Indonesia's PNPM Generasi Program : final impact evaluation report

 72509 Indonesia’s PNPM Generasi Program: Final Impact Evaluation Report June 2011 Benjamin A. Olken, M.I.T. Department of Economics Junko Onishi, The World Bank Susan Wong, The World Bank i Table of Contents Acknowledgments ........................................................................................................................... iv Glossary ......................................................................................................................................... vi EXECUTIVE SUMMARY ............................................................................................................ 1 1 INTRODUCTION .................................................................................................................. 6 1.1 Background ...................................................................................................................... 6 1.2 The Generasi project ........................................................................................................ 8 1.3 Experimental Design ...................................................................................................... 12 1.4 Survey Design and Implementation ............................................................................... 14 2 EVALUATION METHODOLOGY .................................................................................... 18 2.1 Regression Specifications .............................................................................................. 18 2.2 Balance Tests.................................................................................................................. 20 3 MAIN RESULTS.................................................................................................................. 22 3.1 Direct Benefits of Generasi Funds ................................................................................. 22 3.2 Program Impact on Main Targeted Indicators ............................................................... 23 3.3 Program Impact on Long-Term Final Outcomes ........................................................... 24 3.4 Impact on Non-Targeted Indicators ............................................................................... 26 4 Where Were Generasi's Effects Largest? .............................................................................. 27 4.1 Heterogeneity by Area Characteristics ........................................................................... 27 4.2 Heterogeneity Over Time ............................................................................................... 30 4.3 Heterogeneity Across Poor vs. Non-poor Individuals ................................................... 30 5 HOW AND WHY DID THE GENERASI PROJECT WORK? .......................................... 32 5.1 Changes in Provider Quantities ...................................................................................... 32 5.2 Changes in Provider Quality .......................................................................................... 33 5.3 Changes in Provider Effort ............................................................................................. 33 5.4 Changes in Community Effort ....................................................................................... 34 5.5 Services and Prices from Providers ................................................................................ 35 6 DISCUSSION ....................................................................................................................... 37 7 POLICY IMPLICATIONS AND CONCLUSION .............................................................. 39 i References ..................................................................................................................................... 41 Annex I. Randomization and Implementation of Generasi in 2007 ............................................. 44 List of Tables Table 1. Performance metrics and weights .................................................................................................................. 11 Table 2 Questionnaire Modules and Sample Size ....................................................................................................... 16 Table 3. Generasi implementation and randomization results ..................................................................................... 45 Table 4. Baseline Regressions, 12 main indicators * ................................................................................................... 46 Table 5. Baseline regressions, long-term final outcomes ............................................................................................ 47 Table 6. Direct benefits ............................................................................................................................................... 48 Table 7. Direct benefits, incentivized vs. non-incentivized ......................................................................................... 49 Table 8. Direct benefits, provincial breakdown ........................................................................................................... 51 Table 9. Program impact on main targeted indicators ................................................................................................ 52 Table 10. Program impact on main targeted indicators, incentivized vs. non-incentivized ........................................ 54 Table 11. Program impact on main targeted indicators, provincial breakdown.......................................................... 56 Table 12. Program impact on longer-term outcomes ................................................................................................... 58 Table 13. Program impact on longer-term outcomes, incentivized vs. non-incentivized ............................................ 60 Table 14. Program impact on longer-term outcomes, provincial breakdown ............................................................. 62 Table 15. Program impact on non-targeted indicators ................................................................................................ 64 Table 16. Program impact on main targeted indicators, interactions with pre-period subdistrict level variables ....... 67 Table 17. Program impact on main targeted indicators, interactions with pre-period subdistrict level variables, incentivized vs. non-incentivized ....................................................................................................................... 69 Table 18. Program impact on longer-term outcomes, interactions with pre-period subdistrict level variables ........... 71 Table 19. Program impact on main targeted indicators, pre-period subdistrict average per capita consumption interactions ......................................................................................................................................................... 73 Table 20. Program impact on longer-term outcomes, pre-period subdistrict average per capita consumption interactions ......................................................................................................................................................... 75 Table 21. Program impact on main targeted indicators, pre-period village access interactions .................................. 77 Table 22. Program impact on longer-term outcomes, pre-period village access interactions ..................................... 79 Table 23. Program impact on main targeted indicators, excludes 49 newly added Generasi subdistricts in Year 2 . 81 Table 24. Program impact on main targeted indicators, comparing incentivized vs. non-incentivized, excludes 49 newly added Generasi subdistricts in Year 2 ..................................................................................................... 83 Table 25. Program impact on longer-term outcomes, excludes 49 newly added Generasi subdistricts in Year 2 ..... 85 ii Table 26. Direct benefits, per-capita consumption quintile breakdown. ................................................................... 83 Table 27. Program impact on main targeted indicators, per-capita consumption quintile breakdown. .................... 88 Table 28. Program impact on longer-term outcomes, per-capita consumption quintile breakdown. .......................... 90 Table 29. Results for service provider quantities ......................................................................................................... 