Selected social safety net programs in the Philippines : targeting, cost-effectiveness and options for reform

77 PSP Discussion Paper Series 19701 November 1995 Selected Social Safety Net Programs in the Philippines: Targeting, Cost-Effectiveness and Options for Reform K. Subbarao Akhter Ahmed Tesfaye Teldu November 1995 Poverty and Social Policy Department Human Capital Development and Operations Policy The World Bank  This Booklet of Abstracts contains short summaries of recent PSP Discussion Papers; copies of specific papers may be requested from Patricia G. Sanchez via All-in-One. The views expressed in the papers are those of the authors and do not necessarily represent the official policy of the Bank. Rather, the papers reflect work in progress. They are intended to make lessons emerging from the current work program available to operational staff quickly and easily, as well as to stimulate discussion and comment. They also serve as the building blocks for subsequent policy and best practice papers.  FOREWORD After efforts at stabilization and structural reform lasting a decade, the Philippines is entering a promising phase in its fight against poverty. To ease transitional problems, and to protect the well-being of the poor, governmental efforts have focused on three major "safety net" programs, viz., food subsidies, public employment programs, and credit-based livelihood programs. Although substantial resources have been devoted to these programs, few studies have reviewed their impact on the poor. This study attempts to fill this gap. Drawing upon our operational support to the Region, the study examines the effectiveness of the safety net programs from the perspectives of targeting, cost- effectiveness, and sustainability. It identifies various problems in the management of safety net programs in the country, and argues that public policy needs to focus on effecting design changes rather than on providing additional resources. The study offers various suggestions for improving the design and outreach of social assistance. The Poverty and Social Policy Department hopes that policymakers, researchers, and practitioners engaged in the design of safety net programs will find this study useful and timely. Ishrat Husain Director Poverty and Social Policy Department  I  ACKNOWLEDGMENTS We wish to express our gratitude to the Government of the Philippines for its support and cooperation in providing information and data reviewed in this study. In particular, special thanks go to the National Economic Development Authority for facilitating our work in every possible way. Valuable comments were provided by Harold Alderman, Christopher Chamberlain, Sanjay Dhar, Jeffrey Hammer, Richard Heaver, Erika Jorgensen, William McCleary, Oey Astra Meesook, Nicholas Prescott, Antoine Schwartz, Hisan -Shishido, Jayasankar Shivakumar, Gurushri Swamy, and J.W. van Holst Pellekaan. The authors also benefited from extensive discussions with Professor A. Balisacan of the Department of Economics, University of the Philippines, Professor Intal and other faculty members of the Philippine Institute of Development Studies, and the Presidential Commission to Fight Against Poverty. Logistical support provided by Marilene Montemayor, World Bank Resident Mission in Manila, is gratefillly acknowledged. Successive drafts of the study were processed by Kari Labrie, Precy Lizarondo, and Gay Santos.  I  TABLE OF CONTENTS Executive Summary v Chapter 1: Recent Economic Performance And The Poor 1 Recent Economic Performance 3 Poverty In The Philippines 4 Characteristics Of Vulnerable Groups 6 Private Transfers And The Poor 6 Sources of Credit 8 The Nature Of Poverty And Safety Nets 9 Chapter 2: Public Transfers In The Philippines 11 Cash Transfers 11 In-Kind Transfers: The Food Subsidy Program 12 The Prevailing Food Subsidy Program 14 Options For A Targeted Income Transfer 20 Conclusions And Recommendations 23 Chapter 3: Labor-Based Public Works: Rationale, Scope, And Effectiveness 25 Profile Of Public Works Programs 29 Food-For-Work Program 30 The Second Rural Roads Improvement Project 33 Conclusions 34  Chapter 4: Livelihood Creation (Self-Employment) Programs 37 Objectives And Operational Details 37 Evaluation Criteria 38 Family-Based Enterprises 38 Grameen Bank Replication Program 44 Medium-Size Enterprises 46 Conclusions And Recommendations 47 Chapter 5: Safety Net Programs: A Reform Agenda 49 References 71 Appendices Appendix l:Estimation Of Poverty Gap And Distnbutionally Sensitive Poverty Indicator 51 Appendix 2: Method Of Calculation Of NFA's Rice Subsidy Costs And Benefits For The Consumers 53 Appendix 3: Ranking Of Provinces For Targeting 55 Appendix 4: Design And Performance Featres Of Selected Public Works Programs In Asia 67 - Design And Performance Features Of Selected Public Works Progams In Africa 69  List of Tables Table 1.1 Macroeconomic Indicators, 1970-93 3 Table 1.2 Changes In The Incidence Of Poverty For Selected Countries 4 Table 1.3 Poverty And Food Poverty Incidence By Region 5 Table 1.4 Household Characteristics Of The Poor 6 Table 1.5 Percentage Of Income Received From Abroad And Domestic Receipts By Decile Group, 1991 (Percent) 7 Table 1.6 Distribution Of Sources Of Credit To Low-Income Families, Since 1991 (Percent) 8 Table 2.1 Poverty Profile Of The Philippines, 1991 16 Table 2.2 Cost-Effectiveness Of NFA' s Rice Subsidy For The Consumers (Pesos Per Metric Ton) 17 Table 2.3 National Rice Balances (Thousand Metric Tons) 18 Table 2.4 Regional Distribution Of NFA Rice And Food Poverty 19 Table 2.5: General And Targeted Subsidies. Jamaica 1988 (Percent) 22 Table 2.6 Probable Cost-Effectiveness Of A Food Stamp Program 24 Table 3.1 Incidence Of Poverty, Unemployment, And Underemployment, And Net Migration By Region (Percent) 27 Table 3.2 Food-For-Work Program: Type Of Rural Works (Percent) 31 Table 3.3 Selected Statistics For Second Rural Roads Improvement Projects In Kabankalan, Sultan Kudarat South Cotabato, And Wao-Banisilan 34 Table 4.1 The Philippines: Regional Distribution Of Livelihood Programs And The Poor (Percent) 39 Table 4.2 Livelihood Programs Of The Department Of Labor And Employment And The Department Of Trade And Industry: A Summary 41 Table 4.3 Loan Defaults In Livelihood Programs (Percent) 42 Table 4.4 Benefit-Cost Ratios For Bureau Of Rural Workers Livelihood Creation Programs 43 Table 4.5 Grameen Bank Replication Program: A Summary (Pesos) 44 Table 4.6 ILO-Japan Revolving Fund-Supported Medium-Size Enterprise Development Program: A Summary 47 ...  :~~~~~~~~~~~~~~~  Executive Suimmary i. Until 1991, success in reducing poverty has been muted in the Philippines as the growth performance was disappointing. In 1986-89, structural reforms were introduced to move the economy toward sustained growth and to give the poor better access to resources to improve their standard of living. But it was not until 1991-92 that economic reform gained momentum and the economy strengthened in 1993 and 1994. The Philippines appears to be entering a phase of sustained growth, raising hopes of long-term reduction in poverty. However, as of 1991, the head count poverty ratio and the poverty gap ratio (which measures the depth of poverty) remained high. H. In response to the pervasiveness of poverty the government implemented three maor safety net programs-a generalized food subsidy, wage employment programs, and livelihood creation (self- employment) programs. The government is spending more than 4 billion pesos (P) on these programs (compared with P7.5 billion on the health sector as a whole). Although substantial resources are devoted to safety net programs, few studies have reviewed their impact on the poor. This report attempts to fi]l that gap. Food Subsidy Program iii. The food poverty gap (the amount of income needed to raise the poor to a level where they can sustain an adequate diet) of poor households is large. Estimates suggest that an income transfer of P16.8 billion (US$671 million) or about 1.4 percent of GDP would be required to eliminate this gap. Given that the need is substantial and resources are scarce, the government needs to carefully target income transfers to the poorest individuals. Estimates suggest that raising the poorest 10 percent of the population to the food poverty line would reduce the severity of food poverty by 25 percent. Raising the same percentage of people who are just below the food poverty line up to the line would reduce the severity of poverty by only 5 percent. This study emphasizes that, within given resource constraints, it is more desirable to reduce the food consumption shortfall of a larger number of the poor than to eliminate the shortfall of a smaller number, particularly because such a limited policy intervention is unlikely to crowd out private transfers. iv. In order to maximnize the impact of transfers on the depth or severity of food poverty, resources (subsidies) should be distributed to reach the poorest income groups and regions. However, analysis suggests that the poorest income groups and regions are not receiving their share of benefits from the food subsidy program run by the National Food Authority (NFA). Even in 1992, when the largest income transfer was made, NFA's food subsidy program raised the (annual) income of the poor, by no more than 1 percent and their calorie intake by no more than a 0.3 percent. Moreover, regional distribution is extremely unequal. During 1991-93, the National Capital Region (NCR) and the Cagayan Valley received 35 percent of NFA-subsidized rice, but accounted for only 3 percent of food poverty. In contrast, five regions accounting for 62 percent of food poverty received 29 percent of NFA rice. v  v. Furher, the small amou= of income ransferred to the poor was achieved at a hbih cost. During 1991-93, the government has spent P3 for every P1 of income transferred to a poor family, assuming no leakage. Bu; a subsntial quanmy of NFA nce has been disibuted to regons with a very low depth of poverty. Allowing for leakage, the governent spent P4 to P6 to tansfer P1 of income to a poor fnily. vL. Based on 1992 conditions, eliminaring the consumer sbsidy wouid save the government about P3 biUion without adversely affeting the poor's numtonal status. viL The resources saved by eliminating the subsidy could be used for a beter-targeted in-ldnd tanser program such as a food stamp program, or a nuitmon program. A simple sinulanon exercise suggests that, asming 12 percent leakage, if the resources currently beng used for NF.Xs general food subsidy are shited to a fod samp program, it can effect an income transfer equialent to 16 percent of an uirapoor smily's ncome, compared with less than 1 percet under curren subsidy program vii Since the food povery gap of very poor households is subsanal, the report recommends that the newly esablished program be targeted to those at greatest mnuinonal risk. Several approaches now exst for reducirg lealkge and ensurmg cost-effectiveness, including self-targeting through commodity selection, geographical targeting, and carefully tareed food stmps. Food stmps could be used for wage payments in the self-targeted public works programs or could be dismibued at schools or health clics. The report thus recommends eliminating the consumer food subsidy and adopting beter desgned and taeted altemanves. Targeting could be effected m two smges. First, the saved resources may be allocated to reDons (provinces) m propom o thei conibudon to the couny's food povery gap. The within the regions (provmces) the poorest 10 percen be targeted for food transfer. The report also recommends that the poor households be identified by the Deparment of Social Welfare and Development (DSWD), which has developed a capacity to ide and taget assistance to the needy. Pablic Works prognms mx. The Phipnes has more than two decades of experience with labor-based public works programs. Yet no synthesis of this experience has been completed and the database is thin. x. The principal problem with public works programs in general and with food-for-works programs in palar, is that the cont of s pooriy understod (and applied). For example, food-for-works programs were located in poor provices, but the poor did not benafit because the effective wage rate (food plus money wages) was set too high; as a result, many nonpoor took advanmg of the progrm. xi. Public works programs can provide insurance to the poor as an employer of last resort, address the problem of seasonal umployment, create asset for the poor (such as schools), or provide disaster relief. However, none of these objecaves were facred into the design of works programs in the Philippines. vi  xii. The most effective form of management for enhancing worker productivity must be determined. Increasingly, private contractors are being hired to execute labor-based public works. But private contractors may not necessarily have an incentive to adopt labor-based methods, even when they are technically and economically viable. It is important to identify which programs (drainage, flood control, reforestation, and so on) can be effectively carred out using labor-based methods, and then monitor the program even if it is executed by a private contractor. ,ii. The govermnent proposes to consolidate all the ongoing works programs under the umbrella program, Kabuhayan 2000. In order to maximize the benefits for the poorest groups, the report recommends wages be carefully set so that the program wage rate is lower than the prevailing market wage rate. The report also recommends that the line departments identify infrastructure projects for which the use of labor-based methods is both economically and technically feasible. These projects should be located in provinces that have poor nfrastucure. To the extent the ultrapoor (those falling below the food poverty threshold) are also located in these regions, public works are most likely to be targeted toward them. Revamping works program would also need to be combined with efforts to maintain the assets created and to monitor and evaluate the program by collecting baseline data on wages, assets, and incremental employment and incomes generated. Livelihood Creation programs xiv Expanding self-employment and opening channels of formal credit for the rural poor are the main objectives of livelihood creation programs. Livelihood programs in the Philippines differ from similar programs elsewhere in three respects: several government line deparmnents implement the programs in addition to performing their usual functions, most line departments rely on NGOs to perform banking functions, and the NGOs are subsidized by the government (in most counties the subsidies are transferred directly to the beneficiaries). If beneficiaries do not repay their loans the goverrment bears the cost of the "transfer". The estmated cost of subsidizing livelihood programs is at least P1 billion. Fifty four livefihood programs presently operate, executed by more than 17,000 NGOs. xv. As in most other countries the programs benefited the nearpoor and the nonpoor more than the ultrapoor. Programs were implemented in regions that were better endowed with infrastructure and whose residents had higher average incomes. Leakage was substantial. With the exception of the program of the Department of Social Welfare and Development (DSWD), and the Grameen Bank Replication Program, the livelihood programs overall have not been able to reach the very poor. xvi. The benefit-cost ratios of major livelihood programs were unfavorable, and the incremental employment and income effects were negligible; even modest gains disappeared in less than two years. The lack of income and employment sustainability undermined the rationale for continuing the programs. Repayment rates were also unsatisfactory, due to an unwillingness, not an incapacity, to repay. vii  xvii. The DSWD-sponsored program, the Small Enterprise Assistane Program, is an exception. The program successfully reached the ultrapoor, partly because it was implemented by community workers. Although the program was originally uitended to ransfer resources to the poor to enable them to become self-employed, the community workers persuaded the ultrapoor to repay at least a part of the transfer, which was then recycled to other beneficiaries. Almost 60 percent of the beneficiaries repaid their loans. Despite this success, the program essentially provided a financial relief rather than long-term assistnce. Only very small loans were made and most beneficianes used them for pressmg consumption needs. A few engaged in petty trading for a brief period of time, but establishing sutainable livelihoods remained a distant goal. xviii. Another exception is the Grameen Bank Replication Program, which like the Small Enteprise Assistance Program, reached the ultrapoor. Beneficiaries invested in projects that raised their incomes. But the cost of operation was very high because the government subsidized NGO operations. The unit costs of operation in the Philippines were four to five times the costs of Grameen Bank in Bangladesh. xix. A few medium-size enterprises were set up with assistnce from the International Labor Organization (ILO) and Japan, intending to further hired labor absorption. The programs successfully raised incomes, but most beneficiaries belonged to nonpoor families. Their level of education was high and little outside labor was hired (instead, family labor was used). xx. For reducing the depth of poverty in the Philippines, better-targeted food and nutrition programs and a revamped public works program have much greater potential than livelihood creation programs. Thus, livelihood programs should not be subsidized and the involvement of line departments needs to be reviewed. xxi. The two exceptions-the Small Enterprise Assistance Program. and the Grameen Bank Replication Program-ed to be modified so that they enable beneficianes to create susaiable livelihoods. Presently, both programs are fragmented. The Small Enterprise Assistance Program needs to increase the loan amount given, serve more families, and regularly supervise and monitor their activities. The GfRB needs to reduce its operational costs, and consequently the subsidy cost. If the program is rendered financially viable and a majority of the ulrapoor recipients are able to sustain livelihoods, the subsidy cost can be justified. xxii. The programs evaluated in this report constite important components of the proposed Social Reform Agenda. With devolution, the programs are likely to be implemented by the Local Govmment Units (LGUs). Considering the scarcity of resources at the LGU-level especially in poor provinces, it is important to reach the poor cost effectively. Revamping the design of the programs as suggested above can enhance the benefits to the poor from the Social Reform Agenda, and ensure that the programs are rendered cost-effective. viii  Social Safety Net Programs