Does International Child Sponsorship Work? A Six-Country Study of Impacts on Adult Life Outcomes Bruce Wydick, Paul Glewwe, Laine Rutledge, 2013 Outline • Background • Study objectives • Related literature • Data and Methodology • Results and Discussions • Conclusions and Implications • Comparison with other papers Background • International child sponsorship represents the most intimate and direct form of involvement of the wealthy with the poor . Background • Estimated current private financial flows to internationally sponsored children exceeds US$3 billion • Compassion International is a leading child sponsorship organization (3rd largest globally) currently serving 1.3 million children in 26 countries Background • Compassion program • benefits sponsored children receive vary somewhat by country and even within countries • these programs are different from the standard Conditional Cash Transfer programs • children receive most of the benefits provided Background • Compassion program • selection of children for sponsorship is done locally based on the following criteria Study objective Research Question/objective • Exploring the impact of child sponsorship on the educational and employment/leadership outcomes of sponsored children Related Literature Literature • Limited research of impact of sponsorship programs • Kremer, Moulin, and Namunyu (2003), use a randomized experiment to analyze the impacts of a Dutch child sponsorship program that funded new classroom construction and provided students a $6 uniform and $3.44 worth of textbooks. They find that student beneficiaries attend school a half year longer and advance to a third of a grade farther in formal education. • Most existing data is on the impacts of conditional cash transfers • Oportunidades program in Mexico usually evaluated • Behrman, Parker, and Todd (2007) estimate that receiving Oportunidades cash transfers for 5.5 years increased grades completed by 0.8–1.0 year • Schultz (2004) estimates that Oportunidades increased formal schooling by 0.66 years 0.72 for girls and 0.64 for boys Data and Methodology Data • Survey from June 2008 to August 2010 • Study Counties Uganda Guatemala Kenya Bolivia India Philippines • Sample 1. formerly sponsored children (1,860) 2. unsponsored siblings of formerly sponsored children(3704 ) 3. individuals from nonparticipating families in program villages (2,136) 4. individuals from similar, nearby non-program villages (2,444) Selection criteria for study • For larger projects, individuals randomly selected to be surveyed from the first 2 or 3 years of enrollment lists. • For smaller projects, data collected from all individuals enrolled in the first 2 or 3 years • Local assistants hired to assist in tracing the households of formerly sponsored individuals who were on the early enrollment lists the sample includes both children who were sponsored for many years and children who dropped out relatively early to avoid bias. Key outcomes Education outcomes Employment and leadership outcomes • Total years of education • Primary school completion • Secondary school completion • Tertiary completion • Formal employment • White collar employment • Community leadership • Church leadership Sample summary Expected estimation issues • Likelihood of spill over of program effects to siblings and non-treated children in both program and non-program villages • More needy children likely to be chosen. • Children more likely to succeed may be chosen Cont’d • The study addressed the bias by using a strict selection criteria which states that 1. sponsored children must be 12 or younger 2. maximum number of children per household between 1 and 3 3. sponsored children must reside within a 30 minute walking distance of program center • The study performs the analysis using data of children (now adults) who were beneficiaries of the Compassion International program against their siblings, peers in project communities and peers outside project communities Methods Difference in difference Estimation techniques • OLS / OLS-FE estimation • IV-GMM / IV-GMM-FE Fixed effects also used to control for unobserved inter household characteristics that can affect selection Instruments used Interactions between dummies of Child age at program and sibling order relative to program start Main equation where yi is the adult outcome of interest for person i, Ch is a dummy variable indicating a household with a sponsored child, Cv is a dummy variable indicating residence in a village with the Compassion program, and X i is a vector of controls that include gender, age, age squared, birth order, number of siblings in a family, and mother’s and father’s education Cont’d IV justification • vector of instruments comprising interactions between dummy variables for a child’s age