Federal Reserve Bank of San Francisco

FRBSF Economic Letter

1996-07 | February 16, 1996

More Economic Letters

Has Job Security in the U.S. Declined?

Has Job Security in the U.S. Declined?

Yes, according to public perception and a number of articles in newspapers and business magazines. They report that workers increasingly feel that their jobs are at risk. They cite cost-cutting pressures, increased use of temporary or contingent workers, and the need for greater flexibility in response to a more competitive and dynamic economic environment as factors that have frayed the ties that bind workers to firms. Even during the past year, when economic growth has been relatively strong throughout the U.S., wage growth reportedly has been held down by workers’ concerns about job security.

In contrast to the popular view, a recent study that attracted media attention (Farber 1995) reports that job security has remained relatively constant since the 1970s. Farber bases this conclusion on the finding that average job duration the typical length of time that a job lasts has changed very little. Such results, however, provide limited insight into job security per se. In particular, they are not centered around a reasonable definition of job security that properly distinguishes between employer-initiated and worker-initiated separations.

This Weekly Letter presents an analysis of job security based on such a definition. The evidence suggests that the popular view of declining job security may be closer to the truth than results regarding stable job durations suggest.

Economic bases for job security

In order to decide whether job security has changed, it is important to identify the bases for job security. Most U.S. employees are not union members, and therefore their employment conditions are not specified by written contracts. Instead, employment conditions for the bulk of U.S. employees are determined by what economists call “implicit contracts” unwritten and generally unspoken agreements between firms and workers. Because these agreements are not explicitly binding, they must be “self-enforcing” in order to be sustained. In other words, implicit contracts must possess features such as mutual gains for involved parties, or “reputation costs” for parties that breach agreements that minimize violations.

An example of mutual gains arises from the turnover costs related to screening, training, and job search activities, that firms and workers encounter as part of the job-matching process. These costs become more significant when workers accumulate skills that are largely confined to specific firms or when a worker’s productivity and satisfaction in a given firm or job is revealed only over time. Under these circumstances, the employment relationship constitutes a shared investment between workers and firms, so it is advantageous to both to agree implicitly on wage profiles that minimize separations through sharing of the costs and benefits of the investment. This in turn generates increasing wages and decreasing turnover with seniority.

An example involving “reputation costs” was first posed by Lazear (1979) and it is based on rising wages as a means of eliciting worker effort and loyalty. Wages below productivity early in a worker’s tenure, followed by wages that exceed productivity later, constitute a conditional performance bond that encourages workers to work hard and stay at the firm. Under these circumstances, however, firms have an incentive to fire highly paid senior workers. They are prevented from doing so by “reputation costs”: firms that prematurely fire senior workers will be forced to pay higher wages to attract new workers, as potential recruits learn of the firm’s bad faith and require higher wages to insure against future violations of the implicit employment agreement.

These features of implicit contracts may adequately ensure that firms retain workers under normal economic conditions, thereby producing job security. However, during periods of economic change, the degree of self-enforcement in implicit contracts may be reduced. For example, changing technology may reduce the value of firm-specific or industry-specific skills, thereby inducing firms to discharge senior workers who have made substantial investments in such specialized skills. Also, the reputation costs that enforce Lazear-type implicit contracts may be reduced by economic dislocation, thereby providing firms with the incentive to dismiss senior workers; Idson and Valletta (1996) find evidence consistent with this behavior.

Job stability vs. job security

Farber (1995) argues that despite popular perception of declining job security, job stability in the U.S. has remained essentially constant from 1973-1993. He examined data on the duration of jobs and found that although job durations declined over the period for some groups of low-wage males, they increased for females, so that on average they remained approximately constant.

Debate over these results continues. But even if they are correct, results on changes in job stability have limited implications for changes in job security. Suppose that due to sharp shifts in product demand patterns across industries, firms engage in widespread permanent dismissals of workers who believed their jobs were protected against such shifts. By itself, this increase in dismissals will reduce measured job durations. Suppose also, however, that workers who in the past might have quit to search for new jobs observe this signal of labor market weakness and decide to retain their current jobs. Then quit rates fall below what they otherwise would have been, offsetting the increase in dismissals and possibly leading to stable job durations. In this case, even though job durations appear to be stable, job security has declined. This example suggests that data on job durations should not be used to draw inferences about job security.

An alternative view of job security is required. A working definition of job security should describe how safe workers’ jobs are with respect to changes in the economic environment, within or outside the firm. Thus, the real issue regarding changing job security is whether variation in factors external to firms, or firms’ responsiveness to those factors, have changed so that employees sense that they are increasingly vulnerable to being permanently dismissed by their employers. Either change changing variation in outside conditions, or greater responsiveness by firms could be interpreted as decreased “job security.” Greater responsiveness by firms, however, is more interesting, because it indicates that firms’ behavior has changed.