in the Philippines: Targeting, Cost-Effectiveness, and Options For Reform Chapter 1 Recent Economic Performance and the Poor Several recent attempts have been made to estimate the incidence and depth of poverty in the Philippines (Balisacan 1992, 1993b, 1994b, and forthcoming; NSCB 1993; World Bank 1993), to identify the household characteristics and geographical location of the poor (Intal 1994; Virtucio 1994); and to assess the poor's access to health and education services (Herrin and Racelis 1994). These studies reveal the importance of three factors for reducing poverty: labor-intensive runal growth. access to social services provisions (health, family planning, nutition, and education) and cost- effective safety net programs. The trends in poverty and two of the poverty-reducing strategies-abor-demanding economic growth and social services provision-have been explored in detail by the World Bank (Country Economic Memorandum 1993) and many Philpine analysts.' However, few recent studies have attempted to assess the complementary strategy: poverty-targeted safety net interventions. This report aims to fill that gap. It builds on a World Bank (1994b) examination of the reform issues confronting the social security institutions in the Philippines2, an evaluation of the access of the poor to health and education services (Herri and Racelis 1994), and a World Bank (1994c) assessment of a variety of child nutriion programs that have been in operation in the country for many years. The present study focuses on cash transfers made by the Department of Social Welfare and Development, the in-kind food subsidy program implemented by the National Food Authority, and wage employment programs and livelihood programs implemented by various line deparanents. The report is based on the findings. of two Bank missions that visited the Philippines in October 1993 and October 1994. Primarily in response to the pervasiveness of rumral poverty, the Philippines Govermnent implemented a variety of safety net programs to relieve the poor, from cash transfers and in-knd income transfers (food subsidies, nutrition interventions) to wage employment programs and livelihood creation (self-employment) programs. In addition to the government, the private sector, bilateral and multilateral agencies, and nongovernmental organizations (NGOs) have been active in financing, plannig, and executg livelihood creation programs. It is estumated that more than 1 to 2 billion pesos (P) are being spent on these programs. More than P2 to P3 billion are being spent on food For a recent comprehensive assessment of the macroeconomic environment and poverty in the Philippines, see de Dios and others (1993), Balisacam (1994a), and World Bank (1993, 1994c). 2 In the Philippines the fonial social secuity program (pensions) is limited to fonnal sector employees who, in general, are not poor. 1  subsidies and other cash transfers. Overall, more than P4 billion are being spent annually on safety net programs, compared with P7.5 billion on health and P45.8 billion on education. The study assesses the possibilityr to improve thie efficiency and effectiveness of the targeting of public resources devoted to selected safety net programs for the poor. (For discussion of the term "safety net' and the main policy issues involved, see Box 1.1). Woi I Sa==i: rr f rgas n a Srwi erwd. opsigal.no3 ui-ae ragmns(iaesft es~ . & i ik hwpus,i oilsci rgm pflelryrlvn o omlsco mlye -igrl.r o h o.sttop.it~ ~lpn _omues _ancp_tyaree _imreiosTsa.bc den'xo.ayhn eze l u .a.-kdtaseLT.bikin.o shn.4dtzs.rs.e of j~~~tb1i~~~ ~~paiing.de ~~~~ike pn~~~~grarns ~~~~isually varies ~~~~~~mss countries. Mlocarion depends on ike strength of~~~~~~~~~~~M..... p::zbW~rcseMpvn. ~g~~1e~an.fporm n t_tfugdg rar~st.aalbe~ca~ 'stx. rcs sbscsea vc-est. uxto, n.do-os wri ,._~~~~~~emtsI~~s t'ell ae~~trni rddtepo The anrbe rasaepieaytagetd h cl ~~Xt~IIpo Vndn~pnd ~~~~~ . _.. .~ n gris atfTfnslomy&*eyit i  Recent Economic Performance Structural reforms began in the Philippines during 1986-89, with the elimination of most price distortions. Economic reform gained momentum in 1991 and 1992, and economic recovery began in 1993 and continued throughout 1994. Real output grew at the modest rate of 2.4 percent during 1993 (Table 1.1), but then reached 4.5 percent in the first half of 1994, according to a recent report released by the National Economic Development Authority (NEDA). Real GDP was expected to grow by at least 4 percent for all of 1994. After gaining strength in 1993 and 1994, the economy seems to be entering a period of sustained growth." Table 1.1 Macroeconomic Indicators, 1970-93 1970-80 1981-85 1986-89 1990-92 1993 Percent Real GNP growth 5.8 -1.7 5.6 2.2 2.4 Real GDP growth 5.7 -1.1 5.2 0.9 2.0 Inflation (% change in CPI) 14.5a 18.2 5.9 13.9 7.6 Export growthb 11.3 -2.0 12.2 3.6 9.0 As percentage of GNP Current account deficit -2.7 -5.4 -0.7 -3.4 -5.9 National government deficit - -2.9 -3.1 -2.2 -1.4 Gross domestic investment 27.3 24.8 18.7 21.4 23.5 National savings 25.9 20.6 18.0 18.0 17.7 - Not available a. 1973-80. b. Exports of goods and nonfactor services in constant pesos. Source: World Bank (1994a). Sectoral trends in output growth are also worti noting. Agriculture accounts for 20 percent of GDP. Many subsistence farmers and landless workers derive their incomes from rice and corn production. NEDA reported that output growth in palay rice, the country's largest crop, was 8.5 percent in 1994 compared with 4.4 percent in 1993. The absence of a major typhoon has favored strong agricultural growth. Manufacturing output grew by 22.3 percent in value-added terms from January-September 1994. Food processing, a labor-intensive industry, registered a substantial increase of 31 percent. 3 For an ovaview of recent economic performance, see World Bank (1994): Philippines: Recent Macroeconomic Developments and Reform Efforts. 3  The improved economic situation is contributing to larger inflows of foreign direct investment, which increased from US$176 million in 1993 to US$658 million in the first six months of 1994; the power sector attracted the highest levels of investment. However, these positive developments should not distract attention from the country's major concern, poverty. Poverty in the Philippines Compared with other countries in the region, the Philippines' long-term poverty record is disappointing (Table 1.2). A recent report on poverty, growth, and fiscal crisis in the country observes: "The record is alarmingly inferior even to that of India, where, despite a much larger population, the absolute increase in the number of poor has been less over a comparable period, and the reduction in poverty incidence much faster. The proportion of the population in the Philippines unable to meet the most basic needs of human existence is now one of the highest in the Asia-Pacific region" (de Dios and others 1993, p. 4). lTo learn how the poor have fared in the wake of recent economic changes, a comparison of headcount poverty ratios in 1985,. 1988, and 1991 is helpful4. The headcount measure of poverty represents the proportion of individuals who are below a threshold poverty line. There has been a substantial increase in the headcount poverty ratio from 1988 to 1991 in eight of the country's fifteen regions (Table 1.3). The national average was 39.2 percent in 1991. In eleven of the regions the poverty ratio was higher than the national average. The National Capital Region, however, registered a dramatic decrease in the incidence of poverty. Table 1.2 Changes in the incidence of poverty for selected countries GNP per Capita Annual Poverty Poverty Incidence (most Countrv (1992 US$) Reduction ()recent estimate) India, 1972-87 310 1.0 38.0 (1987) Indonesia, 1970-87 670 2.3 17.0 (1987) Thailand, 1962-86 1840 1.4 26.0 (1986) Philippines, 1971-91a 770 0.3 56.0 (1991) 1985-95' 0.7 39.2 (1991) a The poverty ratios, estimated by Balisacan (1992), represent individuals below the poverty line. bThese are estimated by the National Statistical Coordination Board (NSCB). The NSCB estimates are based on a lower poverty threshold that excludes nonessential nonfood items. Moreover, the proportion represents househokds, not individuals. Since the family size of the poor households is larger in the Philippines, the ratio estimnated for individuals tends to be higher than the ratio estimated for households. For details, see NSCB (1993). Sources: World Bank (1990, 1994); Kakwani and Subbarao (1993); Balisacan (1992, 1993a); de Dios and others (1993a); NSO (1993). Fra review ofda sources and trends in poverty duning 1971-88, see World Bank (1993). Since then the results of the Famih, income and Expenditure Survey 1991 have become available. 4  Table 1.3 Poverty and Food Poverty Incidence by Region Food Food Poverty Ratio Poverty Poverty Region (households) RatiOa Ratio (households) (individuals) 1985 1988 1991 1991 Philippines 44.2 40.2 39.2 19.7 25.8 NationalCapitalRegion(NCR) 23.0 21.6 13.2 2.1 3.7 AONCR 47.5 43.1 43.3 22.8 - Cordillera Autonomous Region (CAR) - 41.9 36.3 18.8 26.3 Ilocos Region 37.5 44.9 48.4 24.6 32.2 Cagayan Valley 37.8 40.4 43.3 20.1 26,1 Central Luzon 27.7 29.3 31.1 11.1 16.0 SouthemnTagalog 40.3 41.1 37.9 17.1 23.6 Bicol Region 60.5 54.5 55.0 31.6 39.1 Western Visayas 59.9 49.4 45.3 21.8 30.9 Central Visayas 57.4 46.8 41.7 23.3 29.2 Eastern Visayas 59.0 48.9 40.1 26.1 33.3 Western Mindanao 54.3 38.7 44.0 23.3 29.9 Northern Mindanao 53.1 46.1 53.0 33.6 41.0 Southem Mindanao 43.9 43.1 46-2 26.2 32.9 Central Mindanao 51.7 36.1 50.2 26.1 32.0 - Not available. 