Skip to content
Back to search results

Cash Transfer Evaluation

Introduced by
PROGRESA/Oportunidades (Mexico); Santiago Levy; World Bank evaluation programmes
Year
1997
Type
Programme impact evaluation
Family
Social protection impact evaluation
Font size

What does it do?

Cash transfer evaluation is the body of impact-evaluation practice used to measure the effects of giving money directly to poor households — conditional on behaviours such as school enrolment and clinic visits (CCTs) or unconditional (UCTs) — on consumption, schooling, nutrition, health, and broader welfare.

Pioneered by Mexico's PROGRESA/Oportunidades programme in the late 1990s, which built a randomised phase-in into its rollout, the field has produced some of the most influential causal evidence in development economics and now spans dozens of countries and hundreds of studies.

The intuition behind the method

Handing cash to poor families raises an obvious empirical question: does it actually improve their lives, and do the strings attached to conditional transfers earn their administrative cost? Because programmes target the poor and families self-select, a naive comparison of recipients with non-recipients is contaminated. Evaluation borrows a clean counterfactual — either by randomising which villages or households receive the transfer first, or by exploiting the sharp eligibility cutoff in a proxy-means test, where families just above and just below the threshold are nearly identical except for the transfer they receive.

History

The modern field began with Mexico's PROGRESA (later Oportunidades, then Prospera), launched in 1997 under Santiago Levy with a built-in randomised phase-in across communities, evaluated by IFPRI — a deliberate design that made rigorous impact evaluation possible and politically durable. Its demonstrated effects on schooling and nutrition inspired conditional cash transfers across Latin America (Brazil's Bolsa Família, Colombia's Familias en Acción) and beyond. The World Bank's 2009 policy research report by Fiszbein and Schady consolidated the global evidence, while Baird, McIntosh, and Özler's 2011 Malawi experiment sharpened the conditional-versus-unconditional debate by randomising conditions directly.

When to use it

Use cash transfer evaluation when assessing whether a social-protection programme delivers its intended welfare gains, when weighing conditional against unconditional designs, or when calibrating transfer size and targeting for scale-up. It is most powerful where the rollout offers an experimental phase-in or a sharp eligibility threshold suitable for regression discontinuity. It is less informative for programmes already at saturation coverage (no comparison group), for very long-run human-capital effects that require multi-decade follow-up, and for general-equilibrium effects on local prices and labour markets that household-level designs cannot capture.

Limitations and considerations

  • Household-level designs typically miss general-equilibrium effects such as local price inflation and labour-market spillovers.
  • Regression discontinuity identifies the effect only locally, near the eligibility cutoff, and may not generalise to the very poorest who are well below it.
  • Short evaluation horizons capture take-up and intermediate outcomes but rarely the long-run human-capital and intergenerational effects that justify the programmes.
  • Conditions' effects are hard to separate from the income effect without an explicit unconditional comparison arm.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Cash transfers are among the most rigorously evaluated development interventions, with experimental and quasi-experimental evidence from many countries.
  • Programme rollouts often supply natural identification — randomised phase-in or proxy-means eligibility cutoffs — yielding credible causal estimates.
  • Direct measurement of consumption, schooling, and health allows transparent cost-effectiveness comparison against in-kind and other interventions.
  • Comparing conditional and unconditional arms isolates the marginal value of conditions, informing a central policy debate.

Applications

  • Evaluating PROGRESA/Oportunidades-style conditional transfers on child schooling and nutrition in Latin America.
  • Testing unconditional transfers (e.g., GiveDirectly) on consumption, assets, and psychological well-being in Sub-Saharan Africa.
  • Comparing conditional and unconditional arms to measure the marginal effect of schooling and health conditions.
  • Assessing the effect of transfers paid to women on bargaining power and child outcomes.
  • Using eligibility-threshold regression discontinuity to estimate impacts of social pensions and poverty-targeted grants.

