Spatial Job Dynamics and Employment Outcomes: The Case of Michigan
Kerwin Charles, University of Michigan
Melvin Stephens, Carnegie Mellon University
Professors Charles and Stephens received CLOSUP funding through the Small Grants Program for this research project focused on spatial job dynamics and employment outcomes.
Where in physical space economic activity occurs is central to many active areas of research in economics and other social sciences. Economic geography asks questions such as where industries locate and how that location affects trade outcomes (Krugman 1989, 1994, 1999). Sociologists have speculated that diminished economic activity in particular parts of the country have contributed to a rise in the number of men who are “un-marriageable” and a reduction in the marriage rate in those regions (Wilson (1987)). Labor economists have examined changes in employment outcomes across different regions (Katz and Blanchard (1990), Bound and Holtzer (2000)). Most relevant for this study, there is a very old idea in economics that some regions are affected by what is termed “spatial-mismatch” – a disconnection between the skills required by employers in particular places, and skills possessed by people in those areas. Unfortunately, almost no recent work has examined this issue.
In this study, funded in part by the State and Local Policy Center, Kerwin Charles of the University of Michigan and Melvin Stephens of Carnegie Mellon University ask two main questions. First, they ask: How do firm-side employment dynamics differ across different regions? Their emphasis on firm-side employment dynamics is novel and important. Virtually all of the small previous literature on employment dynamics has focused on dynamics measured from the worker's perspective. Yet, most of the interesting questions on why particular regions have fared less well than others has to do with the behavior of firms. For example, the lament is often heard from politicians and scholars alike that there is persistent poverty in the nation's central cities because jobs – which is to say, employment opportunities offered by firms – have fled those areas. But it is impossible to get a handle on this question by looking at employment outcomes among individuals. Information on whether firms are growing or shrinking, being born or dying is the only way to learn about the issue of jobs fleeing, whether or how the quality of jobs is changing, or any similar issue. Using firm level data, they examine how four dimensions of firm dynamics – birth, deaths, expansions and contractions – changed between 1980 and the late 1990's across different regions, and whether particular regional characteristics are related to these dynamics.
The second topic is how these firm-level dynamics affect individual level labor market outcomes. For example, they ask: Across all counties experiencing net firm growth between 1980 and 1990, how was the employment rate of young black men affected relative to those of young whites? Evidence of firm dynamics moving one way and individual level outcomes moving another would provide the first known evidence of spatial mismatch not forthcoming from a case study.
The “regions” on which this work focuses are counties in the state of Michigan. Michigan is ideal laboratory in which to examine the questions of spatial dynamics which are our focus because of its size, the diversity of its population, and the number of different industries to which the state is home. The study's information on dynamics is drawn from the complete ES202 unemployment insurance records of virtually all firms in the state from 1980 to the late 1990's. Like every other state in the country, all firms which employ more that 20 people must report, among other things, their level of average level of employment by quarter, and their total compensation to a state agency. In Michigan, these data are collected by the Michigan Employment Security Council (MESC). In the records MESC records, firms are identified by id numbers, names, and detailed information on location.
The research described in this project is currently in progress. Results of the analyses will be available from the authors upon completion.