The Protected Class?
Friday, November 19, 2010
Every worker will be asked to make sacrifices as governments put their fiscal houses in order. Private-sector workers should be confident they are treated the same as public-sector workers.
At least four recent studies concluded that state and local government workers are no better paid than their private-sector counterparts. These studies come from the Center for State and Local Government Excellence (CSLGE), the Economic Policy Institute, the Center for Economic and Policy Research, and the Center on Wage and Employment Dynamics at the University of California-Berkeley. They use very similar methods and, in my view, suffer from similar errors. I have written a number of blog posts arguing that these studies inadequately account for the pension and retiree health benefits public-sector workers receive, thereby understating the true level of public-sector compensation.
The Roosevelt Institute’s Mike Konczal, writing at Rortybomb, cites a memo, countering my arguments, by University of Wisconsin-Milwaukee Professors Keith A. Bender and John S. Heywood, who authored the CSLGE paper. I believe (and will argue below) that Bender and Heywood are mistaken. But, in their defense, the issues I raised are not front and center in the public versus private pay debate, they are tricky to explain (meaning, I surely explained them badly and generated estimates of varying quality as I worked through them), and some of the limitations of the compensation data relied on by Bender and Heywood are not well-known.
The state/local pay studies rest on pretty solid ground when it comes to salaries alone. They use detailed data from the Current Population Survey (CPS) to compare state and local salaries to private salaries after controlling for differences in workers' education, experience, and other factors. These methods estimate that state and local employees receive salaries 4 to 11 percent below those paid to similar private-sector workers, depending on the study. While these results differ based on how the researchers control for the effects of unionization, firm size, and other factors, the basic approach is sound.1 Jason Richwine of the Heritage Foundation and I applied similar techniques to federal government pay and calculated a federal pay premium of around 12 percent, a finding consistent with the academic literature. The message: the same methods that show state and local employees to be underpaid show federal employees to be significantly overpaid.
If you properly account both for retiree health coverage and pension benefits, the 4 to 11 percent salary penalty reported for state/local workers is easily pushed into pay premium territory.
But the state/local studies only partly account for public-sector pension and retiree health benefits. To begin, it is important to know that the CPS data used to compare salaries exclude information on benefits. As a result, analysts must infer benefits received by workers using Bureau of Labor Statistics (BLS) data on how much employers pay to fund these benefits. For employee benefits consumed today, that method is fine: a dollar of health coverage, vacation, or sick time received today is worth the same whether it is in the private or the public sector.
But for benefits consumed in the future—specifically, pension and retiree health benefits—this approach can significantly understate state/local compensation.
Retiree health coverage
Most public employees receive free or subsidized health insurance from the time they retire (often in their 50s) through age 65, when they qualify for Medicare. After age 65, this coverage often pays their Medicare premiums and provides "Medigap" insurance. Most private workers do not receive such retiree health coverage, which can be worth several hundred thousand dollars and will increase in value as health costs rise.
Because retiree health coverage is not provided to active workers, however, the BLS excludes it from the employee compensation data that the four studies use. Nevertheless, retiree health coverage is obviously deferred compensation and should be included in public-private pay comparisons.
There is no comprehensive data on how much retiree health benefits are worth. But government disclosures required under the "GASB 45" rule provide some case-by-case illustrations. In New York State, for instance, the "normal cost" of retiree health benefits—that is, the value of future benefits accrued by employees in a given year—equals around 7.8 percent of employee wages. Benefit accruals equal 10.5 percent of pay for California state employees and exceed 14 percent of payroll for Los Angeles County employees.
This compensation is absent in the Bender-Heywood and other public pay reports, but it is not unreasonable to conclude that an average value may be around 10 percent of pay. That is enough to tip most of the studies toward pay parity or even a state/local pay advantage.
Retiree health coverage is obviously deferred compensation and should be included in public-private pay comparisons.
Most public-sector employees are also eligible for defined-benefit pensions. Since we lack data on actual pension benefits, which in any case will occur years or even decades in the future, pay studies infer benefit levels based on what employers contribute to fund those benefits today. The problem is that, due to differences between public and private pension accounting, public employers contribute significantly less per dollar of future retirement benefits than do private employers. Failing to control for these accounting differences will understate public-employee pension benefits.
In deciding how much to contribute to funding pension benefits, public-sector employers generally assume that their pension contributions earn an 8 percent annual rate of return. Since benefits in most states are effectively guaranteed, this is equivalent (in a fully funded system) to guaranteeing employees an 8 percent return on pension contributions.
