diversified (5.5 crops per farm), the leading crops being corn (aver- age of 29 percent of crop area), tea/mate (21 percent), cassava and tobacco (10-14 per- cent), and a small percent of area in most other crop categories. In Salta, the contrast is between low productivity farms with 2.0 crops per farm, emphasizing corn (66 percent), 41 Argentina Poverty Assessment Background Paper No. 6 fodder crops (16 percent), potatoes, other vegetables and "other crops" (4-5 percent each; and high productivity farms (3.5 crops per farm) with corn, other cereals and fod- ders in roughly similar proportions (21-27 percent each), condiments and tobacco (8 percent each), and a variety of minor crops. 87. Overall, it is apparent that reduction of area in low value-added crops is an impor- tant part of the explanation of productivity differences in Misiones, but not nearly so in Salta. In Misiones, also, diversification into fruit, vegetables, and miscellaneous minor crops, albeit on a minor part of crop area, plays a role in productivity increase. Indeed, as the poorest farmers in Misiones already have high average crop productivity and are not overly-specialized in low value-added crops, the growing of high value-added minor crops on a relatively small part of their cropping area may be all that crop choice in itself can contribute to their productivity (compare the lassoed figures in Tables 19 and 20). 88. But does a higher productivity crop mix result in higher total household incomes? If diversification and selection of higher value-added crops entail greater on-farm labor requirements at the expense of a sacrifice of off-farm income, it might not be worthwhile for the household. However, tabulations of the survey data indicate that: (a) overall the proportion of income derived from non-farm activities only declines from 50 percent to 43 percent as crop productivity increases from less than $250 to over $3,500 per hectare, implying that farm and off-farm income are positively correlated (possibly because off-farm income. provides the working capital neces- sary for intensification); (b) there is no consistent relationship between reported hours of farm work per household and net crop productivity, in either Misiones or Salta; however, (c) wage income in Misiones tends to increase, but in Salta to decrease, in absolute terms and as a percentage of total net income, between lowest and highest crop productivity classes. In Salta, it seems that households with substantial off-farm income tend to manage their farms as ranches (with small area in crops and low crop productivity), and this creates the illusion of a tradeoff between crop productivity and off-farm income. Therefore, overall the survey data suggests that diversification out of lower value-added crops is a part of the solution to the poverty of farmers; but also that off-farm income may help make this pos- sible. 89. Farm Productivity. One element of a strategy for addressing rural poverty is to identify those among the poor with substantial potential for increasing farm productivity and provide them with access to the technical assistance and other resources required to realize that potential. But how can we identify those with such potential, and what they require to realize it? Working from information provided by a sample of farmers, it may be possible to isolate factors associated with already-realized higher productivity, and infer 42 Argentina Poverty Assessment Background Paper No. 6 answers to both questions. The model and estimation techniques employed and some of the more interesting results are described in Annex A. A diagram of the model is included in the text (Figure 7) for reference. The following is a non-techical interpretation of the results of the analysis of the model. 90. General Findings. There are two distinct types of agricultural household, ranchers and farmers. Households with large areas of land, especially in Salta, tend to become ranchers, as do secondary school graduates. Although ranchers usually also grow crops, techniques are less intensive and crop yields much lower than those of farmers. The asso- ciation of ranching with secondary school graduates suggests that individuals may become ranchers for the social prestige rather than income and, as most secondary graduates have lucrative non-farm jobs or enterprises to rely on, that crop cultivation isn't managed to maximize income in such cases.29 It is also likely that an unmeasured or latent variable, land quality, is inversely associated with both ranching and crop yields, hence helps ac- count for the negative relationship between ranching and productivity. 91. Management ability (a latent, or hidden, variable in the model) is apparently re- flected in aggressive pursuit of advantage through access to technical assistance, partici- pation in coops and farmers associations, and attendance at meetings of the same. Good farm managers tend to have completed a primary (but not secondary) education; to have borrowed in the past and have savings now. Good managers tend toward diversity in cropping rather than specialization, use higher levels of current inputs, and thereby obtain higher crop yields for any given level of inputs. As a result, most are able to support their households without relying heavily on off-farm income. 92. Fully half of the variance in receipt of technical assistance among farm households is accounted for by its pursuit by good farm managers (more detail on this below). Re- ceipt of TA in turn influences the decision to use fertilizer and other inputs (along with completion of primary education and being a farmer rather than rancher). However, most of the variance in the decision to use fertilizer is not explained by these influences. Given the decision to use fertilizer, availability of liquid assets is the main influence on the amount of current inputs purchased (with an elasticity of 1.49), and most of the variance in the latter can be accounted for.30 93. Possession of land title appears to be strongly associated with fixed capital invest- ment, but not current input use (due to the disfunctional credit market). However, levels of fixed capital use are not among the significant determinants of crop yields. Nor is household labor, which as used in crop cultivation, appears to be in surplus (marginal product in crop cultivation not significantly different from zero). It is presumed that this is 29 There are only five secondary school graduates in the sample of farmers. Three of these are ranchers and lose money at it. Of the two remaining, one has much higher and the other much lower crop productivity than the average for other farmers. 30 Note that all of the above statements implying a direction of causation are interpretations of the author. The possi- bility that the direction is the reverse of that stated should be considered. 43 Argentina Poverty Assessment Background Paper No. 6 due to a labor market which rations limited numbers of off-farm jobs. Among the "nor- mal" factors of production, the scale of crop area contributes negatively to yields, ceteris paribus--small farms have a distinct land productivity advantage, ceteris paribus. Level of use of current inputs, on the other hand, has a significant positive impact on yields. 94. Additional findings were that existence of a local coop or farmer's association of- fice or branch has a positive impact on membership, but that the act of joining either of these and especially the receipt of TA are also influenced by management ability of the farmer. Indeed, memberships, meeting attendance, savings, and past borrowings are col- lectively indicators of management ability. An (endogenous) outcome of good manage- ment, but also a useful indicator, is the relatively low dependence of the better managers on non-farm sources of income. Able managers tend to receive TA, because they seek it out and/or the technicians target them. 95. To fur- the dmontre Percent of Farms Receiving Technical Assistance by Management ther demonstrate CabiftCls the last point, it Cabntyls Quintile is possible to Mg Capability Q1 Q2 Q3 Q4 Q5 Total estimate the in Low 0 5 0 3 0 2 dices of the la- Medium 75 71 36 50 36 50 tent manage- High 100 100 100 100 100 100 ment ability of Total 21 29 18 26 31 25 each farm unit from a set of factor loadings Table 21. Management Capability and TA Receipt computed from the solution matrices (analogous to solving the reduced form equations for Mi as an unknown). The values tend to cluster into three ranges, which we arbitrarily call "Low", "Medium", and "High" management ability, which account for 65, 23, and 12 per- cent of farmers respectively. Table 21 indicates how management ability, thus defined, is distributed over households in various quintile groups, and also the percentage of house- holds in each cell which received agricultural TA. Technical assistance appears very well targeted, inasmuch as all fanners with "high" management ability, half of those with "me- dium" ability, and almost none of those with "low" ability received TA. This is not so re- markable, considering all the visible indicators of management ability (as described above). Also neither quintile status nor farm size (not shown) are factors predisposing toward re- ceipt of TA when management abilities are taken into account. Whether technical assis- tance, as currently constituted, and from whatever source, is a key requirement for pro- ductivity increase in farming is moot, however--as seen above, its main role seems to have been to get farmers, who otherwise have the necessary liquid assets, to purchase commer- cial inputs. 44 Argentina Poverty Assessment Background Paper No. 6 Towards a Strategy for Addressing Rural Poverty 96. Before turning to questions of anti-poverty strategy, it may be helpful to summa- rize what have been found to be the main differences between the poor (including the indi- gent) and the non-poor (alternativAy, characteristics associated with low per capita in- come/expenditure levels). Compared with the non-poor, pool rural households in Misio- nes and Salta: * have larger families, the difference being mostly children * are not at all subsistence-oriented in their production; indeed, buy a higher percent of their food * derive a smaller proportion of their income from farm production * are slightly less likely to be farmers than wage workers, self-employed, or unem- ployed * (if adults) are more likely to be illiterate or to have dropped out of school earlier, and obtain lower marginal gains in income per year of schooling, especially for non-agricultural jobs, probably due to lower quality educational facilities * (if children of indigent households) represent the majority of drop-outs before sec- ondary school * are less likely to have received individual training or technical assistance or to re- ceive it through group meetings * (if indigent) have less access to cooperatives and farmers' associations, and are less likely to participate when access exists, but often do participate in labor exchange groups * are less likely to have savings, and rarely hold them in financial instruments * are less likely to have ever borrowed money and, when they did so, borrowed rather small amounts * have higher unemployment rates, are more likely to be active job seekers if unem- ployed, or to have given up in discouragement * are more likely to have experienced migration of family members, yet received the lowest remittances per migrant * if self-employed in non-farm occupations, receive less compensation than even sea- sonal farm laborers * cannot meet the standard of minimum daily food consumption set for Buenos Aires residents * do not seem relatively disadvantaged in access to social infrastructure by their geographical location (but only relative to other rural populations in the same provinces) * tend to have access to free or subsidized health care, albeit of low level or quality, but also have a disproportionate need (due to large family sizes) * lack modem plumbing or sewage, internal or nearby water sources, fuels other than wood and kerosene for cooking and lighting respectively * have less land access or cultivated area per capita, although they manage to pro- duce almost the same total income per household and usually higher productivity 45 Argentina Poverty Assessment Background Paper No. 6 per hectare by intensive production together with diversification of both agricul- tural products and farm/non-farm income sources * lack land titles (if owner-cultivators) or are squatters (if pastoralists) This adds up to a formidable list of disadvantages relative to non-poor elements of rural Argentina, although most of the rural non-poor may also seem relatively deprived in com- parison with their counterparts in other regions or in urban centers. As the root of rural poverty lies in lack of sufficient combinations of land, physical, financial, and human capi- tal, the prospects for an improvement in the condition of the rural poor without outside assistance are not favorable. 97. Throughout the analysis above, possible points of intervention have been evaluated in the light of the survey data. Distinguishing cause and effect and deducing what policies or programs might be effective merely on the basis of cross-sectional comparisons of poor and non-poor households is an exercise fraught with peril, but it does help focus the search and illuminate the potential limitations of any single measure. Possible elements of a strategy of interventions to address rural poverty, their potentials and qualifying condi- tions or limitations, which emerge from the statistical analysis are summarized in Table 22, and further discussed below. 46 Argentina Poverty Assessment Background Paper No. 6 Measure Beneficial? Effect Conditions for Net Benefit Land distribution Doubtful Most of the poor own their land; Natural conditions permitting switch those who don't are mainly from ranching to crop cultivation ranchers, and productivity might not increase Land titling program Possibly May provide access to finance; If titles of small farmers are registered otherwise may not increase pro- and can be used as collateral for fi- ductivity nance Labor market reform Uncertain Might improve off-farm earnings If minimum wage requirements or opportunities gender discrimination can be shown to effect local labor demand Reform policies favoring large Likely Maximize land and probably total If capacity to respond is not limited by farms factor productivity cash unavailability, natural condi- tions, and/or weak management abil- ity of farmers Promotion of diversification and Likely Improve household food security If financial constraints can be over- intensification through exten- and productivity come and extension can be focussed sion on poor but capable farmers Measures to attract investment Likely Generate off-farm employment Uncertain what measures are required; to backwards areas and demand for farm products; probably area-specific reduce migration pressures; pro- vide finance for farms Family planning program Likely Raise per capita income by re- Unclear what influences child bearing ducing the denominator attitudes, other than completion of post-primary education; won't help the current generation Improved quality of rural public Likely Contributes to earnings ability of Subsidies to reduce post-primary drop- education, primary and above poor in all occupations; may fa- out rate? cilitate off-farm occupations, suc- cessful migration Organizational efforts in poor Likely Raises likelihood of membership, Increased appeal to poor farmers areas, with TA targeted to poor and an efficient vehicle for provi- through rural organizations sion of TA Job or skills training (and anti- Likely Benefits the groups most at risk of If the main constraint isn't the ab- discriminatory measures) fo- unemployment sence of demand for non-agricultural cussed on females and youth labor Improved access of the rural Likely Availability of finance would If the drawbacks of almost all rural poor to unsubsidized, short-term promote cash input use and adop- credit programs (mainly mis-targeting finance tion of better technology, leading and low recovery rates) could be to higher farm productivity avoided. Table 22. Potentials and Limitations of Anti-poverty Measures 98. Agriculture versus Non-farm Activities. For the half of rural dwellers in these provinces who are not farmers (and these proportions are higher in richer areas), it goes without saying that a focus on agricultural interventions is not directly relevant--though it is still arguable that agricultural policies which favor agricultural growth may have a spill- over stimulus on the demand for non-farm services and products. For farm households, the importance of non-farm income is indisputable, but farm income remains more than half of total income except for the poorest and richest. Thus it is not a question of 47 Argentina Poverty Assessment Background Paper No. 6 whether to focus on the farm or non-farm sectors--there is need and scope for improve- ment of incomes in both, and each is needed to sustain the other. For a given household, financial resources, along with managerial ability, represent the most severe constraints on either intensification of production or acquisition of additional land. Off-farm opportuni- ties contribute income which relieves financial constraints and facilitates investment in farm production as well as boosting consumption. If only we knew how to stimulate de- velopment of off-farm opportunities, there would be less need to worry about the state of the rural financial system. 99. It was found that ahnost all non-farm entrepreneurial activity of rural households in these provinces currently was of such trivial scale and so poorly paying that it should be viewed as part of a survival strategy, and a symptom of underemployment, rather than an opportunity for escape from poverty. Off-farm wage earnings were, in contrast, more at- tractive, even though apparently discriminatory against females. The survey data gives us no further clues as to how to augment off-farm earning opportunities, so we must look elsewhere for solutions. Perhaps the most plausible line of argument is that lack of such opportunities is attributable to absence of incentives to invest in backwards areas, which relates to weak provision of social services and infrastructure in such areas. This in turn is related to the limited ability of local governments to supply these services. However, what it takes to induce more local investment, from inside or outside, needs to be answered by other investigations. 100. Family Planning. Household size was identified as the single factor most closely correlated with rural poverty status. While correlation is not equivalent to causation, and a larger family labor force does facilitate agricultural intensification, it is clear that varia- tion in total household income between the poor and the near-poor is less important than size of family and the dependency ratio in distinguishing the two on average. Should fam- ily size be regarded as a matter of choice, influenced by education, culture, religion, and attitude, and availability of the means of birth control? Or would it be influenced by avail- ability of off-farm jobs for women, for example? Aside from the age of female spouse, the only variable in the survey data found to be statistically associated with smaller families was completion of secondary education; and off-farm occupation of female spouse was not associated with smaller family size. The findings therefore do not really shed much light on the utility of family planning as an instrument for poverty reduction. If it can be demonstrated to be cost-effective, it is surely the most direct way to address poverty for the next generation. 101. Education. The return to education is another tricky issue, not least because quantity rather than quality was measured. Completion of primary education--perhaps a correlate of an individual's ability or determination--was found to be one of the indicators of management ability, contributing to productivity in crop cultivation. Years of educa- tion contributes to earnings ability, but the marginal contribution for additional education after primary school appears less than opportunity cost for the poor. Indeed, secondary education of household heads was negatively correlated with crop productivity, perhaps because those with good off-farm jobs were not attracted to intensive agriculture as a 48 Argentina Poverty Assessment Background Paper No. 6 source of income; and perhaps also because rural secondary education hasn't provided skills useful in farming. If low marginal returns together with family poverty explain high post-primary drop-out rates and if educational levels also convey strongly across genera- tions, then the ability to intervene in rural poverty solely through promotion of education may be quite limited. Perhaps the initial need is for an off-farm labor market which re- wards skills. Not to imply that education is irrelevant even to agriculture: managerial ability is hugely important to farm productivity, as also the willingness to use commercial inputs, and there remains much unexplained variance in these characteristics among sur- veyed farmers. Differences in the quality of education between the poor and non-poor may be responsible for much of the financial disincentive to pursue further schooling, and this is probably the most effective focus for educational reform in the context of an anti- poverty program. 102. Organization. It is natural to think of addressing the needs of the rural poor via various kinds of rural organization. The survey found that farmer access to typical rural organizations was rather limited in these provinces, especially in Salta, and access was the most important determinant of membership. Access currently strongly favors the less- or non-poor, hence a need for organizing efforts in poorer areas is indicated. But various organizations also attracted different types of farmers--cooperatives the more mechanized farmers, and farmer associations those who appeared to be better managers. Coop or farmer association currently seems to be a gateway for receipt of technical assistance, though not credit. However, there are unresolved questions of the direction of causation here: does joining gain the farmer access to TA, either because these groups link their members to TA or because provision of TA is targeted through such groups? Or is it sim- ply that certain types of farmers are both "joiners" and seekers of TA, leading to coinci- dental correlation? In either case, how does one assist the 40-odd percent of farmers (88 percent being poor) who did not join even though organizations were locally accessible? Aside from making such organizations locally accessible, it is apparently necessary to im- prove their attractiveness to poor farmers. 103. Rural Finance. The survey revealed a paucity of rural savings and the statistical analysis showed that availability of liquid assets was the most important factor in the ex- tent of use of commercial inputs (correlation coefficient of 0.45), which in turn was a sig- nificant determinant of farm productivity--findings which are evidence of a serious finan- cial constraint. This constraint is also reflected in the low level of use of commercial in- puts among the sample farmers: some 58 percent used no fertilizer and 18 percent pur- chased no cash inputs of any sort. Those who did purchase inputs apparently used the cash surplus derived from on- and off-farm activities to finance their next year's farming. 104. In 1995-96, so few farmers in these provinces (under one percent) were receiving cash loans for productive purposes that the link between TA and (subsidized) credit, common in other developing countries, appears to have been broken. Since farmers' cur- rent receipt of TA and past borrowing were positively correlated, this link may have ex- isted in the past--i.e., tecnicos served as gatekeepers to subsidized credit. This may be 49 Argentina Poverty Assessment Background Paper No. 6 why TA currently seems so well targeted on superior managers, and has some positive effects on crop productivity.31 105. In the absence of the creation of an effective rural financial system offering access to the poor, it is difficult to conceive of major improvements to the farm incomes of the poor in Argentina. What measures need be taken to promote such a system is the subject of another study--mis-targeting and poor credit recovery are the major problems of official efforts the world over. However, the survey indicated that overcoming fear of failure, and of complicated processes and paperwork, would be a necessary part of opening access to the poor, as would substituting for the absence of sufficient collateral on the part of most poor farmers in these regions. 106. Land Distribution and Tenure Improvements. Larger farm size in Misiones and Salta does not necessarily imply either greater wealth or productivity: there are poor households with farms over 500 hectares, although these are likely to be occupied (squat- ted upon) ranches in low-fertility areas; and the highest productivity is found on the small- est farms.32 Thus policies favoring (or not discriminating against) small farms in these ar- eas would be economically justified, and the size of the productivity differential even sug- gests some potential for voluntary land transfer from large to small farms," if the financing arrangements could be made. Viability of small farms, defined as the ability to provide a living standard above poverty levels, would be facilitated by smaller household sizes and/or more adequate off-farm earning opportunities--some 20 percent of farms may be classed as small but already viable and productive at present, and the policy objective should be to enlarge this group. 107. Two major tenure issues were highlighted: lack of ownership among poorer ranchers in Salta, where it is unclear whether ownership or formal tenure would improve productivity34; and lack of title in both provinces, where possession of title is associated with greater fixed capital investment. However, a statistically-significant relationship be- tween the amount of fixed capital investment and crop productivity was not found, so jus- tification of a titling program would have to depend on the ability to use titles as collateral for financing short-term (working capital) investment. Titling, without remedying the ex- isting deficiencies of the rural financial market, might thus be ineffective. 31 If tecnicos cannot offer access to subsidized credit, their services are presumably sought only by those looking for technical advice, and that advice is taken only if it is productive. 32 The last has been demonstrated for land productivity, and would hold for total factor productivity if the latter were calculated based on shadow wages which fully reflected the high rates of unemployment. With the caveat that part of the the productivity differentials by farm size probably reflects differences in natural conditions among farms in two different provinces. Switching from ranching to crop cultivation would apparently improve productivity greatly, but it is not clear whether this would be possible under local natural conditions. 50 Argentina Poverty Assessment Background Paper No. 6 108. Labor Market Reform. The issue is whether there is anything in the institutional or legal framework of the Argentine labor market (such as an "excessive" legal minimum wage) which prevents the rural wage labor market from clearing on the basis of price. If so, a reform of such practices might stimulate greater off-farm employment, albeit at lower wage rates. The survey indicated a high rate of unemployment, particularly among youth and females, and a gap of nearly 50 percent between the average reservation wage and actual wages received by full-time workers, suggesting that the rural labor market cleared through rationing rather than price. However, most of this gap may be a matter of gender discrimination: only in the case of women was there a substantial potential supply willing to work for less than the "going wage" for their gender, experience, and educational status. Thus the institutional issue which may deserve further attention is the apparent barriers to employment of females in various non-farm jobs now accepting mostly young rural males.35 109. Food Security. For most of the rural poor in these provinces, who are net buyers of food and overwhelmingly dependent on cash income for their survival, attaining food security is a matter of total income (and not only cash income), and increasing that should be the clear focus of policies and interventions. On the whole, however, the usual pre- scriptions of food security programs--diversification, less reliance on industrial crops, more vegetable crops and fruit trees--are also routes to higher value-added and hence con- sumption. Also, promotion of organic farming and soil nutrient and moisture conservation techniques can only be helpful, so long as it is recognized that use of cash inputs is also part of the solution. 