The Heritage Foundation

Backgrounder #2539 on Labor

March 31, 2011

March 31, 2011 | Backgrounder on Labor

Public-Sector Compensation: Correcting the Economic Policy Institute, Again

Key Points

  1. While federal employees are clearly overpaid in both wages and benefits compared to their private-sector counterparts, pay comparisons for state and local workers are more ambiguous.
  2. Several recent reports, many published by the Economic Policy Institute (EPI), argue that state and local employees receive compensation that is less than or equal to that of comparable private-sector workers—but all of these reports omit important public-sector benefits.
  3. After correcting the omission of retiree health benefits, the undercounting of pensions, and the failure to value job security, state and local compensation can be considerably higher than in the private sector—by 30 percent in California, for example.
  4. EPI’s dismissal of these corrections is a disservice to policymakers and American taxpayers.

Abstract: Previous public–private pay comparisons at the state and local levels, including numerous reports published by the Economic Policy Institute (EPI), significantly undercount public-sector pension benefits, omit retiree health coverage, and ignore job security. Using California as an example, an analysis by economists Andrew Biggs and Jason Richwine showed that these errors can produce a substantial underestimate of public-sector compensation, leading to the erroneous conclusion that public workers receive compensation at or below market levels. In its rebuttal to the Biggs-Richwine analysis, EPI repeats the same errors, necessitating this response.

In recent months, a number of studies from left-leaning think tanks have argued that state and local government employees are underpaid compared to similar workers in the private sector.[1] The Economic Policy Institute (EPI) in particular has released a series of state-specific reports authored by Jeffrey Keefe of Rutgers University, including reports on workers in Wisconsin, Michigan, Ohio, Minnesota, Missouri, New Jersey, and other states.[2] We (the authors of this Heritage Foundation Backgrounder) criticized these reports in a recent Wall Street Journal op-ed[3] and an accompanying White Paper,[4] explaining in detail that these reports ignore the critical issue of job security, significantly understate public-sector pension benefits, and omit the value of retiree health coverage.

For California, the subject of our own recent study,[5] we found that properly accounting for these missing benefits adds approximately 30 percent to public-sector compensation, moving public workers well ahead of comparable private workers.

EPI recently released a rebuttal authored by Keefe titled, “Desperate Techniques Used to Preserve the Myth of the Overcompensated Public Employee.”[6] Our analysis is anything but desperate: Authoritative evidence from academic and official government sources confirms that job security, the guaranteed high return on public pension plans, and retiree health benefits are substantial components of public-sector compensation. Failure to consider them invalidates most public–private pay comparisons.

Job Security

Much of the observed public–private difference in unemployment rates is attributable to the ability of unions to promote job security in the public sector.
—Steven G. Allen, Unions and Job Security in the Public Sector[7]

A basic theory in labor economics is that workers with a comparatively low probability of being discharged should, all else being equal, accept lower wages than workers with a greater chance of being discharged.[8] The theory is especially important for measuring compensation in the public sector, where job security is quite high by private-sector standards. For California public employees, we estimated that the value of their added job security is equivalent to about 15 percent more compensation.

In his rebuttal, Keefe acknowledges that “most labor economists believe that compensating wage differentials do exist.” Yet, he says, “Empirically, if job stability is as highly valued as Biggs and Richwine claim, we should be able to observe its effects on earnings across industries.” He then purports to show that no value for job security can be found in earnings data. But, as Keefe himself admits, parsing out differentials from survey data is “notoriously difficult.” One academic study showed that traditional approaches can understate true wage differentials by a factor of 10, sometimes even mistaking the direction of the effect—meaning, for instance, that jobs with less security would deceptively appear to pay lower wages.[9]

For those reasons, we simply started with the highly plausible assumption that job security has a value, and then estimated that value using a stylized model. The model produces a baseline public-sector job-security value of 6 percent, which hardly seems implausible.[10] This base premium then increases to 15 percent for California employees because of their higher assumed risk aversion, combined with the salary and benefit premium they enjoy compared to similar private workers. Perhaps 15 percent is too high or too low—we are open to alternative estimates. So far, neither Keefe nor the authors of any other recent pay studies have offered any.