at program introduction (ACI) and dummy variables for sibling order relative to program roll-out (SORR) • dummy variables have strong predictive power for the children chosen by parents for the program because of parents’ tendency to choose the oldest age-eligible siblings for sponsorship • satisfy the exclusion restriction. Authors see no reason why, after controlling for characteristic variables, a child’s age at the time of program rollout interacted with SORR should affect adult life outcomes except through its effect on the probability of sponsorship Estimation equations • IV fixed effects • First stage • IV-GMM-FE • First stage • Second stage • Second stage OLS FIXED EFFECTS program impact • With Spillover assumptions • Intra-household spillovers • Assuming no spillovers • Intra-village spillovers program impact with intrahousehold spillovers program impact- no spillovers Results and discussion Summary statistics Difference-in-difference table (education outcomes) There exists significant differences in educational outcomes between sponsored children and their unsponsored siblings as well as all non-sponsored individuals in general Difference-in-difference table (employment and leadership outcomes) There exists significant differences in employment and leadership outcomes between sponsored children and their unsponsored siblings as well as all nonsponsored individuals in general Education outcomes Education outcomes Education outcomes The Compassion child sponsorship program significantly • increases years of completed schooling by 1.03–1.46 years over a baseline of 10.19 years • increases the probability of primary school completion by 4.0–7.7 percentage points (baseline 88.7 percent), • increases secondary school completion by 11.6–16.5 percentage points (baseline 44.9 percent), • Increases university completion by 2.1–2.4 percentage points (baseline 4.3 percent). • Results show little to no significant effect of intra family and intra village spillovers Employment and leadership outcomes Employment and leadership outcomes The Compassion child sponsorship program significantly Employment and leadership outcomes • Increases the probability of formal employment in adulthood by 5.1–6.3 percentage points • Increases the probability of white collar employment in adulthood by 6.5–6.7 percentage points • Increases the probability of community leader position in adulthood by 1–2.2 percentage points (OLS estimates) • Increases the probability of church leader position in adulthood by 3.5–6.03 percentage points (OLS estimates). IV estimates show negative outcomes (largely insignificant) Country estimates – educational outcomes Results for educational outcomes for individual countries show that program effects are larger for all outcomes in Uganda for both OLS & OLS household fixed effects Country estimates – years of schooling by gender Results show that program effects in terms of years of schooling are larger for the gender with the lower baseline value. program effects are higher for boys in Philippines and India with lower baseline values. In Uganda, Guatemala, and Bolivia where girls have lower baseline years of schooling, program effect is higher for girls. Robustness checks • for the education and employment outcomes, estimates that limit the sample to those over age 25 yield similar results and significance • Results from using different instruments yield nearly identical estimates for the educational outcomes. • Estimates for the education and employment variables on sponsored children who had no older siblings yield coefficients that are generally similar to those in the data. Conclusions and Implications Conclusions & implications • Compassion International sponsorship program has significant implications on the adult educational and employment/leadership outcomes of sponsored children. • increases secondary school completion by 11.6–16.5 percentage points • increases years of completed schooling by 1.03–1.46 years over a baseline of 10.19 years • Increases the probability of formal employment in adulthood by 5.1–6.3 percentage points • Intra-household and intra-village spill overs are generally statistically significant. Conclusions & implications • Generalization? • Different sponsorship programs have different targets hence the results from this study cannot be generalized for all sponsorship programs • Policy? • Sponsorship seems to work but with no spill-over effects, can government implement this? Comparison with other papers 2. INCENTIVES TO LEARN Michael Kremer, Edward Miguel, and Rebecca Thornton (2009) Secondary Papers 3. Role of cash in conditional cash transfer programs for child health, growth, and development: an analysis of Mexico’s Oportunidades Lia C H Fernald, Paul J Gertler, Lynnette M Neufeld (2008) INCENTIVES TO LEARN Michael Kremer, Edward Miguel, and Rebecca Thornton Summary background • Girls’ Scholarship Program (GSP) carried out by a Dutch nongovernmental organization (NGO) ICS Africa, in two rural Kenyan districts, Busia and Teso. • NGO awarded scholarships to highest-scoring 15% of grade 6 girls in program schools within each district • Scholarship winners selected based on their total tests score on districtwide exams Study objectives • To estimate the impact of a merit scholarship program for girls in Kenyan primary schools Data and Methodology Data • Randomized control trial of over 5,000 students • Two main program districts • Test score data obtained from the District Education Offices (DEO) in each program district • school participation data based on unannounced checks-one in September or October 2001 and one in each of the three terms of the 2002 academic year. • Survey data collected in 2002 Data and Methodology Methods • Ordinary Least squares (Differences in differences) • Empirical strategy • Reduced form equation • β1 captures the average program impact on the population targeted Main Findings positive program impacts on academic performance: girls eligible for scholarships in program schools had significantly higher test scores than comparison schoolgirls Teacher attendance improved significantly in program schools positive program externalities among girls with low pretest scores, who were unlikely to win in both districts boys in one district (Busia) experienced significant test score gains even though they were ineligible for the scholarship Role of cash in conditional cash transfer programs for child health, growth, and development: an analysis of Mexico’s Oportunidades Lia C H Fernald, Paul J Gertler, Lynnette M Neufeld Summary background • Over 200 million children under less than 5 years are not fulfilling their potential for growth, cognition, or socioemotional development • conditional cash transfer (CCT) programmes is a means to address the larger issue of poverty alleviation by developing country governments • CCT programs are aimed at improving poor families lives through interventions in health, nutrition, and education • Families enrolled in CCT programmes receive cash in exchange for complying with certain conditions Study objectives • to examine the impact of cash transfers on child health, growth, and development outcomes in Mexico's conditional cash transfer program, known as Oportunidades • determine whether receiving more money (higher cumulative transfers) in Oportunidades was associated with improvements in child growth, health, and development outcomes. Data Quasi-experimental design Data and Methodology Survey data from 2 periods- 1997 and 2003 Sample restricted to children who had been Oportunidades beneficiaries their whole lives. Analysis not based on beneficiaries vs non-beneficiaries but rather relationship between transfers accumulated and child outcomes Data and Methodology Methods • Ordinary least squares • Logistic regression • controlled directly for the number of household members and proportion of children in each age category at baseline • controlled for a wide range of household-level and community-level variables Main Findings A doubling ofcumulative cash transfers resulted in a statistically significant increase in heightfor-age Z score A doubling of cumulative cash transfers resulted in a statistically significant reduction in stunting prevalence as well as being overweight A doubling of cumulative cash transfers resulted in a statistically significant improvements in endurance, long and short-term memory, visual integration and language development There was no association between increased cash transfers and the number of sick days in the 4 weeks before the survey, with the skill component of motor development, or with haemoglobin concentration. INCENTIVES TO LEAR N ROLE OF CASH IN CONDITIONAL CASH TRANSFER PROGRAMS FOR CHILD HEALTH, GROWTH, AND DEVELOPMENT PAPER DOES INTERNATIONAL CHILD SPONSORSHIP WORK? Program Compassion International Girls’ Scholarship program Program (GSP) Oportunidades CCT prog ram Focus School materials, health care Merit scholarships for program schools Cash transfers to poor families Experiment type Randomized control trial Randomized control trial Quasi-experimental design PAPER DOES INTERNATIONAL CHILD SPONSORSHIP WORK? INCENTIVES TO LEAR N Methods DiD (OLS, OLS-fixed effect DiD (OLS) IV-GMM, IV-GMM fixed effects) ROLE OF CASH IN CONDITIONAL CASH TRANSFER PROGRAMS FOR CHILD HEALTH, GROWTH, AND DEVELOPMENT OLS Logistic regression Analysis focus Beneficiaries vs non beneficiaries Program schools vs non program schools Focuses on transfers accumuulated Spillovers Significant spillover effects Not covered Largely no spillover effects