When jobs appear in one place, who actually gets them?
Commuting, Migration, and Local Employment Elasticities
Ferdinando Monte, Stephen J. Redding, Esteban Rossi-Hansberg (2018) · American Economic Review 108(12)
Article ↗ · Replication data ↗ (openICPSR)
The archive requires an account we do not hold, so the files were not opened. What follows rests on the published abstract and on this platform's own observations.
What the authors found
Monte, Redding and Rossi-Hansberg show that how much local employment responds to a local labour demand shock depends on how open that place is to commuting. Somewhere well connected, the new jobs are filled substantially by people who travel in and the local employment rate barely moves; somewhere isolated, the same shock has a much larger local effect. Estimates of local multipliers that ignore commuting are therefore averaging over economies that behave very differently.
What our data adds
The openness the paper turns on is now measurable here, and it varies enormously. Across 131 NUTS-2 regions in 2025, the median region has 93.6 per cent of its resident workers working inside it. But Walloon Brabant has only 48.0 per cent — half its workforce commutes out, mostly to Brussels — and Pest, ringing Budapest, has 56.1 per cent. At the other end Catalonia is 98.9 per cent self-contained and Lazio 98.8. Belgian Luxembourg is a third case again: 29.5 per cent of its workers cross a national border every day. On the authors' argument, an identical labour demand shock would produce a very different local employment response in Walloon Brabant than in Catalonia, and a regional multiplier estimated across both is averaging two different economies.
Absolute counts. The note's argument is about shares, which are in the table below.
| Figure | Value | Indicator |
|---|---|---|
| Median share working in their own region | 93.6% | COMMUTE.INR |
| Walloon Brabant, least self-contained | 48.0% | COMMUTE.INR |
| Catalonia, most self-contained | 98.9% | COMMUTE.INR |
| Belgian Luxembourg, working abroad | 29.5% | COMMUTE.FOR |
Every figure here is computed from observations in this database and can be charted, downloaded or queried.
What this cannot show
This is not the region-to-region matrix an earlier version of this note promised, and the correction matters. Eurostat's dissemination API publishes the destination as four categories — same region, another region, a foreign country, no response — not as a named region. So it can say half of Walloon Brabant's workers leave it and cannot say they go to Brussels; that inference is geography, not data. Coverage is also thinner than the GDP series: 131 regions with a full breakdown against 276 with GDP.
The data behind it
Million euro
NAMA_10R_2GDP.MIO_EUR
ESTATPersons employed in the population of active enterprises in t - number
BD.V16910