2 Proportion of population whose income does not enable them to sustain an adequate diet. Note: For methodology, see Chapter 3 and Annex 1. Source: NCSB (1994); authors calculations. In 1991, the food poverty ratio (the proportion of the population whose income does not enable them to sustain an adequate diet) was 19.7 percent for Philippine households and 25.8 percent for individuals. The size of households below the food poverty threshold (the ultrapoor) is larger than the size of households below the poverty threshold and that of nonpoor households. The food poverty ratios are higher for the Philippines than for countries with comparable per capita incomes. However, the ratios are smaller in the high-growth regions (the NCR, the Central Luzon) than elsewhere, reflecting the favorable impact of growth on food poverty. The recent favorable growth performance may have lowered the food poverty ratio in other regions as well. A World Bank Poverty Assessment (forthcoming) will analyze these and other aspects of poverty. 5  Characteristics of Vulnerable Groups About 70 percent of the poor live in rural areas. The majority of the heads of poor households are between the ages of 30 and 49. Family size also influences poverty-while .only 24 percent of three-member households are poor, 56 percent of six (or more) member households are poor (Table 1.4). Subsistence farmers (growing mostly rice and corn), farm and forestry workers, and fishermen account for three-fifths of the poor and seven-tenths of all households under food poverty threshold (Intal 1994). Members of poor households tend to have multiple jobs, presumably as a survival strategy. Table 1.4 Household Characteristics of the Poor Percentage to total number of poor families Heads of households with less than elementarv education 71.5 Family size of three or less 24.0 Family size of six or more 56.3 Heads of poor households between the ages of 30 and 49 56.9 Rice and com farmers, farm workers, forestry workers. and fishermen 61.5 Source: Intal (1994). Private transfers and the poor Remittances from abroad and domestic private transfers play an important role in the Philippines. Earlier studies have shown that for 1988, posttansfer income shares of the poorest households were much higher than their preransfer shares (Cox and Jimenez 1993; World Bank 1993). These analyses did not examine regional variations in the impact of transfers, however. Also, remittances from abroad and from within the country were not disaggregated. After doing so, we found two important trends (Table 1.5). First, remittances from abroad are a significant source of household income in the more-developed regions of the country (NCR, Central Luzon, Southern Tagalog, Western Visayas). The Ilocos region (a poor region) is an exception because it has a long tradition of emigration, primarily to the United States. Second, upper-income groups receive most of the remittances from abroad. Thus, remittances from abroad accentuate income inequality among regions and households. 6  Table 1.5 Percentage of Income Received from Abroad and Domestic Receipts by Decile Group, 1991 (percent) Decile 1st 2nd 3rd 4th 5th 6th 7th 8th 9th 10th Abroad Philippines 1.1 1.1 1.7 1.8 2.6 3.9 5.9 8.0 10.8 12.2 National Capital Region 3.0 4.9 6.6 8.7 7.8 8.5 13.3 13.3 14.8 8.6 CordiDera Autonomous Region 2.4 1.4 1.4 6.7 2.2 2.0 4.3 6.0 3.4 11.2 Dlocos Region 7.0 6.3 5.9 5.7 5.5 10.5 1.5 12.8 18.9 21.4 Cagayan Valley 1.4 - 1.2 2.1 1.9 3.1 2.1 4.7 10.4 9.6 Central Luzon 2.4 2.1 3.6 6.1 7.2 7.5 8.7 14.8 17.6 15.9 Southem Tagalog 1.5 1.6 2.4 2.0 4.9 6.6 9.1 9.7 11.6 12.2 Bicol Region 0.5 0.9 0.3 1.1 1.5 0.5 1.8 3.1 4.3 10.9 Western Visayas 0.4 0.6 1.9 0.8 1.2 1.0 1.2 4.2 7.4 15.2 Central Visayas 0.8 1.4 0.3 1.9 0.7 2.4 2.7 2.0 5.0 14.1 Eastern Visayas 0.1 0.9 0.9 1.4 3.3 1.6 0.8 2.4 6.4 8.8 Westemr Mindanao 0.3 - 0.1 1.7 0.1 2.3 3.5 2.4 5.2 6.0 Northem Mindanao 0.3 0.3 0.3 1.7 1.1 1.8 1.2 2.5 1.3 3.9 Southem Mindanao 0.3 0.3 1.5 0.6 0.2 1.5 0.8 1.3 3.0 9.4 Central Mindanao 0.7 - 0.3 0.3 0.5 2.6 1.6 1.6 1.9 Autonomous Region of 0 0.7 0.5 1.4 2.2 0.6 2.3 3.6 3.6 8.4 Moslem Mindanao Domestic Philippines 8.0 5.2 4.8 3.6 3.5 2.8 2.4 2.1 1.8 1.0 National Capital Region 5.0 3.2 2.6 *2.2 3.1 1.7 2.6 1.7 1.9 0.8 Cordillera Autonomous Region 5.1 1.8 2.3 2.3 2.0 1.5 1.2 0.7 1.3 0.4 Ilocos Region 9.9 6.1 7.1 3.3 3.7 2.6 3.0 2.5 1.6 1.3 Cagayan Valley 7.4 4.3 3.9 2.2 2.6 1.9 2.4 2.5 0.3 0.5 Central Luzon 9.5 5.3 5.8 3.4 3.5 2.5 2.5 2.4 1.7 0.9 Southem Tagalog 7.0 5.3 4.4 3.7 2.7 2.0 2.1 1.6 1.7 0.6 Bicol Region 10.7 7.7 6.7 5.7 3.9 4.7 4.4 3.9 3.7 1.9 Westem Visayas 8.5 5.2 6.4 5.7 3.0 4.9 4.2 4.0 3.3 1.4 Central Visayas 16.5 9.5 8.1 6.9 5.8 3.7 4.5 2.6 4.0 0.8 Eastemrn Visayas 10.5 10.3 8.5 5.7 5.9 3.4 4.2 4.2 3.2 1.4 Western Mindanao 3.1 1.6 1.8 1.9 1.0 0.9 0.6 2.2 0.4 0.5 Northem Mindanao 4.7 4.3 3.3 3.0 2.4 2.7 1.2 2.0 1.6 0.6 Southem Mindanao 4.6 4.1 1.3 2.7 3.5 2.1 1.2 0.9 0.8 0.5 Central Mindanao 2.8 4.3 2.8 3.0 2.9 1.6 1.4 2.1 1.0 1.1 Autonomous Region of 0.5 0.9 1.4 0.7 0.5 0.2 0.7 0.5 0.1 - Moslem Mindanao* - Not available * ARC Source: NSO. 7  By contrast, domestic transfers are significant among the poorer population (Intal 1994). The Telative importance of remittances from abroad for upper-income groups is not surprising. Generally, the educated and the skilled migrate abroad, most of whom are not the poor. Moreover, the costs of emigration (travel, visa) may also be too high for the poor to bear. Finally, domestic remittances accrue disproportionately to the lowest deciles in the poorest regions. The poor generally migrate from less-endowed regions to cities and richer regions in search of employment. For the very poor domestic migration a survival strategy. These fmdings have implications for safety net programs. For example, targeting transfer programs to poorer regions is unlikely to crowd out private transfers from abroad, which are the largest source of private transfers in the Philippines. However, it is important to maintain public transfers at a low level, and target them only to the poorest households to avoid crowding out private transfers from domestic sources. Sources of credit The 1992 Socioeconomic Survey reveals that despite the proliferation of credit-based livelihood programs, low-income fizilies' share of fonnal credit, especially from govenment-sponsored credit programs, was negligible: the share of credit obtained from NGOs was 2.9 percent in urban areas and 1.5 percent in rural areas (Table 1.6). Nearly two-thirds of the poor's credit needs were met by private moneylenders, relatives, and employees. Private banks, govemrnment banks, and cooperatives provided about 20 percent. Table 1.6 Distribution of Sources of Credit to Low-Income Families, Since 1991 (percent) Source Urban Rural Total Relatives/friends 36.1 39.0 38.2 Employer/landlord 4.9 6.4 6.0 Private moneylender 26.4 15.4 18.2 Private bank 2.1 2.0 2.0 Government bank 3.2 5.2 4.7 GSIS/SSS 0.0 0.7 0.5 Cooperative 12.3 13.3 13.1 NGO 2.9 1.5 1.9 Other 7.2 12.0 10.8 One source 95.2 95.6 95.5 Two sources 3.6 3.7 3.6 Three sources 1.2 0.8 0.9 All sources 100.0 100.0 100.0 Number who availed of credit 99,789 290,073 389,862 Source: NSO and NEDA, 1992 Socioeconomic Survey of Special Group of Families. 8  The nature ofpovery and safety nets The mostly mral character of poverty, the predominance of marginal farmers and landless laborers engaged in rice and com cultivation among the poor, and relatively high poverty and food poverty ratios in many regions of the country suggest a need to accelerate labor-demanding rural growth The pervasiveness of food poverty also underscores the need to provide cost-effective social protecton for the poorest households, especially those receiving no private transfers. 9  l.0 0  Chapter 2 Public Transfers in the Philippines The government makes trsfers to marginalized groups in two ways.' First, it provides cash transfers to the ultrapoor (families whose per capita expenditures fall short of the food threshold) for emergencies and cash assistance for self-employment. Second, the govermnent runs a generalized food (in-kind) subsidy program. Cash transfers The Department of Social Welfare and Development (DSWD) offers a range of services to the ultrapoor. Budgetary support for these services, however, is very limnited. The department's total budget in 1993 was P445 million (compared with P33.7 billion for the Department of Education and P6.9 billion for the Departnent of Health). Of this amount, about P150 million is used for administrative expenses, leaving a balance of about P300 million to finance direct transfers to the poor. In other words, the government is incurring an administrative cost of one peso for every two pesos of income transferred to the poor. About 5.8 million people receive assistance from the DSWD, obtaining on average, P52 (US$2.0) per year. The budget for 1994 was P756 million. The DSWD has been working closely with nongovermnental organizations in delivering welfare services. The socially disadvantaged and the physically and mentally handicapped are targeted. The Department has a cadre of grassroots community workers who regularly screen potential beneficiaries for assistance. The Department's outreach, however, is limited to a few of the most depressed barangays (wards or villages) in the municipalities. The barangays are selected if their poverty incidence is 51 percent or higher, if slum areas are prevalent, or if a calamity has struck. In 1991, the DSWD classified only 18.5 percent of the barangays as "socially depressed " and targeted them for assistance. The DSWD organizes community volunteer resources (for socially disadvantaged communities); cash assistance for self-employment (for socially disadvantaged families); self- employment assistance (for socially disadvantaged women); care, protection, and rehabilitation of children in difficult circumstances; kindergartens and day care centers, supplemental feeding, and so on; cash assistance and self-employment assistance for rehabilitation of the disabled; and relief and rehabilitation of victims of natural calamities. 