The Bowles-Simpson chairman's mark strongly implies that government employees are overpaid (see here). But the evidence from any number of sources is that this is completely false, even when you include benefits and control for everything imaginable. Reihan Salam brought up an corollary argument, Andrew Bigg's argument that you have to include unfunded benefit liabilities in these comparisons. This was taken apart months ago by the National Institute on Retirement, so let's review that.
To get there, I'm going to first run through EPI's Debunking the Myth of the Overcompensated Public Employee (pdf), a pretty fantastic paper. Lots of people have also done this research at the state level: Jeffrey Thompson and John Schmitt of CEPR for New England, California from the Center on Wage and Employment Dynamics at the University of California-Berkeley, etc.
We can break the entire argument about government employee compensation into three steps.
First step: education level
The first is an apples-to-apples comparison of workers among education levels. The government's workforce is more educated than the private workforce. For instance, the government's "college plus" level is 54%, while all private workforce is 35%, for instance. "Some college" is 14% of government workers, 19% of the private workforce. Here is the penalty government workers take when you include all benefits across both categories:
On average, government workers make 3% less total compensation when you control for education levels. As someone who once considered doing regulatory work with his professional Master's degree work, I completely agree that there's a 30%+ pay gap between the private and public sector. Tom Ferguson and Rob Johnson had a good paper at INET comparing the evolution of regulatory salaries and industry salaries -- the results were in line with the real incentive for regulators to be joining the industries they regulated.
Some groups make more in government. This appears to be more the result of the declining wages and benefits over the past 30 years for non-college educated workers in this country than runaway government compensation. The paper notes: "The public sector appears to set a floor on compensation particularly improving the compensation of workers with high school educations, when compared to similarly educated workers in the private sector. This result is due in part because the earnings floor has collapsed in the private sector," citing Lee 1999. But on average, there's a compensation penalty for working for the government.
Second step: gender and race
On the first approximation, government workers are underpaid. So are we done? No. There are two more steps. Beyond education, there are other things we need to control for. State and government workers are "slightly less experienced (21 years compared to 23 years); are more likely to be female (57% to 43%); work fewer hours (42.6 to 43.3); are more likely to be black (14% to 12%); are less likely to be Asian (3% to 6%); and are less likely to be Hispanic (10% to 13%). [p. 9]"
We need to control for all of these factors, as they impact the comparison of wages. Government workers leave 6 minutes earlier each day, women make 80 cents on the dollar as the result of systemic gender discrimination men make $1.25 on the dollar because they are compensated for their evolution-derived risk-taking, aggression and promiscuity, older workers make more than younger workers, etc.
So the paper controls for all that (they put some time into this), and this is what they find:
Once again, controlling for everything imaginable, government workers make less in total compensation than regular workers. The result is less pronounced at the local level (1.8% less rather than 7.5% at the state level), though still significant.
Third step: cost of benefits
Now there is a third critique, led by Andrew Bigg of AEI, which says these two steps don't get the entire costs of providing benefits because "most state and local employees also become eligible for defined-benefit pensions and health benefits in retirement. But state and local governments haven’t come close to fully funding these obligations. That means that the amount government employers spend today may be well less than what employees will actually receive when they retire." He believes this is not the case for private employers, hence government workers are likely overpaid.
Let's hand the microphone to the National Institute on Retirement, which is specifically addressing Bigg's claims (my bold):
Some have argued that because many public pension plans around the country are not fully funded, the entire cost of defined benefit (DB) pension benefits is not recognized in the data we used in our study, which comes from the National Compensation Survey (NCS). While the NCS, like any survey, does have some limitations, it remains the definitive source researchers use for assessing the cost to employers of non-wage benefits...
In other words, if an employer fails to fully pay for the benefits accrued in any given year, then it is possible that the cost of the pension benefit -- and therefore, the full cost of employee compensation -- may be slightly understated in the data. The corollary to this is that if an employer’s contributions to a pension plan exceed the cost of benefits accrued in that year, the NCS will overstate the cost. Because employer contributions can vary with the funded status of the plan, which in turn, is driven by macro-economic factors like the performance of stock and bond markets, there may be a cyclical bias in the data. Over time, though, these over- or under-estimates should average out, which is why we used an average (from 2004 to 2008) of benefit costs, rather than a point estimate in our analysis. Moreover, it is important to note that any bias (positive or negative) would apply equally to public or private sector employers.
So, claims that our analysis systematically understates costs for public employers are invalid on this basis.
Similarly, claims that our study should have added the value of the entire unfunded liability (of state and local government DB plans) onto a single year’s compensation costs are completely off base. Any analysis that does so will reach conclusions that are equally inappropriate and flawed.
First, an unfunded liability represents all accrued benefits, from all years past, that are not currently funded. Yet the point of the analysis is to compare the cost of annual compensation. Therefore, it is inappropriate to include the entire unfunded liability from all prior years into the calculation of the cost of benefits in a single year.
Second, even if one felt that incorporating the cost of unfunded liabilities served a purpose (notwithstanding that this is not how economists define current compensation), one would have to apply the same standard to the private sector side of the analysis, too. Currently, many private sector DB plans have unfunded liabilities -- a result of the recent stock market downturn that affected investors of all stripes. If the idea is to compare public and private pension costs in a fair, “apples to apples” manner, then the unfunded liabilities of private sector pensions should be calculated as well.
Third, adding the expected annual cost of paying off unfunded liabilities (even under the worst-case scenario) would not change the result.
The Center for Retirement Research at Boston College recently calculated that it will cost states on average just 2.2% of payrolls to pay off their entire unfunded liability over 30 years. In the Out of Balance report, we found that total compensation is 6.8% lower for state government employees and 7.4% lower for local government employees than for comparable private sector workers. So, even if we were to add 2.2% of payroll to state and local employee compensation -- which would pay off the entire unfunded liability -- state and local workers would still be paid 4.6% and 5.2% less, respectively, than their private sector counterparts.
In conclusion, the results of the Out of Balance study stand up to scrutiny. Even acknowledging the additional contributions that will be required to restore pensions to full funding does not alter the ultimate findings of the study. State and local employees still receive significantly less total compensation than their counterparts in the private sector.
The idea that private market pensions are in tip-top shape doesn't strike me as accurate, and piling on the entire unfunded liability into a single year is, of course, going to distort the scales and not focus on the specific hypothesis testing that is being carried out. And yes, even if we were expecting to pay off the entire unfunded liability, we are talking about 2.2% of payroll, still less than what NIR, EPI and others found as the pay gap.
Again, there's simply little fat to cut here. Government workers are less compensated than private workers -- more cuts will simply mean a weaker workforce with less human capital, which will lead to weaker government services. Which will discredit the idea of government, which I guess for some people is the point.
Mike Konczal is a Fellow at the Roosevelt Institute.