Colombia - Poverty assessment report (Vol. 2 of 2) : Annexes

Report No. 12673-CO Colombia Poverty Assessment Report (In Two Volumes) Volume II: Annexes August 8, 1994 Country Department III Country Operations Division I Latin America and the Caribbean Region Document of the World Bank  Report No. 12673-CO The World Bank Colombia Poverty Assessment Report August 8, 1994 VOLUME TWO - ANNEXES Country Department in Country Operations Divsion 1 Latin America and the Caribbean Region  Table of Contents Annx 1 I. The Household Surveys ................ 1 Coverage ................ 1 Income Adjustments ...........................1....... II. Poverty Indices ....... ................................... 3 Poverty Lines ............... ................... 3 Estimated Poverty Indices ............................... S The Index of Unsatisfied Basic Needs (NBI) .................... 8 m. Poverty Analysis ....................................... 11 Decomposition of chainges in poverty ......................... 11 A. Sectoral decomposition ............................ 11 B. Growth and inequality components ..................... 14 IV. Probability Analysis ............................... ....... 16 V. Main Social Indicators . ..................................... 20 VI. International Compaions . .................................. 26 A.cx 2 1. A Computable General Equilibrium Model of Colombia .................. 33 Strucue of the Model .............. .................. 33 A. Short-Rim Model ............ ................... 33 B. Long-Run Model ............ ................... 41 Sources of Data and Parametffs ............................ 43 Calibration of the Model: 1990-1992 ......................... 43 II. Income Determination: The Role of Policy Reforms and Economic Shocks, 199092 ................................ 49 Strucural Policy Reforms ............................... 49 A. Trade liberalization .......... .................... 49 B. Tax reforms . ................................. 53 C. Minium Wage Policy and Labor Reform ................ 58 The l992 Crisis in Agriculture ............................ 62  - iv - m. Magnitude of the New Oil Discoveries: An Overall Perspective .... ........ 73 IV. Adjusting to the Oil Boom . .................................. 78 Base Case Scenario: Prudent Fiscal Policies and Social Security ............... .................... 78 Adjusting to the Oil Boom: Proactive Policies to Enhance Rural Incomes ................. 79 Annex 3 ... 83 Annex 4 ... 91 Annex5 ... 100 Annex 6 ... 103 Distribution Formula for National Tax Allowance Transfer ( ado Fiscal) .................................. 103 Distribution Formula for National Investment Transfer (Transferencia de Inesin Social) ........................ 104 Annex7 ......7.......................... 107  v - List of Tables and figures Tables 1.1 Departments Included in the Rural Regions ....................... 2 1.2 Adjustment Coefficients for Income Under-Reporting. 3 1.3 Poverty Lines, Selected Years. 4 1.4 Poverty Indices, 1992. 6 1.5 Poverty Indices - Comparable Samples. 7 1.6 Poverty Under Selected Inflation Rates. 8 1.7 Unsatisfied Basic Needs (NB) .9 1.8 Sectoral Decomposition of Change in Poverty, 1978-92 .12 1.9 Sectoral Decomposition of Poverty Change, 1978-92 .13 1.10 Sectoral Decomposition of Change in Poverty, 1978-92 .14 1.11 Income Inequality - Gini Coefficients ........................... 15 1.12 Growth and Redistribution Components of Changes in Poverty 1978-88. 15 1.13 Growth and Redistribution Components of Changes in Poverty, 1978-88 .15 1.14 Growth and Redistribution Components of Changes in Poverty 1978-88 .16 1.15 Results of Probit Regressions for Urban Areas, 1992 .18 1.16 Results of Probit Regressions for Rural Areas, 1992 .19 1.17 Life Expectancy at Birth, 1950-90 .20 1.18 Infant Mortality Rates, 1955-90 .20 1.19 Evolution of Fertility Rates by Zone, 1960-90 21 1.20 Life Expectancy for Men and Women by Region and Department, 1985-90 .22 1.21 Infant Mortality Rates by Region and Department, 1985-90 .23 1.22 Fertility Rates by Region, 1987-90 .24 1.23 Fertility Rates by Mother's Level of Education, 1987-90 .24 1.24 Net Primary Education Coverage Rates, 1990. 