from EVIDENCE to POLICY Learning what works, from the Human Development Network February, 2013 75655 Do Grants to Communities Lead to Better Health and Education? Conditional cash transfer programs have proven to be an effec- working towards meeting the United Nations Millennium tive tool for helping reduce poverty. As use of such programs Development Goals. In Indonesia, the World Bank worked grows, development experts and policymakers are considering with the Government of Indonesia on a community grant how to successfully promote better health and education in program to boost the use of health and education services. large countries with diverse and often remote communities. The impact evaluation built into the program found that In some cases, they are rethinking traditional cash transfer cash transfers to rural communities led to positive impacts programs to create grant programs that target communities on average across health and education indicators, with a instead of families. One question is whether linking grants to strong decline in malnutrition. Communities whose grants performance, similar to conditional cash transfer programs, were linked to performance-based incentives did even better can promote better results than giving grants that are not than those whose grants weren’t linked to incentives. This Social Protection performance-tied. suggests that conditioning grants can produce positive re- The World Bank is at the forefront of efforts to reduce sults. Because of the measured impacts, the Government poverty and create shared prosperity. As part of this, the of Indonesia is expanding the community grant program World Bank is helping countries create and implement pro- to eight provinces from five and focusing attention on grams to reduce maternal and child mortality and increase combating malnutrition. And a portion of all grants will school enrollment, both of which are crucial for countries now be based on performance. Context Indonesia, like many middle income countries, has difficulty geted communities. In both cases, the transfers were designed to providing universal access to education and adequate access to encourage families to meet basic health and education indicators, healthcare, particularly in poor and rural areas. While nation- including prenatal visits for pregnant women, childhood immu- al government programs have been effective in raising prima- nization, regular weight monitoring, and school attendance. ry school enrollment, other health and education indicators The first, called the Hopeful Family Program, known as still lag, and huge geographical disparities remain. In 2007, PKH, provides conditional cash transfers to extremely poor the year that the government embarked on its sweeping anti- households with children or pregnant women. Households in poverty programs, maternal mortality was 228 per 100,000 the program have shown significant positive gains in multiple live births, among the worst in Southeast Asia, while close to health areas, such as significant increases in pre-natal, post- 20 percent of children under the age of five were underweight. natal, and delivery care; in vaccinations and treatments for Other surveys found that enrollment rates dropped precipi- diarrhea; and in child growth monitoring and health service tously between primary and middle schools, going from 94 provider visits in general. While the effect on education has percent to 65 percent. been more muted, the government is expanding the program To tackle these problems, the Government of Indonesia from 810,000 households in 2010 (itself an increase from launched two large-scale programs in 2007. The programs both 432,000 households when the program was launched) to 3 relied on cash transfers, but one targeted households and one tar- million by 2014.* The second poverty alleviation program, known as the Na- To push communities to focus on the most effective policies, tional Community Empowerment Program—Healthy and a portion of subsequent year grants is based on how well com- Smart Generation, or PNPM Generasi, gives block grants to munities do in meeting the previous year’s health and education poor, rural communities. The program takes the idea of condi- targets. In this way, the program takes aspects of conditional tional cash transfers and redesigns it to enable communities and cash transfer and pay-for-performance programs and reformu- local health and education providers to work together to decide lates them to encourage community-wide performance and ac- what needs to be done to increase schooling and use of critical countability. In order to test the effectiveness of linking grants health services. Trained facilitators help communities decide to the previous year’s performance, a second version of the pro- how to best address bottlenecks using the community grants gram was carried out in which communities received the money from the project to improve the targets indicators. Communi- irrespective of the previous year’s performance. The grants have ties can focus on stimulating demand by giving people cash ranged from an average of $8,500 in 2007 to $18,200 in 2009. transfers or scholarships to use certain services, or they can fo- This World Bank-supported program now reaches about 5.4 cus on supply problems that might be limiting access, such million people. as too few health clinic workers or overcrowded classrooms. *For further information see: “Program Keluarga Harapan, Main Findings from the Impact Evaluation of Indonesia’s Pilot Household Conditional Cash Transfer Social Protection Program�, World Bank Office Jakarta, June 2011 Evaluation The World Bank-supported Generasi program included a built- linked block grant group. With over 2,100 villages randomized in impact evaluation component partially funded through the to receive either the incentivized or non-incentivized version of SIEF trust fund. The Generasi program was initially focused on the Generasi program (plus some 1,000 villages in control sub- rural areas in five provinces chosen by the government. After re- districts), and over 1.8 million target beneficiaries in treatment searchers eliminated the wealthiest 20 percent of districts, based areas, to the best of our knowledge, this represents one of the on malnutrition rates, school transition rates and poverty rates, largest randomized social experiments conducted in the world 20 districts were randomly picked from the remaining eligible to date. districts. Within these, 300 target subdistricts were picked based Surveys were conducted at baseline, prior to the program on their eligibility for a previous anti-poverty program or be- being implemented after August 2007; one year into the pro- cause they were classified as less than 67 percent urban. Using a gram in October through December 2008, and two years into lottery system, these were equally divided between the control the program in October through December 2008. Resurvey group, the performance-linked block grant group and the un- rates were 95 percent and better. The Findings Giving communities block grants to help them cially in the second year of the program. The program showed devise programs to boost use of basic health a statistically significant impact across all 12 indicators when services and increase school enrollment and compared with the control group. Changes included a 6.8 per- attendance works. cent increase in weight checks for children, and a 4.7 percent increase in iron supplements given to pregnant women. Overall, the Generasi program improved use of crucial health Likewise, the program overall boosted primary school at- services and boosted school enrollment and attendance, espe- tendance by 0.8 percentage points, raising total enrollment