In the past 6 months, conservative think tanks have been filling the inter webs with studies of public employee pensions. These studies, intended to inflame legislators and the public, have had varying effects. One group, the Arnold Foundation, has funded research and funded documentaries on PBS, has attacked the very foundation of public employee pensions and have advocated for a reduction in public employee benefits. Another group, the American Enterprise Institute, has followed suit in trying to point out that public employee pensions are so generous to defy rationality. These sensational studies have taken hold with the public and with state legislatures. =
Today, I want to share with you a critique of the latest study published by the American Enterprise Institute. I don’t usually do “guest posts” here, but I wanted you all to read this first here so that you can have ready ammunition to undermine the dubious statistics used by the AEI to attack public employee pensions, My colleague, Val Burris, Professor Emeritus at the University of Oregon, has generously shared his critique of the latest AEI study of public employee pensions. Val shows that everything about the study is “cherry picked” and that the conclusions do NOT follow logically from the data. The next time someone throws these data in your face, you can throw them back because the AEI’s study does not follow logically from any data published. Neither Val nor I can speak to other pension systems, but when it comes to Oregon’s PERS, the AEI document is badly flawed and misleading. This does not bode well for interpretations of other states. What follows is Val’s analysis of the latest screed from the AEI.
Not So Honest: The AEI’s Latest Attack on Public Employee Pensions
Val Burris, Professor Emeritus, University of Oregon
This month the American Enterprise Institute (AEI) published a report titled “Not So Modest: Pension Benefits for State Government Employees.” The AEI is a conservative think tank in Washington DC that is financed mainly by large corporations, too-big-to-fail banks, and assorted right-wingers within the American plutocracy. Andrew G. Biggs, the author of the report, is a public policy huckster who has spent most of his career in the employ of right-wing think tanks like AEI and the even more far right Cato Institute, spreading disinformation in the interest of big business and the super rich. For several years he also served as a low-level official in the George W. Bush administration promoting the privatization of Social Security. The AEI study is representative of the kind of the ideologically-driven, pseudo-scientific analysis that spews forth from the well-financed network of right-wing think tanks – promoting tax breaks for the rich, cuts in public services, deregulation of corporate malfeasance, and climate change denial. The study presents comparative data on public employee pension benefits in all 50 states, but my focus in this comment will be specifically on the findings presented for Oregon PERS retiree benefits in each of the report’s four main figures.
Figure 1. In this figure the author uses data from Comprehensive Annual Financial Reports (CAFRs) of various states to estimate pension benefits of full-career retirees. The crucial issue here is what counts as a “full-career” retiree. As we shall see, the author’s definition is both arbitrary and bears little resemblance to the data he presents. On page one the author specifies a full-career retiree as someone retiring “after 30-34 years on the job.” However, there is considerable variability in how this definition is applied across states. A few states report benefits for a specific category of retirees with 30-34 years of service, but more often they collapse retirees into broader categories like 30+ years of service, or 31+ years of service. Such categories typically include a quarter to a half of retirees who have put in 35 or 40 or even 45 years on the job. Therefore, in many if not most cases, the data in Figure 1 do not correspond to the definition of a “full-career” retiree that is stated in the text. Moreover, as one might expect, retirees with 35 or 40 years of service typically receive higher pensions than those with lesser service. If this is true, then the average benefit reported for many states will be considerably higher than the benefit of a worker retiring with 30-34 years of service, which is the only definition of a “full-career” retiree that the author provides anywhere in the report.
As I have just explained, data for different states are not comparable because states do not report the years of service of their most senior retirees in the same categories. For Oregon the highest category reported is 31+ years of service, which is the category that the author actually used in his report. Therefore, despite what one might be led to believe, when the author computed his estimates of retiree benefits shown in Figure 1, an Oregon public employee retiring with 30 years of service was not counted as a full-career retiree, whereas an employee retiring with 35 or 40 years of service was counted. By comparison, for New York State, which reports its data in more detailed categories (and which the author disingenuously uses to illustrate what the data supposedly mean and how they were assembled), a retiree with 30 years of service was counted as a full-career retiree, but one with 35 or 40 years was not. Hence, the author is mixing apples and oranges, and it would be unwise to rely on his data for comparing one state with another.
Returning to the New York State example will give us a clearer sense of just how non-comparable the data for different states may be. In Figure 1 the author cites an annual benefit of $43,020 for a full-career New York State government retiree. This figure refers only to those retirees with 30-34 years of service – nothing higher. The first thing to note is that these retirees account for less than half of New York State retirees in 2012 with 30 or more years of service. Second, if the author had followed a method more closely corresponding to that applied to Oregon (but never mentioned in the text) and counted all of those with 30+ years of service, the average annual benefit would have been $51,291, or nearly 20% higher than reported.
