ONE of the most persistent and troublesome characteristics of the U.S. labor market has been the high unemployment rates of blacks and women relative to white males. These observed differentials have been studied by economists, who have focused primarily on analyzing the flows into and out of unemployment. In addition, the increase in labor market mobility, which has accompanied the shift in the labor force towards more women and youths, has been used to explain increasing levels of unemployment.' But these studies are inadequate on two grounds. First, for policy purposes, turnover that is voluntary must be separated from that which is involuntary. Furthermore, the impact of turnover on unemployment cannot be ascertained simply by looking at duration of unemployment. The interrelationship of duration, number of unemployment spells, and participation decisions require a more comprehensive approach. This article incorporates these points in exploring the impact of voluntary turnover on adult unemployment rates through a micro simulation of the labor market. The observed differences in group mobility patterns are decomposed into movements that reflect the unequal opportunities each group faces in the labor market and movements that reflect variation in the choices made in response to the same job opportunities. As the study shows, the relationship between turnover and unemployment depends upon the extent to which race and sex groups have equal access to jobs. As an interesting side-light, the simulation shows that white female participation rates vary inversely over the cycle because of the large increase in the stock of the unemployed rather than because of increases in the probabilities of leaving unemployment or employment or decreases in the probability of entering the labor market. The article concludes that voluntary turnover may explain a large part of the relatively higher unemployment rates for white females but not for blacks. The caveat is added that policies directed solely at reducing voluntary turnover may result in increasing the duration of unemployment.

As women's labor force participation rates have continued to increase, it has become commonplace to argue that the targets that policy planners set for average unemployment rates should be adjusted upwards to correct' for the fact that women's unemployment rates have historically been higher than men's.1 In this paper, we argue that most corrections are based on an oversimplified approach to labor market realities. This approach has tended to promote the attitude that the high unemployment rates of women are an incurable and unregrettable fact of nature, and has also tended to bias policy discussions in a direction that leads to the toleration of overall slack and puts low or zero emphasis on the labor market problems of women. Abstracting from issues of age and race, we may characterize the usual target correction methodology as starting with the estimation of a simple relationship between women's and men's unemployment rates (U, and U,,) such as