110. Specialization, Diversification, and Intensification. Economists often see di- versification as a means of protection against risk, with a corresponding tradeoff of in- come, and see specialization as the best way to exploit comparative advantage and maxi- mize income. These generalities do not hold empirically in the survey data, probably be- cause the technical basis for achieving economies of scale through specialization are lack- ing for farms of this scale and low degree of capital intensity. Surprisingly, larger-scale producers of a crop, however specialized, did not even obtain better prices for most types of crops, which undercuts a common rationale for marketing assistance to smallholders. On the other hand, intensification--particularly of labor use and current inputs such as fer- tilizer--is clearly advantageous for farmers with relatively abundant labor and scarce land and fixed capital. There doesn't seem to be any serious tradeoff between intensifying farming and increasing off-farm income, given the extent of surplus labor and shortage of cash. However, the rural poor, as a group, already behave as if they understood these points of development strategy, which raises the question of what technical assistance really has to offer. Winter (1994), summarizing the findings in six Latin American countries other than Argentina, shows that gender "discrimination" largely accounts for the gender wage gap found in all of these countries. 51 Argentina Poverty Assessment Background Paper No. 6 111. Technical Assistance. If an extension program is most efficient in focusing its efforts on farmers best equipped to take advantage of it, then the technical assistance pro- gram in these provinces cannot be faulted for inefficiency, as those farmers identified as highest on an estimated management index are receiving TA. However, it is unlikely that all potentially good managers have been identified. For example, capable farmers, mainly among the poor, who lack access to, and therefore are not recorded as participants in coops or farmer's associations, would not score so high on the management index. For that reason, it cannot be asserted from the findings of this study that TA is reaching all who can benefit from it, or that the percent of farmers who could benefit can be readily estimated from the sample. On the other hand, if such organizations were to become more accessible and attractive to the rural poor, membership and participation would be the most useful criterion for targeting technical assistance. 112. The main impact of TA detected was its association with the use of fertilizer and other commercial inputs. Otherwise TA was found to have no independent effect on pro- ductivity. This gives reason to take a hard look at the quality and relevance for these cli- ents of both agents and messages. Finally, as the scale of use and therefore the impact of commercial inputs on productivity and income were found to be largely determined by the farmer's financial resources, it is arguable that TA and access to financial resources need to be combined for the former to have much effect. This throws us back to issues of re- form of the rural financial system. 113. Differentiation of Clients. Of all the ways to differentiate among the rural poor, the most obvious is by farm management ability, as the indicators of this ability have been identified and are rather visible. Should, therefore, assistance in agriculture technique be narrowly targeted on the most able farmers, and other strategies be applied to the rest? This entails doing no more than Argentine extension agents are now doing in these two provinces. 114. The point of targeting is either to obtain the maximum economic gain at the mar- gin for a given extension effort, or else to avoid committing public resources to individuals or groups which can afford to purchase technical assistance on their own (such as wealth- ier farmers). Targeting the most able farm managers unfortunately does not seem to ac- complish either objective. Insofar as good management is partially correlated with wealth, targeting public resources on good managers somewhat biases public service provision against the poor and subsidizes those who can afford to pay. But, equally important, there is no reason to believe that the marginal economic gain from targeting good managers is higher than for others. Considering that receipt of TA is mainly associated with use of fertilizer and other commercial inputs, it is noteworthy that the average increases in crop productivity associated with the use of fertilizer are 28, 80, and 31 percent respectively for farms with "low", "medium", and "high" management indices. Since the majority of better managers are already using fertilizer, whereas the majority of the 65 percent of farmers evaluated as of "low" management ability are not, the highest marginal economic gain would likely come from targeting of the latter, provided they can be convinced to use commercial inputs at reasonable cost and provided the purchase can be financed. 52 Argentina Poverty Assessment Background Paper No. 6 115. Obviously, in specific contexts, these conditions will not obtain, nor will the effects of fertilizer and commercial inputs meet average expectations in every situation. Discrimi- nation among clients is justified to that extent, if not as a matter of broad strategy. 116. A strategy of client differentiation in the promotion off-farm income sources for the rural poor may also be inappropriate, for two reasons: the interventions most needed may be those which stimulate regional economic growth and therefore labor demand as a whole, rather than training exercises or artisanal projects targeted on particular individu- als or groups; and virtually all types of poor rural household could benefit, as almost all depend heavily on off-farm income sources to supplement farm income. On the other hand, focussing on youth and females is certainly advisable for skills development and anti-discriminatory programs. 117. Going beyond matters on which the survey sheds direct light, it is very unlikely that either technical assistance or finance can be delivered at reasonable cost to the rural poor except through rural organizations. Differentiation of the poor into those who are members and participants in rural organizations and therefore eligible for a program of targeted interventions, and those who are not and must be organized as a precondition, is likely to be an efficient strategy. 118. A Multi-Instrument Strategy. Overall the study has suggested that an effective strategy to address rural poverty needs multiple instruments or components going well beyond the narrow focus on agricultural productivity characteristic of traditional rural programs in Latin America. The components summarized in Table 22 (with the omission of rural infrastructure, insufficiently documented by the survey) would deal with all of the correlates of low-income status. However, most would also not be effective in isolation. There is a need for balanced attention to agricultural and non-farm elements of livelihood, and to long-term solutions, such as improved education and family planning, as well as short-term measures, such as rural organization and delivery of training, technical assis- tance, and finance. Attention to quality and relevance, e.g., of education and technical as- sistance, may be at least as important as increased scope of service delivery. And attention to institutional development problems, in the areas of finance, land tenure, labor market, and regional development, may be as relevant as targeted projects. 53 Argentina Poverty Assessment Background Paper No. 6 6. BIBLIOGRAPHY Benencia, R. 1995. "Diagnostico de las Pobreza Rural en la Argentina. Propuesta para la Deteccion y Seleccion de Beneficiarios entes Auspiciantes e Iniciativas en un Proy- ecto de Desarrollo para Pobres Rurales." Buenos Aires: SAPyA - PROINDER. Fiorentino, R. 1996. "Informe Tecnico: Acceso a Credito y Otros Servicios Financieros por los Pequefios Productores Agropecuarios Argentinos (Borrador para Dis- cusion)." Buenos Aires: SAPyA - PROINDER. Gallacher, M. 1996. "Impactos de Reformas Macroeconomicas y Sectoriales." Buenos Aires: SAPyA - PROINDER. GonzAlez, C. C. 1966. "Anilisi y Propuesta de Investigaci6n, Validaci6n y Transferencia Tecnol6gica a Pequefios Productores." Buenos Aires: SAGyA - PROINDER. GonzAlez, M. del C. 1996. "Estructura de Tenencia y Mercados de Tierra: Borrador para Discusion." Buenos Aires: SAPyA - PROINDER. Juv6, L.C. 1996. "Estrategias de Mercadeo para Pequefios Productores en un Programa de Desarrollo Agricola." Buenos Aires: SAPyA - PROINDER. Lopez, R. and A. Vald6s. 1997. Rural Poverty in Latin America. Forthcoming. Wiens, T. B. 1995. "Agriculture and Rural Poverty." In D. Dollar, P. Glewwe, and J. Litvack, eds. Household Welfare and Vietnam's Transition to a Market Econ- omy. Forthcoming. Winter, Carolyn 1994. Working Women in Latin America: Participation, Pay and Public Policy. Washington, D.C.: The World Bank, Latin America and the Caribbean Region, Technical Department. World Bank 1995. Argentina's Poor: A Profile. Washington, D.C.: The World Bank, Latin America and the Caribbean Region, Human Resource Division, Country De- partment 1. Norman Hicks L:\ARGENT\BP6.doc 01/31/00 11:51 AM 54 Argentina Poverty Assessment Background Paper No. 7 Poor People in a Rich Country A Poverty Report for Argentina Background Paper No. 7 Household Structure, Gender, and the Economic Determinants of School Attendance in Argentina By Walter Sosa Escuderol Universidad Nacional de La Plata La Plata, Argentina and Mariana Marchionni Universidad Nacional de La Plata, La Plata, Argentina April 4, 1999 1 Corresponding author: Walter Sosa Escudero, Departamento de Economia; Universidad Nacional de La Plata; 6 e/ 47 y 48; La Plata, Argentina. Tel/Fax: (54-221)-422-9383. E-mail: wsosa@feedback.net.ar Argentina Poverty Assessment Background Paper No. 7 Abstract This paper explores the role played by household structure and gender on the decision to send children to school in Argentina. We explore the determinants of the education patterns of boys and of girls based on the gender of the individuals who make (or influence) the school attendance decision, and the economic and social household structure in which this decision is evaluated. The results of this paper suggest that differences in family structure influence the way children are educated beyond their effect on family income. Gender considerations like the gender of the head of the family or the sex of the child are found to have important implications regarding the schooling decision. 2 Argentina Poverty Assessment Background Paper No. 7 1 Introduction The relationship between education and the welfare of a society has a relevance that exceeds the scope of a single discipline, and it is consequently studied by practically every social and human science. From an economic standpoint, the traditional Beckerian approach (Becker, 1965, 1981) to the decision to study sees education as an investment that allows an individual or family to increase its stock of human capital. Within this framework, the individual (or his/her family) faces the direct costs of acquiring education (fees, books, travel expenses, etc.) and accepts a temporary reduction in his/her potential earnings while he/she studies. Nevertheless, the decision to study is essentially a more complex phenomenon, influenced by other social, cultural and psychological features beyond the strictly economic ones. Even though a thorough understanding of the determinants of education is an essential input for the efficient design of policy measures aimed at poverty alleviation, the analysis of such an important topic has long remained unexplored for the Argentine case, mainly due to the high economic instability that characterized the Argentine economy in the past decades, which forced most of the applied research to concentrate on macroeconomic issues aimed at controlling inflation and stabilizing the economy. The low levels of inflation experienced since 1991 and the structural changes adopted thereafter were followed by a marked and sustained increase in the unemployment rate, which went from 6% in 1991 to 17.4% in 1997. This phenomenon is likely to have introduced important changes in the social or economic structure of households, mainly due to the increasing participation of women in the labor force. In this paper we examine the role played by household structure and gender on the decision to send children to school in Argentina. The study will be based on the Permanent Household Survey (Encuesta Permanente de Hogares, EPH) collected bi-annually by the National Statistical and Census Institute (INDEC). The EPH contains abundant information at the individual and household level. We explore the education patterns of boys and of girls based on the gender of the individuals who make (or influence) the school attendance decision, and the economic and social household structure in which this decision is evaluated. The paper is structured as follows. Section 2 defines the economic and sociological background of the paper, and advances relevant empirical questions to be explored in subsequent sections. Section 3 describes the data set used in this study and the methodology utilized to handle family structures and gender issues in the EPH. Section 4 presents estimates of binary choice school attendance models, and the results of several comparative statics and simulation exercises. Section 5 concludes and presents some policy recommendations. 2 Gender and Family Structure in the Economic and Sociological View of the Decision to Study The standard economic approach to the decision to study relies heavily on Becker's (1965) classic work. Education arises as the result of a utility maximization problem, whose solution is a function of the household production function and the family investment function. The optimal level of education equates marginal benefits with marginal costs; so, with perfect capital markets, 3 Argentina Poverty Assessment Background Paper No. 7 parents act as if they face a perfectly elastic supply of funds to finance education, implying that family economic resources should not be relevant in the determination of the educational attainment. Consequently, within this framework, any observed difference in education between two children should be associated to factors other than family financial resources. In general terms, these "other factors" reflect economic, psychological and social restrictions faced by the family who makes the schooling decision, and the different social and cultural characteristics that define the way in which different educational alternatives are evaluated against each other through the family utility function. The sociological point of view2 does not necessarily contradict the economic point of view, but the emphasis is put on parental ability to provide children with the motivation and skills necessary for school achievement (Garasky, 1995). The socialization theory of education focuses on different views about the way families contribute to the educational process of their children. One view stresses the role of the family in providing a hierarchical structure of authority, which is a key aspect in succeeding in institutions strongly characterized by such structures, like the educational system or the labor market (Nock, 1988; Weiss, 1979). A second view focuses on the family as a provider of role models for children (Hess and Camera, 1979). Other versions emphasize the role of the family in controlling stress and conflict, factors which may inhibit children's development significantly (Loh, 1996). From an empirical point of view, it is difficult (and maybe unnecessary) to exploit these theories to come up with a complete and orthogonal list of relevant factors that determine the decision to study. It is natural to expect that many of these factors interact, complicating the process of statistically identifying the effects on education of each of them separately. For example, most studies find a strong positive association between educational attainment and parents' education. This may reflect both the fact that better educated families are less likely to be credit constrained, and that these families have stronger "tastes for education"; so, other things being equal, families with higher education provide more and better education to their children. Similarly, the effect of marital dissolution may affect education through a reduction in income, and through some other channels not directly observable by the analyst (reduction in time spent with children, increase in family stress level, etc.). In this study we emphasize two groups of factors that may affect the educational process: household structure and gender. In the Beckerian economic point of view, household structure affects education through altering the cost-benefit scheme where optimal schooling is decided. For example, marital dissolution should have a negative impact in investment in both financial terms and in the amount and quality of time spent with children in schooling age. Within the socialization framework, different family structures provide different educational environments for their children. From this point of view, for example, the number of children also affects the schooling decision by inducing the family to assign different roles to each child. There are several ways gender issues are involved in the school attendance decision. In general terms, they reflect the prevailing (or perceived) social norms which determine differences between males and females. Gender issues arise in the schooling decision through two channels. 2 There are several reviews of the so called "socialization theory" of education. Ours is based on Loh (1996) and Garasky (1995), and we refer to these authors for more details and references. 4 Argentina Poverty Assessment Background Paper No. 7 The first refers to the gender composition of the family who is involved in the schooling decision. The second is related directly to the gender of the person whose education is being considered. By gender composition of the family, we refer to how the gender of the different members of the family alters the way different factors affect the decision to study. This is closely related to the different roles played by the members of the family in the educational process. For example, Leibowitz (1974) suggests that while both parents' education is relevant, father's education tends to represent genetic factors, while mothers' education reflects both genetic and home-specific factors, since mothers usually spend more time caring for their children than do fathers. Also, some studies (Pitt and Khandker, 1997) have recently found that mothers make a more efficient use of resources relevant in family decisions, implying that, other things being equal, single parent families headed by women may educate their children differently than those headed by men. In summary, this discussion suggests that it is worthwhile to explore whether several factors typically thought to influence education, such as income and parental education, have a differential impact on children's education depending on the gender of the persons who are involved in the schooling decision. The second route to explore gender issues appears when parents are thought to have different preferences for their sons and daughters, or, due to prevailing social and economic conditions, parents perceive that boys and girls will perform differently in institutions like the educational system or the labor market (Deaton, 1997, pp. 224-225). From this point of view, independently of the gender composition of the household, children would be edacated differently simply as a result of assigning scarce resources efficiently, according to the prevailing social rules. For example, faced with a decrease in permanent income, parents might decide that the boy quits school and enters the labor market, and that the girl stays at school, as a result of the perception that wages are higher and job opportunities are better for boys than for girls. Previous work on how family structure and its gender composition affect the decision to study lies in the limits of economics and sociology3. Loh (1996) studies how observed changes in family structure affect schooling attainment and poverty status, and finds that changes in family structure lower completed schooling, but the effect is difficult to identify since it mimics other regressors, like income. Garasky (1996) explores how changes in family structure experienced by children affect the probability of finishing high school, emphasizing whether the family effect varies with the age at which the change took place. He finds that family structure effects on educational attainment vary according to whether it is either the biological mother or father that lived with the child, and also according to the age of the child; and that these changes are more detrimental when they take place early in the educational life of the child. Tansel's work (1997) measures the importance of parent's education on school attainment and studies differences by gender in two African countries; Tansel finds that, in contrast to high income countries, father's education seems to be more important than mother's. Studies specific for the Latin American case are scarce. Rodriguez and Abler (1998) study school attendance in Peru, finding that there are not obvious gender differences and that, quite 3 A detailed review of empirical findings on the subject exceeds the goal of the paper. We will limit the review to recent work specific to household structure and gender. Tansel (1997) has abundant references and useful results for international comparison. 5 Argentina Poverty Assessment Background Paper No. 7 surprisingly, income has a minor effect. For the Argentine case, Gasparini et al. (1998) study school attendance in the province of Mendoza, but focus on the impact of income policies on education. 3 Data and Methodology This section describes the data set used in this study and how the previous discussion translates into observable indicators of household structure and gender composition. In general terms, the empirical strategy adopted in this paper is the result of the considerations advanced in the previous discussion and the limitations imposed by the availability of data for the Argentine case. The analysis is based on the Argentine Permanent Household Survey (Encuesta Permanente de Hogares, EPH), collected and processed by the local National Statistical and Census Institute (INDEC). The survey is conducted bi-annually in May and October and covers 25 urban centers, representing 70% of the urban population of the country and 98% of the population living in centers with more than 100,000 habitants. The EPH is implemented through a family questionnaire as well as a personal one. The first one includes information on household characteristics such as dwelling attributes, family composition and other demographic characteristics. The personal questionnaire contains information related to occupational, educational and income characteristics at the individual level. The survey has a rolling panel structure: once a household is chosen, it remains in the sample for four waves, that is, for two years. Each period, 25% of the families are replaced so that between two waves 75% of the families remain in the sample4. This study is based on the October wave of 1997 covering Greater Buenos Aires (GBA), La Plata and Mendoza . There is a fundamental difference between this and other studies on the subject, which is a direct consequence of the type of information available in the EPH. Several studies concentrate on school attainment and its determinants, which require the availability of either a panel long enough to cover all the educational history of the individual, or "historic" questions. For example, if the goal is to measure the effect of family dissolution on educational attainment, it is essential to know not only whether events like father's death or divorce took place during the educational process but also when they occurred. In a very long panel, it is possible to recover the latter from questions regarding whether parents divorced or the father died at the moment the survey was conducted. Alternatively, the same information can be obtained from historical questions asking, for example, for how long parents have been divorced. The panel coverage of EPH is only two years long and it does not contain historical questions on demographic or personal characteristics, but they refer specifically to the moment when the survey is collected. For example, at a specific moment we know whether the family is 4 For detailed information on the EPH, its historical development and perspectives see INDEC (1997) ' The choice of these agglomerates is a compromise between homogeneity and sampling variation. Any single agglomerate (like GBA) at any period guarantees regional homogeneity across individuals but leaves few observations of interest. Attempts to expand the sample incorporating panel information do not produce an obvious gain in information due to the very small time span of the panel and to the high correlation between observations induced by the panel structure. On the other hand, La Plata (for geographical proximity) and Mendoza (being another important urban area), add more variability to the sample, while allowing us to control for intergroup heterogeneity. 6 Argentina Poverty Assessment Background Paper No. 7 headed by a woman, but we do not know for how long this situation has prevailed. Given this restriction, we must concentrate the analysis on school attendance, that is, on the determinants of the decision to send children to school. For all individuals in the sample, the survey indicates the maximum level of education attained or, if the individual is still attending school, the educational level of the individual at the moment of the survey. The population of interest will be composed of all children who are observed to have decided to attend school or not, and whose families are involved in that decision. Under these considerations, the estimation will be based on all boys and girls who, at the moment of the sample, were between 13 and 19 years old, single, classified as sons or daughters in the EPH, and have completed elementary school6. This leaves a sample of 1,548 children: 729 girls and 819 boys. The variable to be analyzed in this study is a binary indicator of whether a child attends school at the moment of the survey. We will estimate binary choice models of the decision to attend school. As is customary with these type of models, the decision to study is the result of a utility maximization problem where the utility of attending school is compared to the alternative of not participating in the educational system. Consequently, the binary indicator takes a value equal to one if the utility of attending school is greater than the utility of not doing so. A conditional model expresses the probability of attending school as a function of several explanatory variables that may affect the decision to study. As discussed in the previous section, gender issues will be reflected in the decision to study in two ways. The first is the way in which factors such as parents' education and family income affect education according to the gender composition of the family, independently of the gender of the child being educated. The second refers to the opposite: how factors affect boys' and girls' education differently, independently of the gender composition of the family. As for the second channel, Table I shows attendance rates by gender. Each cell in the table has three numbers. The first one refers to the number of individuals of the sample in each cell; the second one to the proportion of individuals according to the classifications corresponding to the rows of the table; and the third one to the proportion of individuals in the classifications implied by the columns. For example, 339 children in the sample do not attend school. Of this group, 37.46% are girls and 62.54% are boys. Of the 729 girls in the sample, 82.58% go to school. 