The Magazine

Yes, They’re Overpaid

The truth about federal workers’ compensation

Feb 14, 2011, Vol. 16, No. 21 • By ANDREW G. BIGGS and JASON RICHWINE
Widget tooltip
Single Page Print Larger Text Smaller Text Alerts

The specific econometric procedure is called “fixed effects,” because it focuses on wage changes for individual workers, who have many characteristics that are fixed from year to year. One of the first economists to apply fixed effects analysis to the federal pay issue was Princeton’s Alan Krueger in 1988. Using a dataset called the Displaced Workers Survey, Krueger found that workers who lost jobs in the private sector and then joined the federal government earned about 12 percent more than displaced workers who found another private sector job. (Somewhat ironically, Krueger would go on to become President Obama’s chief economist at the Treasury Department.) 

A forthcoming Heritage Foundation report updates Krueger’s analysis using a much larger and more representative dataset known as the Survey of Income and Program Participation (SIPP). The SIPP follows tens of thousands of people over several years, carefully documenting their wages and labor force status on a monthly basis. Combining the SIPP waves that began in 2004 and 2008, the fixed effects analysis indicates a federal wage premium of at least 8 percent.

A similar approach confined to postal workers reached a similar conclusion. In the late 1990s, the Postal Service surveyed all new hires, asking them how much they were paid in their previous job. Overall, new postal hires received salaries over 28 percent higher than what they had been paid in the private sector, which University of Pennsylvania law professor Michael Wachter and his co-authors called “enormous wage increases over their previous wages in full-time private sector jobs.” 

If fixed effects analysis works so well, why use the human capital method at all? Because fixed effects analy-sis has its own limitations. For one thing, the smaller samples make measurement error more of a problem. If some private workers are incorrectly identified as federal workers or vice versa, then the federal wage premium will appear smaller than it really is. Furthermore, the SIPP covers a relatively small part of a worker’s life cycle. We know from the human capital studies that the federal premium tends to get larger as experience grows. Since the SIPP data capture pay in only the first year a worker switches to federal employment, the observed 8 percent pay premium probably underestimates the overall pay gap. 

Human capital and fixed effects models tell us a lot about wages in the federal versus the private sector, but they tell us nothing about nonwage compensation. To supplement the findings on wages, analysts commonly estimate the value of pension and health benefits offered by each sector and then add them to the wage results from the human capital model. But even this is incomplete, because benefits come in many forms that can be hard to quantify. Even a low-salary job without a 401(k) or a health plan could be relatively attractive if it offered other forms of compensation, such as generous sick leave, lengthy vacation time, reliable job security, and flexible scheduling. Federal employment offers all of these, but how do we incorporate every perk into the federal-private comparison?

One method is to use quit rates. Federal workers quit their jobs at less than one-third the rate of private workers, which suggests federal employees don’t feel they can get a better combination of salary, benefits, and job perks in the private sector. Just as fixed effects naturally accounts for many hard-to-measure skill differences, quit rate analysis automatically encompasses the full range of compensation in each sector. 

For years, defenders of federal pay have attributed low quit rates to the fact that federal employees receive traditional defined benefit pensions, which reward long job tenure and discourage midcareer employees from leaving. Richard Ippolito, the author of a 1987 study that made this claim, suggested what he called a “litmus test” for his theory: Switch federal employees from traditional defined benefit to 401(k)-type defined contribution plans, then see if quit rates change. “If federal workers are paid too much relative to their quality level,” Ippolito wrote, “the quit rate will not change much; if their pay is too low, the quit rate will increase markedly.”

Recent Blog Posts

The Weekly Standard Archives

Browse 19 Years of the Weekly Standard

Old covers