How it works

How it works

1

The evaluator characterises the transfer programme: its size and frequency, whether it is conditional or unconditional, the conditions imposed (school attendance, health check-ups), the targeting rule (geographic, categorical, or proxy-means tested), and the recipient within the household (often the mother). These design features define both the expected behavioural channels and the identification opportunities the rollout offers.

2

A credible counterfactual is selected. Randomised phase-in (as in PROGRESA) assigns the transfer to some communities before others; a regression discontinuity exploits the eligibility threshold of a proxy-means score, comparing households just above and just below the cutoff to estimate the local treatment effect at the boundary. Difference-in-differences or matching may be used where neither experiment nor sharp cutoff is available, at the cost of stronger assumptions.

Formula
\lim_{x\to c^{+}}\mathbb{E}[Y\mid X=x] - \lim_{x\to c^{-}}\mathbb{E}[Y\mid X=x]
3

Outcomes spanning the programme's theory of change are measured through household surveys and administrative data: per-capita consumption and food expenditure, school enrolment and attendance, child anthropometrics and health utilisation, labour supply, and — increasingly — women's empowerment and psychological well-being. Both intended outcomes and possible unintended effects (dependency, local price inflation, fertility) are tracked.

4

The transfer's causal effect is estimated within the chosen design — difference-in-means for the randomised arm, a local polynomial at the cutoff for the RDD — typically distinguishing conditional from unconditional arms to isolate the value of the conditions. Heterogeneity by baseline poverty, gender, and age is examined, and cost-effectiveness is computed to compare the programme against alternatives.

Common pitfalls

  • Comparing recipients to non-recipients without a valid counterfactual confounds the transfer with the targeting that selected recipients.
  • Manipulation of the proxy-means score around the eligibility threshold invalidates the regression-discontinuity design if not tested for (e.g., with a density test).
  • Attributing all measured gains to the conditions when an unconditional arm shows the income effect alone drives much of the impact.
  • Ignoring local price responses, so that nominal consumption gains overstate the real welfare improvement in saturated markets.

Frequently asked questions

What is the difference between conditional and unconditional cash transfers?

Conditional cash transfers (CCTs) pay households on the condition that they undertake specified behaviours — typically sending children to school and attending health check-ups — while unconditional cash transfers (UCTs) impose no such requirements. CCTs aim to build human capital and overcome behavioural or informational barriers, but carry monitoring and exclusion costs. Evidence such as Baird, McIntosh, and Özler (2011) shows conditions can raise schooling more than pure cash, while UCTs may better protect the most vulnerable who fail conditions; the optimal choice depends on objectives and capacity.

Why is regression discontinuity well suited to cash transfer evaluation?

Many transfer programmes assign eligibility using a continuous proxy-means score with a sharp cutoff: households below a poverty threshold qualify, those above do not. Families just on either side of the cutoff are nearly identical in characteristics, so comparing their outcomes isolates the causal effect of the transfer locally at the threshold. The design requires that the score cannot be precisely manipulated around the cutoff, a condition that can be tested by examining the density of the running variable.

Do cash transfers make people work less or become dependent?

A large body of evidence finds little support for the fear that transfers reduce adult labour supply or create dependency; reviews of multiple programmes (including a well-known multi-country study) show no systematic reduction in work among working-age adults, with reductions concentrated among the elderly and child labour. Recipients commonly invest part of transfers in productive assets and businesses. Concerns about local price inflation and spillovers are more empirically grounded and are increasingly addressed with saturation-design evaluations.

Related methods

Sources

  1. Fiszbein, A., & Schady, N. (2009). Conditional Cash Transfers: Reducing Present and Future Poverty. World Bank Policy Research Report. Washington, DC: World Bank.

    ISBN: 9780821373521
    Copied.
  2. Baird, S., McIntosh, C., & Özler, B. (2011). Cash or Condition? Evidence from a Cash Transfer Experiment. Quarterly Journal of Economics, 126(4), 1709–1753.

    Open DOI reference
    Copied.