6 Elementary school is mandatory in Argentina and attendance rates are almost perfect, leaving little (if any) room for conditional analysis. Estimations for this subsample give trivial results (available from the authors) and were not included in the paper. The range 13-19 years covers most of the interesting cases. The earliest age to start high school in Argentina is 13 years by June 30th of the corresponding year; so, in October of a given year, every high school student must be at least 13 years old. High school lasts 5 to 6 years, depending on the specialization (5 for commercial and the baccalaureate, 6 for technical and some University high schools). 7 Argentina Poverty Assessment Background Paper No. 7 TABLE 1: SCHOOL ATTENDANCE AND GENDER Number of observations and percentages (by column and row) Attendance Gender Female Male All 127 212 339 Do Not Attend 37.46 62.54 100 17.42 25.89 21.9 602 607 1209 Attend 49.79 50.21 100 82.58 74.11 78.1 729 819 1548 All 47.09 52.91 100 100 100 100 These numbers show some interesting facts. The gender composition of those who do not go to school is largely dominated by boys (62.54%), whereas the gender composition of those who attend school is practically equal. Attendance rates are higher for girls (82.58%) than for boys (74.11%) Regarding the first channel, at this point it is relevant to discuss what will be understood by family and household structure for the purposes of this paper. The election depends not only on the goals of the investigation but also on the characteristics of the available information. In any case, when defining the unit of analysis that will be taken as a family, it is important "to delimit.. .homogeneous subsets according to some essential dimensions in the analysis of household units" (Torrado, 1998). The definition of household is usually goal-directed and is, in general terms, associated with a group of people living and eating together (Deaton, 1997) who share resources for the purpose of pursuing their mutual well-being (Bryant, 1990). By family we understand two or 7 more individuals of a household related between them by blood, adoption or marriage . For the purposes of this work, it is relevant to consider what are known as multiperson conjugal households (Torrado, 1998). A multiperson household is defined as a group of two or more people sharing the same dwelling, and nutritional or some other basic needs. By conjugal it means that there exists a conjugal link, that is, a family in one of the following forms: couple (married or not) with no children, couple with one or more unmarried children, or mother or father alone with one or more unmarried children. In each of these households one person plays the role of the head of the household. In the EPH, the head is "the one recognized as such by the other members of the households" (INDEC, 1997). Within this category it is possible to find characteristics that delimit homogeneous subsets, such as the presence or absence of the spouse and/or the gender of the head. We will distinguish between complete (mother and father, or head and spouse) and incomplete (mother or father, or head only) households, and between households headed by a woman and households headed by a man. Figure 1 summarizes these categorizations. ' Recent discussions on different definitions of household and family which are relevant for economic analysis can be found in Bryant (1990) and Deaton (1997). From a sociological perspective and with special reference to the EPH, Torrado (1998) is particularly informative. 8 Argentina Poverty Assessment Background Paper No. 7 FIGURE 1: HOUSEHOLD STRUCTURES RELEVANT FOR THE SCHOOLING DECISION Multiperson Conjugal Households Complete Incomplete Headed by Headed by Headed by Headed by women men women men Source: Torrado (1998) Within complete households, we will distinguish between (legally) married and non- married couples; and within incomplete families we distinguish between divorced, widowed and those who do not fall in the previous categories. Each of these five groups can be headed by women or men. In order to capture the effects of these different structures on the schooling decision, we constructed the binary indicators married and couple for the complete families (1 for legally married and not married parents, respectively), and widowed, divorced and single for the incomplete ones (1 depending on the civil status of the family head). Another indicator that characterizes the family structure isfemhead (1 if the family is headed by a woman and 0 if headed by a man). In Table 2 we present the evolution of households headed by women HHW's as a proportion of the total of households, for Greater Buenos Aires. Within the HHW, we distinguish between single person (e.g., female senior citizens living alone) and multiperson households as defined above. In this category, we include non-conjugal (e.g., two girls sharing an apartment), complete, and woman only households. TABLE 2: EVOLUTION OF HOUSEHOLDS HEADED BY WOMEN PARTICIPATION (% of total households) Headed Single Multiperson by women person Non-conjugal Complete Woman only 1980 17,1 6,9 3,5 0,8 6,4 1982 19,6 7,3 3,8 1,0 7,5 1986 19,9 7,4 3,6 0,9 8,0 1988 20,7 8,6 3,7 0,8 7,8 1990 21,1 8,5 3,9 0,6 7,9 1992 22,2 9,1 3,6 1,3 8,2 1994 23,6 10,4 3,6 1,l 8,5 1996 24,8 8,8 4,0 1,6 9,3 Source: Torrado (1998b) based on EPH for Greater Buenos Aires In the last decades, the number of HHW's has increased steadily in Greater Buenos Aires, going from 17% of the total of households to 24,8%. As stressed in Torrado (1998b) it is interesting to observe that an important proportion of the HHW's corresponds to women living alone, which 9 Argentina Poverty Assessment Background Paper No. 7 face different needs and characteristics. In any case, these figures reveal that a significant part of the population lives in households where important decisions are considerably influenced by women. Table 3 shows school attendance patterns for the different household structures relevant for this work. The attendance rates for boys do not seem to be altered by changes in family structure; for girls, on the other hand, attendance rates are lower for those living in incomplete families. Within this group, attendance is lower for those living with their fathers compared to those living in a HHW. TABLE 3: SCHOOL ATTENDANCE AND HOUSEHOLD STRUCTURE Girls Boys Family structure Total Attend in % Total Attend in % All 729 82,58 819 74,11 mother and father 608 84,54 679 74,37 only father 19 63,16 39 74,36 only mother 102 74,51 101 72,28 Corresponds to family structure in the estimation sample: children from multiperson conjugal households, 13 to 19 years old, single and with primary school completed. Many studies (Rodriguez and Abler (1996), Loh (1996), Garasky (1995)) find that the number of siblings is relevant in the decision to study. Once we control for per capita income, the number of siblings should reflect the effect of family size which is not captured by changes in financial resources available for the family. That is, we allow for the possibility that an increase in the family size that leaves per-capita income unaltered, has an effect on education which is not related to the financial resources available for the family. We also considered the possibility that child's birth order play a role in intrahousehold resource allocation. We included several dummy variables to control for this effect. Income is measured as the log of per capita income (lpci), which is the total household income divided by the number of members. As discussed in the previous section, current income should reflect the resources available to finance education. We also constructed a variable labeled incomeh, which is the proportion of total family income earned by the head of the family. This indicator measures the degree of concentration of family income by different sources. incomeh close to one might identify wealthy families in which a high income by one member of the family has a strong income effect, inducing other members of the family to quit the labor market. It may also be an indicator of risk associated with the source of income: low incomeh should indicate that total income come from many possible unrelated sources, indicating more stable income. We will allow this indicator to interact with femhead (incfem) and we will measure whether the concentration effect has a differential impact if the family is headed by a woman. Table 4 shows attendance rates by income deciles. Attendance rates for girls are higher than for boys in all income deciles, though these differences are minor in the upper two deciles. 10 Argentina Poverty Assessment Background Paper No. 7 TABLE 4: SCHOOL ATTENDANCE BY INCOME DECILE Girls Boys decile* total attend in % total Attend in % 1 137 67,15 136 55,15 2 83 71,08 118 67,80 3 86 76,74 101 65,35 4 66 80,30 83 71,08 5 88 87,50 94 80,85 6 69 91,30 78 80,77 7 67 92,54 60 83,33 8 57 94,74 70 87,14 9 42 100,00 50 98,00 10 34 100,00 29 96,55 * households ranked by household per capita income Figure 2 shows this information graphically. The solid line corresponds to attendance rates for girls and the dashed one for boys. Income has a positive effect on education, but this effect is slightly stronger for boys than for girls, as revealed by the steeper slope of the boys' curve. FIGURE 2: INCOME, EDUCATION AND GENDER 100- a,80- 60 , . 40 20- 0 (0) In r- 0) Income deciles Parent's education is a relevant factor affecting children's education. We will allow father's and mother's education to contribute differently to the stock of family human capital as it influences child's education. It is important to recognize that the identification of this differential effect may be complicated by the expected correlation between parent's education (matching). For our sample, the correlation between parents' education is 0.4084. Unemployment of father or mother may affect school attendance by the implied reduction in income or by introducing uncertainty or stress in the family. We measure if it is relevant whether particular members like mother or father are unemployed, or if it just matters whether the head of the family is unemployed. 11 Argentina Poverty Assessment Background Paper No. 7 Age is typically included in attendance models. It can either reflect increasing opportunity costs as time passes, or it may reveal inertial effects. With the information in the EPH it is difficult to predict its effect, since we do not observe grade level, we only know whether she/he attends school. Take the case of a 16 year old child still in school. If he is close to finishing, then age reflects that he spent considerable time in school, which may induce him to complete the degree. On the other hand, if he is in the first year of high school, his opportunity cost will increase dramatically, possibly inducing him to quit school. 4 Empirical results This section presents estimation and inference results of logit models of school attendance. The set of explanatory variables is listed in Table 5, and the Appendix describes how these variables were constructed from the EPH. TABLE 5: VARIABLE NAMES AND DESCRIPTION Variable Description age age in years age2 age squared Ipci log of per-capita income eduf education of the father in years edum education of the mother in years funempl 1 if the father is unemployed, 0 otherwise munempl I if the mother is unemployed, 0 otherwise siblings number of siblings femhead I if household is headed by a woman married 1 if child lives in a married complete couple 1 if child lives in an unmarried complete family single I if child lives in an incomplete family with single head widowed 1 if child lives in an incomplete family with widowed head divorced 1 if child lives in an incomplete family with divorced head incomeh proportion of household income earned by the head of the family incfem lpci * femhead unskh 1 if head has not completed high school laplata 1 if the city of residence is La Plata mendoza 1 if the city of residence is Mendoza eldest 1 if eldest child. Table 6 presents basic descriptive statistics for the whole population and for the sample, divided into two groups: those who attend school and those who do not. Table 7 presents the same descriptive statistics for boys and girls separately. 12 Argentina Poverty Assessment Background Paper No. 7 TABLE 6: DESCRIPTIVE STATISTICS BY SCHOOL ATTENDANCE MEANS (PROPORTIONS) FOR ALL VARIABLES ALL ATTEND SCHOOL DO NOT A TTEND Variables mean std. dev. mean std. dev. mean std. dev. Age 16.27 1.83 16.01 1.84 17.19 1.46 Lpci 5.20 0.89 5.33 0.86 4.73 0.84 Pci 264.11 351.41 295.95 387.31 150.56 111.77 Eduf 8.04 4.81 8.68 4.94 5.76 3.45 Edum 8.88 4.26 9.54 4.28 6.51 3.19 Funempl 0.12 0.48 0.09 0.41 0.24 0.65 Munempl 0.10 0.43 0.10 0.44 0.08 0.38 Siblings 2.15 1.51 2.02 1.43 2.62 1.68 Femhead 0.15 0.36 0.14 0.35 0.20 0.40 Married 0.73 0.44 0.76 0.43 0.62 0.49 Couple 0.10 0.30 0.08 0.27 0.17 0.37 Single 0.01 0.11 0.01 0.10 0.01 0.12 Widowed 0.05 0.22 0.04 0.20 0.08 0.28 Divorced 0.09 0.29 0.09 0.29 0.09 0.29 Incomeh 0.67 0.32 0.71 0.30 0.55 0.35 Incfem 0.10 0.27 0.10 0.28 0.08 0.24 unskh 0.69 0.46 0.62 0.49 0.95 0.22 laplata 0.15 0.36 0.16 0.36 0.14 0.35 mendoza 0.24 0.43 0.25 0.43 0.22 0.42 eldest 0.43 0.49 0.43 0.50 0.42 0.49 obs 1548 1209 339 13 Argentina Poverty Assessment Background Paper No. 7 TABLE 7: DESCRIPTIVE STATISTICS BY GENDER MEANS (PROPORTIONS) FOR ALL VARIABLES MALE FEMALE ATTEND DO NOT ATTEND ATTEND DO NOT ATTEND variables mean std. dev. mean Std. dev. mean std. dev. mean std. dev. age 15.87 1.79 17.25 1.45 16.15 1.89 17.09 1.47 lpci 5.34 0.86 4.78 0.87 5.32 0.87 4.63 0.80 pci 290.91 266.51 161.07 121.25 301.03 479.43 133.01 91.62 eduf 8.88 5.02 6.09 3.34 8.49 4.85 5.20 3.56 edum 9.50 4.33 6.74 3.24 9.58 4.24 6.13 3.09 funempl 0.08 0.40 0.28 0.70 0.09 0.42 0.17 0.56 munempl 0.10 0.43 0.06 0.33 0.10 0.44 0.11 0.46 siblings 1.99 1.45 2.69 1.72 2.05 1.42 2.50 1.61 femhead 0.14 0.35 0.17 0.38 0.14 0.35 0.24 0.43 married 0.74 0.44 0.65 0.48 0.78 0.41 0.58 0.50 couple 0.09 0.29 0.17 0.38 0.07 0.26 0.16 0.37 single 0.01 0.09 0.01 0.10 0.01 0.11 0.02 0.15 widowed 0.05 0.21 0.06 0.24 0.04 0.19 0.12 0.32 divorced 0.09 0.29 0.10 0.31 0.09 0.28 0.08 0.27 incomeh 0.72 0.31 0.51 0.35 0.70 0.30 0.63 0.33 incfem 0.10 0.29 0.05 0.18 0.10 0.28 0.14 0.31 unskh 0.60 0.49 0.94 0.24 0.64 0.48 0.96 0.20 laplata 0.14 0.35 0.12 0.33 0.17 0.38 0.17 0.38 mendoza 0.24 0.42 0.19 0.40 0.26 0.44 0.27 0.44 eldest 0.43 0.50 0.39 0.49 0.43 0.50 0.47 0.50 obs 607 212 602 127 Tables 8a and 8b present logit estimates of several specifications for the school attendance model. In all cases we estimated models for boys and girls separately. As mentioned before, it is expected that indicators capture more than one effect, since there are interactions between them. For this reason, we estimated several versions using different sets of explanatory variables, in order to see how results change with the introduction or exclusion of additional explanatory variables. Tables 8a and 8b present five specifications. Other less informative cases will be mentioned but not presented in the paper8. The five different specifications start with a "basic specification" (Model 1). The second adds family structure variables (couple, single, widowed and divorced). The third one explores income effects by gender adding incomeh and incfem. The fourth specification includes head's work qualification (unskh). The last specification controls for city of residence (adds dummies for La Plata and Mendoza) and a binary variable indicating whether the child is the eldest in the family. All specifications estimated provide reasonable econometric and economic results, and the null hypothesis that no variables are significant is strongly rejected in all cases. In general terms, the set Additional specifications can be obtained from the authors. 14 Argentina Poverty Assessment Background Paper No. 7 of explanatory variables included seems to explain education for boys more than for girls, as reflected in the pseudo R2 coefficients. 4.1 Significant factors affecting the decision to study With 10 models and 19 explanatory variables involved, it is impossible to comment on every estimated coefficient. To organize the interpretation, first we will comment on the different groups of variables, and then we will summarize differences and similarities by gender. Except for some specific cases, interpretations will be based mostly on the last two specifications, which includes most of the covariates. To begin, age and age2 are always significant across all specifications and genders. In the relevant range, age has a negative and quadratic effect on education, and its effect is stronger and more significant for girls. This suggests that the second effect mentioned in the previous section prevails: age has a strong detrimental effect on education, meaning that the inertial effect is more than compensated by the increase in the opportunity cost. Log income (lpci) always has a significant and positive effect on education for both boys and girls. From the Beckerian perspective, we should be inclined to conclude that limited access to the credit market makes current income relevant for school attendance, but one should be cautious with this interpretation, since income usually captures more than lack of financial constraints. As is the case with age, coefficients are higher and more significant for females, but the null hypothesis of equal coefficients for boys and girls cannot be rejected, indicating that the marginal contribution of income is equal for both sexes. Mother's and fathers' education levels increase the probability of school attendance. In contrast to Tansel's (1997) findings, and in agreement with findings for high income countries (Schultz, 1993), mothers' education seems to contribute to the probability of school attendance more than fathers'. Its coefficient is always greater and more significant. Conversely, father's education is significant only for boys only at the 8% level. For the case of boys, the father being unemployed reduces the likelihood of going to school, and the mother being unemployed increases this probability, though significantly only at around 10%. Results related to fathers' status are likely to be linked to more stress, reductions in permanent income not directly captured by current income, and other negative social or psychological consequences of being unemployed. Mothers' unemployment is somewhat surprising, but it could reflect the fact that in terms of education, and other things being equal, the beneficial side of being unemployed (more time to spend with her children) more than compensates any negative aspect directly associated with her being unemployed. The number of siblings in the family is an interesting case: across all specifications it has a significant negative effect for males, whereas for girls the effect is still negative but never significant. This result sheds light on the way in which boys and girls are substituted in family tasks as a reaction to changes in family size. The result indicates that as the number of siblings gets larger, families are more likely to take boys out of the school system than girls. The family structure variables (couple, single, widowed and divorced) present interesting results. The excluded (base) category is a married couple. Children living under non-married 15 Argentina Poverty Assessment Background Paper No. 7 couples (couple=1) are negatively affected when compared to those living under married ones, and this effect is significant for both boys and girls. The other variables are not significant, with the exception of divorced, which has a positive effect on girls but no significant effect on boys. The proportion of income earned by the head of the family has a significant positive effect for boys but not for girls. This indicates that families whose source of income is concentrated in a few people or one person face better conditions for school attendance. Femhead has a negative effect, and is significant only in specifications that control for differential effects on income, and at 13% of significance. On the other hand, the positive coefficient of incfem suggests that woman heads make a different use of resources than male heads. The positive coefficient suggests that additional income increases the probability of sending children to school in a more significant way if financial resources are controlled by a woman. Compared to the previous result obtained byfemhead, this means that for the case of boys (and less emphatically for girls), the positive effect on education of being in a household headed by a woman appears only through its interaction with income effects. Attempts to control for parents' work qualification are complicated, since this variable is difficult to measure in the EPH, resulting in many missing observations. Instead, we have tried to proxy work qualification by looking at the head's education level. We included a dummy variable indicating if the head of the family has not finished high-school. The variable unskh (1 if the head has not finished high school, 0 if he/she has) has a negative and significant effect on education for boys and girls, suggesting that low levels of family education have detrimental effects on children education not captured by the variables controlling for education. The coefficient of eldest suggests that for the Argentine case, being the eldest child does not have any effect on education9. Similarly, the city where the family is located does not influence the probability of attending school. In terms of the classification proposed in the previous section, we can summarize the findings of this section as follows: 1) Family structure: Its effect on education goes beyond altering income. Once controlling for the latter, children living with unmarried couples are less likely to go to school. Also, girls living with a single divorced mother or father are the least likely to attend school. The number of siblings is an important negative factor affecting boys' education but not girls'. 2) Gender issues: a) Gender composition of the family: The effect of being a woman head appears as significant when it is interacted with log income. This gives some support to the idea that households headed by women make more emphatic (though not necessarily efficient) use of their resources in providing more education to their children. The model presents some evidence that mothers' unemployment may have a positive effect on boys education. Finally, mothers' education seems to be more important than fathers' in inducing education. 9 We also tried to control for birth order, with similar results. 16 Argentina Poverty Assessment Background Paper No. 7 b) Gender ofthe child being educated: Boys' education is better explained by the covariates as reflected in higher pseudo R2 and more significant variables. Age, parent's education and income affect boys and girls alike. Also, being in a non-married couple household affects education in a gender-neutral way. The number of siblings, the degree of concentration of family income, and whether the family is headed by a woman affect boys' education but not girls'. Being in a divorced family affects girls' but not boys' education. TABLE 8a: LOGIT ESTIMATES (MODELS 1-3) (z-values in parenthesis) Model 1 Model 2 Model 3 Variable Female Male Female Male Female Male Age -4.4414 -2.6581 4.6657 -2.7634 -4.7457 -3.0347 (-2.971) (-2.102) (-3.019) (-2.167) (-3.06) (-2.329) age2 0.1218 0.0618 0.1272 0.0648 0.1296 0.0737 (2.726) (1.629) (2.758) (1.691) (2.800) (1.881) lpoi 0.5764 0.3152 0.5958 0.3203 0.6389 0.3560 (3.795) (2.399) (3.824) (2.414) (3.971) (2.533) eduf 0.1234 0.1465 0.1402 0.1465 0.1488 0.1461 (2.856) (4.280) (3.157) (4.202) (3.302) (4.088) edum 0.1769 0.1219 0.1815 0.1412 0.1709 0.1266 (4.978) (4.483) (4.845) (4.685) (4.474) (4.051) funempl -0.1045 -0.6126 -0.0979 -0.5596 -0.0825 -0.3817 (-0.507) (-3.487) (-0.464) (-3.124) (-0.361) (-1.927) munempl 0.1314 0.3769 0.2056 0.3879 0.2335 0.4530 (0.518) (1.428) (0.777) (1.441) (0.875) (1.549) siblings -0.0279 -0.2544 0.0190 -0.2379 0.0152 -0.2056 (-0.367) (4.013) (0.242) (-3.69) (0.193) (-3.124) Femhead 0.1499 0.6473 -0.3825 -0.0027 -0.9607 -0.7374 (0.376) (1.820) (-0.807) (-0.006) (-1.485) (-1.251) Couple -0.7865 -0.5347 -0.7475 -0.5125 (-2.249) (-1.81) (-2.119) (-1.743) Single 0.9323 1.7700 0.7392 0.7598 (1.106) (1.712) (0.865) (0.692) widowed 0.1219 0.6115 -0.0170 0.4324 (0.245) (1.227) (-0.033) (0.809) divorced 1.3519 0.7495 1.1265 0.3939 (2.567) (1.703) (2.082) (0.838) Incomeh -0.0649 0.6639 (-0.149) (1.782) Inefem 1.2304 2.1498 (1.451) (2.670) Const 36.4285 24.9236 38.3474 25.6721 38.8808 27.0964 (2.927) (2.375) (2.979) (2.427) (3.009) (2.507) Pseudo R2 0.2436 0.283 0.2655 0.2935 0.2692 0.3166 Log-likelihood -255.03374 -335.78923 -247.64132 -330.86561 -246.40541 -320.0841 Observations 729 819 729 819 729 819 17 Argentina Poverty Assessment Background Paper No. 7 TABLE 8b: LOGIT ESTIMATES (MODELS 4-5) (z-values in parenthesis) Model 4 Model 5 Variable Female Male Female Male age -4.7651 -2.8312 -4.7780 -2.8745 (-3.078) (-2.183) (-3.076) (-2.203) age2 0.1303 0.0677 0.1309 0.0685 (2.820) (1.735) (2.823) (1.747) lpci 0.6161 0.3572 0.5885 0.3902 (3.783) (2.557) (3.549) (2.726) eduf 0.1174 0.0785 0.1160 0.0784 (2.348) (1.743) (2.302) (1.730) edum 0.1445 0.1061 0.1467 0.1018 (3.478) (3.261) (3.509) (3.077) funempl -0.0700 -0.3779 -0.0865 -0.3705 (-0.308) (-1.898) (-0.379) (-1.852) munempl 0.2325 0.4775 0.2202 0.4734 (0.868) (1.640) (0.818) (1.613) siblings 0.0181 -0.1889 0.0051 -0.1640 (0.231) (-2.88) (0.062) (-2.377) femhead -0.9611 -0.9129 -0.9748 -0.9475 (-1.487) (-1.538) (-1.507) (-1.582) couple -0.7498 -0.4966 -0.7354 -0.5026 (-2.134) (-1.696) (-2.046) (-1.698) single 0.5519 0.4746 0.5034 0.3518 (0.628) (0.425) (0.568) (0.313) widowed -0.2207 0.1173 -0.2305 0.1730 (-0.415) (0.213) (-0.432) (0.310) divorced 0.9992 0.1346 0.9855 0.1618 (1.802) (0.277) (1.772) (0.330) incomeh -0.0278 0.6465 0.0179 0.5981 (-0.064) (1.728) (0.040) (1.542) inefem 1.0057 2.0047 1.0031 2.0573 (1.166) (2.469) (1.160) (2.517) unskh -0.9387 -1.0758 -0.9577 -1.0844 (-1.656) (-2.512) (-1.687) (-2.51) laplata 0.0715 0.2684 (0.228) (0.866) mendoza -0.1460 0.1690 (-0.52) (0.657) eldest -0.1423 0.2347 (-0.561) (1.026) const 40.3645 26.9824 40.6425 27.1416 (3.122) (2.515) (3.129) (2.514) Pseudo R2 0.2737 0.3237 0.2748 0.3261 Log-likelihood -244.87732 -316.72217 -244.50022 -315.62573 Observations 729 819 729 819 18 Argentina Poverty Assessment Background Paper No. 7 3) Economic factors: Age has a strong detrimental effect on education. Income is also a fundamental determinant of education. Unemployment in itself has a detrimental effect that appears significant only for boys. The effect seems to be captured by the consequent reduction in income induced by unemployment. 4.2 The impact on education of changes in socioeconomic variables The previous analysis is based strictly on the significance and signs of the factors affecting education. It is also of interest to examine to what extent these variables affect the probability of attending school. Being the logit a non-linear model, the differential effect of each factor varies with the point at which it is measured. This subsection presents several exercises aimed at quantifying the effect of socio-economic factors on education. Again, all exercises will be based on Model 4 of Table 8b, which presents all the relevant covariates. First we explore the effect of income on education. Figure 3 shows the predicted probability of school attendance for different levels of per capita income. The solid line shows this probability for individuals with covariates set at the averages of individuals who are observed to attend school, and the dashed line corresponds to covariates set at the averages of individuals who do not attend school. The left pane corresponds to boys and the right one for girls. FIGURE 3: EFFECT OF INCOME ON EDUCATION BOYS GIRLS 1.0 ~1.0 Atn AAttend 0.8 0.8 --.- .. .... -.