1 The Causal Question
Do conditional cash transfers (CCTs)—payments to poor families contingent on children attending school and visiting health clinics—improve human capital accumulation and well-being in developing countries? This question matters enormously for policy. In the 1990s, Latin America faced persistent poverty, high child labour, low school enrolment, and poor nutritional outcomes. Traditional social assistance programmes provided income support without incentivising investment in children's human capital. Could a CCT break the inter-generational poverty trap?
Mexico's PROGRESA programme (later renamed Oportunidades and then Prospera) was the first large-scale CCT to be rigorously evaluated using randomised assignment. Its evaluation design became the template for dozens of similar programmes worldwide, and the resulting evidence base is among the richest in development economics (Parker and Todd, 2017).
2 The Programme
PROGRESA (Programa de Educación, Salud y Alimentación) was launched by the Mexican government in 1997. It targeted poor rural households in marginalised villages across Mexico. Eligible households received three types of transfers, all conditional on behavioural requirements:
- Education grants: monthly cash payments for each school-age child enrolled and attending school (at least 85% of classes). Transfers increased with grade level and were larger for girls at the secondary level, reflecting higher gender gap in secondary enrolment.
- Health transfers: fixed monthly payment conditional on family members attending regular health check-ups and nutrition supplements.
- Food supplements: in-kind nutritional supplements for young children and pregnant or lactating women.
Transfer amounts were substantial—approximately 20-30% of household consumption for a typical beneficiary household. The programme combined demand-side incentives (transfers) with supply-side investments in schools and clinics in targeted villages.
3 The Identification Strategy: Randomised Phase-In
The evaluation of PROGRESA exploits a randomised phase-in design, a quasi-experimental approach that exploits the sequential rollout of a programme as a source of random assignment.
Design. The evaluation covered 506 poor rural villages in seven Mexican states. Due to budget constraints, not all villages could receive PROGRESA simultaneously in 1998. The government randomly assigned:
- 320 villages to receive PROGRESA starting in 1998 (treatment group)
- 186 villages to receive PROGRESA starting in late 1999 (control group)
Random assignment was at the village level, ensuring that spillovers within a village do not contaminate comparisons (SUTVA holds at the village level). Baseline data were collected in 1997 before any transfers were made; follow-up surveys were conducted in 1998 and 1999.
ITT and LATE. The primary estimator is the intention-to-treat (ITT) effect: the difference in outcomes between villages assigned to early treatment and villages assigned to late treatment. Because programme take-up is not universal among eligible households, the ITT understates the effect on actual recipients. The LATE—the average effect on households that actually participated due to early assignment—is estimated by 2SLS, instrumenting household participation with village-level assignment:
where compliance rates are high (roughly 97% of eligible households enrolled once offered the programme), so LATE ≈ ITT in most specifications (Imbens and Rubin, 2015).
Balance and validity. Baseline characteristics—household assets, parental education, consumption, health outcomes—are balanced across treatment and control villages, confirming that randomisation succeeded. Pre-treatment trends in enrolment and health indicators are also parallel (Schultz, 2004).
4 Key Findings
Education. Schultz (2004) estimates that PROGRESA increased school enrolment by 3.4-3.6 percentage points for primary school children and 7.2-9.3 percentage points for secondary school students. Effects were larger for girls, particularly at the secondary level where pre-programme gender gaps were pronounced. Conditional on being enrolled, attendance also improved.
Nutrition and health. Behrman and Hoddinott (2005) document that children in treatment villages grew 1-4% faster in height-for-age over the evaluation period. Stunting rates fell and nutritional status improved, especially for children under three years old—a particularly sensitive developmental window. Preventive health visits increased substantially for both children and adults.
Household consumption. Treatment households increased total consumption by 6-8%, with a larger share directed toward food. The transfer was not fully spent on current consumption; some was saved or invested in productive assets.
