Despite the rise in the feminist movement, the enactment of equal opportunity legislation, and the rapid increase in women's labor force participation, a substantial sex differential in wages has persisted in the United States for at least 4 decades. Measured by hourly earnings of year-round workers, this wage gap was 31% in 1955, widened to a 35%-37% range by the 1960s and early 1970s, and then narrowed to 33% by 1982. This paper examines the extent to which changes in the characteristics of men and women in the labor force and other factors can account for the observed pattern. The main finding is that increases in women's labor force participation were initially associated with a declining skill level of employed women relative to employed men, where skill is measured by years of schooling and job tenure. More recently the work experience of employed women has been increasing, which helps account for the recent narrowing trend in the wage gap. During the 1970s the wage gap probably woul have narrowed more significantly had it not been for high levels of unemployment combined with the downward pressure on wages of less skilled workers arising from rapid increases in the supply of young workers and women. Gains by younger women in work expectations, work experience, and college enrollment and other work-related investments point to a further narrowing of the wage gap in the next decade.

Since 1976 the gender gap in wages on average declined by about 1% per year. This article focuses on identifying the factors underlying this trend. Three data sets are analyzed-the Current Population Survey, the Panel Study of Income Dynamics, and the National Longitudinal Survey. We find that convergence in measurable work-related characteristics (schooling and work experience) explains one-third to one-half the narrowing. The remainder is attributable to a relative increase in women's returns to experience as well as to declining wages in blue-collar work and other factors.

Probability distributions for the duration of welfare spells are estimated. The principle guiding the work is that a recipient will not exit from welfare if the expected utility on welfare exceeds the expected utility of welfare. The analysis indicates that while the majority of welfare spells are of short duration, a nontrivial minority of spells are quite long. Those recipients with long spells are found to differ in predictable ways from those experiencing brief spells. This suggests that strategies to move women off welfare are unnecessary in many cases, and should be targeted on those most likely to be long-term recipients.