-. ..Do not Attend fDonotattend 0.6 0.6 ,* 0.4 0.4 0.2 0.2 0.0 0.0- 0.0 0 200 400 600 800 1000 0 200 400 600 800 1000 Per cpita ino Per month Per capita inrone per month Though statistically significant, income variations seem to have a minor effect on school attendance of both boys or girls. The predicted probability is for all levels of income always above 0.8. On the contrary, and especially for girls, income has a very strong positive effect on the schooling decision of those who do not attend school. Other things equal, in the case of girls, the probability of attendance rapidly raises with income. This effect is less powerful for the case of boys. As an illustration, Table 9 measures the attendance probabilities for the different deciles of the distribution of household incomes: 19 Argentina Poverty Assessment Background Paper No. 7 TA3LE 9: PREDICTED ATTENDANCE PROBABILITIES BY INCOME DECTLES Deciles Mean Male Female of pci Pci Attend Do not attend Attend Do not attend 1 39.39 0.82 0.51 0.83 0.45 2 76.22 0.86 0.57 0.88 0.55 3 106.95 0.87 0.60 0.90 0.60 4 138.40 0.88 0.62 0.91 0.64 5 171.46 0.89 0.64 0.92 0.67 6 211.89 0.89 0.66 0.93 0.70 7 258.49 0.90 0.67 0.94 0.72 8 318.33 0.91 0.69 0.95 0.75 9 433.15 0.92 0.71 0.95 0.78 10 891.64 0.93 0.76 0.97 0.85 In all cases, it is important to remark that although the effect is powerful, it requires a substantial effort in terms of income to compensate for the detrimental factors faced by those who do not attend school compared to those that attend. Families should receive (for each boy who does not attend school) more than $90 per capita per month for the boy to have the same attendance probability as an average boy already going to school. On the other hand, this amount is less than $20 to make the same compensation for a girl. This important difference in part reflects the stronger influence on boys of family structure, education and employment. These income compensations could also be interpreted as a summary of the effects of other variables influencing the attendance probability. In order to quantify these effects, we computed the changes in the school attendance probabilities induced by several changes in the socioeconomic variables studied in this work. For each change, we also computed how much income is necessary to give to (take from) the family so each effect does not alter the probability of attending school. The comparison will be made starting with a family with $200 of per-capita income per month, in which both parents are present and legally married and have three children. Both parents have completed primary school education and are both employed. We will compute the effects for 13 and 17 years old children. Table 10 shows some results. TABLE 10: MARGINAL EFFECTS AND INCOME COMPENSATIONS BOYS GIRLS Income Marg Effect Income maral Effect Changes Compensation 13 years old 17 years old Compensation 13 years old 17 years old Imother has 12 years of schooling -154.71 0.00862 0.10894 -138.12 0.00503 0.05180 father has 12 years of schooling * -193.35 0.01624 0.23825 -175.77 0.00434 0.08869 child has 3 siblings 139.43 -0.00430 -0.04398 -5.79 0.00017 0.00323 family headed by the woman -196.72 0.02120 0.34751 -186.46 0.00970 0.22969 family headed by the woman 2 -196.72 0.02906 0.42390 -186.46 0.00957 0.22727 parents legally married -150.21 0.01317 0.11920 -140.78 0.01074 0.15853 * Note that this effect interacts with the unskh effect. 1. Note that this effect interacts with the incfem effect In this case, the woman earns 75% of family income. 2. In this case, the woman is still the household head, but now she earns only 25% of family income. 20 Argentina Poverty Assessment Background Paper No. 7 In the first row we study what happens to the attendance probability if, starting with the "base" family, we change the mother's education from completed primary to completed secondary. As expected, this change increases the probability of attendance. The effect is particularly strong for 17 years old boys (almost 10%). In terms of income, once the mother's education is increased from primary to secondary, it would be necessary to subtract 154.71 per month to compensate this effect. The effect with father's education is more complex, since in most cases he is also the head of the family. When we switch his education from elementary to completed high school, the father also becomes "skilled". The joint effect is valued at 193.35$; again, for 17 years old boys, it represents a substantial increase in the attendance probability. The effect of having an additional sibling decreases the probability of attendance, and the effect is stronger on boys' education. The probability of attendance would be unaltered if the family is compensated with $139.43. Regarding effects of family structure, we explored what happens when the woman starts heading the family. As a result of the model specification, this effect comes through a binary indicator (femhead) and its interaction with income (incfem). Also, some assumptions must be introduced regarding how her participation in income (incomeh) results when the change is introduced. We computed the effects for the case when the woman head makes 75% of income and when she makes 25% of total income. The results are very significant for boys, for whom the attendance probability increases dramatically. The income value of this effect is $196.72. Note that even though the effect on the probability is substantial, the income compensation is not as strong since the introduced change alters the marginal effect of income through the interaction term with femhead. We also explored the effect of having legally married parents as opposed to unmarried ones. For this, we started with the "base" family but with unmarried parents, and computed the change in the attendance probabilities resulting from considering legally married parents. As expected, this increases the probabilities, especially for girls. In terms of income, this amounts to a change in $150.2. Age effects are not treated directly, since initially this variable is included as a control for having to deal with heterogeneous individuals. In all cases, the decrease in probability as the child gets older is substantial, which, as seen in the previous results, amplifies substantially the effect any change in other variables may have on the attendance probability 5 Conclusion Every political or social sector in Argentina agrees that the government should do more to induce children and their families to acquire more education. The focus on educational policies is a key element for the current government and for every political party in the opposition. So the relevant question is not whether more should be invested in education but how to invest more in education. This paper contributes to the literature by providing empirical results on the economic and social determinants of the decision to send children to school for the Argentine case. As expected, for a child of a given age, income and parent's education play a key role in the decision to send boys and girls to school. The paper explores the role played by family structure and gender issues. The issue is not trivial, since there are a priori reasons to suspect that many of these 21 Argentina Poverty Assessment Background Paper No. 7 effects are captured by variations in income. The results of this paper suggest that differences in family structure influence the way children are educated beyond their effect on family income. In particular, the effect of living in a family with non-married parents has a detrimental effect for both boys and girls. The paper presents interesting results regarding the much-debated subject of women heading families. Being in a woman-headed household does not increase or decrease the likelihood of education per se, but when faced with additional resources, women heads assign a larger part of this increment to education than men would. This has relevant policy implications regarding the structure of income subsidies for education. Gender considerations appear in several ways in the process of determining education. In addition to the previous result regarding women heads, mothers' education is more significant than fathers', which is consistent with higher income countries. The proposed set of covariates seems to account for boys' education better than girls'. This suggests that changes in the economic and social environment affect boys' education more than girls', which may help explain the observed disparities in attendance rates between boys and girls, the former being higher than the latter. The paper goes beyond the qualitative analysis of what factors are relevant and what are not. Our results provide a quantitative exploration of the effects of several changes in family structure and its gender composition. As expected, income has a powerful effect in inducing education, but if there are no changes in the negative socioeconomic aspects that affect education, it requires a substantial investment to compensate for the effects of these factors. It also suggests that it is important to consider social factors (like the number of siblings, parental education and family structures) in designing efficient subsidy policies aimed at keeping children at school, since the effects of these factors vary considerably by income and age groups. 22 Argentina Poverty Assessment Background Paper No. 7 References Becker, Gary S., 1981, A Treatise on the Family, Cambridge: Harvard Univesity Press Becker, Gary S., 1965, "A Theory of the Allocation of Time", Economic Journal, 75: 493-517. Bryant, W., 1990, The Economic Organization of the Household, Cambridge: Cambridge University Press. Deaton, A., 1997, The Analysis of Household Surveys, Baltimore: The Johns Hopkins University Press for the World Bank. Garasky, S., 1995, The Effects of Family Structure on Educational Attainment: Do the Effects Vary by the Age of the Child?, American Journal ofEconomics and Sociology, 51, 1. Gasparini, L., A. Fassio y D. Bonari, 1998, El Impacto Distributivo del Gasto en Educaci6n en Mendoza, mimeo, Ministry of Economy and Public Works, Argentina. Hess, R. and K. Camera, 1979, Post Divorce Family Relationships as Mediating Factors in the Consequences of Divorce for Children, Journal of Social Issues, 35(4): 79-98. Instituto Nacional de Estadistica y Censos (INDEC), 1997, Encuesta Permanente de Hogares: Manual del Entrevistador, Publicaciones del INDEC, Buenos Aires. Instituto Nacional de Estadistica y Censos (INDEC), 1997, Encuesta Permanente de Hogares: Desarrollo Actual y Perspectivas, Publicaciones del INDEC, Buenos Aires. Leibowitz, A., 1974, Home Investments in Children, in T.W. Schultz, ed., Economics of the Family, Chicago: University of Chicago Press, 1974. Loh, Eng Seng, 1996, Changes in Family Structure, Attained Schooling, and Adult Poverty Status, Social Science Quarterly, 77, Vol.1. Nock, S., 1988, The Family and Hierarchy, Journal ofMarriage and the Family, 50(4): 957-966. Pessino, Carola, 1995, Returns to Education in Greater Buenos Aires 1986-1993: From Hyperinflation to Stabilization, CEMA Working Paper 104. Buenos Aires, Argentina. Pitt, M. and Khandker, S., 1997, The Impact of Group Based Credit Programs on Poor Households in Bangladesh: Does the Gender of Participants Matter?, mimeo, World Bank. Rodriguez, J., and Abler, D., 1998, Asistencia a la Escuela y Participacion en el Mercado Laboral de Menores en el Peru entre 1985 y 1994, mimeo, The Pennsylvania State University. Tansel, A., 1997, Schooling Attainment, Parental Education and Gender in Cote D'Ivore and Ghana, Economic Education and Cultural Change, 45(4), pp. 825-856. Torrado, S., 1998a, Familia y Diferenciacion Social. Cuestiones de Metodo, Eudeba, University of Buenos Aires, Argentina. Torrado, S., 1998b, Las Jefas de Hogar Piden Politicas Especificas, Diario Clarin, Buenos Aires, 17 December, pp. 25. Weiss, R., 1979, Growing Up a Little Faster: The Experience of Parental Structure During Childhood and Adolescence", Demography 29(1): 59-68. 23 Argentina Poverty Assessment Background Paper No. 7 Appendix Construction of variables from the EPH Dependent variable attend. Is a binary variable with value 1 is the individual is currently at school, finished secondary education.or started college. Explanatory Variables * sex: =1 for men * age and age2: age and age squared. * pci and pci2: per-capita family income (variable ipcf in the EPH). All income measures come directly from the EPH, excluding incoherent observations. * lpci: log of per-capita income * eduf father's education. In the EPH " father" is either the male head or the husband. In the EPH education is a qualitative variable (nivel) which (following previous work on the EPH (Pessino, 1995, for example), was translated into a quantitative variable according to the following convention: 3.5 years if nivel=12 (incomplete elementary) 7 years if nivel=1 1 (comlete elementary) 9.5 years if nivel=22 (incomplete high school) 10 years if nivel=42 (incomplete technical high school) 12 years if nivel=21 (complete high school) 13 years if nivel=41 (complete technical high school) 14.5 years if nivel=32 (incomplete superior or college) 17 years if nivel=31 (complete superior or college) * edum: mother's education. "Mother" is either the female head or wife. Her education is constructed in the same way as eduf * funempl: =1 if the "father" is unemployed. The base category is "employed or inactive" * munempl: =1 if the "mother" The base category is "employed or inactive" * siblings: number of brothers or sisters the individual has. * femhead: =1 if female head * married: =1 both parents are present and the couple is legally married * couple: =1 if both parents are present but not legally married * widowed: =1 if the head is single and widowed * divorced: =1 if the head is single and divorced * single: =1 if the head is single and is not widowed or divorced. * incomeh: proportion of total family income earned by the head * incfem: lpci*femhead * unskh: =1 if the head has education less than complete high-school * laplata: =1 if the observation corresponds to La Plata * mendoza: =1 if the observation corresponds to Mendoza * eldest: =1 if the individual is the eldest child in the family. 24 Argentina Poverty Assessment Background Paper No. 8 Poor People in a Rich Country A Poverty Report for Argentina Background Paper No. 8 Marginalization of Men in Argentina: Evidence from Changes in Wage Distributions By Omar Arias The World Bank LCSPR January 2000 Argentina Poverty Assessment Background Paper No. 8 1. Introduction After completing economic stabilization in the 1980s, Argentina as most of the countries in Latin America and the Caribbean have been immersed in a process of economic restructuring driven by the so called "second generation" reforms including the privatization of public enterprises and the provision of some social services, trade, tax and financial reforms, and deregulation of the economy to allow prices to be more aligned with market forces. In the transition to a more productive economy, jobs in traditionally protected sectors are being replaced by jobs in new more competitive sectors. This has likely brought the disappearance of a large number of jobs, obsolescence of existing skills and an increase in structural unemployment and therefore have induced significant changes in the structure of wages. While female workers have also been affected, there is an increasing concern of the possible impacts that this process may have on the capability of certain group of men to fulfill their traditional socially prescribed role of "providers" and thus on their self-identity and role in the family. For instance, Barker (1998) indicates that "men delay family formation or maintain looser ties to their children in part according to their employment and earnings." Moreover, qualitative research by sociologists have suggested that the inability of young men to live up to their socially prescribed "bread winner" role may underlie their increasing involvement in gangs and the significant increase in violence and substance abuse in the region (Barker, 1998). Macro-level research in Mexico suggests that economic restructuring has negatively affected certain groups of men (Cunningham and Maloney, 1998) the most. Although often focused on non-randomly-selected vulnerable male subpopulations (e.g., low-income young men), this sociological research alludes to overall trends of declining male labor force participation, and increasing unemployment or wage deterioration of these specific groups as evidence of economic intra-male marginalization. However, it has not been carefully ascertained whether or not this is an economy-wide phenomenon. Unemployment, which has become critically important in Argentina, has been analyzed recently by Pessino (1997) and others while changes in the structure of wages have received less attention. Being the prices for different kinds of labor, wages are a key determinant of income potential and thus affect current and future levels of consumption through their effect on accumulation of assets (human capital) and the return to those assets. Therefore, changes in the wage structure have important implications for the incidence, persistence and the patterns of poverty in the population. Although a complete analysis of these questions would then consider both trends in unemployment and wages, here we focus on changes over time in wage distributions as our key indicator. The objective of this study is to use evidence from household surveys in Argentina for the period 1988-1997 to document any existing process of intra-gender male wage marginalization in the context of the reform process that started in the late 1980s-early 1990s. We shall seek evidence to answer the following questions: 1) Is there evidence of a deterioration in wages of certain groups of men?, 2) Have certain groups of men lagged behind in accruing the wage growth of economic reforms?, 3) What are their socio-economic and labor market characteristics?. We examine changes in the whole distribution of wages and not just average wages trends. Specifically, we characterize changes in the wage distributions of distinct groups of workers using kernel density estimation and quantile methods. We shall be especially interested 1 Argentina Poverty Assessment Background Paper No. 8 on the evolution of real wages of those "typical" individual males located at the bottom quantiles of the wage distribution, particularly, workers who earn wages lower than granted by their demographic characteristics and observable levels of skills. A focus on "average" wage trends may obscure the particular conditions that affect these men. Our main findings indeed suggest that unobservable heterogeneity that varies across groups of workers (e.g., differences in ability, labor market connections, quality of schooling) affects non-trivially the capacity of male Argentine workers to adapt to economic restructuring and generates differences in wage performance. The paper is organized as follows. In section 2 we briefly outline the major aspects of the reform process and changes in the Argentinean economy that took place over the period. In section 3 we highlight the main features of the methodology and the econometric analysis employed. In section 4 we discuss the data and empirical model specifications. Section 5 presents the findings and section 6 concludes with some policy implications. 2. Overview of the 1988-1997 Period Here we briefly outline the major changes in the macroeconomic context in relation to the Argentinean labor market during the period of analysis.' The cornerstone of the economic policy program during the period has been the so called convertibility plan. The key to this plan, introduced in 1991, is a new monetary system based on a fully convertible exchange rate regime which fixes the rate of the Argentinean peso at 1:1 to the US dollar. It also comprises a wide variety of structural adjustment measures including fiscal, public sector, trade, and financial reforms. The initial success of the plan in quickly moving the economy from historical levels of hyperinflation (1989-90) to remarkable price stability was followed by remarkable sustained economic growth between 1990 and 1994 (around 35%), was halted by the 1995 recession (-4%) triggered by the Mexican crisis (the "tequila effect"), but resumed in steady growth between 1995 and 1997 (around 13%) with remarkably low levels of inflation. According to Pessino (1997) the major developments in the Argentinean labor market associated with the implementation of the convertibility plan are: 1) an increase in the relative price of labor with respect to capital motivated by the elimination of tariffs on capital goods and the real appreciation of the peso; 2) an increase in unemployment and underemployment mainly due to a slow employment growth but aggravated by the increase in labor force participation, particularly of women; 3) technological change that favors increases in the demand for skilled labor; 4) an increase in average labor productivity which however has not been sufficient to reduce unit labor costs; and 5) very slow progress in the reform of labor market regulations that block labor mobility and impose high fixed labor costs that constrain labor demand. The behavior of key indicators of labor market outcomes in Argentina during this period have been well documented.2 Although real incomes increased and poverty rates declined after the introduction of the convertibility plan in 1991 (World Bank (1995)), 1 See World Bank (1996), Report No. 15402-Ar and Pessino (1997) for a more comprehensive and detailed review of changes in economic policy and the economy during the period. 2 Some studies have documented more specific characteristics of poverty and labor market outcomes (e.g., Altimir and Beccaria (1997); CEPA, several years; Pessino (1997)). 2 Argentina Poverty Assessment Background Paper No. 8 unemployment has increased substantially to reach historical levels (around 16% in 1997 up from 7% in 1991) and inequality also increased (Gasparini, 1998). Fewer studies exist that investigate the link between trends in labor market outcomes, particularly changes in the structure of wages, and characteristics of the labor force during the recent years. Pessino (1996) studies changes in the structure of wages during the years of very high inflation (1986- 90) and the posterior years of stabilization and growth (1991-93). She specifically estimates average returns to education for this period and links their behavior to macroeconomic performance. She finds that the average returns to education, particularly for tertiary education, increased during the high inflation period (1986-1989), declined during stabilization, and continued to increase slightly thereafter. As a consequence, average wage differentials between skilled and unskilled labor increased. Our work aims to contribute to this discussion. Below we discuss the results of an analysis that extends considerably the investigation of the factors underlying real wage trends in Argentina by focusing on changes in the whole wage distribution rather than just on trends in average wages. 3. Conceptual Framework and Methodology The concept of "marginalization" in development work is often associated with various manifestations of poverty such as a high proportion of the population with unsatisfied basic needs (e.g., insufficient dwelling space, lack of or poor access to water and public services) or with an insufficient income to meet a basket of basic goods and services (e.g., under the poverty line). Without questioning the validity of such indicators, we consider changes in male income generation capability which depends on the ability to secure stable and good quality employment. Although a complete analysis should consider both trends in unemployment (or more generally the number of hours of work) and wages, we focus on tracking changes over time in male wage distributions to examine the hypothesis of intra-male economic marginalization. Unemployment, which has become critically important in Argentina, has been analyzed recently by Pessino (1997) and others while changes in the structure of wages have received less attention. Being the prices for different kinds of labor, wages are a key determinant of income potential and thus affect current and future levels of consumption through their effect on accumulation of assets (human capital) and the return to those assets. Therefore, changes in the wage structure have important implications for the incidence, persistence and the patterns of poverty in the population. We shall interpret as key evidence of wage marginalization: 1) a systematic decline in the level of wages of a given group of workers relative to overall wage trends, and 2) a change in the wage distribution, not compensated by an increase in wage levels, that increases the proportion of workers (in the population or within a given group) in low-wage jobs.3 The former could result from a shift to the left of the whole or a portion of the wage distribution. The latter, seemingly less obvious, can occur if the distribution of wages become more Given the limited time data points, we do not emphasize secular changes of wages over time, but rather focus on any discernable systematic trends. 3 Argentina Poverty Assessment Background Paper No. 8 concentrated around low-wage levels (ie., skews to the left) or if the left tail grows longer without the distribution shifting "significantly" to the right.4 This work is then related to a vast literature that tries to explain trends of increasing earnings inequality in the U.S. and other countries (e.g., Bound and Johnson, 1992; Levy and Murnane, 1992; Juhn, Murphy and Pierce, 1993). The approach has been to account for changes in the mean or the variance of wages as a byproduct of changes in labor supply, labor demand, demographic and labor market institutional factors. However, trends in average wages, even after controlling for observable worker characteristics through conventional regression analysis, are unlike to provide a complete picture of variations in wage outcomes. Unobservable heterogeneity that varies across groups of workers (e.g., differences in ability, family labor market connections, quality of schooling) may play a nontrivial role in explaining wage outcomes and the capacity to adapt to economic restructuring. For instance, some types of economy-wide shocks may have different effects on the wages of workers (of a given skill) employed in low-wage jobs compared to those in the better paid jobs. For example, the labor demand for unskilled workers could be more sensitive for jobs at the top of the unskilled wage distribution than at the bottom so that among the unskilled, workers in low-wage jobs may suffer relatively greater wage falls during recessions than those in the better paying jobs, and conversely, may benefit less from economic expansions. Others have looked at how these factors have affected the whole distribution of wages rather than a few summary measures of such distribution (e.g., Dinardo, Lemieux and Fontin, 1996). A related approach tries to accomplish this by estimating regression lines at different quantiles of the conditional distribution of wages rather than just at the mean of that distribution. This can be accomplished by estimation of quantile regressions (Koenker and Bassett, 1978) and is the approach we follow in this work. This line of work has indeed provided mounting evidence that the wage structure can be hardly characterized by single mean relationships.' Although we are not interested in inequality per se, the approach allows us to identify those groups of male workers who are losing or lagging behind in wage or earnings growth, that is, those affected by wage marginalization. A focus on average wage trends may obscure the particular conditions that affect these men. We shall be especially interested on the evolution of real wages of those "typical" individual males located at the bottom quantiles of the wage distribution, or more generally those workers who earn wages lower than granted by their observable levels of skills and demographic characteristics. As we shall discuss below, our findings indicate that unobservable heterogeneity plays an important role in the response of wages of different types of workers to changes in the Argentinean economy. 4 Essentially we rule out cases in which all workers are substantially better off because of unprecedented economic growth (the distribution shifts to the right) but there is a significantly higher fraction of workers earning wages that are low relative to the new wage levels in the economy. Attempts to provide more precise definitions quickly encounter the technical difficulties associated with ranking distributions in the applied welfare literature. 5 Buchinsky (1994), Mwabu and Schultz (1997), Arias, Hallock and Sosa (1998), Fitzenberger (1998), Mata and Machado (1999), Maloney and Ribeiro (1999). 