92 Table 30. Results for service provider quantities, incentivized vs. non-incentivized .................................................. 94 Table 31. Results for service provider quality (health and education infrastructure availability) ............................... 96 Table 32. Results for service provider quality (health and education infrastructure availability), incentivized vs. non- incentivized.................................................................................................................................................................. 98 Table 33. Results for service provider level of effort ................................................................................................ 100 Table 34. Results for service provider level of effort, incentivized vs. non-incentivized ......................................... 102 Table 35. Results for community efforts at service provision, monitoring, and participation ................................... 104 Table 36. Results for community efforts at service provision, monitoring, and participation, incentivized vs. non- incentivized ...................................................................................................................................................... 107 Table 37. Service prices and supply .......................................................................................................................... 110 Table 38. Service prices and supply, incentivized vs. non-incentivized .................................................................... 114 List of Figures Figure 1. Village funding allocations, 2007 and 2008 ...................................................................................................9 Figure 2. Timeline of project and surveys ................................................................................................................... 15 List of Boxes Box 1. Generasi Program Target Indicators ..................................................................................................................7 iii Acknowledgments The authors wish to thank the members of the PNPM Generasi Team, including Sadwanto Purnomo, Gerda Gulo, Juliana Wilson, Yulia Herawati, Gregorius Endarso, Gregorius Pattinasarany, Joey Neggers, Lina Marliani, Scott Guggenheim, Robert Wrobel, John Victor Bottini, Threesia Mariana Siregar, Sentot Surya Satria, Christine Panjaitan, Soenoe Widjajanti, Suhartini B. Rianto, and Erni Yanti Siregar. Special thanks go to Yulia Herawati, Gregorius Endarso, Joey Neggers, and Lina Marliani for their excellent and tireless support in survey preparation, oversight, and data preparation. The government of Indonesia—through the Ministry of Planning (Bappenas), the Coordinating Ministry for Economy and Social Welfare (Menkokesra), and the Ministry of Home Affairs — has provided tremendous support to the program and its evaluations over the past three years. Special thanks to Sujana Royat (Menkokesra); Prasetijono Widjojo, Endah Murniningtyas, Pungky Sumadi, and Vivi Yulaswati (Bappenas); and Ayip Muflich, Eko Sri Haryanto, and Bito Wikantosa (Ministry of Home Affairs) for their generous support of the PNPM Generasi program. The University of Gadjah Mada (UGM), Center for Public Policy Studies, implemented the field surveys over the last three rounds. SMERU, the Indonesian independent research organization, collaborated with the program to provide complementary qualitative studies in 2007 and 2009. In total, some 860 enumerators and researchers from these two Indonesian organizations contributed to this tremendous data collection effort. The authors are grateful to both institutions for their diligent work over the past three years. This final evaluation report drew from the 2007 baseline survey report written by Robert Sparrow, Jossy Moeis, Arie Damayanti, and Yulia Herawati. This report benefited from the comments of peer reviewers, including Ana-Maria Arriagada, Emanuela Galasso, Lisa Hannigan, Jack Molyneaux, Rebekah Pinto, Joppe Jaitze De Ree, Emmanuel Skoufias, and Sudarno Sumarto. The team is also grateful for the collaboration with the Program Keluarga Harapan evaluation team from the World Bank, consisting of Vivi Alatas, Jon Jellema, and Edgar Janz. Editorial assistance for this report was provided by Robert Livernash. Anju Sachdeva and Elizabeth Acul provided valuable administrative support. Financial support for the overall PNPM Generasi program and the evaluation series has come from the government of Indonesia; the World Bank Decentralization Support Facility; the Netherlands Embassy; the PNPM Support Facility, which consists of donors from Australia, the United Kingdom, the Netherlands, and Denmark; and the Spanish Impact Evaluation Fund. iv The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors and should not be attributed in any manner to the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors, or the countries they represent. The World Bank does not guarantee the accuracy of the data included in this publication and accepts no responsibility for any consequence of their use. v Glossary ANC Antenatal care ARI Acute respiratory infection Bappenas Ministry of Planning Buku KIA Mother and child health book CCT Conditional cash transfers CDD Community-driven development GIS Geographical information system Gotong royong Semi-volunteer public labor service at the village or community level KDP Subdistrict (Kecamatan) Development Project KIA Kartu Insentif Anak (Child Incentive Health Card) Menkokesra Coordinating Ministry for Economic and Social Welfare MIT Massachusetts Institute of Technology NTT Nusa Tenggara Timur province PKH Program Keluarga Harapan (Hopeful Family Program) PNC Postnatal care PNPM Program Nasional Pemberdayaan Masyarakat (National Program for Community Empowerment) PNPM Generasi PNPM Healthy and Smart Generation Sehat dan Cerdas PNPM-Rural Program Nasional Pemberdayaan Masyarakat Perdesaan (National Program for Community Empowerment in Rural Areas) PNPM-Urban Program Nasional Pemberdayaan Masyarakat Perkotaan (National Program for Community Empowerment in Urban Areas) PODES Village potential statistics Posyandu Village integrated health post (monthly community weighing post) Puskesmas Community health center SUSENAS National Socioeconomic Survey SD Sekolah Dasar (primary school) SD Standard deviations vi SMP Sekolah Menengah Pertama (junior secondary school) SPADA Support for Poor and Disadvantaged Areas Project UGM University of Gadja Madah Notes: All $ = U.S. dollars, unless otherwise noted vii EXECUTIVE SUMMARY Indonesia has made remarkable strides in key human development indicators over the past few decades. Primary school enrollment is close to universal for both boys and girls, and the child mortality rate has declined rapidly. Nevertheless, infant mortality, child malnutrition, maternal mortality, junior secondary school enrollment, and educational learning quality have all remained problematic in Indonesia compared to other countries in the region. Furthermore, achievements in these indicators reveal large geographical disparities, with poorer outcomes in rural and remote provinces and districts. Improving access to basic quality health and education services is a key component of an overall poverty reduction strategy for Indonesia. In 2007, the government of Indonesia launched two large-scale pilots of programs designed to tackle these issues: (1) conditional cash transfers (CCTs) to households, known as the Hopeful Family Program (Keluarga Harapan Program or PKH), and (2) an incentivized community block grant program, known as the National Community Empowerment Program— Healthy and Smart Generation (Program Nasional Pemberdayaan Masyarakat—Generasi Sehat dan Cerdas, or PNPM Generasi). These two complementary pilot projects began in six provinces and are designed to target the same health and education indicators. They are consistent with both the Indonesian government‘s priorities and the Millennium Development Goals: to reduce poverty, maternal mortality, and child mortality, and to ensure universal coverage of basic education. PKH focused more on supply-side ready areas, predominantly urban and in Java, while PNPM Generasi operated in rural areas. This study reports on the final evaluation of the incentivized community block grant program, PNPM Generasi. A separate report has been prepared by the World Bank on the results of the PKH program. The Generasi project began in mid-2007 in rural areas of five Indonesian provinces selected