Developing Educational and Vocational Aspirations through International Child Sponsorship: Evidence from Kenya, Indonesia, and Mexico
Interactive Results — Ross, Glewwe, Prudencio & Wydick (2021)
Research Question
Can child sponsorship change how children see themselves, and their future?
Compassion International's program in Kenya, Indonesia, and Mexico, 2,022 children, causal identification, three psychological outcomes, and schooling attainment.
The Question
Does international child sponsorship change how children see themselves, and their future?
Compassion International, the world's largest child sponsorship organization reaching over 2 million children across 29 countries, pairs material inputs (school fees, meals, uniforms, healthcare) with personal mentoring and regular letters from sponsors. We test whether this psycho-social investment drives outcomes independently from the material support.
Why It Matters
Poverty constrains not just resources, it constrains aspirations.
If low self-esteem and low aspirations cause children to under-invest in education, then programs that raise aspirations can break the poverty trap in ways material aid alone cannot. This is the poverty-of-mind hypothesis, and it is testable.
How We Know
A natural experiment using Compassion International's age-eligibility rule.
Children just old enough to be ineligible serve as the comparison for children just young enough to enroll. The age cutoff is our instrument, it predicts sponsorship without directly causing outcomes.
Study Design: Age-Eligibility Instrument
Outcomes: self-esteem · optimism · aspirations · years of schooling
Figure 1. Discontinuity in program take-up at the age-eligibility cutoff (Ross et al., 2021).
Effect Sizes
Each dot is a causal estimate. Horizontal bars are 95% confidence intervals. Effects are in standard deviations (σ): a gain of 0.26σ corresponds to moving from roughly the 50th to the 60th percentile (under normality, Φ(0.26) ≈ 0.60). Use the dropdown to switch between countries and estimation methods.
Results by Country
The chart below shows all four groups simultaneously so you can compare across countries. The pattern is striking: Kenya dominates in self-esteem and aspirations; Indonesia dominates in optimism; Mexico is imprecise throughout. Hover over any point for details.
Why do results differ so much across countries?
Two factors explain the heterogeneity (see paper, Section 5.3):
Baseline conditions. Effects are largest where children face the greatest constraints at baseline. Kenya has the lowest baseline education levels, worst employment prospects, and lowest income. Sponsorship has more “room to grow” when aspirations are most depressed. This is consistent with Wydick et al. (2013), who found the largest adult impacts in Sub-Saharan Africa.
Duration of sponsorship. Kenya and Indonesia children had been sponsored for an average of 7.0 and 7.3 years respectively at the time of the survey. Mexico children averaged only 3.98 years, shorter exposure limits cumulative psychological impact.
How the Schooling Gain Works: A Mediation Path
Sponsored children complete 0.567 more years of school on average (SE = 0.131, p < 0.01). The program does this through two routes: direct inputs (fees, meals, tutoring) and an indirect psychological channel, children who develop higher aspirations study harder and drop out less.
The diagram maps these pathways. Hover over any element for the underlying estimates.
Mediation estimated pooled across all three countries (Table 10). The indirect effect via aspirations has a bootstrap 95% CI of [0.010, 0.105]; via self-esteem [0.001, 0.048]. Optimism shows a positive but insignificant indirect effect.
Citation: Ross, P.H., Glewwe, P., Prudencio, D., & Wydick, B. (2021). Developing educational and vocational aspirations through international child sponsorship: Evidence from Kenya, Indonesia, and Mexico. World Development, 140, 105336.
Methodology note: Effects estimated by 2SLS using an age-eligibility rule as an instrument. OLS shown for comparison in the first chart. Standard deviations (σ) use the pooled comparison-group distribution. Confidence intervals are 95% (±1.96 SE).