Keefe also argues that the value of job security may be countered by other differentials, such as “a lack of control, autonomy, authority, or flexibility” in government employment. These are characteristics of large firms in general, which is one reason why we (and Keefe) already control for firm size in all pay comparisons. But even if other significant differentials remain unaccounted, it hardly follows that job security has no value. It would imply that Keefe’s analysis is even more incomplete than we have already documented.


A private-sector employee would have to contribute a higher percentage of salary to a defined contribution plan to get the same benefit a State employee would receive from the defined benefit plan.

—California Department of Personnel Administration, October 7, 2010[11]

Public–private pay comparisons face a challenge in measuring future pension compensation, which cannot be observed directly. Moreover, defined-benefit plans and defined-contribution plans in the public and private sectors have different funding standards and strategies, meaning that any given level of current funding could imply very different levels of future benefits. Any measurement of pension compensation must account for these differences.

Our method is simple: If two employees are going to receive the same pension benefits from their employers, we assign their pensions the same value. Keefe ignores future benefits and considers only current employer contributions, a method that systematically undercounts the value of public-sector pensions.

Consider the pension plans for North Carolina and Louisiana teachers. Assume that participants in those plans will receive the same benefits, down to the penny. However, North Carolina assumes a 7.25 percent return on plan investments, while Louisiana assumes an 8.25 percent return. Louisiana’s more aggressive funding strategy means it would make smaller annual retirement contributions than North Carolina, at the cost of higher contingent liabilities on Louisiana taxpayers should those investments fall short. According to Keefe’s methodology, which focuses only on current employer contributions, Louisiana teachers would receive lower compensation than those in North Carolina—despite receiving exactly the same benefit. This alone should be a signal that the methodology is severely flawed.

These problems become more pronounced in public–private pension comparisons, where funding strategies differ even more significantly. Private-sector defined-benefit plans discount benefit liabilities at the yield on corporate bonds, currently around 5.5 percent, requiring higher current contributions per dollar of future benefits than public-sector plans, which generally assume 8 percent returns. Defined-contribution plans—401(k) pension plans, for example—make no such calculation, but employers who want to guarantee their employees a future income would discount the desired benefit at the return on riskless U.S. Treasury bonds, currently yielding around 4 percent, and contribute to the 401(k) plans accordingly. By adjusting for different discount rates, our approach measures employee compensation based on what employees will receive. By eschewing such adjustments, Keefe’s estimates are skewed by differences in pension funding strategies.

Keefe responds to this argument with two of his own, neither of which is valid. First, he states that:

In discussing defined-benefit pension plans…Biggs and Richwine lose their focus. They incorrectly assert that ‘employer contributions to pensions are only a proxy by which we infer the value of an actual future pension benefit.’ This is blatantly wrong. The employer contributions are the cost of the employees’ compensation whether they are invested poorly or wisely.

Keefe is correct for a defined-contribution plan, where the employer’s obligation begins and ends with the contribution. But for defined-benefit plans, Keefe is himself blatantly wrong. Defined-benefit pension compensation is not a contribution today but a benefit tomorrow, the value of which derives from the formulas used to calculate benefits and from the government’s legal obligation to pay them. How much a pension plan currently contributes and how aggressively those contributions are invested have literally nothing to do with the actual benefits workers will receive. The employee’s entitlement to future benefits, an entitlement that is protected by law, exists even if zero employer funds were put aside in a given year.