by the government: West Java, East Java, North Sulawesi, Gorontalo, and Nusa Tenggara Timur. The project builds on the Indonesian government‘s existing community-driven development program, known as the National Community Empowerment Program in Rural Areas (PNPM-Rural) or formerly, the Kecamatan Development Program (KDP). The Ministry of Home Affairs Community Development Department implements both Generasi and PNPM- Rural. In 2007, Generasi covered 1,605 villages in 129 subdistricts, with a total budget of $20 million. By 2009, Generasi covered just over 2,000 villages in 164 subdistricts in the same five provinces with a total annual budget of $40 million. The program operates as follows: each year, villages receive a block grant. With the assistance of trained program facilitators and service delivery workers, villagers undertake a social mapping and participatory planning exercise to decide how best to use the block grant funds to reach 12 education and health targets related to maternal and child health behavior and education behavior. These 12 targets relate to prenatal and postnatal care, child immunizations, and primary and junior secondary school enrollment and attendance. To give communities incentives to focus on the most effective policies, the government bases the size of the village‘s Generasi block grant for the subsequent year partly on the village‘s performance on each of the targeted indicators. The Generasi project thereby takes the idea of performance incentives from 1 conditional cash transfer programs and applies it in a way that allows communities the flexibility to address supply constraints, demand constraints, or some combination. To the best of our knowledge, the Generasi project is the first health and education program worldwide that combines community block grants with explicit performance bonuses for communities. To allow for a rigorous, randomized evaluation of Generasi, the government of Indonesia incorporated random assignment into the selection of Generasi locations. Within the districts selected by the government for the program, entire subdistricts (kecamatan) were randomly allocated to either receive Generasi or to be in a control group. Each Generasi location was further randomly allocated to one of two versions of the program: one ―incentivized‖ treatment with the pay-for-performance component (treatment A) described above, and a second, otherwise identical ―non-incentivized‖ treatment without the pay-for-performance incentives (treatment B). This document describes the findings from the three-wave evaluation series carried out from 2007 to 2010. The baseline survey took place from June 2007 to August 2007. The second wave was conducted from October 2008 to January 2009, after 15 to 18 months of Generasi implementation. The third and final evaluation survey was implemented from October 2009 to January 2010 after 27 to 30 months of project implementation. Over 45,000 household members, village heads, and school and health facility staff were respondents for the third and final round of survey. The evaluation series also included a qualitative component. To the extent possible, the authors have incorporated findings from the complementary qualitative study in 12 villages in two provinces; this qualitative component was conducted in 2007 and 2009. The qualitative study—using focus group discussions, in-depth key informant interviews, and direct observation—provided deeper insights into processes, causal chains, and villagers‘ values, motivations, and reactions. The main findings of the Generasi impact evaluation are as follows: 1. After 30 months of program implementation, Generasi had a statistically significant positive impact on average across the 12 indicators it was designed to address. The strongest improvements among the health indicators were in the frequency of weight checks for young children. The program also increased the number of iron sachets pregnant mothers received through antenatal care visits. These improvements were supported by dramatic increases in mothers and children participating in village health post (posyandu) activities to receive the targeted maternal, neonatal, and child health services. Education indicators also saw improvements in the final evaluation, reversing the zero or negative impact found at the interim evaluation. The improvement in education indicators was most notable in the increased school participation rate among the primary school-age group. 2. The main long-term impact was a decrease in malnutrition. The latest Wave III survey shows that childhood malnutrition1 was reduced by 2.2 percentage points, about a 10 percent 1 Childhood malnutrition was measured by weight-for-age of children under three. 2  reduction from the control level. This reduction in malnutrition was strongest in areas with a higher malnutrition rate prior to project implementation, most notably in the Nusa Tenggara Timur (NTT) Province, where underweight rates were reduced by 8.8 percentage points, a 20 percent decline compared to control areas; severe underweight rates were reduced by 5.5 percentage points, a 33 percent decline; and severe stunting was reduced by 6.6 percentage points, a 21 percent decline compared to control areas. Surprisingly, in Java, there was a negative impact on stunting and severe stunting which needs to be explored further. Although reductions in infant and child mortalities were observed in the interim evaluation, the same levels of reduction in mortality were not sustained in the final evaluation. In terms of the longer-term education learning outcomes, the program did not improve childhood test scores as yet. 3. Making grants conditional on performance improves program effectiveness in health but not in education. On average, the incentivized group outperformed the non-incentivized group in improving health indicators, particularly in increasing antenatal care services. On net, between 50-75% of the total impact of the block grant program on health indicators can be attributed to the performance incentives. However, for education indicators the incentivized group did not appear to perform better than the non-incentivized group. There may be several reasons for this. The data shows two results: first, the impact of incentives became weaker over time; and second, the positive impacts on education only occurred in Year Two of the program, probably due to time lags in implementation of the education interventions. Thus, by Year Two, the incentives were less strong just when the program was beginning to have impacts on education. Another factor may have been that gains in health were easier to attain than gains in education. Baseline levels for health indicators were lower than for education, making it perhaps easier to make gains in health. In addition, education targets may have been more difficult to achieve since those targets involved more people and involved school enrollment and attendance every day, as opposed to once-a- month for health targets, with fewer villagers involved. The qualitative report also suggests that communities favored giving school assistance directly to the greatest number of students, rather than out-of-school children, and that motivation may have dampened any effects from the incentives. Lastly, the qualitative report indicates that the incentives rules were sometimes difficult for communities to understand. 