Identifying Sorting

Abstract

We develop a methodology to quantitatively study sorting between workers and firms based on their unobserved (to economists) productive characteristics. It is often hypothesized that sorting on unobservables is an important determinant of output, productivity and wage differences at the micro and macro levels. Sorting on unobservables is thought to be substantially more important than sorting on observable attributes that are typically found to account for only a relatively small amount of variation in variables of interest. For example, sorting on unobserved productivity differences is offered as a possible explanation for the persistent differences in wages across employers of different size, industry affliation, or exporting status. Key questions in macro involve measuring the degree of mismatch between worker skills and technology and the evolution of this mismatch over the business cycle. Restricting the analysis to observed skills provides at best only an incomplete set of answers. In the growth literature the distribution of observed and unobserved skills is an important determinant of technology adoption and the extent of misallocation is thought to be an important determinant of ag- gregate TFP. Yet, despite the manifest importance of these questions, the literature currently lacks a theoretically consistent way to measure the unobserved productivities of workers and firms and the patterns of sorting between them. We fill this gap in the literature. This allows to provide answers to the questions mentioned above and many others, discussed in the main text.