Second, Keefe assumes (contrary to almost all financial economists[12]) that public pensions’ more aggressive investment strategies reflect efficiencies inherent in government plans. This is irrelevant, even if true. If state and local pensions could really generate higher benefits at lower costs than private pensions, those higher benefits should not be excluded from the tally of public-sector compensation. The savings from the alleged efficiency could just as well be devoted to other government outlays or tax reductions rather than higher employee compensation. Thus, the need for accurate measurement remains.[13]

The state of California’s Department of Personnel Administration appears to agree with us wholeheartedly. In its “Total Compensation Survey” it includes an exhibit showing that “a private sector employee would have to contribute a higher percentage of salary to a defined contribution plan to get the same benefit a State employee would receive from the defined benefit plan.”[14] While the specifics of the department’s illustrations differ from ours, it reinforces a critical point—that including only employer contributions masks important differences in actual benefits.

Retiree Health Benefits

State employees receive generous retiree health benefits, the costs of which are often overlooked as a portion of their compensation.

—California Department of Personnel Administration, October 7, 2010[15]

It is easy to overlook the value of retiree health benefits. Most private-sector employers do not offer retiree health coverage to their employees, and it is not included in the Bureau of Labor Statistics data series commonly used to measure benefits. Consequently, as best we can tell, neither Keefe’s studies nor other similar ones make any mention of retiree health benefits.

The omission is significant. The California Department of Personnel Administration provides information on annual expected employer contributions to retiree health care for employees retiring at age 60. (See Table 1.) These average annual employer payments begin at slightly above $9,600 in the first year of retirement, rising to $21,000 in the 10th year and to nearly $50,000 in the 20th year of retirement. Over the course of retirement, the department points out, the typical state employee would receive $493,851 in retiree health coverage—obviously not a trivial value.

A more precise valuation is possible. As the California department points out:

[B]y the time the next survey of total compensation is conducted, public employers will have implemented new standards from the Government Accounting Standards Board (GASB) that require future health costs to be included as outstanding liabilities. This change will allow the survey to more fully compare the “worth” of retiree health benefits.[16]

This is exactly the method we used in our report. California is required by GASB Rules 43 and 45 to publish the “normal cost” of Other Post-Employment Benefits (OPEB)—principally retiree health coverage—which “can be thought of as the cost for OPEB being earned by employees in exchange for [their] services now.”[17] The normal cost of retiree health care for employees under the California Public Employees’ Retirement System (CalPERS) is 10.2 percent of wages, a rather substantial sum to omit from a pay comparison.[18]

Moreover, the 10.2 percent normal cost reflects the cost to the government, not the full value to employees. Lacking retiree health coverage, a retired public-sector employee would purchase coverage in the individual market, where costs are on average 25 percent higher than under group coverage.[19] Thus, the value of retiree health coverage to California state employees would be approximately 12.75 percent of annual wages. Put another way, a California public employee would be more or less indifferent to choosing between retiree health coverage and a 12.75 percent salary increase.

In his response to us, Keefe measures effective compensation through retiree health benefits by dividing the cost of benefits paid to current retirees by the level of wages paid to current workers. This is an invalid approach—it measures effective compensation neither for the currently retired nor for the currently employed. As the GASB points out, current benefit expenditures and future entitlements may be “vastly different.” The GASB states that retiree health benefits “are a part of the compensation that employees earn each year, even though these benefits are not received until after employment has ended.” The normal costs expressed in GASB rules 43 and 45 statements are the proper measure of this compensation.[20]

While comprehensive data on public and private retirement coverage is lacking, certain generalizations are possible. According to the Center for State and Local Government Excellence, 92 percent of states offer their employees retiree health care, while 61 percent of localities do. According to the Pew Center on the States, 82 percent of state and local government employees in governmental units larger than 200 were eligible to receive retiree health care.[21]

In the private sector, retiree health coverage is indeed becoming scarcer. As of 2009, only 6 percent of private firms and 34 percent of the largest firms (1,000+ employees) offered coverage, down from 11 percent and 52 percent, respectively, in 1997.[22] Even among relatively large employers (100–999 employees) only 7 percent offer retiree health coverage. Around 21 percent of all employees were eligible for retiree health care as of 2007, down from 40 percent in 1993.[23]

