Research

We study how firms impact the dynamics of earnings inequality. Using linked employer-employee-job title data from Portugal, we show that the compression of firm pay premiums accounted for 86 percent of the large decline in wage inequality over the past decades despite an increase in worker heterogeneity. A decline in job title heterogeneity and a decreasing concentration of high-paying jobs in high-quality firms account for the remainder of this decline. The compression in firm pay premiums resulted from two factors: (i) a decrease in the pay gap between otherwise similar workers working in more or less productive firms, and (ii) declining returns to working with highly skilled co-workers. Using a novel mediation analysis, we show that the effects of firm productivity and skill composition on wage inequality are broader than previously estimated using conventional AKM variancedecompositions. While previous studies assume that firm pay policies fully mediate the relationship between firm characteristics and wage inequality, we show that 76 percent of firm productivity affects wage inequality through factors other than firm pay premiums, such as worker quality and job title pay premiums.

We study the determinants of job destruction among companies with 50-plus employees. We use French firm-level data from 2007 to 2018 and propose two approaches. The first is the identification, with an original algorithm, of the number of firms resorting to mass layoffs, legally defined as being in a position to have to implement a \textit{Plan de Sauvegarde de l’Emploi} (PSE). We find that firm workforce composition (age and occupation) influences the probability of a firm adopting a PSE. Then, we find strong evidence that lower economic and financial performance is associated with a greater likelihood of implementing a PSE. The second approach consists of quantile regression analyses of the annual variation in jobs. Differences in job growth are mostly attributable to sociodemographic variables and to indebtedness. Overall, we find consistent but not identical results on job destruction and on the probability of being in a position of adopting a PSE.

In this paper we investigate the role of firms when they face a mass layoff. We investigate a simple question: `when things go bad, who gets fired first?'. Using administrative and survey data for France, we first show that the companies that had a mass layoff used this event to restructure their workforce. We observe a small but significant increase in the composition of social skills and a decrease in manual skills and cognitive skills within the firm, as compared to the control group. The restructuring of the workforce demonstrates that firms use layoff strategically to recompose, and then selective displacement plays an important role. We also investigate the factors that determining who is fired. The results indicate that firms strategically chose which workers to displace, and that demographic characteristics, perceived cost and skills mismatch are determining factors.

This paper gives a new answer to an old question in labor economics, ``Who matches with whom?'', by introducing a setting where firms and workers are different in many dimensions and we allow workers to be over and under qualified for the jobs they end up occupying. I present a random search model with two sided multidimensional heterogeneity in which firms choose and post a wage with commitment i.e. maintaining the posted wage, independent of the productivity of the new worker. Posted wages determine the set of acceptable jobs for each worker and a unique applicants pool for each firm. The composition of these sets varies in size and composition across workers and firms. The optimal posted wage level takes into consideration the requirements of each firm and the characteristics of the applicants pool. In equilibrium, sorting is assortative but mismatches can occur across all skills dimensions. Using French data on workers observed skills and matches, I estimate the structural parameters associated with the model for France. I find that the disutility of non cognitive skills is higher when mismatched, while employers value more highly good matches on cognitive skills. I also find that the number of dimensions plays an important role, since it is another source for frictions.

This paper models the labor market matching process when skills are multidimensional and workers are naive about the strategic behavior of their competitors. Using supply and demand side data on multidimensional skills from Colombia, the paper numerically solves for the equilibrium allocation of workers to jobs that solves the naive worker problem and finds that the allocation is inefficient, in that workers over-weight job availability at the expense of matching to jobs for which they are over-qualified, leaving less qualified workers to match to jobs with higher skill demands. Three counterfactual simulations suggest training of long-term unemployed can be the most effective at improving the efficiency of the allocation of workers to jobs by both making them better matches for medium skilled jobs and by reducing the likelihood that high skilled individuals will be hired for jobs for which they are overqualified, providing them an incentive to apply increasingly for high-skilled jobs.