To accurately track pension benefits, we need to account for both how much employers contribute and the effective rate of return paid on contributions. Bender-Heywood and other studies fail to factor in the latter, causing them to understate public pension benefits.
These issues are tricky to explain and understand, so I will use a highly simplified illustration. Imagine that two employers each guaranteed a worker a lump sum $100,000 payment in 20 years. A public-sector employer would fund that payment with a single contribution today of $21,455, invested at 8 percent (that's 100,000/1.0820, for those who like math). A private-sector employer with a defined-contribution plan would need to invest $45,639 in U.S. Treasury bonds to produce the same $100,000 in 20 years. If we knew only the employer contributions and not the assumed rate of return, we would infer that the private-sector worker was more highly compensated even though we know that both workers receive exactly the same payment.
Roughly speaking, a given dollar of public-sector contributions implies risk-adjusted retirement benefits around 2.4 times as high as in private defined-contribution plan.
Alternately, if two employers contributed the same amount to their workers' pensions, the public employee would be entitled to a significantly larger benefit than the private worker because he is guaranteed a higher rate of return. Roughly speaking, a given dollar of public-sector contributions implies risk-adjusted retirement benefits around 2.4 times as high as in a private defined-contribution plan. So, if the typical public-employer pension contribution equals 7.2 percent of worker pay, the actual compensation generated is equal to a private-sector contribution of around 17.3 percent. So the other approaches miss approximately 10 percent of pay. The precise amount differs from plan to plan, but is too large to ignore.
In the past, I expressed this point in terms of public pensions being underfunded, which they are if you use private-sector accounting standards. In retrospect, that may confuse more than enlighten. The issue here is not whether public pensions recently lost money on their investments or whether governments failed to pay their required pension contributions in the past. Both factors would underfund a pension, but would not affect compensation in future years. The point is that, in any given year, pension accounting differences mean that a public employer will put aside less to fund a given level of benefits earned that year than will a private-sector employer. This difference causes the Bender-Heywood and other state/local pay studies to understate public-sector compensation.
The value of retiree health and pension benefits differs between public employers, and there is also diversity of benefit types and availability among private-sector workers. Both these factors complicate analysis and make it difficult to say precisely what the average public-sector pay premium or penalty is.
Nevertheless, if you properly account for both retiree health coverage and pension benefits, the 4 to 11 percent salary penalty reported for state/local workers is easily pushed into pay premium territory, in some cases significantly so. In California, for instance, a 20 percent public-sector pay premium is not hard to imagine.2
I am continuing to work with Jason Richwine to get a better handle on pay and benefits state-by-state, but I hope this demonstrates at least conceptually what existing state/local pay studies are missing.
These issues are important not only because the state/local workforce is large—almost 15 million workers receive almost $850 billion in annual pay—but because everyone will likely be asked to make significant sacrifices as the federal, state, and local governments put their fiscal houses in order. If private-sector workers and retirees are to make such sacrifices, they should be confident that there is not a protected class treated better than themselves.
Andrew G. Biggs is a resident scholar at the American Enterprise Institute.
FURTHER READING: Biggs has answered the question “Why Does Government Grow and Grow and Grow?” discussed Social Security in “America the Progressive,” evaluated "New Blood for Social Security" from an inflow of immigrants, and questioned whether we should "Tax to the Max" to pay for Social Security. Biggs also writes that "Personal Accounts Are No Cure-All," and he offers "A Conservative's Take on Social Security Reform." Biggs and Jason Richwine maintain, "Federal Employees Are Not Underpaid 22 Percent."
Image by Rob Green/Bergman Group.
1. Because many public-sector employees are unionized, unionization has a predictable positive impact on wages, and state governments have control over whether employees may collectively bargain. It is unclear whether unionization should be considered a separate variable in a wage regression. If not, meaning that allowing collective bargaining is considered a government policy choice, the state/local pay penalty found in Bender-Heywood should shrink. The penalty also should shrink if the regressions do not control for firm size. Firm-size controls assume that state/local employees would work in similarly large firms in the private sector, which tend to provide more generous pay and benefits. But if state/local workers would otherwise work in smaller companies, such as private schools, sanitation removal companies, security firms, etc., then a firm-size control may understate public-sector compensation.
2. The CWED study finds that California workers have a slight overall compensation advantage even before properly accounting for pension and retiree health benefits.