Working papers
How much power an employer has over wages depends on which other employers its workers can reach, and distance shapes that set. I build a model of wage setting in which two firms compete to the extent that they hire from the same locations, so competition is measured pair by pair rather than within drawn markets. Cheaper distance has two opposing effects. Workers reach more firms, which weakens their employers, but productive firms reach more workers, which can strengthen them, so lower distance costs need not lower market power. Using the commuting flows of every French establishment, I show that the measure predicts where job switchers go, and accounting markdowns, better than drawn markets do. I then study the rise of work from home in French manufacturing between 2016 and 2023. It lowers markdowns at most establishments, but its main effect is to reallocate workers toward productive ones. Welfare rises by 3 percent of output, mostly through the direct value of work from home to workers.
Joint with Micole De Vera and Javier Garcia-Brazales.
Do unemployed workers know their own chances of finding a job, and do their misperceptions matter for aggregate unemployment? Using two decades of Australian panel data, we estimate objective job-finding probabilities with machine learning and compare them with elicited beliefs to measure individual bias. Job seekers are over-optimistic on average, but the average masks substantial heterogeneity: 61% overestimate their prospects and 39% underestimate them. Measured optimism is greatest among the young, while most other group differences are accounted for by differences in objective job-finding prospects. We embed the estimated distribution of biases in a search model that allows workers to misperceive both the arrival rate and the wages of offers. Correcting all misperceptions would raise unemployment by about 2.2 percentage points, primarily because wage pessimism makes job seekers insufficiently selective. This result remains positive across alternative specifications. The increase in unemployment is not a welfare loss: correcting beliefs raises accepted wages by about one-quarter and output by about one-fifth.
Draft coming up soon!.
Publications
Labour Economics 98: 102843, 2026.
Joint with Micole De Vera and Javier Garcia-Brazales.
What determines long-term mental health and its intergenerational correlation? Exploiting variation in unemployment rates upon labor market entry across Australian states and cohorts, we provide novel evidence that the mental health of daughters is affected by the labor market entry conditions of their parents. In particular, a one standard deviation shock to the unemployment rate upon parental labor market entry worsens daughters’ mental health during adolescence by 13% of a standard deviation. This effect is accompanied by lower levels of satisfaction with their health, home, financial situation, and overall life. A mediation analysis suggests that a sizable proportion (22%) of the impacts on the descendants’ mental health is explained by the worse mental health of their parents at mid-life. We do not detect any impact of parental labor market entry conditions among sons.