25 1.25 International Indicators .................................... 27 1.26 Statistical Results ....................................... 31 1.27 Divergence from Predicted Level ............................. 32 2.1 Classification of the Sectors in the Model ........... ............. 35 2.2 Structure of Demand and Factor Use ............................ 36 2.3 Basic Specification of the Model .............................. 38 2.4 Income Distribution ................................. 39 2.5 Differences between the Short-run and Medium-run Models ................................. 42 2.6 Major Macroeconomic Variables, 1990 ....... ................... 44 2.7 Exogenus Variables and Parameters for 199&-92 Simulation ............................ 45  - vi - 2.8 Trade Liberalization's Estimated Impact on the Economy ............... 51 2.9 Trade Liberalization's Estimated Effects ......................... 53 2.10 Tax Reform, 1992 ...................................... 54 2.11 Tax Reform's Estimated Impact on the Economy .................... 56 2.12 Tax Reform's Estimated Income Effects ......................... 57 2.13 Minimum Wages - Estimated Impact on the Economy ................. 60 2.14 Minimum Wages - Estimated Income Effects ............... ....... 61 2.15 Agriculture Production and Valuation, 1990-92 ..................... 63 2.16 International Agriculture Commodity Prices ....................... 65 2.17 Agriculture Commodity Prices' Estimated Impact on the Economy . ........ 68 2.18 Agriculture International Commodity Prices - Estimated Income Effects ... ... 69 2.19 Agriculture Commodity Prices' Estimated Impact on the Economy ......... 71 2.20 Stock Management and Domestic Agriculture Pricing's Estimated Income Effects .................................. 72 2.21 Oil Production and Exports, 1993-2000 ......................... 74 2.22 Total Cusiana Investment . 74 2.23 Net Value and Distribution of Cusiana's Income Flows, 1993-2005 ...... ...... 75 2.24 Cusiana's Balance of Payments ............................... 77 2.25 Structural Effects of the Adjustment to the Oil Boom - Prudent Fiscal Policy and Social Security Reform ................... 78 2.26 Income Effects of the Adjustment to the Oil Boom - Prudent Fiscal Policies and Social Security Reform .................. 79 2.27 Proactive Policies to Enhance Rural Incomes ...................... 80 3.1 Social ExpenditUres and Their Role in Total Public Expenditure .......... . 82 3.2 Social Expenditures and Their Role in Total Public Expenditure .83 3.3 Social Expendites and Their Role in Total Public Expenditure .84 3.4 Social Expenditures and Their Role in Total Public Expenditure. 85 3.5 Expenditures in Education .86 3.6 Expenditures by Educational Levels .87 3.7 Public Expenditure in Health .88 3.8 Per Capita Expenditure in Health by Department. 89 3.9 National Public Health Service - Household Subsidies, 1992 90 3.10 Public Health Service - Household Subsidies by Region, 1992 ............ 90 4.1 Household Distribution by Socioeconomic Stratum, Urban Region . ........ 91 4.2 liousehold Distribution by Income Decile, Urban Region .91 4.3 Stratification and Income Distribution - "Ideal" Stratification and Type I and II Errors .92 4.4 Comparison Between National Plan and Survey Figures .93 4.5 Population by Income and Age, 1992 .94 4.6 Housing Financing in Colombia ................... ........... 95 4.7 Composition of Eligible Low-Income Housing Programs .96 4.8 Beneficiaries of the New Housing Program in Colombia .97 4.9 National Per Capita Household Income Distribution - Housing Program Target Population. 9 4.10 National Per Capita Household Income Distribution - Housing Program Approved Subsidies .......... 98  - vii - 4.11 National Per Capita Household Income Distribution - Housing Program Disbursed Subsidies .......................... 99 S.1 DRY: Distribution of Subsidies,191 ...........................1 . 