In labor market equilibrium, sectoral differences in "natural" rates of unemployment generate a conformable distribution of wage differentials that compensate workers for bearing unemployment risk. This paper offers new empirical evidence on the determinants of this equilibrium. The analysis consists of two stages. First, I estimate a three-state model of employment and unemployment that identifies the determinants of individuals' rates of entering and leaving unemployment spells. Sectoral, demographic, and policy-induced differences in unemployment probabilities evolve naturally from this framework. Second, I estimate the impact of these differences on the distribution of wages. An important finding is the powerful impact of the unemployment insurance (UK) system both on unemployment and on equalizing wage differences. The evidence is strong that the availability of UK increases unemployment, while simultaneously reducing the magnitude of compensating wage differentials. Most of the effect of UK on unemployment is due to an increased probability of entering spells of unemployment, mainly temporary layoffs, though the duration of spells is also affected. Neglect of the role of UK as a substitute for wages partially accounts for the small compensating differentials estimated in previous research. In the absence of UK, each point of anticipated unemployment raises an individual's wage by about 2.5%.

From 1972 to 1990, the United States experienced the largest increase in wage inequality of any developed country. The spread in log wages between the 90th and 10th percentiles of the male wage distribution increased by about 60 points, and inequality increased within virtually every demographic or skill category of the labor force (see Chinhui Juhn et al., 1993). Various explanations have been offered, including changes in the relative demand for skill-intensive goods, international trade, immigration of less-skilled workers, and increased labor-force participation of women. Yet attempts to quantify the relative importance of these factors have not met with much success. This leaves skill-biased change as the residual claimant that rationalizes the data. This paper uses regional differences in the evolution of wage inequality to provide new evidence on the determinants of relative wages. Using a model of factor demand in geographic markets, I isolate contributions of (i) changing skill ratios in the labor force; (ii) increased participation of women; (iii) technical change; and (iv) changes in the industrial composition of labor demand. The data indicate that changing skill ratios and women's labor supply have affected the wages of low-skilled men. In the West, the data indicate that increased immigration of less-skilled Hispanic and Asian workers has adversely affected the wages of natives, causing a greater increase in inequality there than in any other region of the country. I. Modeling Wage Inequality in Regional Markets

Unemployment reminds economists that we have a lot to be modest about. Twenty years ago, Robert Hall (1972) asked Why Is the Unemployment Rate so High at Full Employment? At the time Hall wrote, male unemployment in the United States had risen to 3.5 percent, and it would peak at 4.4 percent in the recession of 1971. Since then, every cyclical low of unemployment has exceeded the one before it; the end of the 1980's expansion reduced unemployment to about 4.5 percent, and unemployment peaked at 7.4 percent of the labor force in the recession that just ended. The natural rate of unemployment has evidently risen through time, which led Lawrence Summers (1986) to place a modifier in Hall's question: Why Is the Unemployment Rate So Very High...? In what follows I pose three questions about the causes and consequences of rising joblessness. First, who are the unemployed? I will show that unemployment and nonparticipation are heavily concentrated among the least skilled and, more importantly, that virtually all of the long-term increase in joblessness occurs among low-wage men. Unemployment and nonparticipation among highly skilled men are unchanged since the 1960's. Second, what factors have caused jobless time to rise? There are many theories, including changes in labor-force demographics, social programs that discourage employment, and changes in the amount of frictional unemployment caused by sectoral shifts. None of these is consistent with the evidence, which points instead to a secular decline in the demand for lessskilled workers. The data also cast doubt on nonmarket-clearing models of rising unemployment. Wages are demonstrably flexible in the long run, and they have fallen dramatically among the unskilled, whose jobless rates have risen. Third, what happens to the unemployed? In contrast to the view that unemployment spells cause transitory reductions in income, recent evidence indicates that job losses have substantial and permanent effects on earning power. Many of the unskilled who are unemployed or out of the labor force appear to have been high-wage workers whose specialized skills have become obsolete.

