Chairwoman McCaskill, Ranking Member Portman, and Members of the Committee:
Thank you for the opportunity to submit written testimony regarding the compensation of federal employees and contractors. Please note that the views we express in this testimony are our own, and should not be construed as representing any official position of The American Enterprise Institute or The Heritage Foundation.
In our testimony we will restrict ourselves, as requested, to the more limited question of the costs of federal employees relative to the costs of equivalent private sector workers. We have written extensively on public-sector compensation in the past and will use this opportunity to review recent research in the area. Tight budgets and a perception that public employees have more favorable salaries, benefits, or job security have prompted research regarding compensation for public sector workers at the federal, state, and local levels. Considerable research on federal employee compensation has been conducted since the mid-1970s, with a resurgence of such work in the past several years.
Comparing Federal Salaries
Public-private pay comparisons are generally conducted using what is known as the “human capital model,” which uses regression analysis to compare pay in government and the private sector, while controlling for differences in worker characteristics that influence earnings, such as education, experience, and a range of other factors. In the human capital model, differences in individual worker productivity are the primary drivers for differences in compensation. Characteristics of the job itself enter the model principally to the degree that they create “compensating differentials” for nonfinancial characteristics of the job that make it particularly desirable or undesirable, such as physical risk, unpleasantness of the work involved, differing levels of job security, and so on. The human capital model is the dominant method by which labor economists analyze public-sector pay.
Our own work begins with an analysis of salaries using the Census Bureau's Current Population Survey.1 Using the standard approach found in the economic literature, we found that federal employees receive salaries about 14 percent higher than those of private sector workers with similar earnings-related characteristics. Consistent with other studies, we found that federal salaries were most generous (relative to private sector levels) for employees with lower levels of education and greater job tenure.
The Congressional Budget Office recently released an analysis of federal compensation that found that federal employees receive salaries on average 2 percent higher than those of similar private sector workers.2 Like us, the CBO utilized the human capital model in assessing federal salaries. But the CBO used a different statistical method that generated a smaller federal salary premium. Our approach is consistent with several decades of academic research, but the CBO's approach utilizes new methods designed to address shortcomings in the past literature.
But regardless of whether the federal salary premium is small or large, what is important here is that both our research and that of the CBO strongly contradict the President's Pay Agent, a bureaucratic body charged with conducting official federal-private salary comparisons each year. The Pay Agent concludes that federal jobs (not workers) pay on average 26 percent lower salaries than similar private sector jobs. This raises a key question: Why do the Pay Agent and the CBO disagree so strongly?
The Pay Agent’s methodology starts with data from the Bureau of Labor Statistics (BLS). As part of the National Compensation Survey, the BLS assigns General Schedule (GS) grade levels to occupations in the private sector and in state and local government. The BLS does not assign GS grade levels to federal jobs based upon its analysis of the job’s work requirements, but instead relies upon the existing GS level assigned by the Office of the Personnel Management (OPM). Using these grade level assignments, and the salaries received by different occupations, the Pay Agent compares federal employee pay to that of supposedly comparable private sector positions. In other words, an actual GS-9 in the federal government is compared to a private-sector job deemed to be equivalent to a GS-9. Based on such comparisons, the Pay Agent concludes that federal jobs pay salaries significantly lower than comparable private sector positions.
There are several reasons to doubt this conclusion, however. Remember that the BLS relies on existing GS grades for federal positions. It is not uncommon for federal jobs to be “overgraded,” meaning assigned a GS level higher than the work requirements of the job merit.3 A 1995 Government Accountability Office study of 1,358 federal positions found that only 230 were correctly graded. Of the remainder, a federal job was over twice as likely to be overgraded than to be undergraded. Moreover, overgrading was significantly more likely at higher GS levels. Overgrading effectively means that federal jobs would be compared to private sector jobs with greater workloads or responsibilities, thus creating or embellishing a federal pay penalty.
Similarly, both the CBO and academic research have found that federal government employees have less education and experience than private sector workers who are performing similar job duties.4 This could be due in part to the overgrading of federal jobs, but it could also happen independently of it. Putting less experienced or educated workers in federal jobs explains how the human capital model might find that a federal worker is “overpaid” even as the Pay Agent concludes that the job itself is “underpaid.” Of course, the government should pay workers for their actual skills, not merely the skills that their job would normally require.
The Pay Agent's approach is further undermined by a recent analysis in the Journal of Economic Perspectives written by two BLS economists.5 The paper compared salaries for state and local government positions to those of private sector positions with similar job requirements, as measured by the GS grade level that BLS had assigned. The authors found that state governments paid salaries about even with private sector levels, while local governments paid salaries around 9 percent above the private sector.
It is difficult to reconcile what the BLS data say about state and local government salaries with the Pay Agent's conclusions regarding federal salaries, for the simple reason that virtually all studies show that federal government workers are better paid than public employees at the state and local level. For instance, our analysis of Current Population Survey data found that federal employees receive salaries 29 percent higher than state and local government workers with similar education and experience.
In summary, the Pay Agent’s salary results are inconsistent with academic research and with analyses published by other federal agencies. The Pay Agent’s methodology is flawed because federal government jobs are overgraded, and federal employees have less experience and education than private sector workers who hold similar positions.