Some additional evidence

In providing a more informative analysis of changing job security as defined here, this study uses data that distinguish between firings by firms and quits by employees. Specifically, I use unemployment data from the Current Population Survey (CPS), which is administered each month to 60,000 households and is the primary source of current U.S. labor market data. The responses of unemployed individuals permit separate identification of those unemployed due to permanent dismissals, temporary or indefinite layoffs (both of which imply an expectation of being recalled to the firm), voluntary quits, and re-entrance or new entrance into the labor market.

Data from the March CPS’s for the years 1968-1993 reveal several interesting patterns in the share of unemployment attributable to different causes. Figure 1 (this file requires Adobe Acrobat) shows the share of unemployment attributable to permanent dismissals each year. The light line shows the unadjusted series, which exhibits sharp changes over the business cycle i.e., the amount of unemployment attributable to permanent dismissals changes disproportionately with the unemployment rate. The solid line shows the series after the cyclical and random components are removed. This line reveals a substantial upward time trend over the period. In other words, conditional on the unemployment rate, the share of unemployment attributable to permanent dismissals increased steadily from 1968-1993; this increase is statistically significant.

Although this result may arise because unemployment durations for dismissed workers have increased, similar analyses of layoff unemployment reveal that the increased dismissal share has been accompanied by a decrease in the layoff share of unemployment. This is shown in Figure 2 (this file requires Adobe Acrobat), which is identical to Figure 1 (this file requires Adobe Acrobat), except with layoffs replacing permanent dismissals. This figure shows a downward trend in the layoff share of unemployment, albeit one that is less pronounced than that for permanent dismissals. A similar analysis reveals no significant time trend in the share of unemployment attributable to quits. Instead, the increasing share of permanent dismissal unemployment is balanced by decreasing shares of layoff and labor market entrant unemployment. Although these results are not entirely consistent with the story in which increasing permanent dismissals are matched by decreasing quits, they indicate that in making their employment adjustment decisions, employers are increasingly substituting permanent dismissals for layoffs.

Further analysis of the data reveals an interesting difference between the pre-1980 and post-1980 periods. In particular, the responsiveness of the dismissal share to the unemployment rate i.e., the cyclical variation in the dismissal share is substantially larger in the later period than in the earlier period. This suggests that in making their permanent dismissal decisions, employers have become more responsive to macroeconomic conditions over time. In conjunction with the upward trend in the share of unemployment attributable to permanent dismissals, these results suggest an expanding role of permanent dismissals in firms’ employment adjustment process, particularly during the period 1980-1993.


Although a recent study finds that average U.S. job duration has been largely constant over the past 20 years, such results provide little insight into changes in job security. In particular, results on job durations ignore the distinction between employer-and worker-initiated separations and the economic bases for long-term attachments between workers and firms. The analysis in this Weekly Letter indicates that despite stable job durations, the share of unemployment attributable to permanent dismissals and the cyclical responsiveness of permanent dismissals have increased. This in turn suggests the increased importance of permanent dismissals in firms’ employment adjustment process. Perceptions of decreased job security may relate to these trends, particularly if the burden of increased dismissals is falling disproportionately on classes of workers who had significant job security in the past.

A more complete analysis of the incidence of permanent dismissals across different categories of workers, and how it depends on outside economic conditions, remains to be performed. Popular perceptions of declining job security may arise from increased variation in relevant economic variables (due for example to accelerated technological change or defense cuts), or from increased employer responses to those variables. The increased cyclical responsiveness of the dismissal share suggests the latter explanation. Additional work along these lines may help to resolve the apparent conflict between general perceptions of decreased job security and research findings of stable job durations.

Robert G. Valletta


Farber, Henry. 1995. “Are Lifetime Jobs Disappearing? Job Duration in the United States: 1973-1993.” NBER Working Paper No. 5014.

Idson, Todd L., and Robert G. Valletta. 1996. “Seniority, Sectoral Decline, and Employee Retention: An Analysis of Layoff Unemployment Spells.” Journal of Labor Economics (forthcoming October).

Lazear, Edward P. 1979. “Why Is There Mandatory Retirement?” Journal of Political Economy 87 (December) pp. 1261-1284.

Opinions expressed in FRBSF Economic Letter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System. This publication is edited by Sam Zuckerman and Anita Todd. Permission to reprint must be obtained in writing.

Please send editorial comments and requests for reprint permission to
Research Library
Attn: Research publications, MS 1140
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120