Was the Downsizing of Older, Well-Educated Males in the 1990s Caused by Firms Reneging on Implicit Contracts?

It is widely believed that in the 1990s the U.S. experienced a period of widespread corporate "downsizing" in which firms lay off workers - often well-educated workers with high tenure - to an extent far greater than in the past. Recent empirical studies have lent some support to this belief. I argue that this phenomenon can be understood in light of the well-known theory developed by Lazear (1979). The theory argues that implicit contracts in which low tenure workers are underpaid relative to marginal revenue product and high tenure workers are overpaid relative to marginal revenue product can be understood as an incentive device designed to combat principal-agent problems in environments where worker monitoring is costly. There is substantial empirical support for the existence of these Lazear-style implicit contracts (LSICs). But there is also evidence of a decline in the formation of new LSICs with recent cohorts of young workers (Balan, 2003). This has important implications for high tenure (overpaid) workers who are party to LSICs. If firms are no longer entering into LSICs with young workers, then they need no longer be concerned with their reputations for trustworthiness and are free to renege on their obligations to high tenure workers by laying them off. I test for this reneging using a multi-year (1984-1998) sample of high tenure workers from the Displaced Worker Supplement to the Current Population Survey. I introduce five proxies for the probability that a particular worker is party to an LSIC. For all five of the proxies, probit regressions indicate that the effect of a one-unit increase in the proxy on the probability of layoff is negative at the beginning of the sample period. In other words, being more likely to be party to an LSIC made a worker less likely to be laid off. For four of the five proxies, this effect diminished over the sample period. For two of the five proxies, the effect of a one-unit increase in the proxy on the probability of layoff is positive by the end of the sample period. Furthermore, multinomial logit regressions provide some evidence that this trend is concentrated among the types of layoffs that can be interpreted as "downsizing." The standard errors on these coefficients are quite large, so these trends are not statistically significant. If, however, the trend is real, then it is consistent with the proposition that in recent years firms have been reneging on their implicit contracts with high tenure workers