An econometric analysis of the mechanisms for increasing human capital through conditional cash transfer programs in Colombia

2017-03-02T00:57:53Z (GMT) by Contreras Suarez, Diana Stella
Using information from a Conditional Cash Transfer (CCT) program in Colombia, this thesis examines some of the mechanisms through which such initiatives contribute to improving human capital in poor households. These programs offer poor households a monetary transfer contingent on children attending regular visits to the doctor and school attendance. First, we look at the causal effect of children’s improved physical development due to the program on their cognitive development. We estimate this effect by instrumenting an intent-to-treat effect of treatment on the anthropometric measures. We find that exposure to the program does not explain gains in physical development one year after the intervention but it reduces the incidence of being underweight four and ten years later. These differentials in physical development explain around one quarter of a standard deviation gain in long-run cognitive development. Improvements in height are found only in children younger than 2. Second, improved child health is one primary aim of CCT programs; we investigate if there are potential spillover effects on the health of other non-targeted adults living in the same household as the child. We use a difference-in-difference model to estimate the intent-to-treat estimate of the spillover effect. We find significant improvements in the health of adults in treatment households, both in terms of incidence of illness (in the short run) and in the severity of illness (in the medium run). The main channel for this effect seems to be the availability of better information and the creation of a health public good within the household as a result of the program. Finally, while CCT programs have been designed to improve long term human capital among poor families through incentives and conditions, we test if they change behaviour through changes in preferences and aspirations rather than, or in addition to, the effect of the conditions and incentives. Using the poverty score for program allocation we estimate the program effect on participation using a regression discontinuity (RD) design. We find no evidence of the program working through time preferences or parents’ aspirations for their children’s schooling in participating households. The positive program impacts on health and nutrition status of children and adults identified in this thesis and the effects on schooling identified in previous studies have important implications in terms of long run human capital development of children in poor households, for example, the receipt of higher wages when adults. However, those impacts appear to be driven by the ongoing receipt of the cash transfers and the associated conditions. Thus if the program was to stop, we would expect investment in children’s human capital to revert to pre-program levels.