How local geography and destinations influence individual effects in the decision to commute across the border. Ruptures and discontinuities along the entire border facade of France

Cross-border commuting is a key functional attribute of border areas with impacts on labour and housing markets, transport flows, and urban development. Cross-border commuting is increasing quickly, yet not all border areas are similar and geographical heterogeneity potentially leads to differences in commuting decisions. These differences cannot be identified from research using data aggregated at regional levels or focused on a specific border. We contribute the first analysis of cross- border commuting for the entire border facade of France, which neighbors eight different destination countries. We use individual data and fine-grained residential locations from a census including all workers in the facade. We estimate several predictive models of cross-border commuting that include individual, household, and activity sector characteristics, as well as geographic attributes at the municipality scale to capture the local environment of the residents in terms of urbanity, relative proximity to jobs within France and the proximity to the border. We interact those variables with the different cross-border zones (France-Spain, France-Switzerland, etc.) to identify specificities.
Our results suggest that in addition to the expected distance effect, the sector of activity is of major importance to understand the probability to commute and that this effect changes according to the different borders. We also find that the local residential characteristics, especially living an isolated commune rather than an urban core or suburb, and having few job opportunities around residences significantly affect the decision to commute across the border with effects varying strongly between zones of the facade. Finally, after mapping residuals from the mode, we identify ruptures along the border facade especially especially where different destinations overlap.