The annual benefit of a full-career Oregon PERS retiree reported in Figure 1 is $58,188, which corresponds to the figure for retirees with 31+ years of service in the 2012 CAFR. But I suspect many readers might think that someone with only 30 years of service should also be placed in the “full-career” category (or 28 years of service or maybe 26). According to the Oregon CAFR, PERS members retiring in 2012 with 26-30 years of service received an annual benefit of $42,096. If we were to define full-career retirees as those with 26 years of service or greater, the average annual benefit (combining the 26-30 and 31+ categories) would be $48,712, or roughly $10,000 less than the figure reported for Oregon PERS full-career retirees. So, once again, one’s findings are highly sensitive to differences in how terms are defined and how data are assembled.
Apart from comparing apples and oranges, the author of the AEI report is also cherry picking to exaggerate public employee pensions by focusing on a limited number of cases. Oregon PERS members retiring in 2012 with 31+ years of service account for only one-eighth of those retiring that year and slightly more than one-quarter of total benefits that were initiated that year.
Figure 2. In this figure the author uses data from state CAFRs and the BLS Survey of Occupational Employment Statistics to compute what he calls the “total retirement income” of state government retirees. This involves combining the questionable estimates of the pension benefits of public employees from Figure 1 with equally questionable estimates of their Social Security benefits. The sum is then reported as a percentile of the yearly earnings of full-time, full year workers in the same state.
How exactly does the author estimate Social Security benefits of retired public employees? On page 3 he says that Social Security benefits were calculated on the basis of a retiree’s final salary at the time of their retirement. This suggests that final salary accurately reflects the average annual income that retirees received over the entire 35 years that were used in calculating their Social Security benefits. This is obviously an absurd assumption, especially in the case of full-career employees who can be expected to have started at a relatively low salary and then moved up in rank and salary over the decades.
A more realistic assumption would be that an entry level public employee might begin working at roughly half of their final salary (in real dollar terms) and move up gradually over the course of their career. Most employees also have a number of low income years, either before they entered into a steady career or because of breaks in employment. So the idea that full-career employees have always earned the same salary they received at the time of retirement exaggerates their Social Security contributions and subsequent benefits by a considerable margin. Using my own case as an example, my average salary for purposes of calculating Social Security benefits was 74% of my final salary; hence, using my final salary to predict my Social Security benefits would grossly exaggerate the amount of those benefits.
Also troubling is the fact that the Social Security benefits calculator used by the author appears to report monthly benefits at the age of “full retirement” (age 66), whereas most people begin collecting Social Security at an earlier age where they receive up to 25% less than the full retirement benefit. Moreover, the youngest age at which one can receive Social Security benefits is 62, whereas the author, elsewhere in his analysis, assumes that full-career public employees are retiring at an age of 60 when Social Security is not even available to them. I will have more to say on this later.
Working backward, it is possible to calculate that the author estimated Social Security benefits for the average full-career Oregon PERS retiree to be $24,302 per year. More realistically, if we were to define full-career retirees as those with 26 or more years of service, estimate that the average salary for their top 35 income-earning years was 75% of their final salary, and assume an average age of 64 for beginning Social Security benefits (the national average is 63.8 years), our full-career Oregon PERS retiree would receive Social Security annual benefits of $19,322 beginning at age 64 – or less than 80% of the estimate in the AEI report.
The author purports to measure total retirement income of public employees as a percentile of the income of “full-time, full-year” workers in their state using the BLS Survey of Occupational Employment Statistics. There are two problems here. First, the BLS survey does not distinguish between full-time and part-time employees; the two are combined in its salary estimates. According to the 2012 Current Population Survey, roughly one-quarter of Oregon employees worked part-time on a regular basis, so the author of the AEI study is comparing PERS pensions to the salaries of a large number of part-time workers. Second, the BLS survey is conducted in a single month and cannot say whether workers will continue to receive comparable salaries throughout the year, especially in a time of recession and high unemployment. For both of these reasons, the earnings of workers in the BLS survey are certain to be lower than those of genuine “full-time, full-year” employees and the percentile ranking of PERS retiree pensions will therefore appear higher than is actually the case.
Based on the multiple fudge factors enumerated above, the author of the AEI study reports that a full-career Oregon PERS member retiring in 2012 received a total income (pension plus Social Security) that ranked at the 90th percentile of Oregon’s current full-time, full-year employees. With more accurate measures of pension and Social Security benefits, as well as a very rough adjustment for the presence of part-time employees in the comparison group of current workers, I estimate that a full-career PERS retiree in 2012 would have earned slightly above the median (50th percentile) based on their pension alone and somewhere between the 70th and 75th percentile based on pension plus Social Security income (if indeed they were of sufficient age to draw Social Security benefits). These results should be treated with caution because of the difficulty of estimating the distribution of salaries of full-time workers from data that conflate both full-time and part-time workers. But, without question, they are more accurate than the figures presented in the AEI report.
