Sections

Commentary

Op-ed

New analysis turns up surprise on long-term wage trends

Workers operate the first assembly line started at the new Bicycle Corporation of America plant in Manning, South Carolina.
Editor's note:

This article originally appeared in Real Clear Markets on May 17, 2017.

Earlier this month four economists released a study of income trends based on an under-used data source, Social Security wage records. These earnings reports are submitted by employers for the purpose of calculating workers’ tax payments and, eventually, Social Security benefits. The income amounts are quite accurate and cover a full career of wage income. The new study confirms that average male earners have seen scant wage gains over the past generation.

The study also turned up a surprise. When we combine the earnings trends for men and women, the rise in inequality appears much slower than when we examine trends among each sex separately.

There is no perfect data set for measuring the progress of American living standards. Household surveys and administrative data files all have their limitations. Many people do not give accurate responses when they’re asked about their income, wealth, or consumption.

Government administrative records, including the Social Security earnings file, do not provide information about all Americans, nor do they give a complete picture of the incomes people receive. One advantage of the Social Security files is that reported wages are unusually accurate. Another is that the records cover workers’ full careers.

Four economists, including University of Minnesota professor Fatih Guvenen, used the Social Security wage records to examine long-term earnings trends. The records give us a snapshot of how labor incomes are distributed in a single year. Not surprisingly, when we look at a series of snapshots from the 1980s to the present, the Social Security records show a pattern of rising earned income inequality, with top earners capturing a large share of wage gains over the period.

The new study by Guvenen and his colleagues gives us detailed tabulations of workers’ career earnings, that is, the sum of their wage income over their prime working ages. The analysts define these ages as 25 through 55. The authors tabulate individuals’ wage income in the period between 1957 and 2013. This allows them to compute the career earnings of workers born between 1932 and 1958. Their results shed light on four key labor market developments.

Adjusting for inflation, the median male worker born in 1958 earned just 1 percent more during his career compared with the median man born 27 years earlier, in 1932.

First, the career earnings of a typical male worker have not improved much over time. Adjusting for inflation, the median male worker born in 1958 earned just 1 percent more during his career compared with the median man born 27 years earlier, in 1932. In fact, the median male born in 1958 earned 10 percent less during his career compared with the median male born 16 years earlier, in 1942. The lack of progress of mid-level male earners is not a surprise, of course. We know the median real hourly wage received by men reached a peak sometime in the 1970s. It has not surpassed that peak in any year since the 1970s, and in many years it has been far lower. At the same time, labor force participation among prime-age men has edged down for the past five decades. This means a career male worker is now likely to work fewer years between ages 25 and 55 than was the case in the 1950s through the 1970s.

Second, the earnings gains of American women have been much faster than those of men. Whereas the median career earnings of men born in 1958 was about the same as it was for men born in 1932, the median career earnings of women born in 1958 was almost 60 percent higher than that of women born in 1932.

The improvement in women’s career earnings was driven by an increase in the number of years a typical woman works and by an increase in women’s earnings in years when they are employed. Guvenen and his coauthors estimate that among career workers born in 1932, women earned just 17 percent of combined men’s and women’s earnings. The other 83 percent of career wages was earned by men. In contrast, among workers born 27 years later, in 1958, women earned 31 percent of combined career earnings. The share of wages earned by men fell 14 percentage points. The trend of faster median earnings gains among women compared with men has persisted for several decades. It is thus likely women’s career earnings will account for a growing share of combined earned income in the future.

Third, career earnings inequality has increased both among women and among men. Men and women near the top of the earnings distribution have seen faster improvements in their career earnings than workers in the middle and at the bottom. While the median male earner has seen negligible earnings gains over the 27-year period covered by the analysis, workers at the 90th percentile of the male career earnings distribution saw their real career wages climb 35 percent. The woman at the 90th percentile of the female career earnings distribution saw her career wages jump 83 percent. The income gains enjoyed by top-end male and female earners have continued up through the youngest cohorts analyzed in this study.

Finally, even though career earnings disparities have increased separately for men and for women, when we combine the earnings distributions of the two sexes, the upward trend in career wage inequality for both sexes combined has been much slower than is the case for either sex alone. The explanation is straightforward. Women who entered the workforce in the 1950s had annual earnings that placed them at the very bottom of the overall wage distribution. In addition, women typically worked fewer years over their careers compared with men who were born in the same year. Their career earnings were therefore below those even of poorly paid men. As women’s earnings have improved and their years of paid work have increased, women’s career wage gains have boosted the relative incomes of many wage earners near the bottom of the combined earnings distribution.

This does not mean overall career wage inequality remained unchanged. Men and women at the top of the earnings distribution enjoyed faster gains in career earnings compared with earners in the middle and at the bottom. In the case of men, top-end earners were the only group of men to see sizable gains in career earnings. As a result, they saw their share of total career earnings increase. However, the rapid career wage gains of women tended to improve the relative position of some of the poorest paid workers—namely, women in the bottom half of the female wage distribution.

No single data set can answer all our questions about trends in living standards. The Social Security earnings records show how career earnings patterns have evolved over time. They do not tell us how many dependents are supported by each earner, nor do they show how individual earners combine their incomes to support one another and other dependents. American family size has shrunk as the birth rate has declined, which suggests that a typical breadwinner now supports fewer dependents than was the case in the past. We also know that married-couple families are nowadays more likely to include two earners rather than just one. Earnings losses among prime-age men may be partially offset by a spouse’s earnings gains. Other data files can shed light on these issues.

The Social Security earnings records confirm a grim message from our household surveys. Prime-age men who earn average and below-average wages have experienced a decline in career earnings in recent decades. Prime-age women, on the other hand, have seen notable gains in career earnings. Even though women continue to earn less than men, the gap has narrowed considerably. Women’s gains have improved the relative position of career earners who are at the bottom of the distribution. These gains at the bottom do not fully offset the career wage gains of top earners, but they have slowed the rise of overall inequality.