This paper uses Young Lives data collected on young families in Peru in 2002 and 2007. Young Lives have discovered that this demographic sample has a fairly high propensity to migrate, and therefore it is interesting to examine the impact of these movements.
We disaggregate migration into rural-to-urban, urban-to-rural, urban-to-urban and rural-to-rural migration and compare the effects of these different natures of migration on household wealth. Difference-in-differences and propensity score matching are used to overcome the bias of time-invariant unobservables, and instrumental variables are used to address endogeneity caused by time-variant unobservables. We also look in more depth at why migrants moved, and the extent of relocation costs, proxied by distance.
The paper aims to test the traditional theory of migration as an investment: That households choose to migrate in order to gain net expected benefits, and that on average they succeed in doing so. Our results for rural-to-urban migrant families support this hypothesis. However, in our Peruvian data there is also a significant number of families moving in the opposite direction, out of urban areas, and this appears to be correlated with a general worsening in household wealth. The result that even the average urban-rural migrant family experiences a substantial decline in wealth is inconsistent with the notion of migration as a rational choice, unless other, perhaps more long-term, benefits of urban-rural migration outweigh the short-term deterioration in our wealth variable, or the counterfactual outcome of remaining in the urban area was expected to have been even worse, due to an unobserved adverse shock. We attempt to address the endogeneity raised by the latter case, by instrumenting for urban-rural migration using previous migration, butconclude that this instrument may in fact serve to reinforce the argument of reverse causality; that former migrants are more likely to suffer from adverse shocks, which 'push' them into return migration.