4 Argentina Poverty Assessment Background Paper No. 8 3. Data description and Empirical Model Specifications We use data from the Encuesta Personal de Hogares (EPH) for greater Buenos Aires, years 1988, 1991, 1993, 1995 and 1997.6 Since the period of analysis covers two recession episodes (1989-90 and 1995) and two episodes of sustained economic growth (1991-93 and 1996-97), we can explore whether wages of particular groups follow systematic patterns over the business cycle which can be informative to anti-poverty policies. We only consider male salaried workers age 15 to 70 for whom salaries are the only source of income (thereby excluding unpaid workers, the self-employed (cuenta propia) and owners of small firms (patrones)y), and who report a positive number of hours of work during the week previous to the survey.8 The analysis is based on hourly wages computed using earning and hours reported in the main occupation. We drop observations with inconsistent income reports and missing data on any of the variables of interest.9 In the wage regression analysis worker characteristics are captured by dummy variables to facilitate isolating changes in counterfactual wage distributions for narrowly defined groups based on human capital attributes such as education, age, occupational skills and tenure, firm size, and other demographic controls (i.e., marital and headship status). We include measures of acquired human capital both general and specific (see Appendix 1 for definitions). Differences in general human capital are measured through five education dummy variables (primary, complete or incomplete high school and college education)."o The sample of men are partitioned into five conventional age groups to inform about specific synthetic cohort changes in the structure of wages. Of particular interest is whether young or older workers face any wage disadvantage that can not be explained by differences in other factors. Accumulated human capital formed either through work experience or "learning by doing" is captured by a worker's reported tenure in his current occupation. This is classified in four levels that refer to new entrants (less than a year), those with some experience (between 1 and 5 years), those with 5-10 years of experience, and those with more than 10 years. These variables would be better proxies of labor market experience for workers with low 6 We merge the May-Oct waves in each of these years to obtain bigger sample sizes. 7 Salaried workers comprise between 60-65% of employed men over this period while the proportion of cuenta propia has remained stable around 20%. Including the self-employed adds considerable unobserved heterogeneity (e.g., unobservable entrepreneurial ability, tastes differences) that is likely to affect individual sectoral choices of employment and can be confounded with any "true" sectoral differences in earnings (See Maloney, 1998a, 1998b). Arias (2000) provides a very similar analysis of wage outcomes that includes the self-employed. 8 The relevant population for inference purposes is the salaried employed male population which may or may not be representative of the whole male population of salaried workers in the labor force (including the unemployed). Biases arising from self-selection into the labor force are typically not as big a concern for men as they are for women (a stable 85% of men participate in Argentina). In any case, attempts to correct for these or for correlation between unobservable determinants of unemployment spells and wage regressors using parametric sample selection methods (Heckman, 1979) are increasingly recognized as unsatisfactory due to their sensitivity to statistical assumptions. 9 Estimates of labor market statistics were not significantly affected by exclusion of these non- responses. '0 In the primary school group we include those with complete and incomplete degrees since the latter represent a very small fraction. 5 Argentina Poverty Assessment Background Paper No. 8 occupational mobility. For workers in a new occupation, general labor market experience would be better captured by the age dummy variables. We also consider interactions between human capital acquired through formal education and labor market experience. We define twenty (education-experience) "skill groups" by interacting the educational attainment and the experience level dummies. This allows for any existing complementarities between education and labor market experience that give rise to wage differentials across workers. For instance, we distinguish between new entrants with a college degree (univ1) and those who only completed secondary education (secunl). Skills that apply to specific tasks (e.g., acquired through "on-the-job-training") are captured with occupation specific dummies, which distinguish between unskilled, skilled, and professional workers." Six industrial dummies are used to capture changes in inter-industry wage differentials. These could arise from emerging productivity differentials and sectoral labor demand shifts typical of transition economies like the Argentinean. We also include a dummy for salaried workers employed in small firms (with less than 6 employees). The low capital/labor ratios that often come with a small scale of operations and particular conditions of product demand can induce wage differentials and different wage responses to economic shocks.12 Many consider such jobs to be part of the informal sector since typically workers do not enjoy benefits such as social security and health insurance but also escape income taxation. We also include demographics to capture wages differences arising from marital status, and whether the worker is the head of a multi-person household.3 We distinguish between married men and those in consensual union to allow for any effect of changes over time in family formation. Previous research has typically documented a wage premium for married males presumably arising from the signal of marriage as a signal of stability and commitment or due to intra-household specialization that allows men to focus more on labor market activities and thus be more productive (Korenman and Neumark (1991)). Differences in wages between household heads and no heads could also arise as a return to unobserved worker characteristics that may have poverty implications. Finally, we include a dummy variable for observations from the October surveys to control for any sort of seasonal difference in measurements across surveys. Table 1 presents summary statistics for these variables in the year covered by each sample. First, note the substantially smaller sample for 1991 due to the much lower sample coverage of the EPH waves that year. Although this creates some problems for comparing estimated sample statistics with other years, for the most part these are not of crucial importance.14 The trends generally reflect an increasingly more educated and somewhat aging male labor force. There is a decline in the proportion of males with primary education and an increase in " This is based on workers' self-reported occupation and tasks in their occupation. We also tried an alternative breakdown closer to the usual blue collar (workers in production, transportation and commerce related occupations) vs. white collar (those in administrative occupations) distinction, with similar results. 12 Indeed, close to 70 percent of small firm salaried workers do not possess neither of three mandated benefits: pension, social security and health insurance. " Our definition ofheads excludes cases where the male worker lives alone or only with his mate. In the survey, headship status is determined by directly asking the members of the household. 14 This is especially the case for the skill groups specification in 1991 since the number of observations is substantially reduced for some categories. In the other years, the number of observations in most of the categories exceeds 50. 6 Argentina Poverty Assessment Background Paper No. 8 the fraction with completed high school. However, by 1997 only 18 percent of employed males had completed no more than high school. Although the age structure appears to be fairly stationary over the period, there is a small increase in the share of workers 25 to 35 years of age. Together with the enhancement in general skills, there has been an improvement in the levels of specific-task skills as reflected by an increase in the fraction of skilled workers that exceeds the decline in the fraction of professionals. There is also a secular reduction in the proportion of employed male workers that are married and in the proportion of household heads, together with an increase in the proportion in consensual unions. There is a process of reshuffling of the Argentinean male employment as indicated by the increase in the proportion of new workers (less than 5 years in the occupation) over time and the decline in the fraction of those with longer tenure in the occupation. Most of the new workers have at least some secondary education. The decline in the fraction of more experienced workers is due to those with incomplete high school and primary education." The change in the sectoral structure of male employment is consistent with the patterns observed during productive transformations that take place during economic restructuring. Although still higher than in most Latin American countries, the share of manufacturing in employment declines while the fraction of employment in services increases. The shares of employment in public instruction, administration and defense, and in low productivity services remain stable over time.16 Employment in construction is somewhat pro-cyclical. There is an increase, although not systematic, in the proportion of workers employed in micro-firms. 4. Real Wage Trends and Changes in Wage Distributions 4.1 Overall Trends Table 2 presents statistics describing the behavior of real wages in Argentina (at 1988 prices) during this period.17 Average real wages increased by 14% between 1988 and 1997 reflecting the positive effect on earnings of the overall significant economic growth during this period. Most of this increase took place during the economic boom of 1992-93. Average wages remained constant between 1988 and 1991 (due to the apparent offsetting of the deterioration during the hyperinflation years and the unprecedented economic recovery in 1991), declined slightly with the 1995 recession and continued to decline during the economic expansion of 1996-97. The wage dispersion, as measured by the standard deviation, has declined during the period, after increasing between 1992 and 1993. The key question of our concern is: Have these overall trends in average wages been uniform across all workers?. We shall try to answer this question in stages. Table 2 tracks the evolution of six different quantiles (bottom, middle and top) of the distribution of wages for each year.18 Clearly, workers at the bottom of the wage distribution benefited the most from the 15 Although the irregular trends in the fractions of new entrants with primary and incomplete high school may be indicative of their little employment stability, they may be also due to sampling errors. 16 Note that the constant share of male public employment is not inconsistent with a possible increase in the share of former public workers among the unemployed. 17 IPCs are used to obtain real values. 1 The pth quantile of wages is defined as the level of wage that is perceived by at most ap fraction of workers. The best known median case refers to the wage that is perceived by at least 50% of all male employed workers. 7 Argentina Poverty Assessment Background Paper No. 8 overall wage bonanza. While the real wage increased by more than 20% between 1988 and 1997 for workers at the bottom half of the distribution, it increased by 3% for those at the top. However, the low-wage suffered more (-13%) than the high-wage (-6%) from the real wage decline in the second part of the decade (1993-97). As a consequence of these trends, there was a reduction in inequality between workers at the extreme of the wage distribution. The 90t to 10 h wage ratio declined from 5.6 in 1988 to 4.6 in 1997. Most of this decline occurred between 1988 and 1991. This tendency has been much less pronounced for the ratio of wages between adjacent quantiles so that wage inequality between workers earning relatively similar wages has changed little. In short, the results so far suggest that: 1) the wages of workers at the bottom of the wage distribution have been more sensitive to changes in the Argentinean economy during the 1990s, 2) wage inequality may have declined among the Argentine male salaried employed population. 4.2 Group Specific Wage Growth Patterns We investigate whether particular groups of the population of employed male workers have been affected any differently. Figures la-id depict, for each year, wage densities for different groups of workers according to age, education, tenure and skill in the occupation.19 These essentially describe the changes over time in the proportion of male workers earning a given wage within each group. The wage densities for all workers are plotted in the first column of each figure for the purpose of comparisons. In each panel the 1988 wage density for any given group (defined by dots) is plotted together with the densities for the first year of each sub-period (the dotted line curve) and the density for the end year (the solid line curve). years. For instance, in all cases, row 2 portrays the 1988, 1993 and 1995 wage densities. Overall and for all the groups considered, the wage distribution shifted to the right between 1988 and 1997 confirming that wages have been rising for male salaried workers in Argentina (column 1, Figs la-id). The bulk of wage increases occurred between 1991 and 1993 and appeared to have benefited every one, especially the low to middle and upper middle wage workers. Wages of male workers at the top of the wage distribution have changed modestly during the entire period, although there is an apparent slight increase in the proportion of high-wage workers. Between 1993 and 1995 no significant changes appear, although workers at the bottom tail of the distribution appear to have lost slightly. Between 1995 and 1997 the wages of workers at the bottom and the middle upper tail of the distribution declined somewhat more. The wage gains of workers at the middle of the distribution in the early 1990s were maintained in the later period. Nevertheless, wage growth has been neither steady nor homogeneous either across groups or within groups over the period. In some cases it is not possible to draw unambiguous conclusions (uniform shifts) about the wage performance of a given group. Although the densities for all age groups generally shifted to the right, wage bonanza was stronger among young and older workers, especially at the bottom tail. This led to a decline in the proportion of young low-wage earners (<$4/hr.) over the period, despite its increase with the 1995 recession. Among middle age and older workers, real wage increases '9 We use density kernel methods with the optimal band width chosen based on the Normal Kernel. 8 Argentina Poverty Assessment Background Paper No. 8 led to a bigger fraction of workers in the middle to upper middle of the distribution.20 Across education groups, shifts in the overall wage distribution mainly reflect the gains of less educated salaried workers (less than secondary). Real wages for the least educated increased uniformly between 1988 and 1993, and only at the bottom for workers with secondary or more. Conversely, workers with less than secondary education suffered most of the real wage declines between 1993 and 1997. With regards to tenure in the occupation, recent workers, particularly new entrants at the bottom tail, are the most benefited. Some of these gains were dissipated between 1993 and 1997, with a significant increase in the fraction of low-wage workers (<$4/hr.). Among the more experienced, real wages increased significantly for workers at the bottom half of the distribution between 1988 and 1993, and changed little thereafter. Finally, the pattern of real wage gains is very similar across occupational skill groups (e.g,, between the unskilled and the skilled/professionals). The shifts mainly occurred between 1991 and 1993.21 Despite the increase in the fraction of low-wage earners (<$4/hr.) among the unskilled between 1995 and 1997, this still remains lower than in 1988. In short, a more detailed analysis of trends in wage distributions confirm that low-wage male workers in Argentina experienced relatively larger real gains after the introduction of the convertibility plan in 1991. It also suggests that: 1) among the low-wage, wage gains were higher for the young, new workers, and those with lower levels of human capital, 2) the low- wage earners in these groups were also the most affected by the backslide of gains in the later years, 3) real wages of high-wage earners with more human capital have been less sensitive to changes in economic conditions and their fraction may have grown over the period. Of course, there is a high overlap between the above groups. More educated workers also tend to be the more skilled, older workers tend to have more accumulated labor market experience, etc. For instance, wage growth could have been higher for workers of the same age and education employed in an expanding economic sector. Occupation specific skills also correlate with general human capital. We would like to isolate wage differentials arising solely from differences in education, experience, age, and all other relevant factors by taking into account their interrelation. More generally, we would like to explore how differences in the distribution of unobservable factors such as individual spunk or school quality affect the wage performance of observationally equivalent workers. In a spirit similar to the above wage density comparisons, we would also like to explore whether existing wage differentials within "counterfactual" narrowly defined groups have changed any differently over the time period. Obviously, we face severe data limitations to control for even some of the possible interrelations by partitioning workers into many different groups for which wage densities (or even average wages) can be estimated. Quantile regressions provide with a powerful tool to come close to such characterization by revealing the relative wage performance of more narrowly defined "archetypical" workers. The results of this multivariate quantile analysis below indeed pinpoint important differences in the wage growth experiences of workers at different points of the conditional wage distribution. 20 Over the period, there is an apparent increase in the fraction of high-wage earners among the old (the right tail becomes thicker) which was robust to changes in the degree of curve smoothing. 21 The likely different wage distribution patterns for professional workers is most likely obscured by the wages of the higher proportion of skilled workers. The small and declining proportion of professionals in the sample creates difficulties in estimating accurate wage densities for this group and in comparisons between groups and over time. 9 Argentina Poverty Assessment Background Paper No. 8 5. Changes in Quantile Wage Differentials For each year we estimate six quantile regressions (quantiles: 0.05, 0.1,0.25,0.5,0.75,0.9) of the log of wages on the independent variables of Table 1 for two different specifications.22 We also estimated conventional least squares (mean) regressions for these two specifications but do not discuss them separately since they are generally similar to median regression estimates. For each worker attribute one category is excluded for proper estimation of the regressions. In the central specification, the comparison (control) groups used to assess relative wage performance correspond to single, non-head, professional workers 45 to 55 years old with college or more, more than 10 years of tenure in the current occupation, and employed in financial/business services in firms with more than 5 employees. In the second specification, based on twenty education-experience (skill) groups, workers with at least college and more than 10 years of experience served as reference the group. Although the precise magnitude of differentials obviously refer to the base group, conclusions about relative wage performance obtained from between group comparisons of changes in estimated wage 23 differentials are clearly independent of the choices of base categories. Two remarks can be made to help interpretation and discussion of the findings. First, each quantile coefficient has the usual regression interpretation. It measures the difference in the log wages of a worker with the particular characteristic with respect to an otherwise similar worker with the excluded category. For instance, the coefficient on complete secondary education measures the wage differential between a worker with completed high school and an otherwise similar worker with completed college. The anti-log of a coefficient thus gives the relative wage gap of high school workers with respect to college, after adjusting for everything else.24 Second, we can think of the different quantiles as pertaining to different "types" of workers located in various points of the distribution of wages for a (counterfactual) group sharing similar worker characteristics in the male employed population. Thus, the low (5tI 0t) quantiles pertain to workers in jobs earning wages lower than granted by their human capital and other observable characteristics. Similarly, the 90h quantile can be thought to correspond to workers in "privileged jobs", that is, those earning wages higher than predicted by their observable characteristics. By tracking the pattern of wage differentials associated with distinct worker characteristics over the period at different points of the wage distribution we can then draw implications about: 1) Differences in wage changes over time within groups, that is, comparisons between low, middle and high quantiles, 2) wage changes between groups in each part of the wage distribution. The raw estimated coefficients (with p-values) together with mean (OLS) estimates are presented in Tables in Appendix III. We now summarize the findings. 22 All the estimations and tests were carried out in Splus V. 4.5. All the reported test statistics are based on related regression rank score procedures. See Appendix II for a brief description of the techniques. Koenker and Portnoy (1997) for a detailed discussion of quantile methods. 23 Our choice of base group seeks direct identification of wage differentials pertaining to workers that are less well endowed with human capital attributes or possess characteristics that make them more likely to be sensitively affected by economic shocks. The wages of these workers are also likely to exhibit relatively stable patterns over time. 24 Minus this coefficient yields the "return" to college education. Although the required conditions are rather strong, it is customary to interpret coefficients on log wage regressions as measures of returns to education. Note that when a coefficient b is bigger than 1-.51 a more accurate measure is exp(b)-1 (Kennedy, 1981). 10 Argentina Poverty Assessment Background Paper No. 8 Intercepts and Demographics . In our formulation, the intercept represents the real (log) wage of workers with the attributes omitted from the regression. Except for their real wage decline during the recession of 1989-90, the upper quantiles within this group show very stable hourly earnings over the entire period, with the 75t and 90 quantiles even gaining slightly between 1994-1995. In contrast, the bottom 5t- 1 Ot quantiles show much higher sensitivity to the business cycle with real wages that are pro-cyclical, ie., wages fall sharply during recessions and increase during expansions. Interestingly, the 25h quantile shows consistent real wage gains throughout most of the period. Adjusting for other worker characteristics, married male workers enjoy a wage premium over single workers, which is often higher at the lower quantiles. After seemingly fading between 1993 and 1995, this rose from roughly 9 to 13 percent between 1988 and 1997, on average, reflecting an increase in the first decile. There are no significant differences in wages of single workers and those in a consensual union. Heads of non-single headed households did relatively worse than non-heads. While the former enjoyed a wage premium similar to that of married workers, this declined consistently over the period. The faster growth of wages of married male workers contrasts with the well documented decline in the married wage premium in the U.S. There, this has been found to be largely due to a decline in the male productivity effects associated with marriage that followed from a reduction in the average degree of specialization across households (Korenman and Neumark, 1991) as women entered the labor force. As reported in Table 1, available data in Argentina indicate a change in the patterns of family formation, with marriages in relative decline and consensual unions becoming relatively more common. Thus, the rising wage premium for married males may reflect a return to the value of marriage as a signal of personal attributes such as stability and commitment. It is unclear whether the decline in the wage premium for heads reflect a loss in the value of the "signal" of this individual characteristic to employees. Age Comparisons of adjusted quantile wage differentials across age groups reveal interesting patterns. On average, real wages are similar between workers age 35 and older workers but these tend to earn higher wages than workers age 25-34 and young workers, other things equal. The wage premium for workers age 35 and older over the young and those age 25-34 is especially higher for the high to middle quantiles and is virtually unimportant for the low quantiles. Workers age 25 to 45 seem to do worse than other workers, on average, although the trends are not systematic. Adjusted wage gaps at all quantiles and across all age groups fade in episodes of recession (1989-90, 1995) except for the young and prime age workers at upper quantiles. The worst performers were workers age 25 to 34 at the high quantiles (wage gaps increased from -9% in 1988 to -20% in 1997). Over time, wages of young workers increased more than wages of older workers at the bottom quantiles so that any wage differentials faded after 1988. In short, it is only for the high quantiles that the real wages of male workers in the early stages of their lives have lagged behind wages of older workers. Workers at the bottom quantiles appear to face similar wage conditions regardless of their age. 11 Argentina Poverty Assessment Background Paper No. 8 Education As expected, adjusting for other factors, education grants a wage advantage that stands the bigger biggest between the more educated and the less educated workers. In 1997, on average, workers with some high school earned 12% more than those with primary, high school workers earned 29% more than those with incomplete high school, workers with some college earned 21% more than high school graduates and the college-high school average wage premium was 30%. These "marginal rates of return" to education are higher for workers at the top of the conditional wage distribution and tend to increase somewhat monotonically over the quantiles, especially for the least educated (less than secondary). In contrast, by the end of the period, there was not significant wage incentive to obtain a college education for high school and incomplete college graduates at the bottom quantiles. Moreover, changes in the economy seem to affect differently the incentives to continuing education for workers within any given education group. Workers with less than primary education show relatively higher increases in their wages than the more educated, particularly at the upper quantiles. Workers with secondary education appear to be the worst wage performers. The upper quantile wage gap between workers with high school and those with primary systematically declined from 90% in 1988 to 52% in 1997 and the premium to high school completion went from 46% to 36% over this period. The college-high school wage premium at the high quantiles went up from 40% in 1988 to 64% in 1997 while their return to college completion went from 22% to 31%. In contrast, adjusted wage gaps for workers in the bottom quantiles of the wage distribution do not show any obvious secular patterns. They appear to be much more sensitive to changes in economic performance, showing a rather systematic counter-cyclical pattern, that is they decline with the recession episodes of 1988-91 and 1995 but increased again during the expansion periods. Given the opposite pattern for wages at the bottom quantiles in the base group (the intercepts), this implies that, at the bottom quantiles, less educated workers in Argentina tend to do relatively better than the better educated during recessions but tend to