11  The DSWD provides capital assistance-a cash transfer-to enable ultrapoor households to operate microenterprises. It also provides soft (interest-free) loans. Unlike other programs, the DSWD's cash assistance program reaches the ultrapoor. According to the Centre for Advanced Philippines Studies study (1993), nearly 75 percent of DSWD beneficiaries are ultrapoor. Clearly, the Department has the institutional capacity to identify and target benefits for the very poor, although its outreach is limited to only 18.5 percent of the barangays in the country. Even within these barangays, the number of beneficiaries is falling. For example, the number receiving cash assistance for self-employment decreased from 360,000 in 1988 to 80,000 in 1991. In-kind transfers: The food subsidy program To improve food security of vulnerable households, those at risk must be identified. Appropriately estimating a food poverty line is an essential first step. Once the size of the poor population is known, it is then useful to determine the amount of resources needed to eliminate the food poverty gap (the amount of income needed to raise all those below the food poverty line up to that level). Finally, a measure of the relative severity of poverty is an efficient mechanism to cost-effectively allocate the scarce resources. Several studies measured poverty in the Philippines [Balisacan (1994a,b); Balisacan and Bacawag (1994); Intal (1994); and World Bank (1993)], but none explicitly estimated the resources needed to close the food poverty gap in the Philippines. Here, we estimate the monetary shortfall on an aggregate national and regional level using the recently published 1991 Family Income and Expenditure Survey data (NSO 1993)5. Then, with a distribution-sensitive poverty measure, indicate how the regional food consumption gaps can be reduced. (The method of estimating the food poverty indicators is given in Appendix 1.) This poverty analysis uses the food threshold to designate the ultrapoor in the Philippines. The proportion of ultrapoor to the total population, or the headcount food poverty ratio, is estimated using the food threshold. The headcount ratio is a simple poverty measure and its interpretation is straight forward. However, the headcount ratio cannot measure the depth of poverty or the equity of income distribution among the poor. 5 This estimat can be done at povincial level, as disaggregated data bcme available. 12  The poverty gap is the estimate of the income that would be required to bring every individual up to the poverty threshold. Unlike the headcount ratio, the poverty gap is sensitive to both the number of poor and the depth of their poverty. For example, a reduction in the income of any poor individual would increase the poverty gap, but the poverty headcount would remain unchanged. The poverty gap relays important information for transfer programs because it indicates the total amount of resources needed to alleviate poverty. However, it is insensitive to the iedistribution of income among the poor. The distribution-sensitive poverty measure formulated by Foster, Greer, and Thorbecke (1984), known as the FGT poverty measure. is also used in this analysis. This poverty measure decreases if income is transferred from a poor individual to a poorer individual. It can be used to estimate the reduction in poverty by targeting groups with different poverty levels. The data for this analysis are provided by the 1991 Family Income and Expenditure Survey, which reports total family expenditures in each expenditure group. We convert total family expenditures into per capita expenditures using family-size data. Although income data are available in the Family Income Expenditure Survey, per capita expenditures are used as a proxy for income for two reasons. First, expenditures are likely to reflect permanent income and are hence a better indicator of consumption behavior. Second, data on expenditures are generally more reliable and stable than income data. The national and regional food poverty thresholds for 1991 are obtained from Intal (1994), who adjusted these thresholds to take into accounft regional differences in food preferences and relative prices based on the latest food consumption surveys of the Food and Nutrition Research Institute. The headcount index indicates that about a quarter of the population in the Philippines were ultrapoor in 1991 (Table 2.1). In the Philippines poorer households tend to be larger. Thus, the incidence of food poverty as a proportion of the total number of families is lower (20 percent) than the incidence of food poverty as a proportion of the total population. The National Capital Region had the lowest food poverty incidence of any region, while the Bicol Region had the highest. Estimates (based on an assumption of perfect targeting) indicate that P16.8 billion, or US$671 million, would have been required to eliminate food poverty in the Philippines in 1991 (Table 2.1). This amount is equivalent to 1.35 percent of GDP. But with imperfect targeting, the goverment would require more than P16.8 billion to close this gap. This amount is substantial. Therefore the government must carefully target income transfers to achieve maximum impact. This goal suggests that the poorest should receive income supplements first, to reduce their large consumption shortfall. 13  The distribution-sensitive PFGT poverty measure can help to target income transfers to where they are most needed. The FGT poverty measure is additively decomposable, making it possible to determine the percentage contribution of any subgroup tO the total depth or severity of poverty. If the poor in that subgroup are made nonpoor, then the severity of poverty would reduce by this percentage. Methodology detailing procedure for estimating the regional contribution to food poverty is given in Appendix 1; results are listed in Table 3.1. The measurement of relative poverty has important policy implications. For instance, the measurement of food poverty in this analysis suggests that raising the lowest 10 percent of the population above the food poverty threshold would reduce the severity of food poverty by 25 percent. In contrast, the severity of food poverty will decline by only 5 percent if the top 10 percent (that is, those poor who are nearest to the food poverty threshold) are assisted. In terms of the headcount index, eliminating the food poverty gap of any decile beneath the food poverty threshold would reduce food poverty incidence from 25 percent to 15 percent. It is, however, more desirable to reduce the food consumption shortfall of a larger proportion of the poor rather than to eliminate the shortfall of a smaller proportion of the poor, given available resources. The govermment does not need to target the poorest groups exclusively. Several studies have pointed to the buoyancy of private transfers among poor households in the Philippines. An intervention that aims to fill only a portion of the food consumption gap has merit in that it is less likely to crowd out private transfers. In sum, a two-step method for reducing food poverty cost-effectively is recommended. First, the total resources earmarked for poverty reduction may be allocated to each region (province) according to its contribution to total poverty. This does not amount to "regional targeting", i.e.. targeting a few regions or provinces exclusively to the neglect of others. In the suggested targeting methodology, every provmce gets a share, but only in proportion to national food poverty gap. Second, at the regional (provincial) level, the food poverty gap should be reduced from the bottom up, restricting the transfer to the lowest 10 percent of the population under the food poverty threshold. The prevailingfood subsidy program The Government of the Philippines created the National Grains Authority in 1972, which- was transformed into the National Food Authority (NFA) in January 1981. In 1987 the NFA was shifted from the Office of the President to the Department of Agriculture. The NFA is responsible for: Stabilizing grain (rice and com) prices. 