تجزیه و تحلیل طولی اصطکاک جستجو و تطابق در بازار کار آمریکا

This paper takes a partial equilibrium on-the-job search model to a decade (1996–2006) of repeated cross-sections from the U.S. Current Population Survey. Each month, a set of parameters ruling worker mobility between labor market states and along the wage ladder is estimated using wage distributions and individual transitions. In particular, job-to-job mobility is decomposed into a voluntary component (on-the-job search) and an involuntary one (job reallocation). The resulting time series of transition parameters are first used in a longitudinal analysis of labor turnover and search frictions. Job reallocations are shown to be key in the acyclical behavior of the job separation rate, and in the procyclical behavior of the probability of changing job. Moreover, an index of search frictions is computed and shown to follow no cyclical pattern. The paper then turns to an estimation of the matching function with both unemployed and employed job seekers. The transition parameters from the job search model are used as weights in an aggregate indicator of labor supply. The inclusion of employed workers increases the estimates of the elasticities of the matching function with respect to its two inputs (labor supply and job vacancies).

Although on-the-job search models have been extensively used in the analysis of labor turnover and wage distributions (see Mortensen, 2002), there have been few attempts to take the search structure to data over a long period. The present paper uses repeated cross-sections from the U.S. Current Population Survey (CPS thereafter) to estimate a partial equilibrium on-the-job search model on a series of monthly intervals covering eleven years, from 1996 to 2006. Its first contribution is thus to produce time series of estimates of structural parameters ruling labor turnover and mobility along the wage ladder. These series are used in two applications: a longitudinal analysis of job separation, job-to-job mobility and search frictions, and an estimation of the matching function accounting for both unemployed and employed job seekers. Before presenting these two applications in more details, the next two paragraphs unveil the main characteristics of the model used for estimation.

This paper is an attempt to merge two branches of the job search literature into a longitudinal analysis of search frictions on the U.S. labor market. First, a prototypical on-the-job search model is estimated on repeated cross-sections of the CPS. In particular, using wage distributions among employed workers, job entrants and job changers allows one to estimate two determinants of job-to-job mobility: voluntary job changes and reallocation shocks. This yields time series of structural indicators showing that the procyclicality of job-to-job transitions pointed out by recent studies (Nagypál, 2008) seems to be driven more by job reallocations than by voluntary job changes. Moreover, introducing reallocation shocks allows one to recover the acyclical behavior of job separations, consistently with the studies of Hall (2005) and Shimer (2005). Lastly, an indicator of search frictions is computed and is shown to have stayed roughly constant between 1996 and 2006.