Within and Beyond National Borders: How Administrative Linked Data Can Revive Internal Migration Studies
Matthieu Solignac, Univ. Bordeaux; INED
Reinventing internal migration studies may require to develop new theoretical approaches bridging the divide between internal and international migration (Ellis, 2012). But such a shift also need to be fuelled by the use of new innovative data. In that perspective, administrative linked data offer promising research avenues. Many information are collected by public authorities and kept in different records such as civil registers, tax records, social security records, the scope of each source being limited to a specific administrative purpose. But if these databases can be gathered, so that each individual can be traced over them while maintaining strict anonymity and confidentiality safeguards, tremendous opportunities are opening up for studying individual trajectories.
Based on multiple quite exhaustive and accurate data collections, administrative linked data are especially well-suited for studying migration. They can be exploited as representative panel data, allowing longitudinal analysis from micro to macro level over long periods of time. Complete enumeration and systematic follow-up give the opportunity to study internal migrations while taking into account the specificities of international migration trajectories. In particular, it can help considering all individuals leaving a specific place over time, regardless of their final destination. Multi-site surveys conducted at both origin and destination helped international migration studies to “look beyond immigration” (Beauchemin, 2015; Willekens et al., 2016) and to extend follow-up of individuals beyond the moment they cross borders. Similarly, administrative linked data can help internal migration studies to fully embrasse all trajectories observed in a given country, not only including those taking place within national borders but also those going beyond.
This contribution draws on an ongoing panel called French Demographic Sample (Echantillon Démographique Permanent or EDP) gathering information from censuses and civil registries since 1968 (Couet, 2006; Solignac, 2015). Covering almost 40 years, EDP contains more than 3 million individual trajectories in 2015. The sampling design is based on days of birth, ensuring representativeness of the sample over time: people born in France enter the panel at birth, the others enter the first time they are observed in a census or in any life-event recorded in French civil registries such as marriage, birth of their children, death. Periodicity of exhaustive census had changed from 7 years to 9 years, before being replaced in 2004 by annual census surveys and rotating samples. Two other types of annual administrative data have been linked: tax-records and information given by all employers about their employees (annual declaration of social data or DADS). Residential location can be studied at the municipality level (over 36,000 areas or communes).
Measuring out-migration from municipalities gives the opportunity to go beyond the traditional cross-sectional retrospective analysis which is restricted to the people that can still be observed at the end of the period. Retrospective analysis excludes all the individuals initially there who left due to death on French soil or emigration. Thus, it only reflects the situation of a selected sub-group of individuals rather than the actual mobility of the people living in France at time t.
First results have already been presented in Solignac (2016). Measuring out-migration with administrative linked data helps correcting the underestimation of mobility ratio among immigrants: they are much more mobile than natives once all trajectories are taken into account. This work is going to be pursued between 1999 and 2015 to study how the absence of exhaustive censuses can be offset by the newly added data sources. Moreover, results about the heterogeneity of out-migration are also expected.
Presented in Session 1233: Internal Migration and Urbanization