Inferring Spatial Wages Distribution from Bilateral Commuting Choices in France

Model predicted wages vs survey wages (workplace)


In this paper, I present a theory-based urban model that relates commuting choices with the spatial distribution of wages, and estimate such relationship using French commuting and socio-economic data at the municipal level. The model implies a commuting gravity equation whose workplace fixed effects are proportional to log labor wages. My results demonstrate that model-predicted workplace wages account for 42 percent of the variation in workplace wages, outperforming traditional predictors. Model-predicted residential wages account for 30 percent of the variation in residential wages, on par with traditional predictors. I further explore under what conditions residential wages have better prediction power and find that the model predicts better for bigger municipalities in terms of employment or density. With a Monte Carlo simulation, I corroborate my empirical findings by showing the model’s prediction power for residential wages is higher for municipalities with a large number of inbound workers.

Jipeng (Tony) Liu
Jipeng (Tony) Liu
Finance PhD Student