14  * Transferring income to farmers by procuring rice and corn at higher-than-market prices. * Transferring income to consumers by selling rice at lower-than-market price; * Operating a buffer stock. * Promoting an efficient grains marketing and postharvest system. Of these functions, the first and last are explicitly assigned by the law, while transferring income to producers and consumers is inferred from the law. Maintaining a buffer stock for emergencies has generally been practiced, but is not required by the law. The govermment provides budgetary support and a guaranteed credit line to the NFA. The procurement prices of nalav (paddy) and corn are determined by production costs, the estimated real income of farmers, expected rice prices, the nitrogen-palay price ratio, parity with the world market price, and support prices in other countries. In recent years the NFA's intervention has been focused almost entirely on the rice market. The NFA procures palay from farmers, especially farmer associations, and monopolizes rice imports. The procured rice (domestic and imnported) is sold at a predetermined release price, which is generally lower than the market price. Rice is sold by registered retailers, totaling 63,180 in 1993, throughout the country. The Authority owns 233 warehouses with an effective capacity of about 1.1 million metric tons of grain. As of September 1994, the NFA had 6.029 regular employees- 1,252 centrally based and 4,777 posted in the field. Cost-effectiveness of NFA's consumer subsidies. One of the NFA's major functions is to transfer income to the consumer by subsidizing grains, mainly rice. The costs of providing consumer subsidies include the net cost of rice (that is, the purchase cost of palay plus the milling costs plus the cost of borrowing, minus the sales receipts) and the administrative cost of delivering rice to consumers (for the method of calculation see Appendix 2.) The subsidy benefits include the income actually received by the consumers (that is, the difference between the market price and the NFA release price). Two most recent reference years for which detailed cost data have been made available by the NFA, 1991 and 1992, are used to calculate the cost-effectiveness of this consumer subsidy (Table 2.2). The NFA distributed 58,171 metric tons of rice in 1991 and 520,561 metric tons in 1992. The two years also happen to be the years when rice distribution by the NFA touched its lowest and highest levels in the last ten years or so. The relatively lower cost per ton of rice in 1992 than in 1991 indicates that considerable scale economies in rice distribution exist. The costs of the Authority's consumer subsidy amounted to P0.96 billion in 1991 and P2.96 billion in 1992 (the net cost per ton multiplied by the total rice distribution in one year). It delivered one peso of income to a consumer at a cost of P2.99 in 1991 and P2.15 in 1992. 15  Table 2.1 Poverty profile of the Philippines, 1991 Area Population Food Poverty Hleadcount Poverty Gap FGT Contribution to Share Line Index Food Total Poverty Poverty (distribution- Measure sensitive measure)a (percent) (peso/person) (percent) (mill. peso) (P2) (percent) The Philippines 100.0 4,865 25.8 16,787 0.0167 100.0 National Capital Region 14.0 5,757 3.7 83 0.0004 0.3 Cordillera Autonomous Region 2.0 4,324 26.3 274 0.0137 1.6 1. Ilocos Region 5.6 5,311 32.2 1,138 0.0143 4.8 II. Cagayan Valley 3.8 4,811 26.1 761 0.0198 4.5 111. Central Luzon 10.1 5,402 16.0 688 0.0041 2.5 IV. Southern Tagalog 13.3 5,422 23.6 2,772 0.0193 15.5 V. Bicol Region 7.1 4,433 39.1 1,989 0.0313 13.4 VI. Westem Visayas 9.0 4,426 30.9 1,495 0.0139 7.5 VI1. Central Visayas 7.1 3,988 29.2 1,451 0.0270 11.6 Vill. Eastern Visayas 5.3 3,894 33.3 754 0.0160 5.1 IX. Western Mindanao 5.1 4,270 29.9 976 0.0209 6.4 X. Northern Mindanao 5.9 4,494 41.0 1,685 0.0292 10.3 Xi. Southern Mindanao 7.0 4,660 32.9 1,784 0.0264 11.1 XIl. Central Mlindanao 4.6 4,499 32.2 937 0.0197 5.5 a Coiltribttioll to total poverly is calculated as: IWu (region poptilatinnil total poptulationi) (reginti P./ | p 16  Table 2.2 Cost-Effectiveness of NFA's Rice Subsidy for the Consumers (pesos per metric ton) 1991 1992 Gross costs 14,027 13,441 Purchase cost (palay) ' 6,000 6,000 Rice conversion cost 3,435 3,331 Borrowing cost 2,657 2,780 Administration cost 776 619 Operation cost 1,157 812 Other costs (net) 2 -101 Sales receipts 7,930 7,750 Net cost 6,097 5,691 Income transfer to consumers (consumer subsidy) 2,040 2,650 Cost/income transferred 2.99 2.15 Note: For method of calculation, see Appendix 2. Source: Computed from data provided by NFA. It is important to note that the NFA operates a general consumer subsidy program, which benefits the entire population. Under this system, at what cost does the NFA transfer income to the poor? The experiences of general food price subsidy programs in other countries suggest that many nonpoor families receive benefits, making'the cost of transferring income to the poor high. This leakage was about 50 percent in India (Jha 1994), between 60 and 70 percent in Egypt and Morocco, and between 50 and 70 percent in Pakistan (Kennedy and Alderman 1987). In our calculations, a conservative estimate of 50 percent leakage is assumed. In other words, the true cost of income transfer to the poor doubles, meaning between P4.30 and P5.98 as the cost of transfering one peso. The Magnitude and Distribuzion of Benefits. The share of NFA rice distribution in total rice consumption over the last five years has been quite small, ranging from 2.8 percent in 1991 to 11.5 percent in 1990 (Table 2.3). To what extent does the NFA consumer subsidy augment the income, food consumption, and nutrition of the poor? Using the 1991 Family Income and Expenditure Survey data, Intal (1994) estimated that about 40 percent of all Philippine families lived below the poverty line. The 1991 data also suggest that the total income of the poorest 40 percent of the population was P103 billion. According to Clarete and others (1992) the income 17  transfer from e rice subsidy was highest at P1.73 billion in 1990 and lowest at P0.41 billion in 1991 over 1986-91. Assuming 50 percent leakage, the NFA income transfers accounted for no more than 0.2 percent of the poor's total income in 1991 and 0.8 percent in 1992. Garcia and Pinstrup-Andersen (1987) estimate the income elasticity of demand for food by the poor in the Philippines to be 0.68 and the calorie-income elasticity to be 0.33. Thus, even during 1992 (when the highest income transfer was made), the one percent increase in the poor's income probably resulted in a 0.5 percent increase in food consumption and 0.3 percent increase in caloric intake. This impact is negligible. Is the regional distribution of NFA rice sensitive to the regional distribution of poverty? Between 1991 and 1993, the National Capital Region and the Cagayan Valley received 34.5 percent of Authority-subsidized rice, but these two regions accounted for only 2.8 percent of the total food poverty in the Philippines (Table 2.4). On the other hand, the Southern Tagalog, Bicol, Central Visayas, Northern Mindanao. and Southern Mindanao regions, accounting for 62 percent of total food poverty, received only 29.4 percent of NFA rice. Table 2.3 National Rice Balances (thousand metric tons) 1989 1990 1991 1992 1993 Rice production 6,186 6,095 6,326 5,971 6,256 Seed, feed, wastage 575 567 588 555 582 Net domestic supply 5,611 5,528 5,738 5,416 5,674 NFA opening stock 297 155 469 630 329 NFA closing stock 155 469 630 329 154 Imports 220 621 0 0 209 Exports 0 0 10 30 0 Total consumption 5,973 5,835 5,567 5,687 6,058 Population 60,097 61,480 62,868 64,259 65,649 Per capita consumption (kg/year) 99 95 89 88 92 NFA rice distribution 472 670 158 521 485 Note: Rice consumption equals net domestic supply plus imports plus change in stock minus exports. Source: National Food Authority (NFA); National Staistical Coordinaion Board, 1993 Philippine Stiscal Yearbook. The large regional mistargeting suggests that 50 percent leakage is not an unreasonable assumption. Clearly, eliminating this consumer subsidy will have a negligible effect on the poor's food consumption and nutrition, and entail considerable savings for the govermment. Based on 1992 estimates, eliminating the consumer subsidy on rice could save the NFA P3 billion annually. The food security and nutrition of the undernourished population can be substantially improved if the resources saved are used for carefully targeted food stamp or nutrition programs. 18  Table 2.4 Regional Distribution of NFA Rice and Food Poverty Distribution of NFA Rice (metric tons) Average Regional Region 1991 1992 1993 Average Regional Contribution Share of to Total NFA Rice Poverty (percent) (percent) The Philippines 158,171 520,561 484,952 387,895 100.0 100.0 National Capital Region 33,234 87,197 138,522 86,318 22.3 0.3 I. Ilocos Region' 14,502 45,285 44,570 34,786 9.0 6.4 It. Cagayan Valley 10,622 47,307 21,282 26,404 6.8 4.5 Ill. Central L,uzon 27,521 65,693 48,445 47,220 12.2 2.5 IV. Southern Tagalog. 12,488 41,075 65,195 39,586 10.2 15.5 V. Bicol Region 8,397 30,299 35,386 24,694 6.4 13.4 VI. Western Visayas 3,882 39,983 29,359 24,408 6.3 7.5 VII. Central Visayas 11,017 14,929 20,873 15,606 4.0 11.6 VIII. Eastern Visayas 14,384 24,657 35,371 24,804 6.4 5.1 IX. Western Mindanao 4,931 25,740 10,575 13,749 3.5 6.4 X. Northern Mindanao 6,233 28,857 11,043 15,378 4.0 10.3 XI. Southern Mindanao 7,637 37,752 11,042 18,810 4.8 11.1 XII. Central Mindanao 3,323 31,787 13,289 16,133 4.2 5.5 a. The poverty contribution of Cordillera Autonomilous Region is added to llocos region. Source: Data on rice distributiotn were obtained from NFA. Poverty shares are takeni from columinl 7, rable 2. 1. 