Labour supply. Programme participation did not significantly reduce adult labour supply, addressing concerns that income transfers might discourage work. Child labour fell modestly, partly because school attendance mechanically competes with work, but the transfers appear to make this trade-off individually rational.
Gender effects. PROGRESA's design deliberately targeted girls through higher educational grants at secondary level. The evaluation confirmed differential impacts: secondary enrolment increased by 8-9 percentage points for girls versus 6-7 for boys in treatment villages (Schultz, 2004).
5 External Replication and Programme Scale-Up
A central virtue of the PROGRESA evaluation design is its replicability. Rawlings and Rubio (2005) documented analogous programmes in Honduras (PRAF), Nicaragua (Red de Protección Social), and Brazil (Bolsa Escola/Família), each with randomised evaluations showing broadly consistent results: school enrolment rises by 3-10 percentage points, health indicators improve, and poverty falls.
The programme has since expanded dramatically. Under the Oportunidades and Prospera names, it eventually covered over 6 million Mexican households. Parker and Todd (2017) review 20 years of evidence, documenting that the education effects translate into long-run earnings gains of 8-10% per additional year of schooling, and that treated children show better labour market outcomes as adults. Mexico's PROGRESA became the model for Brazil's Bolsa Família (the world's largest CCT), Colombia's Familias en Acción, and dozens of similar programmes in Africa and Asia. The rigorous evaluation design was as influential as the programme itself.
6 Limitations
External validity. The evaluation covered poor rural villages in 1997 Mexico. The effects may not generalise to urban settings, more developed contexts, or countries with different institutional capacity to monitor conditionality compliance.
General equilibrium effects. The evaluation sample comprises 506 villages out of thousands in Mexico. With only a small fraction of villages treated in the evaluation period, no general equilibrium effects (e.g., changes in teacher supply, local price levels, or labour market competition) are detectable. Large-scale rollout may produce different effects.
Conditionality. The programme's conditions (school attendance, health visits) require monitoring and enforcement. In settings with weak administrative capacity, conditions may not be enforced, potentially reducing programme effectiveness. Recent evidence (from Parker and Todd 2017) suggests that in Mexico, enforcement was reasonably effective.
Long-run effects. The evaluation period (1997-1999) is short. The programme's full benefits—higher lifetime earnings, intergenerational effects on children of beneficiaries—take decades to materialise and require long panel follow-up.
7 What We Learn
PROGRESA demonstrates the randomised phase-in design as a powerful evaluation tool: when a programme must be rolled out sequentially due to budget constraints or logistical limitations, random selection of which units receive the programme first creates a valid comparison group at no additional cost. This design has since been used in hundreds of evaluations.
The programme also illustrates the complementarity of demand-side transfers and supply-side investments. PROGRESA combined cash incentives with investments in local schools and clinics. The identification strategy is clean; the effects are large and credible; and the external replication record across multiple countries gives unusually strong grounds for generalisation.
References
- Behrman, J. R. and Hoddinott, J. (2005). Programme evaluation with unobserved heterogeneity and selective implementation: The Mexican PROGRESA impact on child nutrition. Oxford Bulletin of Economics and Statistics, 67(4):547-569.
- Fiszbein, A. and Schady, N. (2009). Conditional Cash Transfers: Reducing Present and Future Poverty. World Bank Policy Research Report. Washington, DC: World Bank.
- Imbens, G. W. and Rubin, D. B. (2015). Causal Inference for Statistics, Social, and Biomedical Sciences. Cambridge University Press.
- Parker, S. W. and Todd, P. E. (2017). Conditional cash transfers: The case of PROGRESA/OPORTUNIDADES. Journal of Economic Literature, 55(3):866-915.
- Rawlings, L. B. and Rubio, G. M. (2005). Evaluating the impact of conditional cash transfer programs. World Bank Research Observer, 20(1):29-55.
- Schultz, T. P. (2004). School subsidies for the poor: Evaluating the Mexican PROGRESA poverty program. Journal of Development Economics, 74(1):199-250.