A Bayesian approach to estimate annual bilateral migration flows for South America using census data

Abstract

Abstract Censuses are an important source of international migration flow data. However, their use is limited since they indirectly reflect migration, capturing migrant transitions over long intervals rather than migration events, whilst also underestimating the number of infants and deaths. Censuses also neglect migration of those who are native-born when they only include questions on country of birth, and have sparse temporal availability. We propose a Bayesian hierarchical model to overcome these limitations and produce a set of robust annual migration flow estimates for South American countries. Our model translates five-year transition data from censuses into annual series, corrects biases that arise due to differences in measurement and census data quality across countries, and is grounded in migration theory to impute missing migration data between censuses.