Yes, They’re Overpaid
The truth about federal workers’ compensation
"Scapegoating,” claimed the American Federation of Government Employees. “Punishment,” said the Federal Managers Association. “Transparently cynical,” declared Paul Krugman. President Obama’s late November announcement of a two-year pay freeze for federal workers has been poorly received by unions and left-wing activists, who see it as the end result of a year-long campaign to reduce federal salaries. Taxpayers should hope it is just the beginning. Fundamental reform of federal pay would save tens of billions of dollars annually, and it would be a strong indication that lawmakers are serious about reducing long-term deficits in all parts of the budget.
Unfortunately, the debate over federal pay has been fraught with extreme claims. Some politicians have accused federal workers of making double what they deserve, while government unions maintain they are underpaid by around 25 percent. The rhetorical back and forth has largely hidden a substantial academic literature, dating back to the 1970s, that compares the pay of federal and private workers. Economists have addressed the issue with a variety of techniques and from a number of different angles. Over the past year, we have worked to update their results with the most recent data, and our conclusions have been the same as theirs: Federal employees do receive a substantial wage premium by comparison with similar private workers.
The standard approach to comparing the salaries of different groups is to employ the “human capital model,” which assumes that workers are paid according to their skills and personal characteristics. If any group differences in wages remain after controlling for age, education, experience, race, gender, marital status, immigration status, state of residence, and so on, then one group is said to enjoy a wage premium over the others. Economists using this approach find that federal workers generally earn wages 10 percent to 20 percent higher than comparable private sector workers. When we ran a similar analysis with 2009 wage data from the Current Population Survey (CPS), the result was a 12 percent premium. James Sherk of the Heritage Foundation found that the federal premium today could be as much as 22 percent, depending on the specific control variables employed. In general, the federal pay premium is very large for lower and middle-skilled employees and shrinks for the best-qualified federal workers.
Because we have datasets like the CPS—large, representative samples of American workers providing abundant demographic details—the human capital method is the best and most widely accepted method of comparing pay across groups. But the method does have some limitations, which defenders of federal pay use to cast doubt on these results. The press often has obliged by portraying the pay debate as a he-said, she-said question that can never really be answered.
It’s useful then to approach the federal pay question from more than one angle. For instance, we might ask whether economists’ human capital model can really account for all of the relevant differences among workers. Perhaps federal workers have some personality trait—greater motivation, for example—that we cannot measure adequately with our standard control variables. Or maybe our “years of education” variable disguises more prestigious degrees held by federal workers. Neither of these hypotheses seems particularly likely, but we can’t falsify them with the standard human capital model. It’s not feasible to measure directly every single human capital trait.
To address this concern, we can change the approach. Rather than comparing different people at one point in time, we can follow the same people through time. Workers frequently change jobs, and sometimes they switch between federal and private sector jobs. Their change in wages when they make this switch can tell us a lot about federal pay. If workers get a much bigger raise when they switch from private to federal employment than workers who switch from one private job to another, we can infer that the federal government overpays.
Following individuals over time builds into the analysis an ideally rich set of control variables. When people change jobs, they bring with them not just their observable skills like work experience and education, but also their intelligence, their motivation, their specific training, and whatever else affects their productivity. We need not directly measure these variables. All are naturally controlled for when we compare the change in wages people experience when switching between the federal and private sectors.