Figure 3. This figure purports to represent the current value of lifetime pension benefits of full-career state government employees using some creative assumptions and methods of analysis. First note that this analysis incorporates all of the questionable data and measures encountered in the previous two figures. In addition, the figure raises several additional problems. As noted at the bottom of Figure 3, the analysis “assumes retirement at 60, survival to 85, 2.5 percent COLAs, and 3.5 percent discount rate.” Let us address each of these as they apply to the estimate of lifetime pension benefits of an Oregon PERS retiree.
The assumption of retirement at age 60 is not too far out of line with the average reported by Oregon PERS, but it is at variance with the assumptions of the previous two figures, which focus exclusively on public employees with relatively extended years of service and who are eligible to draw on Social Security (available only at age 62) at the same time they begin to receive their pensions. Clearly, the author is altering his assumptions about age of retirement to fit whatever biased message he wishes to promote in each individual figure.
According to the Social Security Administration, the life expectancy of a 60-year-old retiree is 81 for males and 84 for females, so by assuming a life expectancy of 85 the author is trying to sneak in a few extra years of expected benefits to boost his predetermined conclusions. Further, the COLA for Oregon PERS retirees is nowhere near 2.5 percent as a result of recent legislation. It is currently 1.25 percent for the average full-career PERS retiree. The assumption of a 3.5 percent discount rate also exaggerates the present value of public employee pensions. Oregon PERS pensions (and those of most other states) assume a long-term market return of between 7 and 8 percent. By using a discount rate of half or less of this value the author is guaranteeing a highly inflated estimate of the present value of a given pension income stream.
The author of the AEI study claims that the average full-career Oregon PERS retiree in 2012 received a pension with a current value of $1,204,776, making them an example of what he calls “pension millionaires.” What happens when we make more realistic assumptions of (i) defining full-career public employees as those who have 26 or more years of service at retirement, (ii) who retire at age 63 (the average PERS retiree in 2012 was age 61 with 22 years of service, so 63 years of age seems like a reasonable estimate for those retiring with 26+ years of service), (iii) whose life expectancy is 83 years, (iv) who receive a 1.25 percent COLA, and (v) where the discount rate is 7.5 percent? Under those assumptions, the average current value of a full-career PERS retiree’s pension is roughly $560,500 – much less than half of the AEI study’s estimate and considerably less than a million dollars.
Figure 4. In this figure the author introduces his own method for computing replacement rates for public employee pensions by returning to the ploy of juicing pension benefits with inflated estimates of Social Security income to make the case that public employee pensions are excessive. By this method the author estimates that “total retirement income” replacement rates are 105% of final salary for full-career Oregon PERS retirees. Compare this with the statement in PERS By the Numbers which says that pension benefits for PERS members retiring in 2012 with 30 or more years of service equaled 70% of final salary. The 2012 Oregon PERS CAFR reports roughly the same replacement ratio for employees retiring with 26 or more years of service.
Even though I consider it illegitimate to conflate pension income with Social Security income, I replicated the author’s method using more realistic estimates of pension and Social Security benefits of full-career PERS retirees and came up with an average replacement rate of 97 percent of final salary for pension and Social Security combined. It is important to remember that this replacement rate applies only to a small minority of PERS retirees with the longest years of service. The comparable replacement rate for the more typical PERS retiree with 21-25 years of service is only 46 percent of final salary for pension alone and roughly 70 percent for pension plus Social Security (although the latter figure could easily be exaggerated because of the somewhat generous assumptions I made about these employees’ pre-PERS or non-PERS income for purposes of estimating Social Security benefits).
Concluding Remarks. There are two more sweeping problems with the AEI study. The first is that it ignores the very different mix of occupations among state and private sector workers and their required skills and qualifications. For example, 52 percent of state or local government employees have a college degree compared with 35 percent in the private sector. Second, and more egregious, is the complete failure to examine, or even consider, the issue of total compensation (salary plus benefits) as the appropriate standard by which to assess economic returns to careers in public service and the relative benefits accruing to public versus private sector workers. As many studies have shown, public employees tend to accept lower salaries in exchange for some degree of job security and the promise of an adequate pension. This is a topic about which right-wing think tanks have also generated a deluge of disinformation. But, according to what is arguably the most reliable and methodologically rigorous study of this question from the nonpartisan Center for Retirement Research at Boston College, state and local public employees experience a wage or salary penalty of 9.5 percent versus comparable private sector employees; whereas, they make up roughly 5.5 percent in terms of more generous pensions, health insurance, and other benefits. This results in a net deficit of 4.0 percent in total compensation for state and local government employees. To focus on the magnitude of public employee pensions in isolation of the wage and salary penalty experienced by public employees (and hence their greater difficulty of accumulating private savings for retirement) is an extraordinary omission of the sort that only an ideologically blinded spokesperson of a politically motivated organization would ever consider reasonable.