100 5.2 PNR: Distributuon of Subsidies, 1991 ........................... 101 5.3 INCORA: Distribution of Personal Income and Subsidies, 1991 .... ....... 102 6.1 Redistribution of Functions under Decetralization - Social Sectors .1........ . 5 6.2 Best Prctice Mecbanisms for Targeting the Poor .106 7.1 Selected Programs of the Office of the President .................... 107 7.2 Budgets of Presidential Programs, 1990-93 ........... 108 1.1 idvolution of NBI Index ........... 10 2.1 Production Structure ..................................... 34 2.2 Urban Labor Income, 1990 ............. 59 2.3 Price and Stock Vaiiations for Selected Agriculture Products, 1986-92 .66 2.4 Cusiana - Distribution of Public Sector Net Income. 75  Annex 1 L. The Household Surveys 1.1 The principal source of information for the poverty profile was the Encuestas Nacionales de Hoga-res (national household surveys) conducted by DANE. These surveys are based on probabilistic, stratified, muldstage sampling. Poverty trends were estimated using urban and mral surveys from 1978, 1988, 1991, and 1992 (ENH19, ENH61, ENH74, and ENhM77, respectively). Labor market analysis was based on urban surveys from 1984 E-d 1990 (ENH44 and ENH68, respectively). This section describes the methodology used to adjust the fonrx surveys used to estimate poverty trends, taking into account differences in coverage and income under-reportng. All tables for which no source is indicated correspond to estimates based on household surveys. Coverage 1.2 The four surveys used to analyze poverty trends covered both urban and rural areas. None covered the Temitorios Nacionales (national territories). The concept of rurality changed between the earlier (ENH19) and later surveys (ENH61, ENH74, and ENH77). To maintain comparability, the municipalities that were classified as urban Cabeceras in 1978 were reclassified using the 1988 DANE methodology to determine ru-ality. As a result, 22 muncipalities that were classified as urban in 1978 are now classified as rural. 1.3 Coverage of urban areas for ENH19, ENH61, and ENH74 was limited to seven large and medium-size cities to maintain comparability between surveys: Baanquilla, Bogota, Bucarmana, Cali, Manizales, Medellfn, and Pasto. An adjustment was required in the sample expansion factors for the estimation of national poverty indices. Coverage of urban areas for ENH77 was broader, including iarge, medium, and small cities. For the section on "Poverty in Colombia: 1992," as well as for the probability analysis, the entire sample was used. For comparisons with previous yeais, only the seven cities mentioned above were included, with corresponding adjustment in expansion coefficients. 1.4 Regional classification is based on DANE criteria. Table 1.1 shows the departments included in each of the four ural regions. Income Adjustments 1.5 Reported income was adjusted in three ways. First, income was imputed to individuals that reported income but had missing information. Imputation w-s based on an earnings equation that included gender, age, age squared, and education as explanatory variables. These regressions were estimated for each metropolitan area, rural region, and employment category. The 1978 survey did not include income information for the inactive population or the unemployed. Income was also imputed to individuals that received  Table 1.1: Departmeuts Included in the Rural Regions ReieA D _pelwal Atlanic Atlntico Bolfvar Cesar C6rdoba Guajira Magdaiena Sucre Oriental Boyaca Cudinamarca Meta N. de Santander Santander Cental Antioquia Caldas Caquet4 Huila Quindo Risuralda l ~~~~ToUims Pacific Cauca Choc6 Narifio Valle del Cauca Source: DANE. pensions or rents. For pensioners, imputed income was calculated as a proportion (obtained from the 1988 housenold survey) of a wage earner's average income for the area. For renters, imputed income was calculated as a proportion (also from the 1988 survey) of non- wage eaner's avirage income for the area. 1.6 The second adjustnt imputed income to rural workers reportng income in kind. Rural wages f workers receiving payment in kind (food) was compared with average ural wages (money) for 1988 >*2d 1991 to determine the value of in kind payment. Information foi 992 was not a... able, so 1991 values were used in this adjustment. 