Baird, S., Ferreira, F.H.G., Özler, B. and Woolcock, M. (2013) Relative effectiveness of conditional and unconditional cash transfers for schooling outcomes in developing countries: a systematic review. Campbell Systematic Reviews 2013: 8, pp. 1–124. DOI: 10.4073/csr.2013.8.Link to Source
Headline Findings: a summary statementBoth CCTs (conditional cash transfers) and UCTs (unconditional cash transfers) improve the odds of being enrolled in and attending school, compared with no cash-transfer programme. However, while interventions with no conditions (or with some conditions that are not monitored) have some effect on enrolment rates (18–25 per cent improvement in the odds of being enrolled in school), programmes that are explicitly conditional, monitor compliance and penalise non-compliance have substantially larger effects (60 per cent improvement in odds of enrolment).
Evidence BaseThe review includes 75 reports, with data from 35 studies, including five UCTs, 26 CCTs and four studies which directly compare CCTs with UCTs.
Implications for policy and practice
The findings of the review suggest that both CCTs (Odds Ratio (OR) 1.41, 95 per cent Confidence Interval (CI) 1.27-1.56) and UCTs (OR 1.23, 95 per cent CI 1.08-1.41) have a significant effect on enrolment. These results indicate that CCTs increase the odds of a child being enrolled in school by 41 per cent, and UCTs increase the odds by 23 per cent. The authors do not find a significant difference when comparing CCTs with UCTs (OR 1.15, 95 with CI 0.94-1.42).
The categorisation of these programmes into CCT and UCT ignores the fact that there is a great deal of variation in the intensity of the conditionality. Thus, the authors group studies into three broader categories (i) no schooling conditions (intensity=1 or 2), (ii) someschooling conditions with no enforcement or monitoring (intensity=3 or 4) and (iii) explicit schooling conditions monitored and enforced (intensity=5 or 6). When grouping the studies in this way the authors find that all three categories of cash transfer programs have some effect on enrolment rates. The results indicate that the effects of cash transfer programmes are largest for the category of programmes with explicit schooling conditions monitored and enforced, with a 60 per cent increase in the odds of a child being enrolled in school (OR 1.60, 95 per cent CI 1.37-1.88), followed by a 25 per cent increase in the odds of a child being enrolled in school (OR 1.25, 95 per cent CI 1.10-1.42) for programmes with someschooling conditions with no enforcement or monitoring, and a 18 per cent increase in the odds of a child being enrolled in school (OR 1.18, 95 per cent CI 1.05-1.33) for programmes with no schooling conditions. The 95 per cent confidence intervals for studies with no conditions and studies with conditions monitored and enforced do not overlap.
Meta-regression indicates that beyond the intensity of the conditions imposed, none of the other measured design elements has a significant effect on moderating the overall effect size. Unlike enrolment and attendance, the effectiveness of cash-transfer programmes on improving test scores is small at best.
Implications for further researchThe review authors note that, in order for systematic reviews and meta-analyses to be a useful tool in economics, researchers need to do a more thorough job of reporting details of the study design, as well as reporting the numbers necessary for effect-size calculation. The risk of bias due to self-reporting also needs to be assessed further, but existing evidence suggests that it may be large, especially in experiments. Additionally, further research is needed on evaluating UCTs, to increase the evidence base for this set of studies and to allow for more confidence in comparing the relative effectiveness of CCTs and UCTs.
Increased educational attainment around the world is one of the key aims of the Millennium Development Goals. Cash-transfer programmes, both conditional and unconditional, are a popular tool for social protection in developing countries that aim, among other things, to improve education outcomes. The debate over whether these programmes should include conditions has been at the forefront of recent global policy discussions. This systematic review aims to complement the existing evidence on the effectiveness of these programmes in improving schooling outcomes and to inform the debate concerning the design of cash-transfer programmes.
To assess the relative effectiveness of conditional and unconditional cash transfers in improving enrolment, attendance and test scores in developing countries. The secondary objective was to understand the role of different dimensions of the cash-transfer programmes, particularly the effect of the intensity of conditions and the effects of priming (with respect to the importance of children’s schooling) in cash-transfer programmes.
To be eligible for this review, studies had to either assess the impact of a conditional cash-transfer programme (CCT), with at least one condition explicitly related to schooling, or evaluate an unconditional cash-transfer programme (UCT). The report had to include at least one quantifiable measure of enrolment, attendance or test scores. The report had to be published after 1997, utilise a randomised control trial or a quasi-experimental design, and take place in a low- or middle-income country.
The authors conducted a comprensive search of the published and unpublished literature written in English, Portuguese and Spanish, covering the period between 1997 and April 2012. They searched 37 international electronic databases, contacted researchers working in the area, hand-searched key journals, and reviewed websites of relevant organisations.Given the one-year delay between the original search and the final edits of the review, the authors updated the review with all new eligible references of which the study team was aware as of 30 April 2013.
The authors constructed a data-extraction sheet to collect data on impacts and characteristics of the report and intervention. Enrolment and attendance were coded using odds ratios, while test scores were coded using standardised mean differences. The authors synthesised and summarised effect sizes within and across reports to one effect size per outcome for each study. Given the heterogeneity of true effects in the population, the authors used random-effects meta-analysis models to estimate the pooled effect sizes. They conducted moderator analysis with six additional variables.
The authors conduct a comprehensive search for relevant literature, and use appropriate methods for screening, data extraction, critical appraisal and synthesis. The only minor limitation of the review is that it only includes literature published in English, Portuguese and Spanish.