Has job stability declined? Evidence from the Panel Study of Income Dynamics

American Journal of Economics and Sociology, The, April, 1999 by Dave E. Marcotte

I

Introduction

Throughout the 1990s, there has been substantial interest in the possibility that Americans' jobs have become less stable. Interest in this topic has largely been due to the implications any such long-term trend would have on matters like workers' economic security, incentives for joint investments in training by firms and workers, and the structure of pensions. While job stability trends have received much popular attention, only recently have social scientists begun studying whether this perceived trend is real. The resulting empirical research has reached mixed conclusions. By all accounts it seems that black workers and less educated workers have experienced some declines in job stability between the 1970s and the early 1990s. However, no agreement has been reached about whether there has been any more general decline in job stability. Much of that disagreement is due to complications associated with using the data on which previous studies have relied: the Current Population Survey (CPS). In this article, information from a different and, for the present purposes, richer data set is used to study trends in job stability. The data are drawn from the Panel Study of Income Dynamics (PSID) and reveal new evidence that can help us make progress in determining whether and how job stability may have changed over the past few decades.

This article first briefly reviews recent empirical work on job stability trends and discusses complications associated with using CPS data for such studies. Attention is then focused on describing the PSID and why it is an important source for developing a better understanding of trends in job stability. Next, the methods employed to analyze job stability are described, relying first on retention rate methods and then specifying multivariate models that allow for a more complete assessment of trends in job stability. Finally, the empirical results of these analyses are reviewed and their implications are considered.

II

Background and Recent Empirical Work

While early interest in job stability trends stemmed from the attention paid to the growth in employment in temporary and subcontracted positions in the 1980s, more direct attention was focused on this issue after the recession of the early 1990s. That recession seemed to hit relatively hard groups of workers who usually suffer less during economic downturns, such as college-educated workers (Farber, 1993). However, even after the recession of 1990-91 gave way to a period of sustained expansion, concern about a decline in job stability did not subside. Recent empirical work has tried to identify whether this perception is an accurate reflection of long-term trends in the labor market or an artifact of that recession.

Diebold, Neumark, and Polsky (1994), Farber (1995), and Swinnerton and Wial (1995) have used the CPS to examine whether there is any evidence of a secular decline in job stability in the United States that is consistent with these popular perceptions. The results of these articles point to declines in the stability of employment for black workers and, to some extent, for less educated workers, but lead to different conclusions about whether there has been any broader downward trend in job stability. Swinnerton and Wial conclude that trends in four-year job retention rates between 1979 and 1991 both across and within age/education/race/gender groups of workers suggest a secular decline in job stability. However, using essentially the same data, Diebold, Neumark and Polsky conclude that there is no strong evidence of a secular decline in retention rates. Farber examined the question of whether job stability has declined by looking not at retention rates, but at trends in the distribution of workers' tenure. Generally, Farber found little evidence of a decline in the median reported job tenure between 1973 and 1993.

While this CPS-based research is revealing about some aspects of job stability during the past few decades, the different conclusions have added to a fair amount of confusion about the subject of job stability trends. Alone, the different conclusions resulting from analysis of CPS data suggest that other data are needed to help develop a clearer picture. Moreover, the importance of employing other data to examine trends in job stability is called for because of two substantial complications associated with using the CPS to study trends in job stability; these may in part explain the differences in CPS-based findings.

The first complication is an inconsistency over time in the information contained in the CPS about workers' tenure with their employers. Information on job tenure is central to the analysis of job stability. Using cross-sectional data, inference on trends in job stability can be drawn either by studying trends in the distribution of job tenure or by comparing a cohort of workers in one year who have various levels of job tenure with a similar cohort m years later who have m more years of tenure. The inconsistency in the CPS tenure data makes using either method to draw inference on stability trends more difficult.