4. Generasi had the greatest impact in areas with low baseline health and education indicators. Areas with lower pre-project health and education indicators have more room for improvements. The greater impacts in areas with lower baseline indicators appear more prominently in the final evaluation survey than the interim results, with stronger improvements found in education indicators in these areas. On average, the program was about twice as effective in areas at the 10th percentile of service provision (very low health and education status) at baseline as it was on average. However, these improvements in health and education indicators in areas with low baseline coverage did not appear to have resulted in improving long-term health and education outcomes in these areas outside of malnutrition. Furthermore, the greater impacts observed in health and education indicators were not simply correlated with pre-project levels of poverty, but instead were driven by the level of health and education indicators in the area. 3 Policy Implications and Conclusion The evaluation results point to several relevant policy implications and conclusions: 1. Generasi is most effective in areas with low health and education status. The impact evaluation found that Generasi impacts are stronger in areas where health and education indicators are low. This suggests that future expansion of Generasi implementation should prioritize areas where these indicators are lagging behind and not necessarily in areas identified as poor. 2. Community incentives had mixed results—health responded more favorably than education. As a result of the two-year project implementation, health indicators responded positively to community incentives, but education indicators saw no positive or negative response to community incentives. Learning from this experience, the government may wish to see how community incentive interventions can work in other lagging areas, e.g., water and sanitation access. The policy implications are that poverty programs may wish to experiment more with embedding incentives into the designs; however, the interventions and incentives will need to be monitored and evaluated over time. One possibility is that the conditionalities may work less well over time, as there may be more ―gaming‖ of the system as the program progresses and the rules become more familiar. Alternatively, the program may work better over time as it continues to incentivize communities to work harder toward the specified targets. In addition, qualitative evidence suggests that simplifying the incentive scheme may make it much easier for communities to understand. 3. The government’s existing national community-driven development architecture and network (PNPM) was useful as a platform for other forms of local assistance. Generasi was started as an experiment in adapting the community participatory planning and block grant process to focus on specific education and health targets that were not being addressed sufficiently in the existing community program. This project has illustrated the flexibility and adaptability of this community model once the architecture and machinery are established. It also serves as a possible vehicle for improving health and education indicators in supply-deficient areas, where the traditional household conditional cash transfer model may not be as effective due to supply constraints. 4. The project should regularly review the appropriateness of the targets. Target indicators must be relevant to communities, yet reflect development priorities of the government. Although it is important not to overload the project with too many target indicators, Generasi should regularly review its 12 target indicators and assess if existing ones should be replaced or added. For example, school participation for children ages 7–12 has now reached nearly universal coverage at 98.5 percent and higher, and the program may wish to add other targets to capture other priority areas that are lagging, such as education learning achievement, early childhood development, or water and sanitation. 4 5. A follow-up evaluation may be needed in the future to examine the longer term sustainability of interventions and impacts. The final round of evaluation took place after 2.5 to 3 years of project implementation. This three-year evaluation series was useful in providing empirical evidence to inform project implementation and learn lessons for the next phase. Should the program continue in the original treatment sites and should the original control sites remain as such, the government may wish to consider the possibility of another evaluation in a few years to examine if the impacts of this program are indeed sustainable over time and if additional progress can be made on learning and health outcomes. 5 1 INTRODUCTION 1.1 Background Over the past decades, Indonesia has made remarkable strides in key human development indicators. Primary school enrollment is close to universal for both boys and girls and the child mortality rate has declined rapidly (World Bank 2006; World Bank 2008). Nevertheless, infant and maternal mortality, child malnutrition, junior secondary school enrollment, school transition rates, and learning outcomes are lower in Indonesia than in other countries in the region (World Bank 2006; World Bank 2008). Furthermore, there are substantial geographical disparities in these outcomes, with poorer outcomes in rural and remote provinces and districts. Improving the health and education of children is considered critical to economic development and forms an important component of the Millennium Development Goals. Faced with these challenges, many developing countries have sought to stimulate demand for maternal and child health services and education through conditional cash transfer programs. For example, Mexico‘s Progresa program (Gertler 2004; Schultz 2004; Rawlings and Rubio 2005) links cash payments to behaviors such as immunization, growth monitoring, school enrollment, and school attendance. However, these types of demand-side interventions may be inappropriate in many developing world contexts, where beneficiaries do not have adequate access to health and education services (Schubert and Slater 2006; Lagarde, Haines, and Palmer 2007). In such environments, programs that address both the supply- and demand-side constraints directly may be more appropriate. In 2007, the government of Indonesia launched two large-scale pilots of programs designed to tackle these issues: (1) conditional cash transfers to households, and (2) an incentivized community block grant program. These two pilot projects began in six provinces and were designed to achieve the same objectives and goals. These goals are consistent with the Indonesian government‘s priorities and the Millennium Development Goals: to reduce poverty, maternal mortality, and child mortality, as well as ensure universal coverage of basic education. The Household CCT—the Keluarga Harapan Program (PKH)—applies the traditional CCT design with quarterly cash transfers to poor individual households identified through statistical means. CCT recipient households receive regular cash transfers through the post office as long as they meet the requirements of using specified health and education services. When it began, PKH focused primarily upon more supply-side-ready urban areas, primarily in Java. The Incentivized Community Block Grant Program, known as PNPM Generasi, differs from the Household CCT in that block grants are allocated to communities rather than to individual targeted households. Unlike PKH, PNPM Generasi focuses primarily on rural areas. This pilot program builds upon an existing Indonesian government community program known as the National Community Empowerment Program in Rural Areas (PNPM-Rural). Under PNPM Generasi, over 1,600 rural villages received an annual block grant during the first year. Each village can use the grant for any activity that supported one of 12 indicators of health and 6 education service delivery (such as prenatal and postnatal care, childbirth assisted by trained personnel, immunization, school enrollment, and school attendance). To give communities incentives to focus on the most effective policies, the government bases the size of the village‘s Generasi block grant for the subsequent year partly on the village‘s performance on each of the 12 targeted health and education indicators. The Generasi project thus takes the idea of performance incentives from conditional cash transfer programs and applies it in a way that allows communities the flexibility to address supply constraints, demand constraints, or some combination. To the best of our knowledge, the Generasi project is the first health and education program worldwide that combines community block grants with explicit performance bonuses for communities. To allow for a rigorous, randomized evaluation of Generasi, the government of Indonesia incorporated random assignment into the selection of Generasi locations. Unlike current evaluations of conditional cash transfer Box 1. Generasi Program Target Indicators programs, which cannot separately identify the impact of the incentives from the impact Health Indicators of the additional cash provided (Gertler 1. Four prenatal care visits 2004), the Generasi evaluation was designed 2. Taking iron tablets during pregnancy to separate out these two effects. Specifically, each Generasi location was further randomly 3. Delivery assisted by a trained professional allocated to one of two versions of the 4. Two postnatal care visits program: (1) an ―incentivized‖ treatment with 5. Complete childhood immunizations the pay-for-performance component 6. Adequate monthly weight increases for infants (treatment A) described above; and (2) an otherwise identical ―non-incentivized‖ 7. Monthly weighing for children under three and treatment without the pay-for-performance biannually for children under five incentives (treatment B). This study focuses 8. Vitamin A twice a year for children under five on the Generasi program. It describes the three waves of evaluation surveys conducted Education Indicators between 2007 and 2010. To the extent 9. Primary school enrollment of children possible, findings from the accompanying 6-to-12 years old qualitative study were also incorporated into 10. Minimum attendance rate of 85 percent for primary this report to provide greater understanding of school-aged children possible causal effects, processes, and villagers‘ perspectives. The authors will have 11. Junior secondary school enrollment of children 13- two other forthcoming papers discussing in to-15 years old greater detail the issues of incentives, cost- 12. Minimum attendance rate of 85 percent for junior effectiveness, and project operations. 