A range of qualitative evidence indicates that the very poorest in a community are unlikely to migrate due to the high transaction costs associated with migration, and therefore migration is not a viable strategy to enable these households to move out of poverty. This paper explores this finding quantitatively. First, we describe the main challenges in the empirical literature and introduce a conceptual model to explore the links between migration and poverty. Using a bivariate probit model, our results show that poverty is a positive and significant determinant of migration. Furthermore, migration can have a significant impact on helping poor people move out of poverty. In accordance with dominant literature on migration, migration choice is determined, among other things, by age, gender, marital status, regional effects and education level. By far the largest determinant of current poverty status for all groups is their past poverty status which highlights the path dependent nature of poverty and the problematic of poverty traps. The fact that the analysis here is able to use a two-period model is an improvement on many existing analyses of migration. Controlling for past poverty status in a sequential model indicates that migration effects that are typically evident in one-period models, are significantly reduced. We also investigate the selectivity of migrants and find evidence of selectivity with respect to human capital for Egyptians in Italy who moved with a visa or work permit. With respect to economic migrants we find selectivity in gender and past poverty status for both Ghanaians and Egyptian migrants, but no differences according to human capital. The findings presented in this paper enrich existing empirical studies by providing a clear estimation of sequential events and enable policymakers to better understand the processes behind migration and poverty. Further analysis will explore similar issues with respect to return migrants.
WP-T14, Sussex, UK, DRC on Migration, Globalisation and Poverty, 48 pp.