From the mid-1950s through the 1960s, public-sector pay rose relative to private-sector pay, while beginning in the mid-1970s, relative public-sector pay fell (Freeman 1985). The relatively high-paid public-sector workers of the early 1970s within the span of a decade lost their real compensation advantage over otherwise comparable private-sector workers, seriously denting if not destroying the picture of the ‘overpaid’ public employee that developed in the early 1970s. The groups of public-sector workers who tend to be most highly paid in the USA relative to private-sector workers are blacks and women, suggesting that the public sector discriminates less than does the private sector (Freeman 1985).
In recent years, however, there have been a plethora of studies comparing public- with private-sector earnings and compensation that have challenged the new portrait of public–private worker compensation comparability, and instead, these studies found that the overpaid public employee had returned. To support their proposed changes in public employee law and pay, conservative policy institutes such as the Mackinac Center in Michigan (Hohman 2010) and newspapers (e.g. Cauchon 2011) began publishing comparisons of private- and public-sector employee compensation, showing large public-sector premiums. These studies made simple comparisons of wages and total compensation averages for public- and private-sector employees, using data from the Commerce Department, Bureau Economic Analysis's (BEA) National Income and Product Accounts. The main problem with these studies is that they do not control for education level. The BEA data are comparing the average private-sector job with a worker who has a high school degree with the average public-sector employee who has a college education, and finds, unsurprisingly, that college-educated employees on average earn more than high school graduates.
Who and What Is to be Compared?
Ideally, research would compare workers performing similar work in the public sector with the private sector, but this is not always possible. There are too many critical occupations in the public sector, for example, police, fire and corrections, without appropriate private-sector analogues. Even private and public teaching is significantly different. Public schools accept all students, while private schools are sometimes highly selective and may exclude or remove any poor performers, special needs or disruptive students. Consequently, comparing workers of similar ‘human capital’ or personal productive characteristics and labour market skills is considered the best alternative and well accepted by labour economists. Analyses based on personal characteristics comparisons capture most of the important and salient attributes observed in the comparable work studies (Killingsworth 2002).
Since 2010, there have been over 50 wage estimates reported in policy papers and peer-reviewed research using human capital models that consistently show that state and local public employee wages are less than private-sector employee wages when holding education, experience, hours of work and other demographic variables constant. The results of these studies are reported in Appendix Table A1. While there is little dispute that state and local public employees earn lower wages, there has been a vigorous debate about how to evaluate their benefits and total compensation in relation to private-sector employees.
While there is a broad consensus on a public employee wage penalty, none the less, there are some disagreements about the appropriate wage equation specification involving whether union membership, employer size or occupation (which will be discussed further below) should be included in the human capital wage equation. Bender and Heywood (2010) include unionization in their specification. No other estimates include unionization; Gittleman and Pierce (2012) argue that it does not account for unobserved labour quality in the specification. While there may be an emerging consensus about union membership, there is much greater disagreement about employer size. Allegretto and Keefe (2010), Keefe (2010a, b, 2011a-j, 2012), Biggs and Richwine (2011a), Richwine and Biggs (2011a,), and Munnell et al. (2011) include employer size, while Schmitt (2010), Bender and Heywood (2010), and Gittleman and Pierce (2012) do not include employer size in their specifications.
The argument to include employer size in the comparisons is that it compensates for unobserved productive characteristics of labour. In the USA, large organizations, both public and private, spend considerable resources recruiting and selecting employees, monitoring employee performance, and designing jobs. Through their human resources departments, large firms and government entities recruit applicants and then follow elaborate procedures that may include conducting or commissioning aptitude and capability tests, physical evaluations, drug tests, medical screenings, background and reference checks, reviews of licenses and certifications, structured assessments and simulations, and a variety of other evaluations. Second, large organizations can realize substantial savings in the provision of benefits, particularly health insurance, where large firms self insure and escape the higher market costs of private insurance. Third, the nature of work is different in large organizations. It requires greater cooperation and teamwork, more submission to rules, and the need for conformance to those rules, being subjected to more performance and behaviour monitoring, and engagement with continuous and uninterrupted work flows. On the other hand, those opposing the inclusion of employer size in these wage equations underscore the traditional explanation that larger employers have greater product market power and that workers capture some of these rents, or in the public sector, these rents arise from taxpayers in return for employee and union political support. There is no empirical resolution to these two perspectives on employer size.
In regard to benefits, there is only one reliable source of benefit information in the USA: the Employer Costs for Employee Compensation (ECEC) survey, which is collected by the US Department of Labor, Bureau of Labor Statistics (BLS) as part of the National Compensation Survey (NCS). The ECEC includes data from both private industry and state and local government employees, and provides data for private employers by firm size. Larger employers, over 100 employees, are significantly more likely to provide employees with benefits, in part, because they can spread administrative costs over a larger group, and for insurance purposes, they can more readily diversify risks over a larger group. State and local governments resemble larger-size private employers. The national compensation cost analysis can control for employer size in making comparisons.