2 secondary school-aged children 2 Olken, B.A., J. Onishi, and S. Wong. 2011. ―Should Aid Reward Performance? Evidence from a field experiment on health and education in Indonesia‖ (forthcoming), Also, ――A Community Approach to Achieving Health and Education Outcomes: A Pilot in Indonesia‖ (forthcoming) 7  1.2 The Generasi project This section describes the Generasi project, the Indonesian community block grant program that is the focus of this study. PNPM Generasi—known in full as the National Community Empowerment Program–Healthy and Smart Generation (Program National Pemberdayaan Masyarakat–Generasi Sehat dan Cerdas) — began in mid-2007 in rural areas of five Indonesian provinces selected by the government: West Java, East Java, North Sulawesi, Gorontalo, and Nusa Tenggara Timur.3 In 2007, the project covered 1,650 villages in 129 subdistricts, with a total budget of $20 million. In the project‘s second year, which began in mid - 2008, the project expanded to cover a total of 2,150 villages in 176 subdistricts, with a total budget of approximately $38 million. The third year, in 2009, was in 164 subdistricts, with a planned expansion to one other province, West Nusa Tenggara in 2010. The Generasi project focuses on 12 indicators of maternal and child health behavior and educational behavior (see Box 1). These indicators are in line with the Ministry of Health priorities and protocols and the government‘s constitutional obligation of ensuring nine years of basic education for all Indonesian children. These indicators were chosen by the government of Indonesia to be as similar as possible to the conditions for the individual household conditional cash transfer program being piloted at the same time as Generasi (but in different locations). They are in the same spirit as the conditions used by conditional cash transfer programs in other countries, such as Progresa in Mexico (Levy 2006). These 12 indicators respond to those seeking health and educational services that are within the direct control of villagers—such as the number of children who receive immunization, prenatal and postnatal care, and the number of children enrolled and attending school—rather than long-term outcomes, such as test scores or infant mortality. In Generasi, all participating villages receive a block grant each year to improve education and maternal and child health in their villages. These village block grants ranged from an average of $8,500 in 2007 up to an average of $18,200 in 2009. Block grants are usable for a wide variety of purposes, including, but not limited to, hiring extra midwives for the village, subsidizing the costs of prenatal and postnatal care, providing supplementary feeding, hiring extra teachers, opening a branch school in the village (kelas jauh or satellite classrooms, or sekolah terbuka or formal part-time junior secondary schooling), providing scholarships or school supplies, providing transportation funds for health care or school attendance, improving health or school buildings, or even rehabilitating a road to improve access to health and education facilities during the rainy season. To decide on the allocation of the funds within a village, trained facilitators help each village elect an eleven-member village management team, as well as select local facilitators and volunteers. Through social mapping and in-depth discussion groups, villagers identify problems and bottlenecks in reaching the 12 indicators. Inter-village meetings and consultation workshops with local health and education service providers allow community leaders to obtain information, technical assistance, and support from the local health and education offices as well as to 3 An initial test of the Generasi concept was run in three villages in Gorontalo province from 2006 to 2008. Those villages are not included in the main Generasi project or analysis. 8 Figure 1. Village funding allocations, 2007 and 2008 Education Activities 56% of Expenditures Financial Training & incentives for behavior change teachers (4%) communication (1%) Infrastruc ture (8%) Financial School assistance and materials, school fees equipment, (31%) and uniforms (56%) Source: Generasi Project management information system data coordinate the use of Generasi funds with other health and education interventions in the area. Following these discussions, the elected management team makes the final Generasi budget allocation. In 2007 and 2008, communities used the block grant funds for education (56 percent of expenditure) and health (44 percent) activities (see Figure 1). Communities chose to use most of the funds for ―individual goods‖ such as school materials, equipment and uniforms, and school financial assistance. On the health side, the majority of expenditures were used for supplementary feeding activities and financial assistance for pregnant mothers to use services (e.g., midwife fees, transport, etc.) Performance incentives are a critical (and unique) element of the Generasi approach. The size of a village‘s block grant depends in part on its performance on the 12 targeted indicators. The purpose of the performance bonus is to increase the village‘s effort at achieving the targeted indicators (Holmstrom 1979), both by encouraging a more effective allocation of Generasi funds and by stimulating village outreach efforts to encourage mothers and children to obtain appropriate health care and increase educational enrollment and attendance. 9  The performance bonus is structured as a relative competition among villages within the same subdistrict (kecamatan). By making the performance bonuses relative to other villages in the subdistrict, the government sought to minimize the impact of unobserved differences in the capabilities of different areas on the performance bonuses (Lazear and Rosen 1981; Mookherjee 1984; Gibbons and Murphy 1990). The fixed allocation to each subdistrict also ensures that the performance bonus system would not result in an unequal geographic distribution of funds.4 The specific rule for allocating Generasi funds to villages within the subdistrict is as follows. The size of overall Generasi allocation for the entire subdistrict is predetermined by the subdistrict‘s population and poverty level.5 Within a subdistrict, in year 1 of the project funds are divided among villages in proportion to the number of target beneficiaries in each village (that is, the number of children of varying ages and the expected number of pregnant women). Starting in year 2 of project implementation, 80 percent of the subdistrict‘s funds continue to be divided among villages in proportion to the number of target beneficiaries; the remaining 20 percent of the subdistrict‘s funds form a performance bonus pool, to be divided among villages based on their performance on the 12 Generasi indicators.6 The performance bonus pool is allocated to villages in proportion to a weighted sum of each village‘s performance above a predicted minimum achievement level. Specifically, each village‘s share of the performance bonus pool is determined by: ShareOfBonusv = Pv / (Σ Pj) where Pv=Σ[ wi  (yvi - mvi)] In this formula, yvi represents village v‘s performance on indicator i, wi represents the weight for indicator i, mvi represents the predicted minimum achievement level for village v and indicator i, and Pv is the total number of bonus ―points‖ earned by village v. The weights for each indicator, wi, are shown in Table 1, and were set by the government to be approximately proportional to the marginal cost of having an additional individual complete that indicator. Generasi uses performance relative to a constant predicted minimum attainment level, rather than 4 As discussed by Gibbons and Murphy (1990) and others, one potential pitfall of relative performance incentives is that agents may have an incentive to either sabotage or collude with other agents. With an average of 12 villages per subdistrict, in this case villages face a much greater return from increasing their own performance than from sabotaging that of other villagers. Nevertheless, this possibility remains, and therefore makes the equilibrium implications of the incentives an important empirical question. 