lag behind during economic expansions. This tendency can also be observed for workers with secondary and incomplete college education. Overall, these findings caution against the usual tendency to treat educational attainment as an homogeneous predictor of labor market performance and thus as an entry point for labor market interventions. Education seems to influence in quite different ways wages of workers along the wage distribution. While wages of the less educated tend to cushion a bit more during recession periods and are less responsive to economic growth, those in the bottom part of the more educated wage distribution tend to benefit more from economic expansions but may also be in more risk to fall into poverty when negative shocks hit the economy. A possible policy implication is that, the less educated require ways to increase their human capital because they do not benefit as much from expansions. Meanwhile the more educated in the less privileged jobs would benefit from social protection schemes that allows them to insure from the higher uncertainty in their wages. Occupational Experience On average, wage premiums are higher for workers with longer tenures in their current occupation. In contrast to the pattern of returns to education, the wage premium of workers with some experience (1-5 years) over new entrants is systematically higher for workers in the bottom quantiles. The reverse is true for workers with 5-10 years of experience compared to 12 Argentina Poverty Assessment Background Paper No. 8 those with some experience, the premiums are systematically higher at the top half of the wage distribution. Over time, more experienced workers did remarkably better between 1988 and 1993, but were outperformed between 1993 and 1997. Workers in the third experience group have lagged somewhat behind the other groups over the period, especially at the bottom quantiles. Wage growth has been greater for new entrants and for workers with some experience, especially at the bottom quantiles. By 1993, wage premiums of workers with 1-5 years of experience over new entrants declined to a range of 2-5% across all quantiles down from 9- 16% in 1988, but climbed again to a range of 16-24% in 1997. This could be associated with the downward pressure of high unemployment rates on wages for new entrants. Workers with some experience (1-5) did considerably better than those with more seniority in their occupation (5-10 and more than 10, respectively). As a result there was a surge of substantial wage premiums for new workers already located in more steady jobs over new entrants, especially at the bottom quantiles. This may be indicative of the instability of starting salaries of new entrants during economic transitions. It also suggests that adjusting for other factors, human capital accumulation acquired through labor market experience is proportionally more compensatory of differentials in starting salaries of workers in the less privileged jobs (bottom quantiles). Education-Experience (skill Groups)25 The results for the skill groups specification confirm the expectation that education and experience have a complementary enhancing effect on wages which is seemingly larger, the higher the education and experience levels.26 Over time, the worst wage performers among the less educated were salaried new entrants and salaried workers with at least 10 years of experience in their current occupation, for whom 1997 wage gaps surpass 1988 levels. Among secondary workers, wages of longer. tenure workers fall behind the most. The relatively good performance of workers with primary education seems to be primarily concentrated in new workers with some experience. In fact, the wage premium between workers with some experience in their occupation and new entrants have increased systematically only among workers with primary education. In addition, the already mentioned apparent counter-cyclical pattern of wage gaps of less educated workers at the bottom quantiles is particularly strong among workers with more experience. Occupational Skills Even adjusting for education and other factors, professionals continued to expand their wage advantage over workers in unskilled (agriculture, blue collar) and relatively skilled (white collar, managers) occupations in Argentina. These wage premiums tend to be higher among the highly skilled workers, particularly at the top quantiles. While in 1997 average adjusted wage premiums, were 40% (professional/skilled) and 14% (skilled/unskilled), at the 25 Aside from a slightly lower precision, the coefficients on the other variables are generally similar and thus not discussed further. 26 The results can be used to compute marginal returns to experience for each given education level as well as marginal returns to education for each experience level across quantiles and over time. Nevertheless, such computations do not reveal any obvious patterns and appear to be somewhat noisy. This is to be expected since the precision of the estimates is reduced in each round of computation (because of the loss in degrees of freedom and the augmented sampling error), making it much more difficult to attach statistical significance to any observed differences. 13 Argentina Poverty Assessment Background Paper No. 8 upper quantiles they were 78% and 11%, and only 14% and 38% at the bottom quantiles in these three skill groups. Over the period, both skilled and unskilled workers did relatively worse at the bottom than at the upper quantiles, although the gaps at the top quantiles tend to fluctuate. The skilled/unskilled wage differential shows a pro-cyclical pattern at the bottom quantiles. This implies that the unskilled with jobs at the bottom quantiles tend to do relatively better in recessions than the skilled but they also tend to benefit less from episodes of economic growth. This is similar to the pattern found for the less educated workers at the bottom quantiles. Thus, unskilled workers in jobs at the bottom quantiles are also less sensitive to changes in the economy and would benefit more from programs that prioritize human capital building, while those at the upper quantiles would benefit from safety-nets to smooth the shocks to their wages. Industry Some sizable wage differentials arise between workers in different industries that are unexplained by human capital and other worker characteristics, most of which become muted by the end of the period. On average, wages in all sectors but retail grew faster than in the control financial/business services sector. Wages soared in construction, social services and public administration/defense so as to become more closely aligned. In the remaining sectors wage growth has been relatively weak in the middle quantiles. Thus, adjusting for worker characteristics, workers employed in manufacturing and some non-tradables (retail, personal services, communication/transportation/public utilities) did relatively worse or show no advantageous wage improvements at these quantiles. Other factors considered, wages remained the lowest for those employed in retail and personal services. Wages of public sector workers at the upper quantiles appear to show a counter-cyclical pattern, cushioning during the recession episodes but lagging behind during the 1991-93 economic expansion, although they appear to keep up well with the economic boom between 1996 and 1997. The findings uncover specific groups of workers that have been the most affected by the recession episodes during the period. Workers in the public sector, construction, commerce, transportation and other services located at the bottom of the wage distribution were the hardest hit by the hyperinflation and recession of 1989-90. Meanwhile workers at the middle to upper quantiles employed in the low productivity services and bottom quantiles workers in the public sector were the most affected by the recession of 1995. Micro-firms Salaried male workers employed in small firms (less than 6 employees) face a sizable wage differential with respect to otherwise similar workers in larger firms. On average, this differential declined from 17 percent in 1988 to 10 percent in 1993, but increased to 15 percent in 1997. In 1988, this gap was rather similar across quantiles. Over time, it has declined at the upper quantiles, and first declined for the middle quantiles between 1988 and 1993 to then increased by 1997 to levels comparable to those of 1988. Wages gaps between small firm/big firm employees at the bottom quantiles are pro- cyclical in the period 1988-93. The latter were especially hit by the hyperinflation and recession crisis of 1989-91 (from -16% to -31% at the 5 h quantile). This fact is masked by a focus on average wage gaps. In fact, on average the wage differential between small firm 14 Argentina Poverty Assessment Background Paper No. 8 workers and those employed in bigger firms declined from -17% to -11% between 1988 and 1991, essentially reflecting the relative good wage performance of small firm workers at the upper half of the wage distribution in this period. Wages gaps between small firm/big firm employees recovered significantly with the 1991-93 boom in the Argentine economy but have continued to increase thereafter. In contrast, the average wage gap was down from -20% in 1988 to -16% in 1997, again, reflecting relative good performance of small firm workers at the upper quantiles. This is a clear example of the pitfall of restricting the analysis of labor market outcomes to average trends when the focus in on identifying asymmetries in the performance of different groups in an heterogeneous population. 6. Conclusions and Policy Implications Taken together, the above findings provide rich empirical evidence to address the questions we posed earlier. Given the impressive wage growth concentrated on more vulnerable men groups that immediately follow the introduction of the convertibility plan and the subsequent economic reforms, there is clearly no indication of overall negative impacts on the wage structure leading to a large deterioration in wages of men in Argentina. There is, however, indication that wage declines in the more recent years, although insufficient to dissipate early gains, have also affected those same vulnerable workers the most. Recent trends have been specially countervailing for workers who share the usual disadvantageous characteristics, that is, the less educated, young new entrants and old workers in unskilled jobs, and among these often for those who tend to be employed in the less privileged jobs. The findings from the (conditional) wage analysis allow us to construct archetypical male groups who have experience real wage losses or at best lagged behind over the period: middle age workers, those with secondary education, non-professionals (especially the blue collar), new entrants and those with longer tenure in their occupations, household heads, and workers employed in retail and personal services. The best average wage performers continued to be the young, workers with primary education, professionals, those employed in construction and social services, and new workers with some experience. The results also uncover the stringent limitations of the standard mean wage analysis. A novel and striking finding is that most of the positive wage performance is mainly concentrated at the higher quantiles of the conditional wage distribution. Very often, the dynamics generating wage gains for workers from particular groups or in the expanding sectors have eluded workers in the bottom quantiles. This stresses the important role of unobservable worker characteristics such as labor market connections, school quality or industriousness in determining variations in wage performance. For instance, the rising real wages for the less educated young Argentine workers may have persisted despite the high and persistent unemployment that abate their less fortunate equals because jobs in the upper quantiles are hardly accessible to the unemployed so that the downward pressure on wages of redundant labor operate only in the lower quantiles. Although challenging, this is a matter that deserves further exploration. As a result, although overall wage inequality appears to decrease over the period, there is evidence of an increasing wage polarization both between and within specific male groups. For example, among the college educated, those in the upper quantile jobs vs. those in the bottom 15 Argentina Poverty Assessment Background Paper No. 8 quantiles and similarly among professionals. The extent to which growing wage disparities within "similar" workers are relevant to the phenomenon of marginalization is an open question. There is a literature that suggests that people's perceptions of equity and "fairness" are formed by looking at how they are doing relative to others in similar conditions (i.e. the same occupation, industry, skill level, etc.,.). Our micro-based results are thus relevant to a currently research question of whether inequality per se is also a leading factor in the increase in violence and criminality in the Latin American region. An important finding which can be informative to anti-poverty policies is the evidence suggesting that relative wages between particular groups follow systematic patterns over the business cycle. Particularly, the finding of a counter-cyclical pattern in the bottom quantile wage gaps between the least educated (skilled) workers and the more educated (skilled) suggests that different policies should be emphasized to address their particular conditions. The less educated and unskilled may need interventions that increase their level of human capital since they tend to benefit less from economic growth, the better educated and skilled may require policies that makes it easier to smooth bad shocks to the economy. Whether the latter policies are called for depends on whether or not the more educated in less privileged college have access to private smoothing mechanisms such as savings or the possibility to borrow their way out of bad economic shocks. Effective design of such mechanisms face the usual moral hazard difficulties. These findings also indicate that within any possible categorization, workers in jobs located at the bottom quantiles of any given group tend to be affected similarly by changes in economic conditions. Those at the upper quantiles exhibit fewer similarities across groups in the way they may be affected by the business cycle. Overall, the findings are consistent with a picture of an economy in transition where productive restructuring involves human capital accumulation to complement physical capital modernization. As new better educated workers take on jobs in the economy one could expect wage differentials between experience groups to decline over time as has been the case in Argentina. The surge after the 1995 recession of substantial wage premiums for new workers already located in more steady jobs over new entrants may be indicative of the instability of starting salaries of new entrants. Re-employment programs targeted to new workers, especially those who tend to have the lowest starting salaries, may offer a very cost-effective means to reduce wage disparities and smooth employment transitions. Continuing economic restructuring may affect the income generation potential of males and result in more subtle negative impacts on social outcomes and perceptions that should not be ignored. Increasing and persistent unemployment and barriers to good quality employment have likely affected the capacity of some young Argentinean men and displaced older workers to effect a "breadwinner" role in their current or prospective families (Barker, 1998). The potential link to increasing violence and criminality in the region calls for policy interventions such as progressive programs that help these and older men reflect "about what it means to be men" and find new ways to reaffirm their roles as fathers and partners (Barker, 1998). These may include public support to counseling and mentoring community services and to the creation of spaces for discussions about fatherhood and domestic violence in community organizations. 16 Argentina Poverty Assessment Background Paper No. 8 References Altimir, 0. y L. Beccaria, 1997. Effectos de los Cambios Macroeconomicos y de las Reformas Sobre la Pobreza Urbana en la Argentina. Proyecto Programa de las Naciones Unidas para el Desarrollo (PNUD)/Banco Interamericano de Desarrollo (BID)/Comision Economica para America Latina y el Caribe (CEPAL). Mimeo. Arias, 0., K. Hallock and W. Sosa, 1999, "Heterogeneity in the Returns to Schooling: Instrumental Variables Quantile Regression using Twins Data", mimeo. Comite Ejecutivo para el Estudio de la Pobreza (CEPA), Niveles de Vida, Pobreza e Ingresos en los 27 Aglomerados Relevados por la Encuesta Permanente de Hogares, Ministerio de Economia y Obras y Servicios Publicos. , 1996, "Caracterizacion de los Jovenes en Situacion de Pobreza del Gran Buenos Aires", Documento de Trabajo No. 11/96. ,1996, La Pobreza en la Tercera Edad: un Analisis sobre los Ancianos del Gran Buenos Aires. Documento de Trabajo No. 8/96. , 1996, Hogares de Bajos Ingresos del Gran Buenos Aires, Los Cambios Operados entre 1991 y 1996. Documento de Trabajo No. 13/96. Cunningham, W. and W. Maloney, 1998, "Heterogeneity among Mexico's Micro-Enterprises: An Application of Factor and Cluster Analysis", The World Bank, Policy Research Working Paper 1999. Barker, G., 1998, "Boys in the Hood, Boys in the Bairro: Exploratory Research on Masculinity, Fatherhood and Attitudes Toward Women Among Low Income Young Men in Chicago, SA and Rio de Janeiro, Brazil". May, 1998, Seminar on Men, Family Formation and Reproduction, Buenos Aires, Argentina. Bound, J. and G. Johnson, 1992, "Changes in the Structure of Wages in the 1980s", American Economic Review, June, 82 (3), pp. 371-392. Buchinsky, M., 1994, "Changes in the U.S. Wage Structure 1963-1987: An Application of Quantile Regression," Econometrica, March, 62 (2), pp. 405-58. DiNardo, John, Nicole Fortin, and Thomas Lemieux, 1996, "Labor Market Institutions and the Distribution of Wages, 1973-1992: A Semiparametric Approach," Econometrica, 64 (5), pp. 1001-1044.. DiNardo, John, and Thomas Lemieux, 1996, "Diverging Male Wage Inequality in the United States and Canada, 1981-1988: Do Institutions Explain the Difference?," Industrial and Labor Relations Review, 50(4), July 1997, pp. 629-51. Fitzenberger, B. and C. Kurz, 1997, "New Insights on Earnings Trends Across Skill Groups and Industries in Germany", mimeo, Universit5t Konstanz. 17 Argentina Poverty Assessment Background Paper No. 8 Gray, J., 1997, "The Fall in Men's Return to Marriage: Declining Productivity Effects or Changing Selection?". Journal ofHuman Resources, 32 (3), pp. 480-504. Heckman, J., 1979, "Sample Selection Bias as a Specification Error", Econometrica, 47, pp. 153-161. Juhn, C., K. Murphy and B. Pierce, 1993, "Wage Inequality and the Rise in Returns to Skill", Journal ofPolitical Economy, 101 (3), pp. 410-442. Kennedy, P., 1981, "Estimation with Correctly Interpreted Dummy Variables in Semilogarithmic Equations", American Economic Review, 71 (4), pp. 801. Koenker, R., and G. Bassett, 1978, "Regression Quantiles." Econometrica, January, 46 (1), pp. 1-26. Korenman, S. and D. Neumark, 1991, "Does Marriage Really Make Men More Productive?", Journal of Human Resources, 26 (2), pp. 282-307. Levy, F. and R. Murnane, 1992, "U.S. Earnings Levels and Earnings Inequality: A Review of Recent Trends and Proposed Explanations". Journal of Economic Literature, September, 30, pp. 1333-81. Maloney, W., 1998, "Are Labor Markets in Developing Countries Dualistic?", The World Bank, Policy Research Working Paper 1941. Maloney, W., 1998, "The Structure of Labor Markets in Developing Countries: Times Series Evidence on Competing Views", The World Bank, Policy Research Working Paper 1940. Maloney, W. and E. P. Ribeiro, 1998, "Efficiency Wage and Union Effects in Labor Demand and Wage Structure in Mexico", mimeo. Mata J. and J. Machado, 1999, "Sources of Increased Wage Inequality", mimeo. Mwabu, Germano, and T. Paul Schultz, 1996,"Education Returns Across Quantiles of the Wage Function: Alternative Explanations for Returns to Education by Race in South Africa," American Economic Review, 86 (2), May, pp. 335-9. Pessino, Carola, 1995, "Returns to Education in Greater Buenos Aires 1986-1993: From Hyperinflation to Stabilization". Series Documentos de Trabajo 104. Buenos Aires, Argentina: CEMA. Pessino, Carola, 1997, "The Labor Market During The Economic Transition", World Bank (1996), "Argentina, The Convertibility Plan: Assessment and Potential Prospects". Vols 1-2. Report No. 15402-Ar. World Bank (1995), "Argentina's Poor: A Profile". Report No. 13318-Ar. 18 Argentina Poverty Assessment Background Paper No. 8 Figure ib: Comparisons of Wage Distributions ovr time, real wages $1988 1988, 91-93, all 1988, 91-93, < secun. 1988, 91-93, > secun. 2 0 0 .0 4 50 1 .0 50 1 2 1988 939,a 0.98 9-5 secu.198935,>scn 0 1 2. 1 2 3 4 50 o0 A.0O 1988, 93-95, all 1988, 93-95, < secun. 1988, 93-95, > secun. 0 1 2 3 4 5 0 1 2 3 4 5 0 1 2 3 4 5 logwage logwage logwage 1988, 95-97, all 1988, 95-97, <secun. 1988, 95-97, > secun. ci 0 1 . 0 1* 0 1 2 3 4 logwage logwage logwage 19 Argentina Poverty Assessment Background Paper No. 8 APPENDIX I: Definitions of Dummy Variables Micro-firm: Micro-firm salaried workers microfi are defined as men working in small firms (less than 6 workers). Age: agei refers to the i-th age group, agel= 15-25 years, age2= 25-35 years, age3= 35-45; age4= 45-55 years; age5= 55-70 years. Tenure in the occupation: less than 1 year in the current occupation (expl), between I and 5 (exp2) years, between 5 and 10 (exp3) years and more than 10 years (exp4). Education-Experience: Education-experience dummies are defined as follows: Educi= educ and expi where educ refers to the last level of education achieved (prima= some or complete primary; secinc-- some secondary education; secunc= completed secondary education; univinc= incomplete college; univ- complete college), and expi refers to the i-th experience group (based on reported tenure in current occupation (see above). For example, secun2 corresponds to workers with complete secondary education and tenure 1-5 years. Industry: These are defined as follows. Manufacturing: food, beverages and tobacco, textiles and clothing, chemical and metallic products, other manufacturing; Serv1: electricity, gas and water, commerce, restaurants and hotels, transportation; Serv2: financial, real estate, and business services, health services; Public: public instruction, administration and defense; Other: repair, domestic and personal services, other social and community services. We distinguish between the services that tend to be more intensive in human capital (serv2) and those who demand relatively less skilled labor (serv1). The construction sector and the other category tend to employ the less skilled labor and may be more sensitive to changes in the level of economic activity. Occupations: Each occupational dummy groups men according to three different levels of skills based on self-reported occupations: profe, skilled and unskilled. The first dummy groups men that reported to be working in jobs that require professional qualifications, skilled groups those that reported to be working in administrative (white collar) jobs, and unskilled refers to men that reported to be working in jobs of low qualification. Miscelaneous: The dummies for marital status married and consensual union c. union are based on self-reports. The household heads dummy head refer to men who report to be heads of households with at least two members. 20 Argentina Poverty Assessment Background Paper No. 8 APPENDIX II: Quantile Regression In this section we present some basic results on the quantile regression methods used in this work.27 Consider the following linear model: Y = XJ+u (1) where u is a vector of independent error terms whose i-th component has an unspecified distribution function F. Given the usual conditional orthogonality assumption on the error term, Ordinary Least Squares regression provides a model for the conditional mean of Y given by: E[Y/X]= X/ (2) In the special case of iid errors this Least Squares estimate together with some measure of dispersion would usually provide a complete characterization of (1).28 Nevertheless, it is increasingly recognized that the iid linear model is not well suited to analyze problems which involve very heterogeneous populations. Quantile regressions provide a very useful alternative by providing models for different percentiles of Fy,, the conditional distribution of Y given X. For (1) the r-th conditional quantile of Y is given by: Q = X 8 (r) (3) where the orthogonality condition on u is now assumed for Q,(u|X), that is, the c-th conditional quantile of the error term is assumed to be zero. This gives rise to a family of (quantile) regression curves, one for each r, which provide a more complete characterization of the relationship between Y and X. Quantile regression also offers the usual robustness properties associated with ordinary sample quantiles, particularly robustness to outlying observations in Y. In the case of the id linear model the conditional quantile functions given in (3) will be parallel vertical displacements of one another, and only robustness arguments would lead one to prefer alternative estimators of location other than OLS. The most interesting case arises when the estimated P(c) coefficients differ systematically across -es, suggesting that the marginal effect of a particular explanatory variable varies across different quantiles of the conditional distribution of Y. Estimation of the 8(r) coefficients known as "regression quantiles" is based on a sample of n observations of Y and p explanatory variables collected in the matrix X. It can be shown that estimates of P(t) can be obtained as solutions to the following linear programming problem: min r 1'n U + (1 - r) 1'. v X,8 + u - v = Y] (,, u, v) E RPx R2n (13) where 1'n is an n vector of ones and u and v are the positive and negative parts of the residual vector. In practice, an adaptation of the Barrodale-Roberts algorithm for the least absolute deviation problem makes computation now comparable to least squares methods for reasonable size data sets. Substantial computational time improvement for very large data sets has recently been accomplished through interior point methods (Koenker and Portnoy (1997)). 27 See Koenker and Portnoy (1997) for a recent comprehensive overview of the topic. 28 If additionally, F is assumed to be Gaussian, then OLS yields the optimal estimator of location for the linear model (1). 21 Argentina Poverty Assessment Background Paper No. 8 Inference on the p(c)'s can be based on the following result. Let bn = (bn(TI) ,... bn(tm)) be a pm vector of p estimated regression quantile coefficients for m different quantiles based on a sample of n iid observations; and let P be its population counterpart. Under some regularity conditions Koenker and Basset (1978) showed that: n (bn - P) --> N(0 n & Qo.) (14) where Q is a m x m matrix with typical element: coij = (min (ti , 3) -t) / [ f( F-'(ti) ) f( F-'())] (15) Qo = plim if' (XX), a positive definite matrix, and 0 denotes the Kronecker product. Confidence intervals can be easily constructed based on this result. General linear hypothesis like HO: Hp=h can be tested using the following Wald-type statistic: Tn = (H bn - h)' [ H (K2 (X'X)-') H']-' (H bn - h) (16) which under the null hypothesis has a X2 distribution with rank(H) degrees of freedom. This approach requires the estimation of the nuisance parameter 1 / f(F-'(t)) (called sparsity) which measures the inverse of the density of the observations around the t-th quantile. This is usually accomplished based on estimates of the empirical quantile function constructed from residuals of the t-th quantile regression and using smoothing techniques. An alternative approach to inference that takes advantage of the quantile regression formulation can be based on rank tests. These tests are robust to outliers in Y and are asymptotically distribution free since they do not require the estimation of nuisance parameters depending on the error distribution, and are not more complicated to compute than those based on sparsity estimation.29. Let X=[1:X,:X2] and suppose we are interested in testing the linear hypothesis Ho: P2 = 0 vs Ho: P2 # 0. The following statistic: W = (Yr'M X2 (X2'M X2)-1 X2M Yr) / A (17) has an asymptotic ,2 (q) distribution under the null hypothesis, where Yr is an estimated vector of ranks of the observations, M= I- X XIX, 1 X, and A is a quantity that does not depend on the distribution of the errors. Visual inspection suggests that the rank-based test is very similar to Lagrange multiplier tests where the yj's play the role of the square residuals. The estimation of the ranks vector can be obtained as a by-product of the computation of the regression quantiles for the linear model under the restricted model. Based on the well known duality between hypothesis testing and construction of confidence intervals, a test to evaluate the significance of a single variable can be inverted to obtain a confidence interval for each coefficient.o We used this approach to construct confidence intervals for the vector of quantile regression coefficients. 