The Pay Agent’s analysis of federal employee compensation critically omits fringe benefits. Benefits can have a significant effect on pay comparisons, since public employee benefits tend to be more generous than those paid to similar workers in the private sector. Table 1 summarizes the results of the CBO’s recent analysis, showing that federal benefits are considerably greater than in the private sector. Based on our own work, the CBO’s figures are, if anything, an underestimate of federal benefits.6
Because of the Pay Agent’s inaccurate comparison of salaries and total exclusion of benefits, Congress should not be guided by the Pay Agent’s work as it re-evaluates federal employee compensation. The CBO’s analysis is a much more useful starting point.
Implications for Federal Government Contracting
In comparing federal and private-sector compensation, we have considered only direct employees of the federal government, not independent contractors hired by the government, for whom much less data exist. However, we can discuss in qualitative terms some of the issues regarding the costs of contracting.
In analyzing the costs of contracting versus the costs of direct federal work, one cannot simply compare the prices contractors charge with the costs of salaries and benefits for federal workers. For one thing, contractors must include the cost of their overhead—say, their purchase of heavy equipment—into the prices they charge the government, but federal overhead is not fully reflected in the salaries and benefits paid to federal employees.
Second, contractors are not guaranteed permanent jobs. Once a federal employee passes through the first several years of service, his probability of dismissal is so low that he has what amounts to a “job for life.”7 In contrast, an individual federal contractor might be dismissed for poor performance at any time, and the composition of the broad contractor workforce can be shifted as the needs of the federal government change over time. This “option value” for federal managers is difficult to quantify, but we can safely conclude that a federal manager would be willing to pay more in exchange for the option to quickly alter the composition of his workforce as needs change.8
Third, federal regulations can prevent the government from selecting the lowest-bidding contractor in the first place. We are not experts in contract law, but documents from the Office of Management and Budget describe a federal mandate that potential contractors not receive an advantage in bidding for government work by offering health or retirement benefits that are less costly than those offered to full-time federal employees.9 It is not clear to us how this regulation is implemented but, as the CBO federal pay report shows, combined federal health and retirement benefits are much higher than those paid to similarly-skilled individuals working at large private sector firms.
We do not believe that any definitive analysis of federal contractor compensation has so far been conducted, especially in light of the three cautionary notes we described above. Lawmakers should consider additional data-gathering and analysis of federal contractors to facilitate higher quality comparisons.
Data and technical issues aside, federal contractors certainly could be overcompensated, just as federal employees are right now. After all, the federal government makes compensation decisions for both classes of workers. The more useful distinction from a policy point of view may not be whether certain tasks and duties should be performed by federal employees versus federal contractors, but whether those functions should be overseen by the federal government at all. In the private sector, competition helps to ensure that workers are paid no more and no less than fair market compensation. Those market forces are much weaker in the public sector, making excessive labor costs likely to persist even with our best efforts to avoid them.
1 See Andrew G. Biggs and Jason Richwine, “Comparing Federal and Private Sector Compensation.” American Enterprise Institute Working Paper 2011-2, June 2011.
2 Justin Falk, “Comparing the Compensation of Federal and Private-Sector Employees,” Congressional Budget Office, January 2012.
3 Government Accountability Office, “Comparison of Job Content With Grades Assigned in Selected Occupations,”
GGD-96-20, Nov 6, 1995.
4 Congressional Budget Office, “Reducing Grades of the General Schedule Workforce,” September, 1984; Melissa Famulari, “Maintaining a Labor Force Under Wage Controls: The Case of the Federal Government,” Working paper, University of Texas at Austin, 1997.
5 This paper is not an official BLS analysis, but its scholarly contribution comes from applying BLS data. Maury Gittleman and Brooks Pierce, “Compensation for State and Local Government Workers,” Journal of Economic Perspectives, Vol. 26, No. 1 (Winter 2011), pp. 217-242.
6 The CBO did not include a subsidy to the Thrift Savings Plan’s G Fund that allows it to pay interest rates about 0.77 percentage points higher than similar government bonds available to private investors. Given the $129 billion size of the G Fund, this subsidy to federal employees totals almost $2.2 billion annually, equal to roughly 2 percent of federal salaries.
We also used a lower discount rate to value future retirement benefits paid to federal employees, which include both DB pensions and retiree health coverage. The interest rate used to value a future benefit amount depends upon the risk of the benefit itself; if the benefit is secure a low interest rate should be used, while if it is risky a higher rate is appropriate. Consistent with the academic literature regarding state and local government pensions, we assumed that accrued federal retirement benefits are effectively guaranteed. This does not mean that Congress could not change the terms under which future benefits are earned, or even fundamentally change benefits for newly hired federal workers. But it does assume that benefits which have already been earned will in fact be paid. The CBO assumed that federal retirement benefits have approximately the same risk as defined benefit pensions in the private sector and thus utilized a higher discount rate. This is not an unreasonable assumption, but we believe it overstates the true risk of accrued federal pension benefits. If our assumptions regarding the risk of accrued federal pension benefits prove to be more accurate, then the true benefits premium received by federal employees is even larger than estimated by the CBO.
7 See Biggs and Richwine, “Comparing Federal and Private Sector Compensation.”
8 See Foote, D.A. and T.B. Folta (2002). “Temporary workers as real options.” Human Resource
Management Review, vol. 12, no. 4, pp. 579-597.
9 See “Competitive Sourcing Requirements in Division D of Public Law 110-16 Memorandum For Heads of Executive Departments And Agencies From Paul A. Denett, Administrator, Office of Management and Budget,” February 20, 2008.