19  Optionsfor a targeted income transfer Programs designed to improve the nutrition of vulnerable groups commonly fall into two broad categories: those that increase household income and those that use mechanisms like direct feeding and growth monitoring of children and influencing the behavior of household members. This study focuses on raising household income. The second category is discussed in World Bank (1994b). The needfor targeting. Ultrapoor families require better employment and higher incomes to ensure adequate food intake. Consequently, any long-run solution to malnutrition will require widespread, labor-using economic growth. In the interim a well-managed and appropriately targeted program can provide needed in-kind transfers to improve food consumption and nutrition. Since the need (P16.8 billion in 1991) is substantial relative to available resources, the government may wish to target income transfers to those most at risk. While targeting to the most needy will reduce the fiscal cost of transferring income, it may increase the cost per beneficiary because precise targeting entails considerably more administrative costs and requires greater managerial skills (Kennedy and Alderman 1987). Hence cost-effective programs should balance the cost of leakage and the cost of targeting. Methods exist, however, to substantially reduce the administrative costs of targeting, including self-targeting programs which discourage nonpoor participation; geographical targeting, which limits interventions to specific depressed areas; seasonal targeting, when nutritional stress is particularly high; and distribution of a commodity that is consumed by the poor but not by the wealthy. Of these methods targeting costs tend to be lowest for self-targeted and geographically targeted programs. A recent study suggests that P2 billion of geographically targeted income can reduce poverty by as much as a universal program (such as the NFA's general food subsidy) costing P18 billion (Balisacan 1994b). Interntonal experience. Targeted income transfers can be provided in cash and in kind. A recent study on Bangladesh (IFPRI 1994) demonstrates that the cash-based Rural Maintenance Program provides a lower-cost safety net than any food-based program; food is a cumbersome resource, and cash is more flexible and less costly to manage. Food transfers immediately raise program costs by 25 percent because of transport and handling costs. Food-based interventions, however, can be more effective in improving food consumption than cash-based programs. Several studies of consumption effects of targeted food interventions in both developed and developing countries indicate that the poor's marginal propensity to consume food is substantially higher with food subsidy transfers than with cash transfers. These studies include a wheat-based vulnerable group development program in Bangladesh (Ahmed 1993); the 20  U.S. food stamp program (Davaney and Moffit 1991; Senauer and Young 1986; Benus, Kmenta, and Shapiro 1976); a rice and cooking oil subsidy program in the Philippines (Garcia and Pinstrup-Andersen 1987); a rice subsidy program in Sri Lanka (Edirisinghe 1987); and a rice subsidy program in Kerala, India (Kumar 1979). One alternative to general food subsidies is a food stamp program, which has merits of cash and food-based programs. Food stamp programs avoid the cost of commodity handling. Also, some evidence indicates that participation in food stamp programs leads to greater food consumption. Moreover, food stamps do not have the allocative inefficiency effects of food price subsidies. Rather, food stamps foster greater consumption, thus stimulating local production by raising demand for basic local products. Experience with food stamp programs in developing countries is mixed. Sri Lanka switched from a ration system to a food stamp system in 1979; total government expenditures fell from 14 percent to 7 percent (Kennedy and Alderman 1987). Targeting was partially successful. The poorest 40 percent of the population received 67 percent of food stamps in 1982 compared with 50 percent of subsidies under the ration system (Edirisinghe 1987). The food stamp program in Sri Lanka was means tested. Stamps were issued to households based on their incomes and number of children. The administrative cost of the program was only 2 percent of the program cost (Cornia and Stewart 1993). Sri Lanka's long experience with food subsidy programs might have contributed to the low administrative costs. However, as Edirisinghe (1987) found, the real value of food stamps fell sharply because of inflation. The Jamaican food stamp program was launched in 1984, directed at pregnant and nursing mothers and young children, regardless of household income (Alderman, 1991). Although the Jamaican program was not specifically targeted at poor families, more than half of the recipient households belonged to the poorest quintile, while only 6 percent of those in the wealthiest quintile obtained stamps (Table 2.5). The stamps could be collected only at the primary health clinics in exchange for immunization of children and pregnant women and nutrition counselling of mothers. Since the nonpoor did not frequent these clinics, the program effectively reached the poor, and the administrative cost was 4 percent of program cost (Grosh 1993). Food stamp programs were pilot-tested in Colombia, but the program never progressed beyond this stage (Uribe 1986). Mexico successfully replaced its urban maize subsidy program with the tortilla program, targeted to urban households with incomes lower than twice the minimum wage. Plastic cards were issued to households for identification. In Honduras, a food coupon scheme proved to be cost-effective in alleviating poverty. Kennedy and Alderman (1987) provide a summary of leakage estimates for various food intervention programs. Food stamp programs also experienced low to moderate leakage, ranging from 3 to 10 percent in Colombia and Indonesia, where the programs were targeted by the health 21  status of vulnerable household members. In the income targeted food stamp programs in Colombia, Sri Lanka, and the United States, leakage ranged from 10 to 30 percent. Further, food stamps have some disadvantages. First, they can be duplicated. In Zambia large-scale counterfeiting led to the virtual elimination of the food stamp program (von Braun and others 1992). However, fraud was prevented in many countries. Second, food stamps are not indexed for inflation and are not a good option if inflation is high. Finally, it is difficult to get people off food stamps once their incomes increase, necessitating periodic means-testing. Table 2.5: General and Targeted Subsidies, Jamaica 1988 (percent) General Targeted Subsidy Item Subsidy (food stamps) Cost as share of government expenditurea 3.0 1.6 Propormon of transfer going to: Poorest quintile 14.0 31.0 Richest quintile 26.0 8.0 Transfer as share of expenditure per recipient: Poorest quintile 2.3 9.5 Richest quintile 0.1 1.0 Proportion of households covered: Poorest quintile 100.0 51.0 Richest quintile 100.0 6.0 a Does not include administrative cost. Source: Jamaica Statistical Institte and World Bank 1988 and 1989 as cited in Imnplementing the World Bank's Strategy for Reducing Poverty and Hunger. A promising option for the Philippines. Experience in the Philippines suggests that a targeted food-mediated income transfer can be successful. The 1983 Pilot Food Subsidy Scheme was very successful, though short-lived. A successful food stamp program must identify the needy in the most cost-efficient manner. Identification can be achieved in two steps. First, food stamps may be allocated to the provinces according to provincial (regional) contribution to food poverty (see Table 2.1). This food poverty measure can be updated periodically (using the latest available Pamily Income and Expenditure Survey data) and completed at the level of provinces to ensure accurate targeting. Second, at the regionallprovincial level, the neediest must be identified to ensure that the poverty gap is filled from the bottom up. Fortunately, the institutional structure for identifying the poor and operating a public safety net program is quite favorable in the Philippines. The DSWD is experienced in providing disaster relief, identifying vulnerable households through community-based initiatives. Community workers are already able to identify the ultrapoor with reasonable accuracy. Though DSWD has been devolved to Local Government 22  Units (LGUs), its experience in community-based beneficiary identification needs to be preserved and used for delivery of food stmps or any other insrunent food transfer. Leakages ranged from 3 to 30 percent in food stamp programs in other countries. An assumption of 12 percent leakage, based on judgments about administrative capabilities and difficulties anticipated in the Philippines, seems reasonable. With this amount of leakage, it would cost the government P1.31 to transfer P1 to a targeted household (Table 2.6). 