Benefits are also allocated differently between private- and public-sector full-time workers in the USA (see Table 1). State and local government employees receive a higher portion of their compensation in the form of employer-provided benefits, and the mix of benefits is different from the private sector. What is important when considering both the employer-provided benefits and direct pay is whether state and local government workers have a total compensation package that costs what they would receive if employed in the private sector. It is the total cost of compensation package — not the mix of cash and benefits — that is important in making a comparison.
Table 1. Comparisons of Private and Public Employee Compensation (Studies between 2009 and 2012)
|Wages and Salaries||69.6%||73.7%||70.2%||66.9%||65.9%|
|Retirement and Savings||4.5%||2.5%||3.4%||4.8%||8.1%|
|Federal Unemployment Insurance||0.1%||0.2%||0.1%||0.1%||0.0%|
|State Unemployment Insurance||0.5%||0.7%||0.6%||0.3%||0.2%|
The ECEC national data reported in Table 1 reveal some obvious differences between private and public employee compensation, when examining the provision of employee benefits. Public employers contribute on average 34.1 per cent of employee compensation expenses to benefits, whereas private employers devote between 26.1 per cent and 33.1 per cent of compensation to benefits, depending on the employer's size. Public employers provide better health insurance and pension benefits. Health insurance accounts for 6.3–8.3 per cent of private-sector compensation but 11.2 per cent of state and local government employee compensation. Retirement benefits also account for a substantially greater share of public employee compensation, 8.1 per cent compared with 2.8–4.8 per cent in the private sector. Most public employees also continue to participate in defined-benefit plans managed by the state, while most private-sector employers have switched to defined-contribution plans, particularly 401(k) plans. On the other hand, public employees receive considerably less supplemental pay and vacation time, and public employers contribute significantly less to legally mandated benefits. The ECEC data are used to mark up wages to generate employee compensation (see Lewin et al. 2012).
There are several factors that have been disputed that influence the appropriate valuation of public total employee compensation. Most studies find no excessive public employee compensation expenses (Allegretto and Keefe 2010; Bender and Heywood 2010; Keefe (2010a, b, 2011a-j, 2012); Munnell et al. 2011). These studies conclude that higher benefit costs are offset by lower public employee wages. However, there are critics of this majority view. One set of criticisms raised by Biggs and Richwine (2011b) and Richwine and Biggs (2011a) argues that the ECEC does not adequately account for the costs of state and local government retiree health benefits, the guaranteed nature of public-sector pensions, and the value of public-sector job security. When they adjust for these alleged omissions, they find that the public employees are excessively compensated by 30 per cent in California and 43 per cent in Ohio when compared with similar private-sector employees. A second line of criticism focuses the human capital model, using a job evaluation method instead or a human capital model with occupational controls. These job evaluation and occupation supplemented models reveal a significant compensation premium for public employees from 3.2 per cent for state employees and 10.5 per cent for local government employees (Gittleman and Pierce 2012). Each of these criticisms will be addressed below.
One set of criticisms focus on several shortcomings of the ECEC benefit accounting, for example, the possible omission of retiree health benefits. While some states pre-fund retiree health benefits, most states have pay-as-you-go retiree healthcare financing. This means that each year, a state must allocate funds from its operating revenue to pay for retiree healthcare. The ECEC does not account for these expenditures because they are not reflected in current employee costs. The US General Accounting Office (GAO) (2007) estimated that retiree health benefits cost states approximately 2 per cent of salary or 1.5 per cent of total compensation. The basic premise of the Biggs and Richwine (2011b) criticism is that retiree health insurance is an irrevocable and unalterable right, which is mandated to be funded by the state irrespective of any changes in the labor force or the state's finances. This premise is false. In most states, public employee retiree healthcare is not a guaranteed benefit. Instead, an accurate assessment of public and private employee benefits does require a small upward cost adjustment where the entity provides retiree health benefits on a pay-as-you-go-retiree health insurance. A second criticism offered by Biggs and Richwine involves the overstatement of the public pension funding ratios. Munnell et al. (2011) summarize the problem as follows:
Comparing ECEC pension data across the public and private sectors involves two problems. First, the ECEC contributions to defined benefit pension plans do not separate the normal cost and the amortization payment to reduce unfunded liabilities. As the employee only earns the normal cost, including the amortization payment overstates public sector compensation. Second, contributions to private sector 401(k) plans and public sector defined benefit plans are not comparable. The public sector contribution guarantees a return of about 8 percent, whereas no such guarantee exists for 401(k)s. Thus, the public sector contribution understates public sector compensation (p. 5).
After making the appropriate adjustments in the ECEC for the proper valuation of pensions and retiree health insurance, Munnell et al. (2011) report that the two roughly balance out. Their estimated difference nationwide for total compensation is a 4 per cent premium in favour of private-sector workers.
