This graph shows the relative distribution of women's to men's earnings. The smooth line encodes the relative frequency of women to men at each level of the earnings scale, the value of this ratio is shown on the vertical axis. The top axis shows the dollar value of the log earnings, the bottom axis shows the rank in the men's distribution. The histogram encodes the fraction of women falling into each decile of the men's earnings distribution. We can see from the histogram that 20% of the women fall in the bottom decile of the men's earnings distribution, and another 18% in the second decile. In all, about 75% of women earn less than the median male (the sum of the first 5 deciles). By contrast, less than 5% of women's earnings reach the top decile of the men's distribution.
The differences between these distributions can be divided into two basic components: differences in location and shape. If the women's earnings distribution is a simple downshifted version of the men's, then after matching the medians (or other location parameter) the two distributions should be the same. Differences that remain after a location adjustment are differences in distributional shape.
The relative distribution can be decomposed into these location and shape components, and the figure below shows the residual shape differences in men's and women's earnings. It is constructed by median-matching the women's earnings distribution to the men's (the median earnings ratio is about 1.4), and constructing the new relative distribution. After adjusting for median differences the relative distribution is nearly flat, indicating that most of the difference between the two groups is due to the median downshift in women's earnings. There are some residual differences, however. Women's earnings have relatively less density in the lower tail (as the bottom decile is below 1) and relatively more density in the upper tail. This probably reflects the dramatic losses experienced by low-earnings men in the last 20 years, and the corresponding gains made by high earnings women. The lower tail difference may also signal a minimum wage effect, as the women's median is lower to begin with, so their lowest earnings are closer to their own median than is the case for men. Readers interested in these trends can find more information in the review article by Morris and Western in the 1999 volume of the Annual Review of Sociology.