Civil War and International Migration from Nepal: Evidence from a Spatial Durbin Model

A growing literature studies microeconomic effects of war on human capital formation and labour market activity. A common research design is to relate spatially aggregated data on conflict rates at the first or second sub-national level to more spatially disaggregated survey data on outcomes of interest. Several studies focus on Nepal’s civil war, that ran for a decade from 1996, and use conflict-related deaths in Nepal’s 75 districts (the second sub-national level). Variation in the conflict-related death rate within Nepal’s districts is more than three times higher than the variation between districts. Consequently, using district-level conflict data creates a measurement error on the right-hand side of regression models, making the conflict seem more widespread, and biases econometric estimates of conflict impacts. Prior studies also ignore spatial spillovers, where local conflict may affect outcomes not only locally but also in surrounding areas. To deal with these biases, we use measures of conflict intensity for Nepal’s 3982 localities in a spatial Durbin model of the change in emigration rates between the 2001 and 2011. We distinguish emigration to India, which is informal and long-standing, from emigration to other countries that is a recent development for Nepal and requires formal recruitment and visa processes. Higher local conflict intensity is associated with slower local growth in the emigration rate between 2001 and 2011. It is mainly indirect impacts, based on the spatial lags, which matter and it is emigration to destinations other than India that was deterred by the conflict. The estimated impacts would be substantially distorted if conflict intensity was measured at the more aggregated, district-level, as in the existing literature.