29 Since we do not attempt to summarize the theory of such tests here we refer to Koenker and Portnoy (1997) and Koenker (1994) for a review and relevant references. 30 Koenker (1994) discusses in detail computational and theoretical advantages as well as montecarlo results in favor of these tests. 22 Argentina Poverty Assessment Background Paper No. 8 APPENDIX III: RAW COEFFICIENT ESTIMATES Table 1: Means (proportions) for All Variables 1988 1991 1993 1995 1997 Married 0.603 0.577 0.550 0.563 0.507 Cons. union 0.071 0.101 0.109 0.110 0.122 Head 0.647 0.640 0.618 0.633 0.585 agel (<25 yrs) 0.235 0.243 0.257 0.224 0.231 age2 [25-35) 0.266 0.276 0.257 0.270 0.290 age3 [35-45) 0.224 0.231 0.227 0.236 0.212 age4 r45-55) 0.163 0.156 0.164 0.178 0.164 age5 (>=55 yrs) 0.112 0.094 0.095 0.092 0.103 Prima 0.497 0.463 0.421 0.445 0.370 Secinc 0.222 0.240 0.247 0.228 0.245 Secun 0.129 0.138 0.164 0.148 0.181 Univinc 0.094 0.088 0.098 0.105 0.127 Univ 0.059 0.071 0.069 0.075 0.076 expl (<1 yr) 0.296 0.298 0.341 0.289 0.362 exp2 [1-5) 0.288 0.298 0.294 0.333 0.304 exp3 [5-10) 0.176 0.159 0.138 0.155 0.143 exp4 (>=10 yrs) 0.240 0.245 0.228 0.224 0.192 Manuf 0.368 0.354 0.351 0.301 0.290 Const 0.079 0.069 0.063 0.067 0.076 servl 0.271 0.298 0.312 0.330 0.317 serv2 0.100 0.095 0.094 0.109 0.135 Public 0.086 0.090 0.090 0.093 0.088 Other 0.096 0.094 0.089 0.100 0.095 Microfirm 0.222 0.279 0.253 0.248 0.269 Primal 0.142 0.139 0.142 0.129 0.134 Prima2 0.138 0.123 0.108 0.135 0.098 Prima3 0.089 0.079 0.063 0.069 0.052 Prima4 0.128 0.122 0.109 0.112 0.087 Secincl 0.075 0.091 0.094 0.073 0.097 Secinc2 0.065 0.066 0.078 0.076 0.084 Secinc3 0.038 0.032 0.028 0.031 0.030 Secinc4 0.044 0.050 0.048 0.047 0.034 Secun1 0.032 0.029 0.054 0.039 0.061 Secun2 0.039 0.050 0.052 0.050 0.051 Secun3 0.024 0.022 0.021 0.028 0.032 Secun4 0.034 0.038 0.037 0.031 0.037 Univincl 0.032 0.028 0.035 0.033 0.050 Univinc2 0.033 0.034 0.036 0.042 0.047 Univinc3 0.015 0.014 0.013 0.013 0.014 Univinc4 0.014 0.012 0.015 0.017 0.017 Univi 0.015 0.011 0.016 0.016 0.020 Univ2 0.014 0.025 0.020 0.030 0.024 Univ3 0.010 0.012 0.013 0.013 0.015 Univ4 0.020 0.023 0.019 0.017 0.017 Profe 0.075 0.077 0.064 0.057 0.058 Skilled 0.651 0.602 0.728 0.731 0.738 Unskilled 0.274 0.321 0.208 0.212 0.203 Sample size 3,883 2,103 2,837 2,640 3,087 23 Argentina Poverty Assessment Background Paper No. 8 Table 2: Summary Statistics for Real Wages, $1988 1988 1991 1993 1995 1997 %93/88 %97/93 %97/88 Mean wage 9.94 9.99 12 11.76 11.31 20.7% -5.8% 13.8% (11.39) (10.19) (12.96) (11.19) (10.75) 0.05 quantile 2.99 3.5 4.27 3.89 3.7 42.8% -13.3% 23.7% 0.10 quantile 3.74 4.33 5.31 4.86 4.62 42.0% -13.0% 23.5% 1st quartile 5.19 5.74 6.82 6.77 6.44 31.4% -5.6% 24.1% Median 7.77 7.83 9.75 9.7 9.5 25.5% -2.6% 22.3% 2nd quartile 12.11 12.6 14.87 14.35 13.73 22.8% -7.7% 13.4% 0.90 quantile 20.76 19.22 22.74 23.23 21.33 9.5% -6.2% 2.7% 90th/10th 5.55 4.44 4.28 4.78 4.62 75th/25th 2.33 2.20 2.18 2.12 2.13 90th/75th 1.71 1.53 1.53 1.62 1.55 25th/lOth 1.39 1.33 1.28 1.39 1.39 24 Argentina Poverty Assessment Background Paper No. 8 Table 3a: Quantile Regression Estimates of Log Wage Gaps, based on EPH 1988- 1997 1988 0.05 0.10 0.25 0.50 0.75 0.90 1991 0.05 0.10 0.25 0.50 0.75 0.90 Intercept 2.201 2.494 2.850 3.487 3.977 4.317 1.907 2.446 2.04 3.242 3.647 4.138 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) married 0.172 0.086 0.083 0.085 0.074 0.028 0.223 0.325 0.140 0.145 0.140 0.189 (0.069) (0.141) (0.008) (0.032) (0.059) (0.679) (0.170) (0.000) (0.001) (0.009) (0.008) (0.014) c.union 0.043 -0.045 0.042 0.077 0.091 -0.045 0,016 0.250 0.091 0.034 0.054 0.162 (0.659) (0.512) (0.370) (0.184) (0.046) (0.583) (0.914) (0.006) (0.086) (0.552) (0.347) (0.094) agel -0.268 -0.278 -0.203 -0.226 -0.267 -0,241 -0.028 0.032 0.010 -0.065 -0.151 -0.261 (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.792) (0.637) (0.860) (0.322) (0.011) (0.001) age2 -0.053 -0.096 -0.060 -0.124 -0.125 -0.094 0.038 0.029 0.045 -0,027 -0.061 -0-151 (0.327) (0.035) (0.039) (0.001) (0.000) (0.030) (0.586) (0.653) (0.362) (0.605) (0.193) (0.045) age3 0:001 -0.019 -0.004 -0.023 -0066 -0.009 -0,062 -0.027 0.007 -0.005 -0.065 -0.088 (0.975) (0.626) (0.878) (0.540) (0.009) (0.868) (0.538) (0.654) (0.894) (0.922) (0.170) (0.175) age5 -0.225 -0.109 -0.002 -0.032 0.008 0.066 0.002 0028 0.(88 0.125 0.070 0.090 (0.003) (0.027) (0.970) (0.460) (0.854) (0.185) (0.978) (0.724) (0,208) (0.064) (0.375) (0.265) prima -0.701 -0.718 -0.774 -0.789 -0.936 -0.981 -0.271 -0.513 -0.543 -0.669 -0,703 -0.851 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.102) (0.001) (0.000) (0.000) (0.000) (0.000) secinc -0.462 -0.506 -0.556 -0.551 -0.678 -0.719 -0.108 -0.338 -0.382 -0.521 -0.518 -0.685 (0.008) (0.000) (0.000) (0.000) (0.000) (0.000) (0.588) (0.019) (0.000) (0.000) (0.000) (0.000) secur -0.347 -0.281 -0.285 -0.255 -0.339 -0.339 0.032 -0.219 -0.162 -0.265 -0.297 -0,416 (0.055) (0.022) (0.000) (0.000) (0.000) (0.000) (0.880) (0.182) (0.124) (0.009) (0.001) (0.000) inivinc -0.137 -0.097 -0.070 -0.092 -0.161 -0.196 0.194 -0.090 -0.139 -0.196 -0.170 -0.319 (0.455) (0.320) (0.324) (0.162) (0.049) (0.000) (0.293) (0.492) (0.158) (0.073) (0.101) (0.002) expl -0.369 -0.347 -0.320 -0.388 -0.334 -0.315 -0.128 -0.281 -0.329 -0.303 -0.216 -0.189 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.121) (0.000) (0.000) (0.000) (0.000) (0.006) exp2 -0.174 -0.197 -0.237 -0.298 -0.266 -0.243 -0.058 -0.177 -0.241 -0.217 -0,190 -0.182 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.437) (0.003) (0.000) (0.000) (0.000) (0,001) exp3 -0.117 -0.138 -0.172 -0.208 -0.139 -0.126 0.044 -0.086 -04167 -0.156 -0.065 -0.005 (0.013) (0.001) (0.000) (0.000) (0.000) (0.021) (0.510) (0.167) (0.000) (0.000) (0.147) (0.949) manuf -0,013 0.036 0.043 -0.080 -0.146 -0.105 -0.165 -0.073 -0.130 -0.138 -0.078 -0.146 (0.866) (0.540) (0.350) (0.047) (0.000) (0.005) (0.042) (0.149) (0.055) (0.005) (0.184) (0.051) const -0098 -0.105 -0.097 -0.216 -0.251 -0.204 -0.245 -0.176 -0.101 -0.034 0109 -0.030 (0.207) (0.142) (0.052) (0.000) (0.000) (0.002) (0.019) (0.030) (0.260) (0.651) (0.230) (0.675) ,serv 0.022 0.019 -4010 -0.106 -0.135 -0.061 -0.077 -0.083 -0.114 -0.065 0.031 -0.106 (0.755) (0.716) (0.819) (0.021) (0.004) (0.132) (0.375) (0,090) (0.153) (0.190) (0.604) (0.103) other -0.047 -0.042 -0.022 -0.095 -0.151 -0.081 -0.203 -0.095 -0.114 -0,106 0.010 -0.067 (0,607) (0.579) (0.712) (0.034) (0.007) (0.043) (0.160) (0.277) (0.177) (0.105) (0.884) (0.350) public -0.198 -0.214 -0.130 -0.198 -0.266 -0.196 -0.152 -0.192 -0.285 -0.179 -0.097 -0.070 (0.086) (0.025) (0.052) (0.000) (0.000) (0.000) (0.260) (0.013) (0.000) (0.014) (0.064) (0.536) head -0.050 0.012 0.053 0.031 0.075 0.154 -0.019 -0.158 0.006 0.044 0.089 0.096 (0.519) (0.828) (0.093) (0.458) (0.042) (0.022) (0.909) (0.114) (0.883) (0.410) (0.126) (0.190) skilled -0-045 -0.132 -0.231 -0.393 -0.401 -0.524 -0354 -0.397 -0.371 -0.422 -0.586 -0.536 (0.780) (0.279) (0.006) (0.000) (0.000) (0.000) (0.003) (0.001) (0.000) (0.000) (0.000) (0.000) unskilled -0.187 -0.264 -0.396 -0.566 -0.594 -0.729 -0.407 -0.491 -0.500 -0.584 -0.796 -0.758 (0.246) (0.028) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) micro -0.178 -0.268 -0.202 -0.180 -0.197 -0.180 -0.373 -0.232 -0.107 -0.073 -0.114 -0.077 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.002) (0.005) (0.000) (0.013) October -0.121 -0.115 -0.084 -0.048 -0.062 -0.095 0.114 0.080 0.068 0.052 0.058, 0.007 (0.000) (0.000) (0.000) (0.017) (0.001) (0.000) (0.002) (0.033) (0.018) (0.035) (0.015) (0.857) p-values are presented in parenthesis. 25 Argentina Poverty Assessment Background Paper No. 8 Table 3b: Quantile Regression Estimates of Log Wage Gaps, based on EPH 1988- 1997 1993 0.05 0.10 0.25 0.50 0.75 0.90 1995 0.05 0.10 0.25 0.50 0.75 0.90 Intercept 2.617 2.659 2.993 3.632 4.019 4.323 1.885 2.382 3.093 3.613 4.102 4.426 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) married -0.111 -0.047 -0.049 -0.04 -0.017 0.034 0.079 0.024 0.033 -0.006 0.049 0.08d (0.198) (0.333) (0.191) (0222) (0.639) (0.597) (0,196) (0.771) (0.525) (0.864) (0.223) (0.279) c.union -0.103 -0.066 -0.097 -0.054 -0.027 0.053 0.005 -0.043 0.01 -0.047 -0.013 0.006 (0.257) (0.312) (0.057) (0.176) (0.635) (0.517) (0.939) (0.563) (0.861) (0.106) (0.754) (0.923) agel -0.043 -0.039 -0.119 -0,227 -0.232 -0.242 0.088 -0.005 -0.056 -0.026 -0.127 -0.261 (0.634) (0.564) (0.019) (0.000) (0.000) (0.005) (0.339) (0.932) (0.333) (0.407) (0.008) (0.000) age2 -0,121 -0.082 -0.05 -0.147 -0.17 -0.145 0.146 0.097 0.026 0.042 -0.048 -0.169 (0.096) (0.243) (0.241) (0.000) (0.000) (0.025) (0.090) (0.147) (0.571) (0.311) (0.292) (0.000) age3 -0.014 -0.002 -0.012 -0.057 -0.065 -0.062 0.227 0.196 0.03 0.051 0.012 0.028 (0.811) (0.977) (0.767) (0.136) (0.122) (0.300) (0.004) (0.000) (0.486) (0.135) (0.777) (0.539) age5 -0.039 -0.042 -0.088 -0.125 -0.112 -0.028 0.274 0,19 0.036 0.026 0.033 0.118 (0.527) (0.500) (0.090) (0.016) (0.028) (0.629) (0.001) (0.001) (0.463) (0.631) (0.487) (0.079) prima -0.560 -0.608 -0.62 -0.705 -0.628 -0.791 -0.317 -0.269 -0.461 -0.659 -0.71 -0.746 (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.029) (0.000) (0.000) (0.000) (0.000) secine -0.443 -0.516 -0.47 -0.544 -0.47 -0.588 -0.303 -0.153 -0.281 -0.512 -0.506 -0.491 (0.007) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.236) (0.000) (0.000) (0.000) (0.000) secun -0.340 -0.418 -0.345 -0.416 -0.313 -0.407 -0.138 -0.042 -0.142 -0.299 -0.316 -0.287 (0.043) (0.000) (0.000) (0.000) (0.000) (0.000) (0.130) (0.733) (0.055) (0.000) (0.000) (0.017) univine -0.186 -0.169 -0.125 -0.141 -0.116 -0.323 0.088 0.142 -0.008 -0.261 -0.19 -0.002 (0.309) (0.103) (0.083) (0.049) (0.087) (0.002) (0.362) (0.234) (0.923) (0.000) (0.015) (0.987) expl -0.131 -0.12 -0.155 -0.205 -0.198 -0.204 -0.347 -0.365 -0.346 -0.345 -0.315 -0.256 (0.016) (0.033) (0.001) (0.000) (0.000) (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) exp2 -0.110 -0.073 -0.134 -0.166 -0.183 -0.155 -0.08 -0.194 -0.229 -0.246 -0.202 -0.141 (0.062) (0.192) (0.005) (0.000) (0.000) (0.014) (0,083) (0.000) (0.000) (0.000) (0.000) (0.002) exp3 -0.059 0.011 -0.068 -0.096 -0.067 -0.07 -0.153 -0.155 -0.188 -0.188 -0.116 -0.076 (0.358) (0.868) (0.255) (0.017) (0.125) (0.283) (0.085) (0.009) (0.000) (0.000) (0.016) (0.180) manuf 0.001 0.057 0.015 -0.108 -0.115 -0.178 -0.062 -0.063 0.038 -0.111 -0.17 -0.131 (0.988) (0.275) (0.757) (0.021) (0.043) (0.009) (0.258) (0.363) (0.429) (0.025) (0.000) (0.008) const 0.034 0.172 0.193 0.121 0.032 -0.045 -0.125 0.073 0.106 0.041 0.022 0.109 (0.685) (0.076) (0.005) (0.058) (0.708) (0.551) (0.500) (0.505) (0.089) (0.486) (0.663) (0.413) servi -0.033 0.068 -0.002 -0.099 -0.102 -0.181 -0.001 -0.08 -0.003 -0.128 -0.1 -0.059 (0.639) (0.213) (0.970) (0.038) (0.077) (0.009) (0.985) (0.346) (0.954) (0.007) (0.016) (0.234) other -0.153 -0.025 -0.07 -0.07 -0.099 -0.089 -0.131 -0.157 -0.034 -0.206 -0.205 -0.192 (0.226) (0.768) (0.287) (0.307) (0.114) (0.224) (0.247) (0.102) (0.479) (0.000) (0.000) (0.025) public -0.221 -0.126 -0.076 -0.165 -0.219 -0.265 -0.137 -0.233 -0.135 -0.232 -0.176 -0.11 5 (0.020) (0.137) (0.233) (0.001) (0.005) (0.004) (0.122) (0.006) (0.009) (0.000) (0.002) (0.065) head 0.233 0.177 0.166 0.151 0.159 0.091 0.105 0.076 0.085 0.139 0.095 0.01< (0.018) (0.004) (0.000) (0.000) (0.000) (0.300) (0.028) (0.220) (0.063) (0.000) (0.016) (0.806) skilled -0.462 -0.419 -0.409 -0.518 -0.657 -0.464 -0.173 -0.397 -0,628 -0.537 -0.71 -0.778 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.300) (0.016) (0.000) (0.000) (0.000) (0.000) unskilled -0.636 -0.595 -0547 -0.648 -0.775 -0.647 -0.29 -0.499 -0.791 -0.732 -0.871 -0.902 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.108) (0.005) (0.000) (0.000) (0.000) (0.000' microrl -0.138 -0.111 -0.096 -0.067 -0.127 -0.094 -0.205 -0.142 -0.169 -0.118 -0.078 -0.04(: (0.007) (0.014) (0.001) (0.003) (0.000) (0.022) (0.001) (0.010) (0.000) (0.000) (0.004) (0.260 october -0.024 -0.015 0.007 0.039 0.02 0.027 -0.023 0.005 -0.007 -0.011 -0.019 -0.022 (0.525) (0.706) (0.748) (0.028) (0.363) (0.463) (0.606) (0.886) (0.776) (0.606) (0.338) (0.411V p-values are presented in parenthesis. 26 Argentina Poverty Assessment Background Paper No. 8 Table 3c: Quantile Regression Estimates of Log Wage Gaps, based on EPH 1988- 1997 1997 0.05 0.10 0.25 0.50 0.75 0.90 hitercept 2.379 2.694 3.194 3.597 4.077 4.319 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) married 0.153 0.195 0. 0.1 16 0.145 0.141 (0.141) (0.027) (0.001) (0.001) (0.002) (0.000) c.union 0.009 0.11 0.028 0.022 0.019 0.002 (0.946) (0.339) (0.539) (0.530) (0.652) (0.975) ag el -0.112 -0.114 -0.155 -0. 191 -0233 -0.306 (0.068) (0.134) (0.001) (0.000) (0.000) (0.000) age2 -0.112 -0.084 -0.097 -0,108 -0.173 -0,215 (0.062) (0.205) 0 016) 0.0m1 10.000) (0.000 aeS -0.134 -0.113 -0 144 -0.085 -U 099 -0 124 (0.069) (0.081) (0 000) 0.012,) (0.024 1 (0.047) dgCS -0.00 -0.102 -0.05 -0 058 -V.053 (0,275) (0.159) (0.374) (0.546) (0,172) (0.405) prim. -0.434 -0.497 -0.701 -0 7117 -0.815 -0 Qj3 <0 000) (0001> 10.1000 0.000) 0.000, ) 000) seeinc -0 36 -0.418 -0.532 -(1.56( -0 72S -0. 81 1 (0.010) (0.004) (0.000) (0.000) (0.000) (0,000) se -0.073 -0.175 -0.336 -0.331 -0.453 -046 (0.550) (0.155) (0.000) (0.000) (0.000) (0.000) uniinc 0.111 0.024 -0.157 -0.125 -0 251 -0.269 (0.447) (0.843) (0 008b (0.10) (10 000) (0 000) exp1 -0.4 -(1.401 -0.358 -0.342 -0 308 -0.293 (0.000) (0.000) (0.000 <(l 000) (0.000) (o 000, C,p2 -0.158 -0.189 -0 183 -0 188 -0.156 -0 098 (0 002) (0.0001 00001 (0.000) (0.00(0 (0.1l.07) ep3 -0.13 -0.144 -u. 15 -0.179 -0 120 -0.054 (0.037 o (0.008) (0.000(0.0001 (0009) 10 335 manuf 0.112 11 109 -0.05 -0.029 -0.01 0024 (0.082) (0.079) (0.273) (0.407) (0.796) (0.565) consi -0.165 0.11 0.047 0.1ö4 0.091 0.122 (0 Wi05 (0 345l (1 4971 10 003) (0 128) (0.1)151 servi -0.016 (025 -0.095 -0.027 -0 018 0.028 (0.835) (0.626) (0.016) 10.420 10.614) (0 567P other -0 173 -0.085 -0.055 -0.02 0 034 0.006 (0.172) (0.354) (0.262) (0.617) (0.545) (0.926) public -0.086 -0.01 -0.101 -O.06 -0.071 -0.039. (0.294) (0.901) (0.081) (0.021) (0.217) (0.555) head 0.066 -0.04 0.047 0.031 -0.007 0.024 (0.583) (0.664) (0.159) (0.283) (0.876) (0.564) skilled -0.46 -0.484 -0.414 -0539 -0.618 -0.577 (0.001) (0.000) (0.000) (0.00) (0.000) (0.000) unskilled -0.562 -0.649 -0595 -0.673 -0.76 4.69 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) mnicrofi -0.241 -0.266 -0,192 -0.178 -0.146 -0.034 (0.000) (0.000) (0.000) (0.000) (0.000) (0.468) October 0.015 -008 -011 -0.02 -.29 -0.062 (0.674) (0.836) (0.621) (0.316) (0.183) (0.031) p-values are presented in parenthesis. 27 Argentina Poverty Assessment Background Paper No. 8 Table 4a: Quantile Regression Estimates of Log Wage Gaps, skill groups specification 1988 0.05 0.1 0.25 0.5 0.75 0.9 1991 0.05 0.1 0.25 0.5 0.75 0.9 intercept 2.312 2:597 2908 3 .443 3.844 4,212 1,624 L967 2471 3.139 3.44 4.019 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) married, 0.16 0.068 <,082 0.063 0.078 0.013 0 .3 376 0.16 0.141 0.153 0.178 (0.100) (0.289) (0.072) (0.108) (0.026) (0.838) (0.045) (0.000) (0.000) (0.009) (0.007) (0.009) c.union 0.018 ,-0.093 0.026 0.064 0.103 -0.053 0.172 0.283 0.101 0.017 0.053 0141 (0.856) (0.191) (0.648) (0.206) (0.018) (0.495) (0.353) (0.002) (0.016) (0.749) (0.479) (0.104) ggel -0.242' -0.299 -0.214 -0.211 -0.259 -0.293 0.078 0.097 0.032 '-0.046 -0.11 -0.194 (0.008) (0.000) (0.000) (0.000) (0.000) (0.000) (0.532) (0.273) (0.568) (0.495) (0.081) (0.013) aäe2 -0.024 -0.098 -0.063 -0.111 -0.136 -0.142 0.032 0.08 0.053 -0.006 -0.056 -0.123 (0.749) (0.046) (0.078) (0.00 1) (0.000) (0.007) (0.672) (0.302) (0.256) (0.919) (0.297) (0.066) 4e3 0.04 -0.032 0 -0.028 -0.064 -0.041 -0.082 -0.012 0.006 0.028 -0.062 -0.039 (0.456) (0.474) (0.999) (0.401) (0.016) (0.393) (0.314) (0.873) (0.894) (0.572) (0.251) (0.583) ageS -0.129 -0.124 0 -0.017 0.008 0.024 -0.048 0,036 0.085 0.164 0,054 0.111 (0.117) (0.047) (0,992) (0.691) (0.889) (0.618) (0.566) (0.729) (0.169) (0.006) (0.501) (0.210) prim -l 272 -I 15 -1.103 -1.123 -1.152 -1.126 -0.326 -0.474 -0835 -0.87 -0.78 -0.989 (0 000) (0.000) (0.000) (0.000 (0.000) (0.000) (0.355) (0 024; (0.0001 <0.000 (0.000) (0 000> prima2 -1.093 -1.064 -1.045 -1.076 -1 099 -1.106 -0.178 -0.347 -0.79 -0.799 -0.749 -0.937 (0.000) (0.0001 (0.000) k0.000) (0.000)1 (0.0001 (0.571) (0.080) (0.000o (0.0001 (0 000) (0.000 prinma3 -0.98 -0.99 -0 983 -0 979 -1.003 -1.026 -0.008 -0.148 -0 626 -0.744 -0.661 -0.846 (0.000) (0.000) (0.000> m.000 (0.000) (0.000) (0.981) (0.465> 10.000) (O.000) (0.000) (0.000) prima4 -0 848 -0.841 -0.801 -0.7(# -0 853 -0.877 0 -0059 -0 478 -0.592 -0.572 -0.774 (0.000 10 000) (0.000) (0.000> (0.0001 (0 000) (1.000) (0.799) (0 000 (0.000) (0.000) <0 000) secincl -1.01S -0.987 -0.914 -0.905' -0.917 -0.921 -0.007 -0.181 -0 667 -0.756 -0 644 -0.807 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.982) (0.336) (0.000) (0.000) (0.000) (0.000) secinc2 -0.816 -4) 793 -0 793 -0.817 -0.851 -0832 -0.121 -0.098 -0.537 -0.651 -0,SW$ -0%'06 10.001) (0.0001 (0.000) (0 000,1 (0.0001(0.000> (0.716) (0.6721 (0.000) (0.000) <0.000) (0.000) seeinc3 -0.799 -0.718 -0.74 -0.808 -0.721 -0.757 -0.109 -0.153 -0.626 -0.569 -0.431 -0.684 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.762) (0.562) (0.000) (0.000) (0.004) (0.000) seeinc4 -0.758 -0,656 -0.621 -0.557 -0.625 -0.628 0.103 0.022 -0.343 -0.357 -0.273 -0.526 (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) (0.755) (0.931) (0.005) (0.018) (0.048) (0.005) secun,l -1.004 -0866 -0.702 -0.764 -0.8 -0.697 -0.115 -0.146 -0.485 -0.424 -0338 -0.682 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.767) (0.625) (0.003) (0.013) (0.0 12) (0.009) secun2 -0.607 -0.589 -0.527 -0.504 -0.516 -0.493 0.048 0.023 -0.363 -0.388 -0.362 -0.642 (0.011) (0.002) (0.000) (0.000) (0.002) (0.000) (0.913) (0.922) (0.005) (0.008) (0.006) (0.000) secun3 -0.509 -0.477 -0.444 -0.405 -0.375 -0.273 0.268 0.107 -0.199 -0.338 -0.258 -0.13 (0.029) (0.007) (0.000) (0.000) (0.003) (0.000) (0.514) (0.701) (0.150) (0.056) (0.304) (0.356) secun4 -658 -0.319 -0.283 -0.165 -0.134 -0.013 0.126 0.16 -0.131 -0201 -0. 183 -9.365 (0.001) (0.223) (0.010) (0.063) (0.406) (0,924) (0.752) (0.476) (0.284) (0.240) (0.138) (0.083) univincl -0.541 -0.557 -0.401 -0.472 -0.444 -0.413 0.243 0.095 -0.488 -0.519 Y0.301 -0.543, (0.033) (0.001) (0.000) (0.000) (0.002) (0.000) (0.607) (0.698) (0.001) (0.001) (0.165) (0.002) univinc2 -0.447 -0.384 -0.381 -0.373 -0.351 -0.324 0332 0.239 -0.315 -0.264 -0.216 -0.402 (0.038) (0.070) (0.000) (0.000) (0.017) (0.000) (0.296) (0.334) (0.040) (0.070) (0.146) (0.028) univinc3 -0.622 -0.653 -0.361 -324 -0.162 0.035 0:333 0.45-0.252 -0.256 -0.101 -0.216 (0.096) (0.002) (0.014) (0.004) (0.286) (0.740) (1.000) (0.662) (0.117) (0.088) (0.664) (0.254) univinc4 -0.711 -0.103 -0.028 0.177 0.038 -0.056 0.36 0:046 0.208 -0.024 0.031 -0.128 (0.092) (0.774) (0.861) (0.151) (0.733) (0.639) (1.000) (0.870) (0.496) (0.874) (0.912) (0.715) 28 Argentina Poverty Assessment Background Paper No. 8 1988 0.05 0.1 0.25 0.5 0.75 0.9 1991 0.05 0.1 0.25 0.5 0.75 0.9 univl -0.781 -0.559 -0544 -0,465 -0,171 0.069 0.472, 0.123 4.089 O97 0.059 -0.028 (0.001) (0.019) (0.002) (0.008) (0.250) (0.741) (1.000) (0.738) (0.771) (0.710) (0.781) (0.929) univ2 -0.658 -0.246 -0.279 -0.267 -0.297 -0.288 0.135 0.256- 4159 -0.151 -0.067 -0.129 (0.028) (0.459) (0.066) (0.006) (0.141) (0.00 1) (0.790) (0.426) (0.256) (0.357) (0.749) (0.726) univ3 -0.252 -0.127 -0.082 -0.047 0.287 0.04 0.516 0.375 -036 0.029 0.171 0.211 (0.364) (0.624) (0.560) (0.732) (0.073) (0.767) (1.000) (0.259) (0.862) (0.889) (0.632) (0.422) manuf 0.034 0.056 0.042 -0.062 -0.142 -0.105 4175 -0.079 -0.055 -0.144 -0.08 -0.165 (0.623) (0.365) (0.428) (0.077) (0.001) (0.004) (0.000) (0.234) (0.356) (0.00!) (0.249) (0.039) const -0.053 -0.079 -0.091 -0.198 -0.27 -0.183 -0.188 -0.121 -0.052 -0.039 0.09 -0.009 (0.374) (0.259) (0.108) (0.000) (0.000) (0.018) (0.148) (0.317) (0.557) (0.588) (0.355) (0.919) serv1 0.071 0.042 -0.011 -0.,085 -0.138 -0.044 -0.05 -0.079 -0.071 -0.071 0.019 -0.098 (0.249) (0.490) (0.804) (0.016) (0.004) (0.317) (0.397) (0.297) (0.260) (0.090) (0.75 1) (0.168) other 0.007 -0.035 -0.038 -0.061 -0.139 -0.065 -0.23 -0.067 -0.065 -0.107 0.03 -0.044 (0.925) (0.613) (0.488) (0.211) (0.022) (0.172) (0.064) (0.516) (0.503) (0.109) (0.703) (0.650) public -0.107 -0.203 -0.112 -0.195 -0.257 -0.208 -0.099 -0.193 -0.213 -0.68 -0.12 -013 (0.216) (0.036) (0.095) (0.000) (0.000) (0.000) (0.446) (0.084) (0.002) (0.018) (0.154) (0.178) head -0.031 0.019 0.041 0.054 0.065 0.159 -0.014 -0105 -0001 0039 0.099 0106 (0.728) (0.755) (0.297) (0.171) (0.032) (0.02 1 (0.927) (0.267) (0.980) 10.428) (0. 107) (0. 120) skilled -0.04 -0.11 -0.249 4,403 -0.365 -0,476 -0.275 -.436 -0.357 -0.42 -0.525 -0.522 (0.831) (0.408) (0.004) (0.000) (0.000) (0.000) (0.021) (0.000) (0.000) (0.000) (0.000) (0.000) unskilled -0.189 -0.23 -0 417 -0.569 -056 -0.694 -0.41 -0 525 -u 464 -0.585 -0.73 -0.746 (0.363) (0.054) (0.000) (0.000) (0.0001 (0.0001 (0.008) (0.000) (0.000 (0.000) (0.000) (0.000) microfi -0.159 -0.22 -0 194 -0.195 -0.196 -0.182 -0.285 -0.245 -0.11 -0.07I -0.123 -0.068 (0 000) (0.000) 10.000) (0.000) (0.000) (0.000) (0.000 1 (0.000) 10.000) (0.015) (0.006) (0.064) october -0 141 -0.121 -0.084 -0.034 -0.049 -0.108 0.117 0.101 0.062 0 052 0.053 0.014 (0.000) (0.000) (0.000) (0.056) (0.019) (0.000) (0.007) (0.028) (0.005) (0.028) (0.061) (0.705) p-values are presented in parenthesis. 29 Argentina Poverty Assessment Background Paper No. 8 Table 4b: Quantile Regression Estimates of Log Wage Gaps, skill groups specification 1993 0.05 0.1 0.25 0.5 0.75 0.9 1995 0.05 0.1 0.25 0.5 0.75 0.9 Interopt 2.65 2.613 2.967 3.6 3,954 4209 1 802 2,361 2.927 3.281 3:942 4216 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) married 0.137 -05 0.058 -0.031 4011 0.056 0.021 0»021 0.021 0.002 0,035 0.087 (0.105) (0.354) (0.145) (0.447) (0.833) (0.454) (0.777) (0.763) (0.660) (0.945) (0.375) (0.202) c.uion 0.144 -0.037 -0093-0.037 -0.017 0.121 -047 -0.036 -0.001 70047 -0.024 0.001 (0.109) (0.532) (0.091) (0.348) (0.760) (0.191) (0.413) (0.622) (0.993) (0.122) (0.532) (0.986) age1 -0.113 -0.071 -0 12 -02 -0 248 -0248 -0039 -0.006 -0.054 -0.034 -0.103 -0 242 (0.236) (0.303) 10.011) l(0 000) 0.000) (0.012) (0.620) (0.943) (0 373) (0.409 (0.029) (0 001) age2 -0.162 -0.13 -0 068 -1.146 -0.178 -1) 164 0 072 0.062 0.034 0.026 -0.048 -0.14 (0.061) (0.013) (0.055) (0.000) (0.000) (0.006) (0.409) (0.331) (0.480) (0.489) (0.207) (0.005) age3 -0.059 -0.014 -0.016 -0.052 -0.073 -0.057 0.207 0.18 0.049 0.049 0.028 0.059 (0.444) (0.808) (0.655) (0.162) (0.082) (0.180) (0.002) (0.001) (0.242) (0.143) (0.492) (0.229) ageS -0.065 -0.076 -0.079 -0.09 -0.093 -0.024 0.283 0.199 0.049 0.028 0.051 0.124 (0.510) (0.196) (0.055) (0.05 1) (0.091) (0.738) (0.000) (0.001) (0.356) (0.582) (0.419) (0.062) primal -0.731 -0.659 -0.693 -0.894 -0.74 -0.896 -0615 0572 -0.698 ,0.586 -0.Š5 4786 (0.001) (0.000) (0.000) (0.000) (0.000) (0.000) 10.000) (.0 001) 10.0001 10 0 <1 «:)0.000) (0 000) prima2 -0.645 -0.586 -0.639 -0.807 -754 -0,837 -0 414 -0.451 -0.588 -0 SI -0.747 -0.633 (0.001 ) (0.0001 1.0.000) (0.000) (0.000) (0.011 (0.003) (0.003) (0.000) (0.025) 10.000) (0.000) prima3 -0.616 -0.604 -0.623 -0.809 -0 645 -0.71 -0 538 -1.415 -0.514 -044 -0668 -0.611 10.008) I0.000) (0.000) 10.000) (0.000) (0.007) (0.002) (0.011) (0 000) (0.051) (0.000) (0.000) prima- -0.644 -0 618 -0 618 -0.77 -0656 -0.721 -0.375 -0.279 -0.377 -0.281 -0.559 -0.504 (0.004) (0.000) (0.00%) (0.000) 10.000 10.00 11 (0.005) (0 095) (0.000) (0.200i (0.000 b (0.00!) secincl -0.541 -0.523 -0.539 -0.755 -(1 636 -0.739 -0.832 -0.4Q7 -0.543 -0.484 -0.681 -0.517 (0.013 (0.0001 10.000 1 (0.000) (0 000) 1(0 002) (0.00) 10.005) (0.000) .0.008,) (0 000) (0.004) secinc2 -0.628 -0.591 -0 565 -0 697 -0 57 -0.584 -0.345 -0.356 -0.413 -0.343 -0.575 -0.45 (0.004) (0.000) (0.000) (0.000) (0.000) (0.019) (0.013) (0.064) (0.000) (0.138) (0.000) (0.007) secinc3 -0.683 -0.629 -0.547 - 648 -0 501 -0 714 -0.47 -0.314 -0.407 -0.323 -0.45 -0.428 <0 003 (0.000, (0.000) (0 000 )0 OJO) (0 003 0.00l ) (0.127) (0.000) (0.125) (0.000) (0.005) secimc4 -0.476 -0471 -0388 -0.458 -0 393 -0.449 -0215 -0.077 -0.112 -0.04 -0.298 -0.21 (0.024) (1.000 > ( 007) 1(1(100) (0.00 I (0.074 1 (0. 183) (0.661) (0.260) (0.824) (0.004) (0.253) secuni -0.578 -0.549 -0.544 -0.637 -0.4S1 -0 531 -0.732 -0.736 4432 -0.233 -0.484 -0.315 (0.010) (0.000) (0.000) (0.000) (0.000) (0.061) (0.000) (0.000) (0.002) (0.282) (0.000) (0.082) secun2 -0.511 -0,416 -0.388 -0.623 -0.455 -0433 -0.134 -0125 -0.28 -0.152 -0.362 -0.184 (0.022) (0.000) (0.001) (0.000) (0.000) (0.104) (0.428) (0.450) (0.011) (0.419) (0.000) (0.306) secun3 -0.366 -0.279 -0.377 -0.42 -0.236 -0.33 0 .011 0»9 -0,207 -0.112 -0.77 0132 (0.133) (0.000) (0.007) (0.000) (0.228) (0.046) (0.944) (0.960) (0.035) (0.579) (0.027) (0.416) secun4 -0.4 -0.217-0.239 -0.262 -0J49 -0.295 -0128 -0.161 0.069 0.132 -0.126 0.05 (0.158) (0.230) (0.053) (0.041) (0.250) (0.240) (0.362) (0.460) (0.533) (0.482) (0.202) (0.715) univinci -0.44 0.n2 -0.236 70.343 -0.261 0.429 36 -05 -0.244 -0.098 0.33 -0.015 (0.074) (0,186) (0.032) (0.010) (0.027) (0.066) (0.386) (0.327) (0.100) (0.653) (0.005) (0.917) univinc2 -0.141 -0 215 -0.234 -.04279i -0.26 -0. 375 407 -0.0 2 -0.111 .0115 -0.28 0,022 (0.484) (0.029) (0.099) (0.011) (0.031) (0.128) (0.586) (0.744) (0.350) (0.563) (0.004) (0.909) univinc3 -0.171 0.06 Ö.134 -0188 -0,153 -0.3 -0.123 0.067 0,094 0.112 0.039 0.468 (0.355) (0.793) (0.347) (0.115) (0.301) (0.089) (0.514) (0.753) (0.560) (0.614) (0.859) (0.012) univinc4 -0.253 0-066 0.06 -0.147 0.062 0,014 -0.07 0,108 0.f07 0-111 - -0432 .074 (0.204) (0.840) (0.610) (0.205) (0.731) (0.955) (0.700) (0.754) (0.394) (0.578) (0.796) (0.769) 30 Argentina Poverty Assessment Background Paper No. 8 1993 0.05 0.1 0.25 0.5 0.75 0.9 1995 0.05 0.1 0.25 0.5 0.75 0.9 tivl -0.049 0.024 0.042 -0.039 0.03 0.O88 -0.086 0.011 0.154 0.074 -0.119 0051 (0.902) (0.829) (0.831) (0.750) (0.849) (0.753) (0.477) (0.972) (0.261) (0.699) (0.326) (0.787) unaD -0.423 -0.009 -0.157 -0.279 -0.186 -0 215 -0.247 -0.267 -U.084 0.218 0.089 0.135 0.187) 10 974) (0.1221 (0.02 3) (0 150) (0.406) (0.150) (0.220) (0.421 ) 10.270) (0.411) (0.492) univ3 -0.17 0.109 0.088 -0.02 0.096 0.058 -0.343 -0.192 0.106 0.143 -0.141 0.136 (0.551) (0.545) (0.491) (0.916) (0.576) (0.786) (0.021) (0.644) (0.456) (0.540) (0.457) (0.614) manuf 0.01 0.045 -0.016 -0.098 -0.133 -0.168 -0.011 -0.023 0.035 -0429 -0.167 -0.092 (0.901) (0.387) (0.758) (0.040) (0.011) (0.005) (0.869) (0.763) (0.493) (0.020) (0.001) (0,124) const 0.026 0.157 0 124 fi.lu 0.018 0.011 -0 103 0 l 14 0 113 0 024 0.019 0.14 (0.840) (0.090 1 (0 077) (0.069) (0.830) (0.876) (0.608) (0.190) (0.100) (0.755) (0.685) (0.271) ser l -0.01 0.066 -0.021 -0.102 -0.104 -0 15 0.021 -0.036 0.003 -0.143 -0.096 4.046 10.903) (0.082H (0.638) (0.057) (0.070) (0.013) (0.722) (0.652) (0.960) (0.012 1 (0.066) (0.445 i other -0 2 -0 062 -0.08 -0.076 -M 095 -0.07 -0.095 -0.134 -0.044 -0.222 -0.206 -0.193 (0 185) (0.454) 10.268) (0.177 i (0.0901 (0.258) (0.283 1 (0.136) (0.469) (0 0000> (0.001) (0.039) public -0 176 -0.112 -0.109 -0.202 -0.235 -0.274 -0.071 -0.198 -0 121 -0.233 -0.137 -0.082 <0.060) (0.135) 10.044) (0.000) (0.000) (0.000) (0.337) 10.001) (0.046) (0.002) (0.0331 (0.227) head 0.208 0.178 0.171 0 142 0.132 0 063 0.092 0.096 0.087 0.134 0.111 -0.008 (0.008) 10.001) (0.000) (0.000) (0.009) (0.464) 10.057) (0.169) 10.0541 (0.000) 10.016) (0.903) skilled -0 403 -0.385 -0.428 -0 508 -o.626 -0.46 -0.076 -0 428 -0.581 -0.587 -0.728 -0.821 (0.000) (0.000) (0.000) (0.000) (0.000 I 10.000) tO 631) (0 045) t0.0001 (o.000) (0.000) (0.000I unskilled -0.574 -p<58 -0.574 -0.642 -0.759 -0.646 -0 134 -0.53 -0.73Q -0.78 -0.892 -0 942 (0 000) (0.000) (0.000) (0.000) (0.000) (0.0001 (0.472) (0.014) (0 000) (0.000) (0.000) (0.000) nicrohi -0.166 -0.113 -0.071 -0.084 -0.115 -0.101 -0.157 -0.159 -0.157 -0.111 -0.081 -0.049 (0.002) (0.002) (0.02 1 (0.000) (0.000 1 (0.004 1 <0.019) (0.000) (0.000) (0.000) (0.015) (0.336) october -0.003 -0.006 0.016 0.43 1)033 0.016 0.025 0.005 0.002 -0.016 -0.011 -0033 (0.947) (0.839) (0.458) (0.034) (0 185) (0.563) (0.596) (0.850) (0.949) (0.417) (0.614) 0.366 p-values are presented in parenthesis. 