Migration, remittances and educational outcomes : the case of Haiti
Using data from three Haitian communities the present paper investigates the effects of migration and remittances on educational outcomes. From a theoretical point of view, remittances can be supposed to have positive effects in poor households, as they alleviate budget constraints. Given the dominance of private facilities in the Haitian ... educational system and the high poverty rate prevailing in the country, budget constraints are likely to play an important role for educational outcomes. Migration in turn is associated with rising duties for children in the household and the loss of parental role models as well as with declining returns to schooling. Thus, migration is expected to affect educational outcomes negatively. The fact that the number of households in the LAMP dataset reporting to receive remittances is considerably higher than the number of migrant households, allows to disentangle the impacts of both phenomena in the empirical model. The results suggest a positive impact of remittances that is restricted to poorer households as predicted by theory. In contrast, no impact of the household head´s absence can be detected. The idea that budget constraints are an important factor concerning educational outcomes due to high poverty and a mainly not-for-free schooling system is supported by the strong influence of the asset index. However, considering the results of the estimate one also has to be aware of problems connected to the chosen empirical approach and the dataset. First of all, the imprecision in the measurement of the schooling variable as well as the relatively small number of observations in the sample lead to extremely vast confidence intervals of the estimators. Furthermore, it is questionable whether the condition of time invariance - that is crucial if the Cox model is applied - holds for all covariates included in the estimations. Especially assuming time invariance of the remittances recipient status seems problematic, albeit remittances flows to Haiti have been found to be relatively stable sources of income. The collection of panel data might be an appropriate way to overcome those difficulties, at least partially. It would allow for the verification of the time invariance assumption and - if the assumption does not hold - for the use of more advanced empirical approaches like a Cox model with time varying covariates as well as the inclusion of fixed household effects. However, collecting such data is supposedly an ambitious undertaking in the Haitian environment. Alternatively, one might collect data that includes the "remittances receipt history" of each household, similar to the migration and life history already included in the LAMP dataset. Finally, besides all concerns linked to the empirical approach, it is important to keep in mind that this paper focuses on the impacts of migration and remittances on the generation following the one of the migrants and the remittances senders. It does not consider the effects of those persons´ exit on the productivity and stock of human capital in their own generation, i.e. the effect of the so-called "brain drain". Thus, one must be aware that the issue this paper deals with only covers one of the multiple aspects playing a role when it comes to assessing whether migration and remittance.