Moreover, broader availability of retiree health coverage in the public sector is buttressed by greater generosity. Using certain approximations, on a population-weighted basis, state governments pay approximately 70 percent of health premiums for retirees under age 65.[24] In the private sector, employers have both tightened eligibility standards and increased cost-sharing through new formulas or explicit global caps on employer subsidies.[25]

The reduced generosity of private-sector coverage is illustrated in Table 2, which makes a selective comparison of the size of per-employee accruing retiree health benefits in the California state government and several large private firms that continue to offer coverage. In 2010, California employees accrued future retiree health benefits with a value of roughly $1.9 billion, while for General Electric, benefit accruals were only $442 million, and only $55 million for IBM. On a per-employee basis, California’s accruals were anywhere from two to 30 times larger than in these selected private-sector firms. Due to the difficulty of finding private-sector payroll figures, comparisons as a percentage of wages were not possible.

We acknowledge that broad data on retiree health benefits is lacking, making a comprehensive national comparison of state and local employees to private-sector workers difficult. But this is not a good reason for simply excluding retiree health benefits from the analysis.


Previous public–private pay comparisons, including numerous reports written for EPI by Jeffrey Keefe, significantly undercount public-sector pension benefits and omit retiree health coverage and job security. Our own report, using California as an example, showed that these errors can lead to a substantial underestimate of public-sector compensation.

Keefe’s response—denying the undercounting of pension benefits, claiming retiree health coverage and job security have little to no value, and characterizing our position as “desperate”—offers little substance and no corrections. It is a disservice to policymakers and American taxpayers who have a justifiable interest in getting public pay right.

Andrew Biggs, Ph.D., is a Resident Scholar at the American Enterprise Institute, and Jason Richwine, Ph.D., is Senior Policy Analyst in the Center for Data Analysis at The Heritage Foundation.

About the Author

Jason Richwine, Ph.D. Senior Policy Analyst, Empirical Studies
Domestic Policy Studies

Show references in this report

[1]For example, see Keith A. Bender and John S. Heywood, “Out of Balance? Comparing Public and Private Sector Compensation over 20 Years,” Center for State and Local Government Excellence and National Institute on Retirement Security, April 2010, at (March 25, 2011), and John Schmitt, “The Wage Penalty for State and Local Government Employees,” Center for Economic and Policy Research, May 2010, at (March 25, 2011).

[2]The methodology is outlined in Jeffrey Keefe, “Debunking the Myth of the Overcompensated Public Employee,” Economic Policy Institute Briefing Paper No. 276, September 15, 2010, at (March 25, 2011). EPI subsequently published a series of state-specific reports.

[3]Andrew Biggs and Jason Richwine, “The Public Worker Gravy Train,” The Wall Street Journal, February 24, 2011, at (March 25, 2011).

[4]The White Paper is now a published report: Jason Richwine and Andrew Biggs, “Are California Public Employees Overpaid?” Heritage Foundation Center for Data Analysis Report No. CDA11-01, March 17, 2011, at

[5]Richwine and Biggs, “Are California Public Employees Overpaid?”

[6]Jeffrey H. Keefe, “Desperate Techniques Used to Preserve the Myth of the Overcompensated Public Employee,” Economic Policy Institute Issue Brief No. 294, March 10, 2011, at (March 25, 2011).

[7]Steven G. Allen, “Unions and Job Security in the Public Sector,” in Richard B. Freeman and Casey Ichniowski, eds., When Public Sector Workers Unionize (Chicago: University of Chicago Press, 1988), at (March 25, 2011).

[8]The theory of compensating wage differentials was introduced in 1776 by Adam Smith in The Wealth of Nations.

