Migration and Mobility
At a Glance
Integrated Modeling of International Migration Flows by Using Multiple Data Sources
Maciej Danko, Dmitri A. Jdanov, Emanuele Del Fava, Domantas Jasilionis, Emilio Zagheni; in Collaboration with Arkadiusz Wisniowski (The University of Manchester, United Kingdom)
To improve our understanding of the causes and consequences of migratory movements, scholars, official statisticians, and policy-makers alike must overcome the limitations inherent in the various data sources that each country uses to produce statistics on migration, especially on flows between countries. The sources used to produce statistics on migration are highly heterogeneous (e.g., population censuses, administrative registers, household surveys, residence and work permits, and registers of foreigners). Even the Internet and social media might be used as a source of information on migration flows. Although all of these sources contain information on migration, they are usually not explicitly designed to measure it. Hence, their use for the assessment of migration patterns is hindered by inconsistencies in availability, definitions, and quality. We might thus expect high variability in the reported numbers, due to, for instance, differences in the nature and the quality of the collected data, the timing criterion underlying the definition of a migration event in a given country, and the coverage by national collection systems. Discrepancies are expected to emerge between data sources within a country, but even more so between countries in the context of estimating bilateral migration flows.
These data limitations restrict our ability to rely merely on single sources to investigate migration flows between pairs of countries. Our proposed solution is to integrate the available data within the same statistical modeling framework, building a ''synthetic database''. We thus extend previous work on estimating international migration by developing a Bayesian model that integrates and harmonizes different migration data sources as well as socioeconomic and demographic information. More specifically, our statistical model includes information on both single countries and the relationships between pairs of countries that may be predictive of migration patterns. The purpose is to obtain new estimates of migration flows between pairs of countries and across time, estimates that would lead to improvements in our understanding of the causes and consequences of migration. Considering that data for some flows in some years may be missing in a data source, either because they are not collected or are not available for various reasons, the data integration approach enables us to account for information available for other countries. If data for a specific flow is missing from all available data sources, our Bayesian framework relies on borrowing information from available data for other countries and periods, as well as from the prior distributions, in order to obtain estimates for those missing flows also.
This project is a first step in the methodological development of an approach for the construction of a Human Migration Database, a synthetic database of migration data for developed countries at the highest possible level of quality and providing detailed, uniformly estimated, and continually updated migration.
Migration, Statistics and Mathematics
Software. https://github.com/MaciejDanko/HMigD_Shiny_App_I: GitHub. unpublished. (2023)
MPIDR Working Paper WP-2023-026. (2023)
SocArXiv papers. unpublished. (2019)