1.7 The tfird adjustment attempted to correct under-reporting of income by makdng it compadble with national accounts information. Three income concepts were considered: wage earnings, non-wage earings (for the self-employed and employers), and other incomes. Reported income for these categories was adjusted upward to make it compatible with information on household income and expenditure accounts from the Colombian national accouns system. The three sources of income considered were  - 3 - compensation of employees, gross operating surplus, and other income (including rents, pensions, and transfers).1 1.8 Mhis procedure was applied to the information from 1978, 1988, and 1991. Table 1.2 shows the adjustment factors for each of those years. The procedure was not applied for 1992 because national income information was unavailable. Adjustment coefficients for 1988 were used b=cause the 1991 rural survey was conducted in December, when incomes tend to be high. 1991 adjustment coefficients could thus be low if applied to 1992. The annual rate of growth in national income (in current prices) between 1991 and 1992 under this methodology is 27.5 percent. For GDP, growth was 26 percent. Table 1.2: Adjustment Coefficients for Income Under-Reporting r -...- = ='* . Earnng Catev :j t978 1988 1991 Wage Earnings .8764 1.282231 1.206740 Other Earnings 2.666985 2.148917 1.941318 Other Income 3.468865 1.936835 2.671760 Total 2.162333 | 1.616042 1.592836 II. Povefy Indies Poveny Lines 1.9 The report uses the "indigence line" esfimated by DANE as its poverty base line. DANE originally calculated the indigence line for 13 cities for August 1984.2 These lines represent the cost of acquiring, at local prices, a nutritionally adequate food basket. For rural areas, the report uses the rural indigence line estimated from a 1981 Income and Expenditure survey used in the most recent study of Colombian poverty.3 1.10 The original poverty lines were adjusted for price changes using consumer price indices published by DANE. For the seven cities included in all three surveys (Barranquilla, Bogota, Bucaramanga, Cali, Manizales, Medellfn, and Pasto), the food component of the consumer price index (CPI) for low-income groups in each city was used. Since rural areas lack a CPI, the food component of the national CPI for low-income groups was used. This could bias estimated poverty indices because rural inflation is not adequately I The approach folowed ases tat under-repordng, by souce, is qually distut among income groups. To the exte ta it varies along di me dsributo the fresults could be biased. 2 DANE (1988). 3 PNUD (1991).  - 4 - captured by this index. The robustness of the indices estimated using the national CPI is discussed below, Lased on alternative issumptions regarding differenitial rates of inflation between urban and rural areas. 1.11 Poverty lines for 1992 were calculated for six additional cities (Cartagena, Cdcuta, Pereira, Monteria, Neiva, and Villavicencio) for which DANE had estinated indigence lines, as well as for 21 other cities for which no indigence line is available. For the first group of cities, the poverty lines calculated for August 1984 were adjusted to September 1992 prices using the food component of the national CP1 for low-income groups. A common poverty line is used for the rest of the urban areas, calculated as a simple average of the lines for the cities of Cdcuta, Manizales, Monteria, Neiva, Pasto, Pereira, and Villavicencio. Poverty base lines are shown in Table 1.3. Table 1.3: Poverty Lines, Selected Years (pesos per capita per month) Region ENH19 ENH61 ENA74 ENH77 l_______________ (1978) (1988) (1991) (1992) Urban Areas Barranquila 914.61 7,647.32 16,047.47 21,031.53 Bogota 859.68 8,038.28 17,292.58 22,086.94 Bucaramanga 800.02 