"Reassessing Trends in U.S. Earnings Inequality" by Robert I. Lerman, in Monthly Labor Review (Dec. 1997), Bureau of Labor Statistics, Washington, D.C. 20212.

That earnings inequality has been increasing in the United States is now conventional wisdom. But just what is "earnings inequality"? The answer is not as straightforward as one might think—and neither is the trend, argues Lerman, an economist at American University.

What data you measure, and how you measure them, goes a long way toward determining what answers you get, he says. Economists often measure inequality as the distribution of annual earnings among fulltime, year-round workers, and even frequently further limit their sample to men or to workers within a certain age range. This may be fine when trying to gauge progress toward some ideal, Lerman says, but it is not the way to assess how large forces such as trade and technological change are altering the overall U.S. wage distribution.

Lerman examined census data from the Survey of Income and Program Participation, as well as the more commonly used Current Population Survey. Defining "earnings" as compensation per hour for all hours worked by all workers in the economy, he got this result: wage inequality increased between 1980 and ’86 (as other researchers have found), but then stayed more or less the same through 1995.

This finding is not necessarily at odds with other, seemingly contradictory trends. For example, the earnings gap between the educated and the less educated appears to have widened since the mid-1980s. But it has been offset by the narrowing wage gaps between men and women, and between blacks and whites.

"Trends in inequality turn out to be highly sensitive to the definition of earnings and the sample of workers used," Lerman points out. An Organization for Economic Cooperation and Development publication shows that between 1979 and 1991, earnings inequality in the United States grew among full-time, year-round workers by nearly 18 percent, but decreased by one percent among all workers, and declined by 11 percent when measured against the working-age population.

Lerman’s conclusion: "Earnings inequality did increase for some groups of workers," and certain forces, such as trade and technology, may have had an impact on the overall situation. But in the U.S. labor market as a whole, the net effect—contrary to the conventional wisdom—has not been higher wage inequality.