The recent debate over the pay gap between men and women, sparked by President Obama and congressional Democrats’ PR push this week, reminds me of a related controversy. When federal compensation first became a hot-button issue a few years ago, Republican politicians and allied media would point out that the average federal salary was roughly double the average in the private sector.
“Nonsense!” was the collective reaction of the Obama administration, federal employee unions, liberal intellectuals, and the mainstream media. “Federal workers are more educated and experienced than the average private-sector worker,” they said. “It’s apples-to-oranges.”
And they were right. That’s why Andrew Biggs and I conducted federal-private pay comparisons that tried to control for as many relevant factors as possible. The ensuing debate focused on the extent to which we had succeeded in creating an apples-to-apples comparison, not on the appropriateness of attempting a controlled comparison in the first place.
But now left-leaning commentators — including the president himself — eschew controlled comparisons of men’s and women’s earnings. They implicitly attribute the entire wage gap to discrimination. When pressed, they say controls can’t explain all of the difference, and then they go on citing the raw difference as if it meant something.
Unfortunately, unchecked bogus claims will spawn more bogus claims. I ran into a couple today, courtesy of the Anti-Defamation League (ADL). Here’s how the ADL introduces the pay issue (emphasis added):
April 8 is Equal Pay Day, marking the number of days the average woman has to work into the new year to earn what a man in an equivalent job earned in the last calendar year alone. Normally it’s not a day to celebrate. Instead, it serves as a stark reminder that women in the United States still earn only 77 cents for every dollar a man receives.
That’s just plain misinformation. There is nothing about “equivalent jobs” in the 77 cents calculation (see table 1 and figure 2). It’s a straightforward comparison of median full-time earnings for men and women, regardless of the jobs they hold.
The ADL continues:
That fact is morally and socially unacceptable. But it is also economically foolish: the World Economic Forum has said that if women’s pay equaled men’s, the U.S. GDP would grow by nine percent.
That struck me as an odd claim. If women earn less because of discrimination, then paying them more wouldn’t increase GDP — it would just redistribute it.
No link was provided, but I found the World Economic Forum report the ADL was presumably referencing. Turns out that “9 percent” is really “as much as 9 percent.” (See page 31.) And it turns out the Forum did not do any calculations about GDP growth. It simply cited a 2007 working paper by the research arm of Goldman Sachs. And that paper speculates about the impact on GDP of an increase in women’s employment, not pay. (See page 5.)
The ADL’s fact-mangling aside, what should we make of the claim that raising women’s employment to men’s levels could raise GDP by 9 percent? From a policy perspective, it’s just as unhelpful as the raw wage gap. It’s derived from the fact that men’s employment in the U.S. is 9 percent higher than the national average. So total employment would go up 9 percent if women increased their labor-force participation to the same level as men. If women had exactly the same distribution of skills as men and took exactly the same distribution of jobs as men, then this 9 percent increase in employment might become a 9 percent increase in GDP in the long run (when capital adjusts).
Of course, the interpretive problem is the same as with the wage gap. Women don’t have the same distribution of skills or job interests. Besides, some women stay home because they have other priorities, such as raising a family. We might also raise GDP by 9 percent if everyone worked 43 hours and 36 minutes per week rather than just 40 hours. But surely there are trade-offs involved!
Calculations like these are not helpful. They make their way into policy debates only because one side has embarrassingly low standards of evidence.