Francesco Armillei (Bocconi University)
Micro-level understanding of on-the-job search remains limited, especially regarding the cost and productivity of search effort. I address this gap by documenting at a granular level the search behavior of workers experiencing job termination within a one-year window before and after separation, exploiting peculiar features of the French Labour Force Survey. This design enables me to study on- and off-the-job search among comparable workers. I find a muted search response in the months preceding termination, followed by a sharp spike afterward. I interpret these findings as evidence of higher costs of searching while employed. With a simple static model of search and additional evidence from Time Use Survey data, I show that these costs stem from the time-consuming nature of job search and its crowding-out effect on leisure, making them highly state-dependent and underscoring the importance of separately modeling and estimating the cost and productivity of search effort. Lastly, I develop a partial-equilibrium search model and estimate it using my main data. The results indicate that, on average, on-the-job search is about 9 times more costly than off-the-job search, while being only 1.4 times more productive.