7,368.13 15,835.12 20,182.33 Cali 898.08 7,841.47 17,254.15 21,706.35 Cartagena n.a n.a n.a 22,863.95 Cdcuta n.a. n.a. n.a 18,043.37 Manizales 825.66 7,691.63 15,586.20 19,145.08 Medellin 839.12 8,141.05 16,366.49 19,903.80 Monteria n.a n.a. n.a 21,156.03 Neiva n.a n.a n.a 21,686.07 Pasto 659.61 6,407.01 15,080.51 18,631.23 Pereira n.a. n.a n.a 20,872.47 Villavicencio n.a. n.a. n.a 21,212.74 Other n.a n.a n.a 20,106.71 Rura Areas 678.20 6,472.64 13,358.84 17,205.19  Esmated Povefty Indkes 1.12 In addition to the poverty base lines, two lines-1.5 and 2 times the original line-were used in the estimation of poverty indices. Three indices were calculated: the head count index (H), the poverty gap (PG), and the P2 measure. All three belong to the Foster- Greer-Thorbecke class of poverty measures, represented by the following formula: p, = 1 q I ( -Yj)l where z is the poverty line, and yj is the income of individual i. The head count index-corresponding to a=O-indicates the proportion of the population falling below the poverty line. The poverty gap index-corresponding to a=l-indirates the mean depth of poverty as a proportion of the poverty line. Finally, FGT2-coffesponding to a=2-is a distributionally sensitive index that weights the poverty gaps of the poorest individuals proportionately.4 1.13 Table 1.4 presents the three indices for the three poverty lines. In addition to the national results, indices are shown for urban and rural areas, and for large, medium, and small cities. Table 1.5 presents the same indices for 1978, 1988, 1991, and 1992. The indices in Table 1.5 for 1992 were calculated with a reduced urban sample (seven cities) to mantain comparability with the previous years. 4 See Ravaiion (1992).  - 6- Table 1.4: Poverty Indices, 1 (percent) Re,Lr Poveuiy Le H Po nG12 Natidonal Line 1 18.8 7.4 4.4 Line 2 34.6 13.8 7.9 Line 3 48.0 20.8 12.1 Total Urban LAne 1 9.9 3.1 1.6 Line 2 23.2 7.5 3.6 Line 3 36.0 13.1 6.6 Lage Urban Line 1 8.2 2.5 1.3 Lie 2 19.9 6.3 3.0 Line 3 32.1 11.2 5.5 MediumUrban Line 1 8.5 2.6 1.3 Line 2 22.3 6.9 3.2 Line 3 35.9 12.5 6.1 Smal! Urban Lne 1 12.6 4.1 2.0 Line 2 28.0 9.4 4.6 Line 3 41.6 15.8 8.2 Rural Line 1 31.2 13.5 8.5 Line 2 50.5 22.7 14.0 Line 3 64.6 31.5 19.8  -7- Table 1.5: Poverty Indices - Comparable Samples (percent) Head Couw Index Poveny Gap FG2 Sate Area 1978 1988 1991 1992 1978 1988 1991 1992 1978 1988 1991 19 Natdona (7 cidis) Line 1 23.6 18.7 16.9 17.7 9.8 7.3 6.4 7.0 6.0 4.3 3.7 4.3 Line 2 40.0 33.8 31.1 32.7 17.2 13.6 12.3 13.1 10.3 7.8 6.9 7.5 Line 3 52.5 46.9 44.6 45.7 24.5 20.4 18.7 19.6 15.0 11.9 10.7 11.5 Urban (7 cities) Line 1 12.1 8.0 7.8 8.0 4.6 2.5 2.5 2.4 2.8 1.3 1.4 1.3 Line 2 26.9 19.9 19.0 19.8 9.5 6.3 6.0 6.2 5.2 3.0 3.0 2.9 Line 3 38.9 32.4 31.1 32.1 15.4 11.3 10.8 11.1 8.4 5.5 5.4 5.5 Ruml Line 1 38.4 33.1 29.0 31.2 16.6 13.8 11.7 13.5 10.1 8.5 6.9 8.5 Line 2 56.9 52.4 41.2 50.5 27.0 23.5 20.6 22.7 16.9 14.3 12.2 14.0 line 3 69.9 66.3 62.6 64.6 36.2 32.5 29.3 31.5 23.5 20.4 17.9 19.8  - 8 - 1.14 The rual poverty line was adjusted for inflation using an index of urban prices. The incidence of poverty in 1978 and 1992 was recalculated assuming that rural inflation was 10 percent higher or lower than the urban one. The head count index is sensitive to these assumptions (Table 1.6). Table 1.6: Poverty Under Selected Inflation Rates (Head Count Index - percent) Inflaton 1978 1988 1992 Rural = Urban 38.4 33.1 31.2 Rural = 1.1 Urban 34.7 33.1 33.6 Rural = 0.9 Urban 41.9 33.1 28.8 1.15 When rural inflation is assumed to be 10 percent higher than urban inflation, the downward trend in unrl poverty diswppears. When rural inflation is assumed to be 10 percent lower than urban iflation, the downward trend is magnified. Unforunately, it is not possible to determine the rate at which rural-urban food prices have evolved during this period. There is evidence the rate of increase in rural food prices might have been lower than the urban increase. Information on agricultural daily wages used to impute rural income in kind offers a way of