[9]Hae-shin Hwang, W. Robert Reed, and Carlton Hubbard, “Compensating Wage Differentials and Unobserved Productivity,” The Journal of Political Economy, Vol. 100, No. 4 (August 1992), pp. 835–858.

[10]For what it is worth, one recent German survey found that individuals valued public-sector job security at around 12 percent of pay. Christian Pfeifer, “Risk Aversion and Sorting into Public Sector Employment,” German Economic Review, Vol. 12, No. 1 (February 2011), pp. 85–99.

[11]California Department of Personnel Administration, “Total Compensation Survey–Benefits,” at (March 12, 2011).

[12]Most economists agree that public-sector pensions employ inappropriately aggressive accounting and funding practices. See Robert Novy-Marx and Joshua D. Rauh, “The Liabilities and Risks of State-Sponsored Pension Plans,” Journal of Economic Perspectives, Vol. 23, No. 4 (Fall 2009), pp. 191–210; Jeffrey R. Brown and David W. Wilcox, “Discounting State and Local Pension Liabilities,” American Economic Review, Vol. 99, No. 2 (May 2009), pp. 538–542; and Andrew Biggs, “An Options Pricing Method for Calculating the Market Price of Public Sector Pension Liabilities,” Public Budgeting & Finance, forthcoming.

[13]Defined-benefit pensions, both public and private, do have certain administrative efficiency advantages over defined-contribution plans. However, if our analysis included defined-contribution plans’ higher administrative costs, this would tend to reduce private-sector pay.

[14]California Department of Personnel Administration, “Total Compensation Survey–Benefits.”

[15] Ibid.

[16] Ibid.

[17]California Department of Education, “Definitions of Key Terms,” at (March 12, 2011).

[18]State of California Retiree Health Benefits Program, Governmental Accounting Standards Board Nos. 43 and 45, “Actuarial Valuation Report,” June 30, 2009. Some other examples: For workers in Massachusetts, the normal cost of retiree health care is 17.2 percent of wages. For New York State employees, it is approximately 8.7 percent of wages. For teachers in the Milwaukee school system, it is 17.3 percent of wages.

[19]Melinda Beeuwkes Buntin, José S. Escarce, Kanika Kapur, Jill M. Yegian, and M. Susan Marquis, “Trends and Variability in Individual Insurance Products,” Health Affairs, September 24, 2003, at (March 25, 2011).

[20]Governmental Accounting Standards Board, “Other Postemployment Benefits: A Plain-Language Summary of GASB Statements No. 43 and No. 45,” at (March 25, 2011).

[21]The Pew Center on the States, “Promises with a Price: Public Sector Retirement Benefits,” December 2007, at (March 25, 2011).

[22]Employee Benefit Research Institute, “Employers Offering Retiree Health Benefits to Early Retirees: How Has it Changed?” EBRI Fast Facts No. 182, October 28, 2010, at (March 25, 2011).

[23]“Mercer National Survey of Employer-Sponsored Health Plans,” in Employee Benefit Research Institute, “Trends in Retiree Health Benefits Offered by Employers,” EBRI Fast Facts No. 156, February 24, 2010, at (March 25, 2011).

[24]Twelve states do not cover health premiums, merely allowing retirees to buy into the employee plan. Even this limited right is worth more than an additional 1 percent of pay per year due to the implicit subsidy of purchasing coverage at the lower rates applicable to the working-age population. Five states pay between 1 percent and 49 percent of retirees’ health costs, with another 19 paying 50 percent to 99 percent of costs. Fourteen states, including California, pay 100 percent of retiree health premiums. See Robert L. Clark and Melinda Sandler Morrill, “The Funding Status of Retiree Health Plans in the Public Sector,” National Bureau of Economic Research Working Paper No. 16450, October 2010.

[25]“Retiree Health Benefits Examined,” Kaiser Family Foundation and Hewitt Associates, December 2006, at (March 25, 2011).