5 In 2007 the average block grant for each subdistrict was $103,000 per subdistrict; in 2008, the average block grant was raised to $178,000 per subdistrict, and for 2009, it was $206,000 per subdistrict. A subdistrict contains roughly between 15,000 and 50,000 individuals and 10 to 20 villages. 6 Starting in year 2, for allocating the non-incentivized portion of the block grant (i.e., 80 percent of the subdistrict allocation in incentivized areas and 100 percent of the subdistrict allocation in non-incentivized areas), the number of target beneficiaries is weighted depending on a village‘s access to facilities. This calculation is identical in both incentivized and non -incentivized areas. 10 Table 1. Performance metrics and weights Weight per measured Potential times per Potential points per Performance metric achievement person per year person per year 1. Prenatal care visit 12 4 48 2. Iron tablets (30 pill packet) 7 3 21 3. Childbirth assisted by trained 100 1 100 professional 4. Postnatal care visit 25 2 50 5. Immunization 4 12 48 6. Monthly weight increases 4 12 48 7. Monthly weighing 2 12 24 8. Vitamin A pill 10 2 20 9. Primary enrollment 25 1 25 10. Monthly primary 2 12 24 attendance >= 85% 11. Middle school enrollment 50 1 50 12. Monthly middle school 5 12 60 attendance >= 85% Source: PNPM Generasi Operational Manual improvements over an actual baseline, to avoid the ratchet effect (Weitzman 1980); the minimums, mvi, are determined based on historical national datasets.7 As noted previously, two versions of the Generasi project are being run to separate the impact of the performance bonuses from the overall impact of having additional financial resources available for health and education: the program with performance bonuses described above (referred to as ―treatment A‖), and an identical program without performance bonuses (referred to as ―treatment B‖). Treatment B is identical to treatment A except that in treatme nt B, there is no performance bonus pool; instead, in all years, 100 percent of funds are divided among villages in proportion to the number of target beneficiaries in each village. In all other respects, the two versions of the program are identical. Even the village‘s annual points score Pv is also calculated in treatment B areas; the only difference is that in treatment B villages the points are used simply as an end-of-year monitoring and evaluation tool, and have no relationship to the allocation of funds. Within a given subdistrict, all villages participate in the same treatment of the program; that is, either all villages received treatment A, or all villages received treatment B. The Generasi project design builds on the Indonesian government‘s existing community- driven development program, known as the National Community Empowerment Program 7 For each of the 12 Generasi indicators i, the project set the predicted minimum attainment level, mvi, in village v to be equal to 70 percent of the average achievement level for villages with similar levels of access to health and education providers and numbers of beneficiaries. These minimum achievement levels were estimated by combining data on levels of each indicator from the 2004 SUSENAS household survey and 2003 PODES census of villages. For all health indicators except monthly weighing, access to providers was divided into three categories: (1) having a midwife practicing in the village, (2) not having a midwife in the village but having a midwife practicing within 4km from the center of the village, or (3) not having a midwife practicing within 4km of the village center. For middle school, access was divided into three categories: (1) having a middle school located in the village or within 4km of the village center, (2) having a middle school located between 5 and 9km of the village center, or (3) having a middle school located 10km or more from the village center. For monthly weighing and primary school, all villages were assumed to have the same level of access, since weighing of children is always conducted in the village at monthly posyandu meetings and since virtually all villages in Indonesia have a primary school. 11 (PNPM), which, along with its predecessor programs (Kecamatan Development Project), have funded over $2 billion in local infrastructure and microcredit programs in some 61,000 Indonesian villages over the past decade. The Generasi project is implemented by the government of Indonesia‘s Ministry of Home Affairs, and is funded through government of Indonesia resources and in part with loans from the World Bank and grants from several bilateral donors. Technical assistance and evaluations have been supported by a multi-donor trust fund with contributions from the World Bank, Netherlands Embassy, Australia, United Kingdom, and the Danish Embassy, and the World-Bank-managed Spanish Impact Evaluation Fund. 1.3 Experimental Design In order to evaluate the overall impact of Generasi, as well as to separately identify the impact of Generasi‘s performance incentives, Generasi locations were selected by lottery to form a randomized, controlled field experiment. The use of randomized evaluation techniques is considered the gold standard for impact evaluation of clinical and public health interventions (Gordis 2004), as well as development programs more generally (Duflo, Glennerster, and Kremer 2007). It has formed the basis of a number of high-profile social policy experiments in the United States (see Newhouse 1993; Kling, Liebman, and Katz 2007) and internationally (see Gertler 2004; Miguel and Kremer 2004; Schultz 2004; Skoufias 2005). The Generasi randomization was conducted at the subdistrict (kecamatan) level, so that all villages within the subdistrict either received the same treatment of Generasi (treatment A or treatment B) or were in the control group. Randomizing at the subdistrict level is important since many health and education services, such as community health centers (puskesmas) and junior secondary schools, provide services to multiple villages within a subdistrict. Increased demand for services from one village within a subdistrict could potentially therefore crowd out the services provided to other villages within the same subdistrict; alternatively, an effort by one village to improve service provision at the community health center could also benefit other villages in the same subdistrict. By randomizing at the subdistrict level, so that all villages in the subdistrict receive the same treatment status, the evaluation design ensures that we capture the total net effect of the program, since any within-subdistrict spillovers would also be captured in other treatment villages.8 This type of cluster-randomized design is common in program evaluations where there might be local spillovers from the treatment (Miguel and Kremer 2004; Olken 2007). The Generasi locations were selected through the following procedure. First, 300 target subdistricts were identified, targeting poor, rural areas that had an existing community-driven development infrastructure.9,10 Each subdistrict was then randomly assigned by computer into 8 Spillovers to other subdistricts are much less likely to be a problem, since the health service providers (Subdistrict Health Centers and midwifes), primary schools, and junior secondary schools that are the focus of this survey primarily provide services within a single subdistrict. Nevertheless, by using GIS information on the location of service providers, we will be able to test empirically for the presence of these cross-subdistrict spillovers. 