Estimating the Demographic Compositions of Immigration and Emigration Flows

Dilek Yildiz, International Institute for Applied Systems Analysis
Guy J. Abel, University Of Hong Kong

Migration, like other components of population change, varies widely according to the age, sex and educational attainment of individuals. To better understand the impact of migration on countries of origin and destination, and to produce effective population projections, it is necessary to measure migration flows by these key demographic characteristics. However, data on international migration flows by age, sex and education are scarce. Education-specific measures of international migration flows do not exist in almost all countries. Without detailed education-specific data on migration flows, our ability to effectively explain and predict human capital gains and losses, and their associated costs and benefits, is severely limited. We present a super learning approach that combines multiple machine learning algorithms to estimate age and education compositions of sex-specific migration flows, drawing on available data from IPUMS-International database census samples. Our empirical-based estimates for 199 countries allow, for the first time, a better understanding of the demographic composition of international migration flows related to human capital losses and gains. They also potentially serve as more reliable base data for the demographic composition of international migration flows used in global population projections which have previously been derived from simple assumptions with scant empirical support.

Keywords: International Migration, Computational social science methods, Data and Methods, Migrant Populations and Refugees

See extended abstract.