Occupational and Job Mobility in the US

Abstract

We propose a new methodology to measure and to study worker mobility across occupations and jobs in US Census data at the monthly frequency. Our approach builds on two main ideas. First, we use the longitudinal dimension of matched monthly CPS files to evaluate each occupational transition in the context of the transitioning worker's employment history over four consecutive months. Second, we rely on the post-1994 Dependent Coding of occupations, and additional filters, to (in) validate potentially suspicious transitions. When we apply our methodology to the 1979-2004 period, we obtain new estimates of the average levels and time series patterns of these labor market transitions. We find that about 3.5% of workers employed in two consecutive months report different 3-digit occupations. This flow is procyclical, mildly rising in the 1980s and falling after 1995, faster after the 2001 recession. Based on the results regarding occupational mobility, we can impute information to the numerous missing answers to the job-to-job (or Employer-to-Employer, EE) survey question. We revise upward current estimates of the average EE rate since 1994, to 3.2% per month. This rate mildly declines in 1994-1997, mildly rises in 1997-2000 and falls significantly and continuously in 2001-2004. This pattern suggests a very persistent negative impact of the latest two recessions on job-to-job mobility.