9 To identify the 300 target subdistricts, we began by eliminating the wealthiest 20 percent of districts (kabupaten) within the five target provinces identified by the government, determined by the district‘s poverty rate, malnutrition rate, and junior secon dary school transition rate. Districts where the PNPM program was not scheduled to operate in 2007 were also ineligible. Twenty districts were randomly selected from the remaining eligible districts, stratified by island group. Within the 20 selected districts, 12 one of three equal-sized groups: treatment A, incentivized (100 subdistricts); treatment B, non- incentivized (100 subdistricts); or control (100 subdistricts). Within a subdistrict, all villages received the same treatment. The randomization was stratified by district (kabupaten), to ensure a balanced randomization across the 20 different districts in the study. The tests for balance confirm that the three groups of subdistricts appear similar on pre-period characteristics (World Bank 2008). Note that a total of 36 out of the 300 subdistricts should not have been included in the randomization, as they were ineligible for Generasi because they had been selected (prior to the randomization) to receive other programs or had had prior implementation problems with previous PNPM programs. Since the eligibility decision was made on the basis of lists determined prior to the randomization, and since we obtained those lists for treatment and control areas, we excluded ineligible subdistricts in both treatment and control groups from our main analysis.11 The Generasi program was phased in over two years. In phasing in the program in the first year (2007), the government for budgetary reasons prioritized those locations that had previously participated in the PNPM rural infrastructure program (denoted group P), since those locations already had the legal infrastructure for distributing PNPM funds and it was easier to re- budget other monies to fund Generasi in those areas. After all group P subdistricts randomized to receive the program had been funded, the government held another lottery to select which group NP subdistricts would begin receiving the program in 2007 and which would begin in 2008.12 By year two of the program (2008), 95 percent of eligible subdistricts—172 out of the 181 eligible subdistricts randomized to receive Generasi—were receiving the program. Of the remaining nine subdistricts, seven received the regular PNPM program instead of Generasi and two received Generasi in year one only.13 Since the randomization results were followed 99 percent of the time in Wave I and 95 percent of the time in Wave II, in the analysis below, we use the original randomization results all 264 eligible subdistricts as the basis of the analysis, and interpret the results as intent-to-treat estimates (Imbens and Angrist, 1994). An important consideration for the analysis is the potential for differential provision of other programs in the pure control groups. The main potential avenue through which this might subdistricts were eligible for Generasi if they had previously received the PNPM program or were considered less than 67 percent urban by the Central Statistics Office. 10 Since Generasi is implemented through the national PNPM program, it could only be implemented in districts that were already included in the PNPM program. Prior experience with PNPM at the subdistrict level also simplified Generasi implementation, since the relevant legal structures for disbursing Generasi funds had already been established in these locations. 11 The determination that these subdisticts would be ineligible had been made prior to the randomization, but was not communicated to the study team, which is why they were included in the randomization. Subdistricts were deemed ineligible if they had been allocated to receive the urban poverty program (UPP), conflict area poverty program (SPADA), or if they had had a previous problem with PNPM implementation. We subsequently obtained the pre-randomization lists used to make this determination, and use these pre-randomization lists to restrict our sample (in both treatment and control areas) to those subdistricts that would actually be eligible for the program. Nevertheless, data collection surveys were conducted in all 300 subdistricts that were initially included in the randomization, regardless of the final eligibility, so as a robustness check we can alternatively estimate intent-to-treat effects using the full 300 subdistricts from the original randomization. 12 Specifically, in 2007 all 105 eligible group P subdistricts were funded. In group NP, in 2007 Generasi was funded in 22 eligible subdistricts. Of these 22 subdistricts, 21 were chosen randomly by computer, stratified by province, in a second lottery among Group NP locations. Group P status was determined prior to randomization. All but 7 of the remaining NP subdistricts were added in 2008. 13 We do not know why these 7 subdistricts received regular PNPM rather than Generasi. We therefore include them in the treatment group as if they had received the program, and interpret the resulting estimates as intent-to-treat estimates. Likewise, we include the 2 subdistricts that received the program in year 1 but not in year 2 in the treatment group throughout. 13 occur is other PNPM programs. Specifically, to ensure a fair allocation of funds, the Ministry of Home Affairs decided that no subdistrict would receive both the Generasi program and other PNPM programs, which typically fund local infrastructure (roads, bridges, etc.) and microcredit. In 2007, 17 (out of 83) eligible control subdistricts received other PNPM programs, as did two treatment subdistricts in the non-priority (NP) areas; in 2008, 31 (out of 83) eligible control subdistricts received other PNPM programs, as did the seven eligible subdistricts that should have been receiving Generasi in 2008 but received regular PNPM-rural instead. Since regular PNPM programs tend to focus on basic infrastructure, not health and education, it is unlikely that the differential provision of other PNPM programs in control areas will have substantial impacts on the Generasi evaluation results. Nevertheless, in interpreting the results, it is important to recognize that some portion of the eligible ―pure control‖ subdistricts received PNPM. 1.4 Survey Design and Implementation The main data for the impact analysis is from a set of surveys of households, village officials, health service providers, and school officials. A detailed list of the contents of each survey module, as well as the sample size for each module, can be found in Table 2. Three waves of the survey were planned as part of the evaluation series. Wave I, the baseline round, was conducted from June to August 2007 prior to Generasi implementation. Wave II, the first follow-up survey round, was conducted from October to December 2008. Wave III, a longer term follow-up survey round, was conducted from October 2009 to January 2010. These surveys were designed by the World Bank and the government of Indonesia and were conducted by the Center for Population and Policy Studies of the University of Gadjah Mada, Yogyakarta, Indonesia. The final evaluation is based on data collected through the Wave I, Wave II, and Wave III surveys, which were funded by the World Bank (through the Decentralization Support Facility, PNPM multidonor trust fund (PSF), and the WB-managed Spanish Impact Evaluation Fund). The sample for the surveys covers each of the 300 subdistricts that were included in the original Generasi randomization. In each subdistrict, eight villages were randomly selected (unless the subdistrict had fewer than eight villages, in which case all were selected). This resulted in a total of 2,313 villages sampled in each of the three survey waves. The sampling design for the household component of the Generasi surveys was chosen to ensure adequate coverage in the key Generasi demographic groups: mothers who recently were pregnant or gave birth, children under age 3, and children of school age. Within each village, one hamlet (dusun) was randomly selected, and a list of all households was obtained from the head of the hamlet. Five households were randomly sampled from that list to be interviewed. These households were stratified so that two selected households had at least one child under age 2, two selected households had a child under age 15 but no children under age 2, and one household had no children under age 15. For some of the analysis (e.g., for examining how the incentives affect the differential targeting of Generasi benefits and increments in service provision), it is useful to have baseline and follow-up characteristics for the same individuals. Therefore, in the follow-up surveys, in 14 Figure 2. Timeline of project and surveys