Finally, Munnell et al. (2011) and Keefe (2011a) conclude that there is no compensating benefit to job stability in the public sector as alleged by Biggs and Richwine (2011b) and Richwine and Biggs (2011a), as the greater job stability in the public sector is consistent with the higher levels of education of the public employee workforce. Higher levels of education are associated with significantly lower rates of unemployment in US labor markets.
The alternative method used to evaluate compensation is the job evaluation method, which scores jobs based on a variety of compensable factors, and then uses a labor market survey of jobs to evaluate compensation. The BLS collects this type of data in the NCS, which is used to evaluate federal civilian compensation in comparison with private-sector pay. The survey collects information using the employer's most narrow occupational classification or job title and on individuals’ earnings, job work schedules and job work levels. NCS interviewers assign a level of work to all jobs in the survey, which ranges from 1 to 15, corresponding to pay levels in the general schedule that sets levels of pay for federal workers. Work levels and occupations serve as a substitute for education level and experience in the job evaluation comparison method.
Using NCS micro data, Gittleman and Pierce (2012) estimate a model that includes detailed occupational and work-level variables which shows that state government employees earn wages 2.3 per cent below private-sector workers, but total compensation 8.7 per cent above private-sector workers, whereas local government employees earn both higher wages (9.2 per cent) and total compensation (17.6 per cent). The NCS estimate excluded uniquely public occupations where there were no matching private-sector workers. The largest shared public–private occupations, however, tend to be low-wage and low-skilled occupations, such as janitors, general clerks, bookkeepers and secretaries. These occupations are better paid with benefits in the public sector, which sets a floor on wages and benefits. The occupation controls therefore biases their estimates, as the higher-wage and higher-skilled occupations are not shared between the public and the private sector, where there is a substantial public employee penalty. Furthermore, the NCS micro data are not publicly available, which prevents an examination of the drivers of the job evaluation model results.
Their research also estimates a hybrid human capital model with occupational controls that produce state government worker wages that were 4.9 per cent below private-sector wages and local government employee wages that were 3.5 per cent higher than private-sector earnings. This model is a hybrid as it controls for both human capital and detailed occupation. The authors reveal, however, that when examining occupations at the two-digit Standard Occupation Classification that are within the education occupational group, employment is relatively concentrated in kindergarten and preschool for the private sector, in primary and secondary teaching for local government, and in postsecondary teaching for state government. Because education accounts for 54 per cent of state and local public employment, this control significantly biases the occupational control estimate. Similarly in protective service occupations, professional police, detectives and firefighters are employed in the public sector, while low-wage private security guards are employed in the public sector. These controls along with social service occupations force equality where there is none, probably biasing their estimates. Another criticism of using occupations in a human capital model is that education and experience already control for occupational selection, and therefore occupation controls are redundant, which would not be a problem if there was a full occupational match across sectors. Hopefully, further research can clarify the discrepancies between these two basic approaches in evaluating compensation.
Relative Wage Compression in the Public Sector
Since 1970, however, there has been a significant relative compression of the wage distribution in the public sector (Borjas 2002). It is well known that public-sector earnings show less dispersion than private-sector earnings do. Thus, individual earnings differentials favour the public sector at the bottom of the earnings distribution and favour the private sector at the top of the distribution (Belman and Heywood 2004). Using quantile regression on CPS data with occupational controls, Gittleman and Pierce (2012) report that below the median wage, there is a public pay premium for state and local government workers, but, at the seventy-fifth and ninetieth percentiles, the private-sector premium is 7.9 and 11.7 per cent, respectively, relative to state government and 3.4 and 9.0 per cent, respectively, versus local government. However, Lewin et al. (2012), using a standard human capital model with size controls, find that the median hourly compensation is 2 per cent lower for state and local public employees, and at the ninetieth percentile, the penalty rises to 8.4 per cent, whereas at the tenth percentile, there is public employee premium of 3.3 per cent. There is no dispute that lower-skilled and less educated workers in the public sector are compensated better than their private-sector counterparts, while, on the other hand, more skilled and better-educated workers in are more highly compensated in the private sector.
This compensation structure creates possible opportunities for cost savings in government by privatizing lower-skilled, less educated and relatively higher-paid public employee work, while in-sourcing higher-skilled, more educated, and relatively lower-paid public employee work. Advocates of privatization tend to focus on lower-skilled work either to use the threat for privatization to reduce public employee compensation or to privatize that work. However, these same politicians rarely consider in-sourcing higher-skilled work either because they believe that the government gets better quality performance from private vendors or because professional services firms in law, accountancy, information technology and engineering are active in politics and provide substantial sources of local political contributions for both political parties (Keefe and Fine 2010). None the less, calls for privatization became widespread in 2011 as a mechanism to reduce public expenditures, solve state and local governments’ fiscal crisis, and eliminate allegedly excessive compensation of public employees.