31 Argentina Poverty Assessment Background Paper No. 8 Table 4c: Quantile Regression Estimates of Log Wage Gaps, skill groups specification 1997 0.05 0.1 0.25 0.5 0.75 0.9 Intercept 2.702 2.927 3.174 3.471 378 4.377 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) married 0.165 0.211 0.102 0.111 0.148 0.142 (0.144) (0.025) (0.006) (0.004) (0.000) (0.000) c.union 0.026 0.096 0.027 0.028 0.007 -0.034 (0.835) (0.352) (0.585) (0.414) (0.852) (0.535) agel -0.167 -0.158 -0.166 -0.181 -0.219 -0.26 (0.064) (0.002) (0.000) (0.000) (0.000) (0.000) age2 4133 -093 -0.11 -0.096 -0.162 -0.173 (0.047) (0.032) (0.008) (0.001) (0.000) (0.004) age3 -0.53 -0.1:39 -0.154 -0.102 -0.097 -0.085 (0.081) (0.005) (0.000) (0.000) (0.014) (0.114) ageS -0. 142 -0.119 -0.084 -0.034 -0.064 -0.029 (0.132) (0.035) (0.136) (0.422) (0.107) (0.701) primal -1.185 -1L244 -1.055 -0.909 -0.829 -1.274 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) prima2 -0.866 -0.956 -0.875 -0.754 -0.714 -1.148 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) prima3 -0.674 -0.917 -0.837 -0.789 -0.639 -1.064 (0.006) (0.000) (0.000) (0.000) (0.000) (0.000) prima4 -0.749 -0.883 -0.759 -0.616 -0.584 -1.003 (0.001) 10.0001 (0.000) (0.000) (0.003) (0 000) secincl -I 022 -1.033 -0.901 -0.82 -0.778 -1.188 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) secinc2 -0.72 -0.908 -0.726 -0.644 -0.596 -0.952 (0.002) (0.000) (0.000) (0000) (0.001) (0.000) secinc3 -0.821 -0.946 -0.7 -0.608 -0.584 -1.049 (0.003) (0.000) (0.000) (0.000) (0.001) (0.000) secc4 -0.655 -0 782 -0.488 -0.403 -0.427 -0.867 (0.008I (0.0001 (0.0001 (0.001) (0.020 1 (0.00 1) secun 1 -0.781 -0.874 -0.728 -0.586 -0.542 -0.978 (0.000) (0.000) (0.000) (0.000) (0.002) (0.000) secun2 -0.552 -0.712 -0,537 -0.402 -0.272 -0.649 (0.008) (0.000) (0.000) (0.000) (0.158) (0.021) secun3 -0.469 -0.695 -0518 -0.393 -0,161 -0.558 (0.038) (0.000) (0.000) (0.000) (0.409) (0.016) secun4 -0.128 -0.352 -0.267 -0.159 -0.165 -0.611 (0.602) (0.023) (0.003) (0.128) (0.317) (0.009) univincl -0.51 -0.637 -0.518 -0.337 -0.248 -0.662 (0.022) (0.002) (0.000) (0.006) (0,169) (0.011) univinc2 -0.316 -0.469 -0.384 -0.284 -0.147 -0.455 (0.231) (0,004) (0.000) (0.008) (0.385) (0.108) univin6 -0.182 -0.289 -0.236 -0.162 -0.151 -0.304 (0.547) (0.181) (0.002) (0.160) (0.435) (0.313) univino4 -0.19 -0,397 -0.3 16 0.025- 0.171 -0.339 (0.618) (0.016) (0.012) (0.846) (0.421) (0.174) 32 Argentina Poverty Assessment Background Paper No. 8 1997 0.05 0.1 0.25 0.5 0.75 0.9 univi -0.605 -0,693 -0.363 0.015 0.05 -0.235 (0.015) (0.000) (0.015) (0.917) (0.764) (0.466) univ2 -0.621 -0,549 -0.179 -0.093 0.038 -0.327 (0.036) (0.010) (0.170) (0.387) (0.843) (0.140) univ3 -0.523 -0.321 -0.216 -0.056 0.158 -0.217 (0.185) (0.274) (0.065) (0.673) (0.560) (0,358) manuf 0.051 0.088 -0.015 -0.02 -0.009 0.016 (0.535) (0.082) (0.721) (0.558) (0.795) (0.722) const -0.119 0.114 0.076 0.154 0.119 0.096 (0.437) (0.304) (0.260) (0.004) (0.014) (0.172) serv -0.078 0.02 -0.07 -0.027 -0.004 0.025 (0.407) (0.652) (0.060) (0.465) (0.912) (0.626) other -0.211 -0.14 -0.029 -0.01 0.038 -0H01 (0.161) (0.087) (0.643) (0.824) (0.409) (0.737) public -0105 -0.074 -0.068 -0.078 -0.03 -066 (0.308) (0.297) (0.355) (0.045) (0.611) (0.395) head 0016 -0.056 0.05 0.03 -0.012 0.039 (0.885) (0.535) (0.154) (0.360) (0.782) (0.347) skilled -0.417 -0.383 -0.405 -0.528 -0.608 -0.577 (0.012) (0.001) (0.000) (0.000) (0.000) (0.000) unskilled -0.521 -0.573 -0.573 -0.667 -0.754 -0.714 (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) microfi -0.255 -0.223 -0.193 -0.184 -0.152 -0.038 (0.001) (0.000) (0.000) (0.000) (0.000) (0.300) oober 0.022 0.016 0 -0.014 -0.016 -0.042 (0.621) (0.607) (0.990) (0.464) (0.444) (0.173) p-values are presented in parenthesis. 33 Argentina Poverty Assessment Background Paper No. 8 Table 5: Mean Regression Estimates of Log Wage Gaps, EPH 1988-97 1988 1991 1993 1995 1997 Intercet 3.409 3.257 3.596 3535 3.564 (0.000) (0.000) (0.000) (0.000) (0.000) married 0.094 0.176 -0.019 0.051 0.129 (0.008) (0.000) (0.629) (0.193) (0.000) c.union >,053 0.097 -0.7 -0.007 0.012 (0.235) (0.064) (0.408) (0.878) (0.752) agel -0.251 -0.096 -0.177 -0.09 -0.22 (0.000) (0.038) (0.000) (0.029) (0.000) age2 -0.118 -0.036 -0.136 -0.01 -0.156 (0.000) (0.335) (0.000) (0.758) (0.000) age3 -0.024 -0.051 -0.044 0.063 -0.123 (0.376) (0.156) (0.160) (0.049) (0.000) ageS -0.032 0.08 -0.083 0.078 -0.067 (0.328) (0.077) (0.034) (0.059) (0.063) prima -0819 -0.666 -0.675 -0.62 -0.733 (0,000) (0.000) (0.000) (0.000) (0.000) seine -0.579 -0.497 -0.509 -0.445 -0,603 (0.000) (0.000) (0.000) (0.000) (0.000) secun -0.274 -0.251 -0369 -0.265 -0.35 (0.000) (0.000) (0.000) (0.000) (0.000) unilvine -0.085 -0.179 -0.168 -0.115 -0.158 (0.079) (0.004) (0.002) (0.034) (0.001) expI -0.352 -0.2.55 -0.179 -0.313 -0.32 (0.000) (0.000) (0.000) (0.000) (0.000) exp2 -0.257 -0.21 -0.156 -0.197 -0.157 (0.000) (0.000) (0.000) (0.000) (0.000) exp3 -0.158 -0.102 -0.066 -0.136 -0.126 (0.000) (0.006) (0.050) (0.000) (0.000) manuf -0.028 -0 117 -0.087 -0.062 0.018 40.366) 10.005) 10.0171 (0.0971 f) 560i const -0.162 -0.047 0.058 o 069 0.096 (0.000) (0.418) (0.263) (0.181) (0.028) servi -0.042 -0.064 -0.078 -0.058 -0.023 (0.191) (0.137) (0.036) (0.115) (0.453) -0.058 -0.056 -0.099 -0.153 -0.019 (0.139) (0.290) (0.034) (0.001) (0.635) public -0.198 -0.19 -0.183 -0.181 -0.069 (0.000) (0.000) (0.000) (0.000) (0.082) head 0.045 0.021 0.129 0 075 0.03 (0.1841 (0.622) (0000) (0048) (0 377) skilled -0 34 -0.448 -0 522 -0 588 -0.509 (0.000) (0.000) (0.000) (0.000) (0.000) unskilled -0.522 -0.614 -0.686 -0.743 -0.659 (0.000) (0.000) (0.000) (0.000) (0.000) microfi -0.187 -0.116 -0.105 .0.122 -0.163 (0.000) (0.000) (0.000) (0.000) (0.000) october -0.086 0.047 0.02 -0.017 -0.02 (0.000) (0.033) (0.292) (0.390) (0.264) p-values are presented in parenthesis. 34 Argentina Poverty Assessment Background Paper No. 8 Table 5b: Mean Regression Estimates of Log Wage Gaps for Skill roups specification, 1988-97 1988 1991 1993 1995 1997 Intercept 3.388 3.128 3.563 3.347 3.567 (0.000) (0.000) (0.000) (0.000) (0.000) Married 0.09 0.18 -0.018 0.046 0.126 (0.012) (0.000) (0.636) (0.247) (0.000) c.union 0.047 0.104 -0.034 -0.011 0.01 (0.295) (0.048) (0.444) (0.798) (0.804) agel -0.253 -0.086 -0.8 -0.087 -0.0)18 (0.000) (0.066) (0.000) (0,035) (0.000) age2 -0.117 -0.034 -0.141 -0.016 -0.55 (0.000) (0.361) (0.000) (0.623) (0.000) age3 -0.025 -0.046 -0.054 0.058 -0.12 (0.373) (0.200) (0.085) (0.068) (0.000) age5 -0028 0.082 -0.001 0.079 -0.064 (0.391) (0.074) (0.039) (0.057) (0.077) primal -1.12 -0.806 -0.2 -0.732 -1.056 (0.000) (0.000) (0.000) (0.000) (0.000) prima2 -1 068 -0.769 -0.764 -0.626 -0.761 (0.000) (0.000) (0.000) (0.000) (0.000) prima3 -0.976 -0.666 -0.712 -0.52 0.65 (0.000) (0.000) (0.000) (0.000) (0.000) prima4 -0.508 -0.539 -0.701 -0.435 -0.771 (0.000) (0.000) (0.000) (0.000) (0.000) secinc 1 -0.902 -0.64 -0.657 -0.59 -0.939 (0.000) (0.000) (0.000) (0.000) (0.000) secinc2 -0792 -0.598 -0.643 -0.467 -0.761 (0.000) (0.000) (0.000) (0.000) (0.000) seeinc3 -0.733 -0.531 -0.631 -0.412 -0.774 (0.000) (0.000) (0.000) (0.000) (0.000) secinc4 -0.598 -0.346 -0.437 -0.179 -0.601 (0.000) (0.000) (0.000) (0.060) (0.000) secun I -0.728 -0355 -0.573 -0.461 -0.733 (0.000) (0.001) (0.000) (0.000) (0.000) secun2 -0.49 -0.373 -0.517 -0.249 -0.507 (0.000) (0.000) (0.000) (0.009) (0.000) secun3 -0.37 -0.175 -0.377 -0.19 -0,472 (0.000) (0.109) (0.000) (0.064) (0.000) secun4 -0.192 -0.164 -0.26 -0.036 -0.307 (0.013) (0.087) (0.003) (0.723) (0.000) univincl -0.439 -0.414 -0.359 -0.205 -0.482 (0.000) (0.000) (0.000) (0.043) (0.000) univinc2 -0.311 -0.213 -031 -0.164 -0.343 (0.000) (0.033) (0.001) (0.093) (0.000) univinc3 -0.259 -0.149 -0.152 0.037 -0,27 (0.005) (0.219) (0.181) (0.757) (0.011) univinc4 -0.004 -0.027 -0.06 0.042 -0.169 35 Argentina Poverty Assessment Background Paper No. 8 (0.964) (0.832) (0.578) (0.702) (0.092) 1988 1991 1993 1995 1997 univl -0.352 0.007 0.008 0.006 -0;237 (0.000) (0.955) (0.940) (0.955) (0.012) univ2 -0.31 0.03 -0.251; 0,57 -0236 (0.001) (0.415) (0.009) (0.559) (0.010) univ3 0.03 0:086 0.042 0.026 4142 (0.764) (0.483) (0.698) (0.825) (0.164) manuf -0.031 -0.107 002 -0,066 0.2i (0.321) (0.012) (0.025) (0.079) (0.506) const -0.168 -0.037 J 005 0.66 0 095 (0.000) (0.528) (0.335) (0.204) (0.031) servl 4042 -0057 073 062 -0021 (0.196) (0.189) (0.049) (0.096) (0.490) other -0158 -0.048 0.094 0 158 -0017 (0.141) (0.371) (0.044) (0.001) (0.675) public -0.199 -0.178 0186 -0 6 -0071 (0.000) (0.001) (0.000) (0,000) (0.075) head 0.05 0,023 0.126 0.078 0.02q (0.145) (0.603) (0.001) (0.040) (0.399) skilled -0.336 -0.45 -0.512 -0.584 -0.501 (0.000) (0.000) (0.000) (0.000) (0.000) unskilled -0,518 -0.612 0678 -0.737 -0.65 (0.000) (0.000) (0.000) (0.000) (0.000) microfi -0.19 -0.117 -0.105 -0.21 -0.164 (0.000) (0.000) (0.000) (0.000) (0.000) october -!.084 04 02 -0.017 0,02 (0.000) (0.029) (0.287) (0.390) (0.257) p-values are presented in parenthesis. 36 Argentina Poverty Assessment Background Paper No. 9 Poor People in a Rich Country A Poverty Report for Argentina Background Paper No. 9 Does the Gender Recipient of Social Programs Matter in Argentina? by Pia Peeters and Wendy Cunningham Argentina Poverty Assessment Background Paper No. 9 Does the Gender Recipient of Social Programs Matter in Argentina? Pia Peeters & Wendy Cunningham Input to the Poverty Assessment Introduction 1. In this short overview paper, we address the questions of whether or not the household structure and/or the gender of the household head is instrumental in the distribution of resources obtained from transfers (in-kind or money) or training programs for the recipient's children. Evidence from England (Lundberg 1996) shows that children whose mothers receive money transfers are more likely to share in some of the wealth than are children whose fathers receive the transfers. Lundberg's work presents a very strong argument for directing money transfers toward wives rather than husbands when the goal of transfers is to benefit children. However, her work is limited to households that are headed by a married couple, so we are unable to conclude whether or not single mothers and fathers have the same tendencies. A recent study from Bangladesh (Pitt 1998) finds that the higher incomes earned by women who participated in job training programs are more generously distributed to their children than are the higher incomes earned by men who accumulate the additional human capital. Again, the results favor targeting wives, but we cannot extrapolate to unmarried individuals. This paper intends to consider both gender and household structure, defined by the marital status of the head, in the context of the allocation of additional resources to children where our outcome variables are school attendance and child illness. Data, Definitions, and Strategy 2. The Encuesta de Desarrollo Social (EDS), surveyed by SIEMPRO in August 1997, is used. We will look at single-headed households and couple-headed households, disaggregated by gender and income. A single- headed household has one, unmarried (or not consensual union) adult as the head. A couple-headed household will be defined as one with two adults at the "head" that are either married or in a consensual union. The "head" is identified by the household members in the questionnaire. Thus, we have four household types: female single-headed household; female couple-headed household; single male-headed household, and male- headed couple households. 3. To measure distribution of the benefits derived from social programs in the household, we will consider school attendance of the children (4-15 years) and children's (0-15 years) illness during the 30 days prior to the suivey date. As a measurement of income, we construct quintiles based on total (labor and non-labor) household income flows. 4. We will define three types of benefit receipt: money, in-kind, and training. Money income includes regular, direct transfers from the government or private organizations. In-kind benefits are non-money goods transfers from the same types of Argentina Poverty Assessment Background Paper No. 9 organizations. An individual is identified as having accumulated training if he or she participated in a firm-level or public training course in the three years prior to the survey date. The benefit collection is summarized for the household, so we cannot identify the gender of the recipient in couple-headed households. The training variable is associated with an individual, though. 5. First, we will map the allocation of resources, by gender, household type and income quintile. Once we understand who receives transfers and training, we will look at the school attendance and health outcomes for children in single- and couple-headed households. Disaggregating by household type and gender will allow us to understand if the household structure has an influence on the distribution of transfers and training programs, and/or if the gender of the recipient has an influence on the distribution of benefits. The differences mentioned below are statistically significant at the 5% level, unless otherwise stated. Allocation of Transfers and Training 6. Who received transfers? Using data on the receipt of money and in-kind transfers during the last 12 months, we find that both gender and the household structure seem to have an influence on the likelihood of receiving transfers. Single-headed households and female-headed households tend to be more likely to receive transfers. This relationship is even stronger when controlling for the presence of children in the household. More specifically, from Table 1: * Single-headed households, regardless of gender, receive more transfers than couple-headed households, especially money transfers. This relation holds even stronger for households with children. * Female-headed households are more likely to receive benefits than are male- headed households of the same structure. This relationship is stronger for couple-headed households then for single-headed households. When controlling for the presence of children in the household, female-household heads are much more likely to receive benefits then male-household heads, both in single- and couple-headed households. * The higher the quintile to which the household head belongs, the less likely that he/she receives benefits. This relationship is especially strong for in-kind benefits. 7. Who followed training during the last three years? The structure of the household does not seem to have an influence on the likelihood of the household head to have followed training during the last three years. For single-headed households, the probability of following training differs between gender, especially for single-headed household with children. The income level seems to be positively correlated with the likelihood of training. Again, from Table 1, 2 Table 1: Social Transfers & Training Single headed Households Couple Headed Households Fmale aeHaFemale Male Total e Male Head Head wI Total a Male Head wl child Head wI Head w Child Child Child a Who collected benefits during last 12 months? 27.22 28.53 24.04 36.86 22.65 17.38 26.74 17.01 20.96 32.86 20.54 money 16.64 17.06 15.64 18.72 7.18 7.31 9.88 7.21 7.63 10.36 7.54 in-kind 15.95 13.51 16.95 25.62 17.13 12.24 19.77 11.95 15.9 27.14 15.51 Who collects benefits, decile disaggregated uintilel 45.33 45.84 43.96 48.3 44.68 34.6 47.41 33.98 36.17 48.31 35.58 quintile2 29.72 30.52 27.05 31.68 17.78 20.57 33.72 20.06 22.47 41.51 21.77 quintile3 29.55 29.91 28.7 26.85 11.54 13.23 18.87 13.07 13.92 5.26 14.07 quintile 4 21.65 23.17 18.21 26.45 11.76 8.01 14.89 7.82 9.69 19.05 9.47 quintile5 11.52 13.04 8.61 19.44 0 4.47 7.55 4.36 5.35 5.88 5.33 Who collects in-kind benefits, decile disaggregated quintilel 36.26 37.27 33.56 40.29 34.04 29.52 42.24 28.91 31.87 43.82 31.29 quintile2 19.45 20.34 16.43 21.29 11.11 14.76 26.74 14.29 18.57 33.96 15.93 quintile3 15.18 15.46 14.5 16.78 11.54 7.58 7.55 7.58 8.74 5.26 8.8 quintile 4 8.74 8.8 8.61 12.4 11.76 3.02 4.26 2.99 4.52 9.52 4.41 quintile5 2.24 2.48 1.78 0 0 1.05 3.77 0.95 1.94 5.88 1.83 Who collects money benefits, decile disaggregated quintilel 19.22 18.76 20.47 19.42 17.02 9.31 9.48 9.3 8.34 8.99 8.31 quintile2 17.02 17.19 16.43 16.83 6.67 8.46 12.79 8.3 8.48 15.09 8.24 quintile3 19.91 19.39 21.15 15.44 0 6.9 11.32 6.78 6.37 0 6.48 quintile 4 17.17 18.04 15.23 15.7 0 5.71 10.64 5.58 6.35 9.52 6.28 quintile5 10.4 12.11 7.12 19.44 0 3.62 5.66 3.54 3.73 5.88 3.67 Sample size 6465 4585 1880 1202 181 13807 516 13291 8307 280 8027 Who followed training during last 3 years? 19.7 21.49 15.41 24.35 7.1 18.5 21.43 18.38 20.43 21.72 20.39 P quintilel 9.11 10.31 5.47 12.01 2.13 7.12 8.42 7.06 7.48 10.13 7.36 quintile2 11.97 14.14 4.92 18.42 6.82 13.32 18.31 13.12 14.91 17,65 14.81 quintile3 17.17 20.71 9.2 33.57 7.69 16.91 12.77 17.04 21.29 16.67 21.37 quintile 4 22.36 26.25 14.04 43.59 33.33 28.51 39.02 28.21 36.2 52.38 35.82 quintile5 39.16 42.89 33.46 63.38 0 37.21 28.57 37.55 48.6 35.29 48.98 Sample size 3614 2550 1064 1039 155 10424 406 10018 7179 244 6935 Argentina Poverty Assessment Background Paper No. 9 * Single-headed households are almost as likely (as couple-headed households) to have followed job training during the last three years. This holds true even when controlling for children in the household. Disaggregated by quintiles, single-headed household heads with children from the top three quintiles are more likely to have followed training than couple-headed household heads with children. * In couple-headed households, male or female household heads are equally as likely to have followed training programs, even when controlling for children. When disaggregating by quintiles, the results are mixed and not statistically significant. * Female single-headed household heads are more likely to have followed job training than male single-headed household heads. After controlling for the presence of children, female household heads are more than three times as likely to have followed training than male household heads. * In general, household heads in the lower quintiles are less likely to have followed training than the household heads of the higher quintiles, with some exceptions for the fifth quintile. School Attendance of Children 8. Whose children attend school? (no control for transfer/training receipt) The patterns of school attendance differ along household structure, gender of the household head, and income lines. According to Table 2, * School attendance of children from couple-headed households is slightly higher than for children from single-headed households. This relationship is stronger for male-headed households than for female-headed households. * In couple-headed households, school attendance is positively correlated with household income. In single-headed households, though, this relationship is not statistically significant. 9. Does school attendance of children differ, controlling for benefit/training receipt? The household structure seems to have an influence on the level of school attendance of children whose parent does not receive transfers or participate in training. However, if household heads do receive transfers or participate in training programs, there is no important disparity between the school attendance level of the children. Furthermore, in single-headed households, the gender of the household head seems to have an influence on the school attendance of children, especially for household heads who do not receive benefits or participate in training programs. According to Table 2, 4 Argentina Poverty Assessment Background Paper No. 9 * The school attendance of children in single headed households where the mother receives benefits or participates in training programs is not significantly different from the school attendance of children in comparable male-headed households. * School attendance of children in households where the single female head does not receive benefits or participate in training is higher than for male single-headed households who do not receive benefits or participated in training programs. * In single-headed households, regardless of gender, receipt of money transfers significantly increases the school attendance of children. This relationship is especially strong for male-headed households. * Participating in training programs significantly increases school attendance of children from both female and male single-headed households. * In couple-headed households, the gender of the household head does not significantly alter the school attendance of children. Children whose household head does not receive transfers tend to be slightly more likely to go to school than those whose parents receive transfers. * The school attendance of children in couple-headed households that receive training tends to be slightly higher than that of couple-headed households that do not receive training. Table 2. Children's School Attendance by Parents Receipt of Benefits Sirgle Headed Hosuseholds Cou pe Headed Households Total Male Head Female Head Total Male Head Female Head Participant in social program 93.33 92.94 93.38 93.72 93.66 94.74 Non-participant in social program 93 88.83 93.73 96.32 96.3 96.81 Participant in SP money transfer 96.3 100 95.99 95.54 95.47 97.22 Non-participant in SP money transfer 92.48 89.02 93.06 95.7 95.69 95.92 Participant in SP in-kind transfer 92.52 91.3 92.67 93.31 93.25 94.16 Non-participant in SP in-kind transfer 93.39 89.71 93.99 96.29 96.27 96.99 Participant in training program 98.13 100 98.02 98.56 98.6 97.37 Non-participant in training program 91.55 88 92.16 95.24 95.21 95.99 School attendance, overall 93.13 90.11 93.59 95.69 95.68 96.03 School attendance, by quintile Quintilel 91.8 90.53 91.96 93.12 93.03 95.27 quintile2 92.09 87.93 92.91 96.17 96.22 95 quintile3 96.33 90 97.95 97.61 97.56 100 quintile4 97.33 94.44 97.73 98.67 98.64 100 quintile5 97.53 100 97.4 99.22 99.19 100 5 Argentina Poverty Assessment Background Paper No. 9 Health of children 10. Do benefits/training affect children's health outcomes? The household structure and/or gender of the household head has little influence on the health of children in the household. According to Table 3, * The likelihood of child illness during the last 30 days, whether disaggregated by gender, household structure or income, do not show clear patterns. * In single headed households that are not recipients, children with only a mother tend to be more sick than children with only a father. This difference, however, is not statistically significant (at the 10% level) for household heads that do receive money or training. In couple-headed households, the difference is not statistically significant. * Children whose household received money transfers tended to be ill more often than children in non-receipt households. * When the household head followed a training program during the last three years, children were more prone to illness. * With the exception of the fifth quintile, the likelihood of illness in the past thirty days is positively correlated with wealth. Table 3. Child Illness During Last 30 Days by Parents' Receipt of Benefits Sin! ile Headed Households Couple Headed Households Total Male Head Female Head Total Male Head Female Head Participant in social program 21.97 20.51 23.46 21.21 21.38 21.03 Non-participant in social program 22.55 20.5 24.11 24.14 24.41 23.85 Participant in SP money transfer 21.58 20.51 22.96 20.17 20.3 20.03 Non-participant in SP money transfer 27.57 28.77 26.55 30.47 31.93 29.01 Participant in SP in-kind transfer 20.77 19.01 22.68 20.46 20.53 20.37 Non-participant in SP in-kind transfer 20.86 18.05 23.08 22.77 22.76 22.77 Participant in training program 22.48 21.54 23.65 20.77 20.99 20.55 Non-participant in training program 25.26 23.85 26.69 21.69 22.01 21.35 Child sick during last 30 days, overall 20.06 18.55 21.61 20.39 20.52 20.24 Child sick during last 30 days, by quintile Quintilel 19.98 18.29 21.56 20.43 21.08 19.74 quintile2 21.79 18.15 25.91 21.69 22.48 20.86 quintile3 25.27 23.57 27.42 21.31 21.4 21.22 quintile4 26.74 26.67 26.8 23.18 21.73 24.54 quintileS 20.00 16.33 23.91 19.05 18.08 20.08 6 Argentina Poverty Assessment Background Paper No. 9 Conclusion 11. The results suggest that children of unmarried women share in the benefits in similar (but a bit lower) proportions than children of married women. However, the statistics suggest that single father's distribution patterns are more similar to those of women (married or not) than of husbands. In particular, the children of those unmarried fathers who received benefits are much more likely to share in the benefits than are those of non-recipient fathers. The school attendance differential between the recipients and non-recipients is particularly striking for the unmarried fathers, as compared to the unmarried mothers, since the unmarried father's children start at lower levels. In couple- headed households, on the other hand, the gender of the head does not seem to significantly school attendance. 13. When one takes the level of wealth into account, school attendance is lower for children in households in the lower quintiles. At the same time, male-headed households in the lower quintiles were less likely to participate in training programs and/or receive training than were unmarried females who were household heads. 14. Despite these differentials and the apparent importance of benefits for children of unmarried men, single fathers are less likely to receive benefits or training in comparison to single mothers. If further research would confirm this relationship, targeting single male-headed households with children (in addition to the single mothers, who already are targeted) for receiving social transfers in money or participating in training programs might have a positive influence on the school attendance level of their children. 15. The health results are particularly questionable given the gender bias in interpretation of "illness". Perhaps the higher level of child illness in female headed households is due to women being more sensitive to their children's complaints than are men. Future work 16. As a measurement of wealth, we have used quintiles. One problem with this measurement is that zero income and missing income levels were coded the same in the survey. Thus, our income measures may be downward biased. Furthermore, such a measure is subject to random employment patterns or income patterns, and may capture an "atypical" period rather than the typical wealth level of a household. An alternative means of identifying wealth is through the Unsatisfied Basic Needs (UBN) indicator to define if a family lives in structural poverty or not. SIEMPRO has calculated and used the UBN indicator in its analysis of social development in the country such that a household is identified as structurally poor if at least one of the following holds true: * More than three persons are living in one room * The household lives in a precarious house * The household does not have its own lavatory * Kids from 6 to 12 years old are not attending school 7 Argentina Poverty Assessment Background Paper No. 9 Having four or more household members per person working and a household head with 2 or less years of primary school However, it was not clear, for the purposes of this study, how to replicate SIEMPRO's measure with the EDS. 17. Finally, the conclusions presented in this paper were drawn from simple cross tabulations that only controlled for the sex, martial status, and transfer/training receipt by household heads. Of course, other household characteristics such as parent's education and labor force status or community (spacial) variables such as proximity to health or school services are not controlled. Furthermore, the tables do not control for selectivity bias. The higher likelihood of school attendance for children of single fathers who receive transfers, for example, may simply be a reflection of the level of concern a father has for his children such that he both sends them to school and seeks financial assistance. Thus these results do not claim to identify causality but rather are suggestive to motivate r the need for a careful study to identify the real effects of social programs on the well- being of the children of the recipients. References Lundberg, Shelly, et. al. (1996) "Do Husbands and Wives Pool Their Resources? Evidence from the United Kingdom Child Benefit" Journal of Human Resources 32(3), 473-480. Pitt, Mark and Shahidur Khandker (1998) "The Impact of Group-Based Credit Programs on Poor Households in Bangladesh: Does the Gender of Participants Matter?" Journal of Political Economy 106(5), 958-996. 8