Source: authors half of the randomly selected villages (four villages out of the eight villages sampled in every subdistrict), the same households sampled in Wave I were contacted again in subsequent waves to form an individual level panel. Teams tracked and re-interviewed migrated or split households who provided information for any of the married women or children modules, as long as they were within the same subdistrict. In panel areas, 95 percent of target households were able to be re-interviewed in Wave 2 and 98 percent of target households were able to be re-interviewed in Wave 3. In the other half of villages, a new cross-section of households was drawn in each survey wave. The combination of panel households and non-panel households allows us to investigate heterogeneous treatment effects based on pre-period income levels and other characteristics, while at the same time ensuring that sufficient respondents with recent births and young children are enrolled in the survey sample in every round. Health facilities and schools were also contacted again to form a panel. For midwives, a randomly selected 75 percent of the midwife sample was re-contacted to form a panel, and 25 percent of the midwives were newly sampled in each wave to ensure the sample captures potential in-migration of midwives in response to Generasi. 15  Data from these surveys were supplemented with detailed administrative data from the Generasi project‘s internal management information system. This included detailed budget allocations for the block grants, performance data on the 12 Generasi indicators, and data on participation levels in Generasi village meetings. Additionally, the Indonesian research organization, SMERU, conducted an accompanying qualitative study in 12 villages to probe deeper into the ―whys and hows‖ of community and service provider processes and motivations. This qualitative work allowed the study team to explore in greater depth some of the decision- making and implementation issues behind the quantitative results. Table 2. Questionnaire modules and sample size Module Contents Sample Panel/Non- Size Panel (Wave (Waves II/III) III) Household core Household roster, deaths in previous 12 months, migration, 12,306 (Respondent: water/sanitation, receipt of government poverty programs, female household participation in non-formal education, consumption, assets, head or spouse of a economic shocks, health insurance, morbidity, outpatient care male household use, social capital, knowledge and participation in head) PNPM/KDP activities Married women Fertility history, use of health services during pregnancy, 11,140 age 16–49 inspection of Generasi coupons (Wave II), family planning, health and education knowledge Children age 6–15 School enrollment, attendance, grade repetition, cost of 9,779 (Respondent: schooling, scholarships, child labor mother of the child) 50% panel, Children age < 3 Growth monitoring (posyandu), immunization records, 6,708 50% (Respondent: inspection of the Generasi coupons (Wave II), motor non-panel mother of the child) development (Wave III), breastfeeding and nutritional intake, weight measurement, height measurement (Waves I & III) Home-based tests Test of math and reading skills administered at home 7,687 for (Respondent: (separate test for ages 6–12 and ages 13–15) math and children age 6–15) (Waves I & III) reading tests, 7,336 for anthro- pometric tests Village Demography of the village, hamlet information, access to 2,315 100% panel characteristics health services and schools, economic shocks, access to (Respondent: media, community participation, daily laborer wage rate, Village Head) development projects in the village (Waves II & III) Community health Head of facility background, coverage area, budget, staff 300 100% panel center roster, time allocation of head doctor and midwife (Puskesmas) coordinator, service hours, services provided, fee schedule, number of patients per service during the previous month, medical and vaccine stock, data on village health post, participation in Generasi (Waves II & III), direct observation regarding cleanliness 16  Module Contents Sample Panel/Non- Size Panel (Wave (Waves II/III) III) Village midwives Personal background, location of duty, condition of facility, 1,177 75% panel, 25% time allocation, income, services provided, fee schedule non-panel (public and private), experiences during past three deliveries, number of patients seen per service during the previous month, equipment and tools, medical supplies and stock, village health post management, participation in Generasi (Waves II & III), structure of subsidies received Principal background, principal time allocation, teacher 1,197 50% panel 50% Primary school roster, school facilities, teaching hours, enrollment records, non-panel (Waves II & III) attendance records, official test scores, scholarships, fees, budget, participation in Generasi (Waves II&III), direct observation of classrooms, including random check on classroom attendance Junior secondary Same questionnaire as for primary school 867 66% panel, 33% school non-panel Village health post Respondent characteristics, health post characteristics, service 50% panel 50% cadre providers, cadre roster, tools and equipment, participation in 2,397 non-panel (Waves II & III) Generasi Sources: Survey questionnaires, UGM reports for Waves I, II and III surveys. Over 45,000 household members, village heads, and school and health facility staff were respondents for the Generasi final Wave III survey. 17 2 EVALUATION METHODOLOGY 2.1 Regression Specifications Since the Generasi program was designed as a randomized experiment, the evaluation is econometrically straightforward: essentially, we compare outcomes in those subdistricts randomized to be treatments with those subdistricts randomized to be control areas, controlling for the level of the outcome at baseline. In implementing our analysis, we restrict attention to the 264 ―eligible‖ subdistricts, as discussed in Section 1.3 above, and use the randomization results combined with the government‘s prioritization rule to construct our treatment variables. Specifically, analyzing Wave II data (corresponding to the first treatment year), we define the GENERASI variable to be a dummy that takes value 1 if the subdistrict was randomized to receive GENERASI and either (a) it was in the priority area (group P), or (b) was in the non-priority area and selected in the additional lottery to receive the program in 2007. In analyzing Wave III data, we define the GENERASI variable to be a dummy that takes value 1 if the subdistrict was randomized to receive Generasi. We define the GENERASI_INCENTIVES variable to be a dummy that takes value 1 if the GENERASI variable is 1 and if the subdistrict was randomized to be in the incentivized version of the program. GENERASI_INCENTIVES thus captures the additional effect of the incentives above and beyond the main effect of having the program, and is the key variable of interest in the paper. Note that by defining the variables in this way, we are exploiting only the variation in program exposure due to the lottery. These variables capture the intent-to- treat effect of the program, and since the lottery results were very closely followed —they predict true program implementation in 99 percent of subdistricts in 2007 and 95 percent of subdistricts in 2008—they will be very close to the true effect of the treatment on the treated (Imbens and Angrist, 1994). In running the regressions, we take advantage of the baseline data by controlling for the average level of the outcome variable in the subdistrict in the baseline survey. Since we also have individual-specific panel data for half our sample, we include the pre-period value for those who have it, as well as a dummy variable that corresponds to having non-missing pre-period values. Since households came from one of three different samples (those with a child under age 2, those with a child age 2–15 but not in the first group, and all others), we include dummies for those three sample types, interacted with whether a household came from a panel or non-panel village. Finally, since many of the indicators for children vary naturally as the child ages, for all child-level variables we include age dummies. To examine the overall impact of Generasi treatment, for each indicator of interest, we estimate the following regressions on the 264 subdistricts that remain after we drop the list of ineligible subdistricts: 18 Wave II data: y pdsi2  αd  β1GENERASI ds2  γ1 y pdsi1  γ21