6 In contrast, NFA's food subsidy program requires P4 to P6 to transfer Pl to a poor household. If P2.96 billion is saved by eliminating the consumption subsidy, how much income can be transferred to the poorest 50 percent of the ultrapoor families? Assuming moderate leakage, the suggested food stamp program could transfer P2.26 billion to the targeted households. In 1991 about 2.42 million families (that is, about 20 percent of the total population) lived below the food poverty threshold. The 1991 Family Income and Expenditure Survey (FEIS) data suggest that the total income of the poorest 10 percent of these families (or the poorest 50 percent of the ultrapoor) was P14.29 billion. If half of the ultrapoor families are given food stamps, an ammal tansfer of P1,900 per family (about 16 percent of family income) can be effected, (compared with less than 1 percent under the current food subsidy program). Determining the exact shape of the Philippine food stamp program is beyond the scope of this study. However, a few elements are worth mentioning. Food stamps can be used as wage payment in public works programs. Following the Jamaican food stamp program, preschool children and pregnant and lactating women could receive food stamps at the clinics, perhaps in exchange for immunization, growth monitoring, and nutritional counseling. To achieve better targeting, clinics located in areas with a high concentration of vulnerable groups could be selected. Food stamps can also be distributed at schools to entice parents to enroll and retain their children in school. An early assessment of a similar program, the Food for Education program, in Bangladesh suggests a high level of success in increasing school enrollment, promoting school attendance, and lowering dropout rates (Ahmed and Billah 1994). A food stamp program is only one of many options for using the resources saved from contracting of NFA's operations. The saved resources could be invested in a targeted child and maternal nutrition program. Conclusions and Recommendations. NFA's food subsidy program has had little effect on the poor's consumption and nutition, and this low impact was achieved at a high cost T These mclud progps costs as well as admninistzve costs. 23  The study reconnmends that NFA not mantain a consnption subsidy on rice. Based on 1992 figures, this policy change could result in a savimg ofP3 billion annually. Further, the saved resources may be devoted to better targeted programs, such as food stamp or nutriton programs. If the resources now going to the NFA food subsidy were spent on a targeted food stamp program, the equivalent of 16 percent of an ultrapoor an*s income could be tansferredt The study recommends that these resources be allocated to each province according to its contrbution to poverty. Then, access may be restricted to the bottom income decile in each province. To identify and target tis group, the services of the now devolved DSWD at the LGU-level may be used, involving communities and nongovernmental agencies. In addition, the government could consider other targeting mechanisms reviewed in the report Table 2.6 Probable cost-effectiveness of a food stamp program Food stamps (pesos) Expected cost' 1.15 Face value 1.00 Printing cost 0.05 Administration and trader commission 0.10 Leakage (percent) None Moderate High Total 0 12 30 Misidentification 0 5 10 Underpay beneficiary 0 5 10 Redeem at less than face value 0 2 10 Income transfer (pesos) 1.00 0.88 0.70 Costlmcome transferred (pesos) 1.15 1.31 1.64 a. Costs are estimated based on acual performance of programs in Jamaica (Grosh 1992) and Sri Lanka (Edirsinghe 1987). 24  Chapter 3 Labor-Based Public Works: Rationale, Scope, and Effectiveness The Government of the Philippines is expanding its labor-based public works schemes, seeking to attain the twin objectives of asset creation and employment generation. The emphasis on employment generation is a conscious attempt to confront problems of labor absorption (growing unemployment, high underemployment, and declining labor productivity), particularly in rural areas where the bulk of the poor live. The two large-scale national programs. the Community Employment and Development Program (1986-87) and Kabuhayan 2000, (currently operating) are examples of such efforts. Labor and Poverty. Central to the problem of labor absorption is the rapid growth of labor supply relative to demand. The labor force grew at an average of 3.7 percent annually in the 1970s and 3.6 percent in the 1980s. A high population growth rate and an increase in labor force participation rate contribute to rapid labor force growth. Population growth averaged 2.5 percent per year in the 1980s, exceeding the rate in Thailand (1.9 percent) and Indonesia (2.1 percent). The period of high population growth coincided with an increase in labor force participation from 59.6 percent in 1980 to 66.4 percent in 1990. The growth in labor force participation was particularly marked for women in rural areas (Sanchez 1994). But given slow overall rural growth and thus slow labor demand growth -demand for agricultural labor grew by only 1.8 percent per year during 1980-91 (Bot 1994)- high participation rates are merely a manifestation of low productivity and high underemployment. The labor force growth rate is much more rapid in urban areas, especially where net in- migration is high (Costello and Ferrer 1992). This trend is evident in the National Capital Region and its adjacent regions, like Southern Tagalog and Central Luzon. The urban labor force grew, on average, by 7.1 percent per year in the 1980s compared with the 1.8 percent annual growth rate of the rural labor force. Such rapid growth translates into large absolute increases in the labor force. For example, 8.4 million new job seekers appeared between 1980 and 1990. During this period the economy as a whole grew at an average annual rate of 3.4 percent, but this rate was not sufficient to absorb all the new labor force entrants. The demand shortfall raised the number of unemployed from 1.4 million in the early 1980s to 2.5 million in 1985 and 2.7 million in 1991. Unemployment rates are very high in rapidly growing urban areas, particularly the National Capital Region and its adjacent regions (Table 3.1) although an increasing number of people are looking for jobs in both urban and rural areas. Unemployment is largely concentrated among the young, especially those with a secondary or college education. 25  The predominant labor absorption problem in rural areas is underemployment. Slightly more than one-quarter of the employed rural labor force was seeking additional work in the 1980s. This rate, which is linked to the rate of economic growth, reached 37.7 percent in 1984, at the height of economic crisis. High labor supply growth and inadequate labor demand contributed to high levels of underemployment and a decline in labor productivity. Data on labor productivity are unavailable, but the decline can be inferred from falling real wages for skilled, unskilled, and agricultural workers (figure 3.1). Real wages for workers in palay, corn, and sugar in 1990 were lower than 1976 levels (see figure 3.2). The shift to nonagricultural employment, which is occurring slowly, is concentrated in the service sector. This shift is a response to diminishing oppornmities in the agricultural sector (for example, because of limited access to land and technology) and poverty. Even in areas where agriculture is growing, the multiplier effect on the nonfarm sector is low; the consumption linkage is weak because the gains from higher productivity are distributed equally (Bautista 1991; Baliscan 1993a). Workers not absorbed in income-earning activities in the agricultural sector (especially those in low-income and resource-poor regions where infrastructure is weak), are mainly self-employed in service sector, earning low pay. In sum, rural poverty is largely rooted in labor absorption problems (unemployment, underemployment. insufficient access to land and improved technology, and low returns to labor). Within rural areas poverty is concentrated in agriculture (farming, fishing, and forestry), especially among the self-employed. These include lessees, tenants, and small cultivators. The poverty intensity is equally high for those that fish, but their relative contribution to poverty is lower (Balisacan, 1993). The relation between poverty and employment varies across regions (Table 3.1). Regions such as Bicol, Northern Mindanao, and Southern Mindanao have a large share of poverty and a high rate of underemployment. Bicol also has a high rate of net out-migration. Some regions have either a high rate of underemployment (for example, the Cordillera Autonomous Region, Western Mindanao, and Central Mindanao) or net out-nigration (Eastern Visayas), but do not make a large contribution to poverty because they have lower population shares. 26  Table 3.1 incidenice of Poverty, Unemployment, and Underemployment, and Net Migration by Region (percent) Region Population Contribution Unemployment Underemployment In-Out Cost Share to Poverty Rate Rate Migration Allocation of (a=2)a Rate Kabulhayan 2000 National Capital Region 14.0 0.3 17.4 7.5 +10.4 18.65 Cordillera Autonomous Region 2.0 1.6 6.0 20.2 -1.7 3.88 I. Ilocos Region 5.6 4.8 8.2 15.5 -0.1 6.16 11. Cagayan Valley 3.8 4.5 6.6 18.0 -3.7 6.01 111. Centtal Luzon 10.1 2.5 12.5 16.0 4.2 11.05 IV. Southern Tagalog 13.3 15.5 9.9 17.7 6.9 12.54 V. Bicol Region 7.1 13.4 6.0 32,8 -15.7 8.07 VI. Western Visayas 9.0 7.5 9.8 25.9 -5.4 5.52 VII. Central Visayas 7.1 11.6 8.5 8.7 -2.5 4.99 VIII. Eastern Visayas 5.3 5.1 7.4 22.4 -16.8 7.43 IX. Western Mindanao 5.1 6.4 6.1 23.6 -3.9 4.39 X. Northern Mindanao 5.9 10.3 8.7 28.0 -2.8 4.30 XI. Southern Mindanao 7.0 11.1 10.6 30.4 1.8 3.64 XII. Central Mindanao 4.6 5.5 7.0 39.6 4.3 3.36 a. Original calculation is in Table 2. 1. 27  FFaure 3.1: Real Wage Indices (19S0-10O0T 120 I60 ----- 40 20 1Q80 1Q85.0 1975 1980 1985 199 -Skilled WVorkera -Unsklild Worksrs - Ar1cultural Workers Figure 3.2: Index Of Real Wage Rates Of Farm Workers, By Crop, 1974-90 (1978 = 100) itO -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 100 1974 1976 1978