This policy note is based the World Bank report: “Indonesia’s PNP Generasi Program: Final Impact Evaluation Report,� June 2011, Benjamin A. Olken (M.I.T.), Junko Onishi (World Bank), Susan Wong (World Bank) for ages 7-12 to 98.5 percent, which is near-universal enrollment. for infrastructure and equipment, 4 percent for health worker When the impact on school enrollment and attendance was bro- incentives and 3 percent on training. ken down by poverty levels, the block grants showed their biggest impact on families in the bottom 40 percent income levels. Among Linking the amount of the block grant to how well these households, there was an overall increase of 2 percentage communities did in meeting health and education points for school enrollment and attendance for children ages 7-12, goals in the previous year did raise health results. and a 7.5 percentage point increase for children aged 13-15. But basing a portion of the grants on performance did not lead to any effects on communities meeting The program’s biggest effect was in cutting malnu- education targets. trition and stunting. When it came to meeting health indicators, communities whose Childhood malnutrition was reduced by 2.2 percentage points, grants were tied to performance generally outperformed communi- or nearly 10 percent over the control group. This was particularly ties whose grants were not linked to performance. Prenatal visits strong in areas where malnutrition was a bigger problem, such as the Nusa Tenggara Timur Province, where severe underweight rates dropped by 33 percent, or 5.5 percentage points, and severe stunting declined by 21 percent, or 6.6 percentage points. In addition, in the first year, there was a drop in neonatal deaths, but this didn’t carry over into the second year. The reason may be that the rate dropped sufficiently that it was difficult to bring it down any further. Although enrollment and attendance rose in many communities that received block grants, there was no corresponding improvement in learning. The evaluation reviewed math and Indonesian tests given to children aged 7-12 and those aged 13-15. There was no measurable impact on test scores, which either means that the project didn’t improve student learning, or that the tests weren’t structured in a way that could capture that achieve- ment. Another possibility is that the program wasn’t under- way long enough to result in better test scores. were 5 percent higher in communities where the money was partly allocated based on performance, compared with the other commu- Communities spent the majority of the grants on nities, and immunization rates were 3 percent higher. education, and most of the money went for But this was only the case in meeting health targets. There was individual assistance, such as school supplies and no gain seen in reaching education indicators. Researchers suggest uniforms. a number of reasons why the incentive-linked grants didn’t work for education targets: among them, health baseline figures were lower The majority of spending, 56 percent, went to education, with a than education ones, making it easier to improve and reaching edu- majority of that money being spent on school materials, supplies cation indicators requires the involvement of more people, from and uniforms. Another 31 percent was used for financial assistance teachers, to parents, to community officials and students, making it for families, 8 percent went to infrastructure, 4 percent on teacher more challenging to improve. incentives and 1 percent on training. Forty-four percent of the grant went to heath activities, with Generasi’s structure pushed communities to develop 41 percent of that used for supplementary feeding activities solutions together, which may be one reason why (such as fortified snacks for children), 27 percent for financial villages that received the grants did show increased assistance for pregnant women to use health services, 26 percent community effort and participation in other ways. Beneficiary villages had a 6.6. percent increase in the number of The program’s biggest impact was in communities health volunteers supporting the health providers. There also was with the most need. some increase in the number of high school (junior second- ary school) parent-teacher committee meetings although In areas where health and education indicators were at the lowest not for primary school. Spillover effects included an average —10th percentile of service provision at baseline—the program increase of 2.7 hours per household—or 7.2 percent—in on average was twice as effective, possibly because there was more the time women spent in women’s community groups. room for improvement. Conclusion Making policy from evidence Indonesia’s decision to give communities block grants to For policymakers and development experts considering improve basic health and education services has proven to or designing similar programs, some issues to consider are: be an effective tool for working in areas where use of ser- • How can block grant programs be structured to ensure vices is constrained not just by demand, but also by supply that impacts are seen in health and education indicators? and access. By linking part of the grant to performance, the • Targets should be regularly adjusted to reflect devel- Generasi program attempted to replicate the conditional- opment priorities and realized gains. ity of cash transfers on a community-wide levels. The gains seen in use of health services in communities that received Generasi Program Target Indicators a portion of the money based on performance show that Health Indicators 1. Four prenatal care visits the positive effects of individual conditional cash transfers, 2. Taking iron tablets during pregnancy at least for some indicators, can be replicated when grants 3. Delivery assisted by a trained professional are community- wide. Indonesia has recognized the success 4. Two postnatal care visits 5. Complete childhood immunizations of the program and is expanding it further. Later results 6. Adequate monthly weight increases for infants from the traditional conditional cash transfer PKH program 7. Monthly weighing for children under three and biannually for children under five will help us learn more about the use of such programs on 8. Vitamin A twice a year for children under five the individual versus community level. For policymakers Education Indicators and development experts considering using block grants to 9. Primary school enrollment of children 6-to-12 years old communities to increase use of basic health-care services and 10. Minimum attendance rate of 85 percent for primary school-aged children encourage school enrollment and attendance, this evalua- 11. Junior secondary school enrollment of children 13-to-15 years old tion shows that in some cases, linking community money to 12. Minimum attendance rate of 85 percent for junior secondary school- performance can work. aged children The Human Development Network, part of the World Bank Group, supports and disseminates research evaluating the impact of development projects to help alleviate poverty. The goal is to collect and build empirical evidence that can help governments and development organizations design and implement the most appropriate and effective policies for better educational, health and job opportunities for people in developing countries. For more information about who we are and what we do, go to: http://www.worldbank.org/sief. This Evidence to Policy note series is produced by SIEF with generous support from DFID. THE WORLD BANK, HUMAN DEVELOPMENT NETWORK 1818 H STREET, NW WASHINGTON, DC 20433 Produced by Office of the Chief Economist, Human Development Network, Communications/Aliza Marcus