quantifying mral food costs. The differential between workers that receive food and those that do not can be used as a proxy for the cost of food and serve as a price index. During 1978-88, the urban price index increased by 854 percent, while the alternative index increased by 838 and 795 percent in cold and warm areas, respectively. Wage information is not available for 1992. But if information for the same period in 1991 is used, the urban price index increased by 106 percent while the wage-based index increased by 81 and 87 percent, respectively, in the two areas of the country.5 These results suggest that the bias implicit in the use of urban inflation to adjust rural poverty lines might underestimate the downward trend in poverty. The Index of Unsatisfied Basic Needs (NBI) 1.16 Table 1.7 and Figure 1.1 present information on the incidence of each of the four NBIs among urban and rural populations since 1973, the first year for which information is available.6 There has been substantial progress in both rural and urban areas. As indicators of social gaps, however, NBIs identify two shortcomings: (a) the improvement 5 These vahles were acuaed using the wage informatton corresponding to the second quaner of 1978 (EH19 was conducted in June of mat year), die fouth quarter of 1988 (EH61 was conducted in November of that year), and the fourth quarter of 1991. lbe CPI for November of 1991 was used for this comparison. 6 lbe infonnation for 1973 and 1985 is from nadonal censuses and was obtained from DANE (1988 and 1989). The information for 1988 and 1991 was obained from dte Encuesta Nacional de Hogpres (ENH61 and ENH74 respectIvely).  of the indicators in urban areas suggests that further progress is unlikely;7 and (b) the NBI index has a strong bias toward housing and housing-related public services that may be inadequate to identify basic needs in mrual areas.8 On both accounts, revision of the NBI indicators is required if they are to identify social gaps. Table 1.7: Unsatisfied Basic Needs (NBI)a (percent) Area 1973 1985 1988 1991 Inadequate Housing Urban 29.4 4.7 2.4 2.2 Rural 33.9 23.4 17.3 14.2 Crowded Households Urban 25.0 15.0 9.5 8.3 Rural 42.6 24.0 25.0 20.0 Inadequate Services Urban 5.5 3.1 1.5 1.0 Rural 52.6 42.2 31.8 24.1 School Absence Urban 18.2 4.3 3.0 2.7 Rural 41.9 18.6 13.8 12.8 a. The definition of urban areas for 1973 and 1985 was made compatible with that for 1988 and 1991, for which only information for seven cities was available. Those cities are Bogota, Cali, Medellfn, Bucaramanga, BarranquiUa, Manizales y Pasto. Thus, to the extent that other urban areas differ from the former a bias should be expected. Source: Estmates based on household surveys (ENH61 and ENH74) and census data (1973 and 1985). 7 in the urban areas, except for the indicamr of household occpaton density, the inide of the N81s seem to be approaching extremely low levels (bween 1 and 2 pe) suggesig that most basic needs have aldy been isfied. In fac such a leveling-off can already be seen between 1988 and 1991. Tbs finding, of coomse, calls into queston dte cna usefulness of the diet indices as measufes of existing socal gaps. 8 Mais shotoming of the NBM had already been identfied in World Bank (1990). In sunmmay, for the rual areas where infasuctua inveens tend to leg behind, the lack of access to electricy. for example, does not necessrly uggest poverty.  S -~~~~~~~~ -.p U.,~~~~~~~~~~~~~~~~~o o X Xc o *g a a c a No No * I~~ I:1fT .0I O O O O X, 0 0 0 OI  - 11 - IIL. Povedy Analysis 1.17 This section presents the results of two sets of changes in poverty. First, sectoral contributions to changes in national poverty are analyzed. Two sub-periods were considered for this exercise: 1978-88 and 1988-92. Second, the contribution of growth and inequality to changes in poverty is assessed. This second exercise was done only for the 1978-88 sub-period. The second sub-period is omitted because analysis of the information from the EH77 indicates that the upper end of the 1992 income distibution might be truncated.'