This paper presents a theoretical and empirical analysis of the short-run employment, layoff, and inventory strategies of firms. On average, more than two-thirds of all layoffs in American manufacturing end in a rehire by the employer of origin (Martin Feldstein, 1975; David Lilien, 1980). These layoffs have recently accounted for about half of all ongoing employer-initiated spells of unemployment in that sector (see Table 1), and have led to a revised interpretation both of the involuntary nature of these separations (Costas Azariadis, 1975; Feldstein, 1976; Martin Baily, 1977) and of the durability of employment relationships. Evidently, a significant proportion of unemployed workers know that the conditions that generated their separations are temporary, and so they expect to be recalled within some reasonable time. The point of departure for this paper is in interpreting these findings as evidence of a privately and socially efficient stock of excess capacity held against contingent future demands; that is, as an inventory of a productive input. This view leads to empirically refutable hypotheses regarding both the role of temporary layoffs and the circumstances under which they will be important. In particular, both excess capacity (idle resources) and inventories are devices by which firms may economize on the costs of rapid adjustments when faced with unstable market conditions (George Stigler, 1939). In such an environment, the demands for layoffs and for other buffer stocks are jointly determined elements of long-run technology: the choice among alternative stocks will depend on relative profitabilities, and firms may choose to hold stocks of inputs as a substitute for inventory accumulation, especially if the costs of varying capacity or of storage are important. This reasoning implies that (i) the structure of short-run employment and layoff decisions will vary across firms and industries and, (ii) differences in these structures will be systematically related to the role of inventories in firms' dynamic strategies. Some perspective is warranted. At least since Charles Holt et al. (1960), economists have recognized the production smoothing role of inventories and the necessity of predicting future demand in making current decisions (David Belsley, 1969; Gerald Childs, 1967; Michael Ward, 1978). Similarly, one of the contributions of recent work on implicit labor contracts is in viewing the exchange of labor services as a bilateral and durable commitment. Both these lines of analysis are concerned with the methods firms use to respond to fluctuating demand, yet the underlying, interrelated nature of employment and inventory decisions has rarely been recognized (Roger Miller, 1968; M. Ishaq Nadiri and Sherwin Rosen, 1973; R. G. Crawford, 1979; Robert Hall, 1972). This paper aims to provide a framework for evaluating these decisions, and in so doing to explain some interesting features of observed labor markets. The notion of substitution between inventories of human and physical capital suggests a more disaggregated look at the data. Table 2 shows the importance of temporary layoffs in the distribution of unemployment among manufacturing industries. It reports layoff and total unemployment rates for twelve representative industries over the period 197376, and average layoff rates and recall frequencies for 1958-76. Among all (21) twodigit manufacturing industries the correlation between layoff and rehire rates is .6, and the table illustrates this. Evidently, industries with larger than normal layoff rates also tend to have larger rehire frequencies from firm-initiated spells of unemployment. This *University of Chicago. I am indebted to Gary Becker, Robert Cotterman, Sherwin Rosen, Jose Scheinkman, Michael Ward, Finis Welch, the managing editor, and an anonymous referee for discussions and comments at various stages. The usual disclaimer applies.

This paper uses longitudinal data to estimate a lower bound on the average return to job seniority among adjustment. The author finds that ten years of current job seniority raise the wage of the typical male worker in the United States by over 25 percent. This is an estimate of what the typical worker would lose if his job were to end exogenously. Overall, the evidence implies that accumulation of specific capital is an important ingredient of the typical employment relationship and of life-cycle earnings and productivity as well. Continuation of these relationships has substantial specific value for workers.

The idea that wages rise relative to alternatives as job seniority accumulates is the foundation of the theory of specific human capital, as well as other widely accepted theories of compensation. The fact that persons with longer job tenures typically earn higher wages tends to support these views, yet this evidence ignores the decisions that have brought individuals to the combination of wages, job tenure, and experience that are observed in survey data. Allowing for sources of bias generated by these decisions, this paper uses longitudinal data to estimate a lower bound on the average return to job seniority among adult men. I find that 10 years of current job seniority raise the wage of the typical male worker in the United States by over 25 percent. This is an estimate of what the typical worker would lose if his job were to end exogenously. Overall, the evidence implies that accumulation of specific capital is an important ingredient of the typical employment relationship and of life cycle earnings and productivity as well. Continuation of these relationships has substantial specific value for workers.