Annuity Options in Public Pensions Plans: The Curious Case of Social Security Leveling

Transcription

1 Annuity Options in Public Pensions Plans: The Curious Case of Social Security Leveling Abstract: Robert L. Clark, North Carolina State University and NBER Robert G. Hammond, North Carolina State University Melinda Sandler Morrill, North Carolina State University David Vanderweide, Fiscal Research Division, North Carolina General Assembly March 2018 * Social Security Leveling is an annuity option allowing defined benefit pension participants to receive a level income throughout retirement by taking a larger pension benefit before Social Security eligibility. This option can enable retirees to access a greater proportion of their employer pension wealth at younger ages and have a smoother income stream in retirement at the cost of a lower pension payment during later retirement years. To evaluate the desirability of this option, a retiree must consider the pricing of the benefit, her own anticipated life expectancy, and her income needs throughout retirement. We show that, due to plan design features, those who expect a shorter than average time in retirement and who have higher personal discount rates would benefit most from selecting this annuity option. This study uses detailed administrative records on all North Carolina state and local government retirees from Among individuals claiming retirement benefits before age 62, over 20 percent chose the Social Security Leveling annuity option. Using multivariate regression analysis, this study finds evidence that North Carolina public sector retirees choose the Social Security Leveling option in a manner that is consistent with the predictions. * An earlier version of this paper was prepared for presentation at the 2016 SIEPR Conference on Working Longer and is available under the title Annuity options in public pension plans: The curious case of social security leveling, NBER Working Paper # This research is part of an on-going project that is being conducted in partnership with the North Carolina Retirement Systems Division and is being funded by the Sloan Foundation, Grant Numbers and G The authors gratefully acknowledge the help and support of Janet Cowell, North Carolina State Treasurer, Steven C. Toole, Director of the Retirement Systems Division, Mary Buonfiglio, Deputy Director of Supplemental Retirement Plans, and Sam Watts, Policy Director of the Retirement Systems Division. The authors would like to thank Nino Abashidze, Bryan Allard, Christelle Khalaf, Siyan Liu, Aditi Pathak, and Emma Turner for research assistance. The opinions and conclusions expressed herein are solely those of the authors and do not represent the opinions or policy of the North Carolina Retirement System or any other institution with which the authors are affiliated. 1

2 For public sector employees, the choice of annuity type upon retirement is a more complex financial decision relative to private sector workers covered by defined benefit pensions. Private sector defined benefit pension plans are regulated by the Employee Retirement Income Security Act (ERISA), which stipulates formulas guaranteeing that payout options are priced to be actuarially fair using standard mortality tables and current interest rates on highquality corporate bonds as published by the Internal Revenue Service. Public sector plans are not subject to these regulations and often do not have default payout options. Moreover, state and local government employees often retire in their 50 s because public sector defined benefit plans typically allow career retirees to begin benefits at relatively young ages. Thus, state and local employees must select the payout method that maximizes their well-being over 30 or more years of retirement. Young retirees must consider the need to finance consumption prior to claiming Social Security benefits as well as the ability to maintain their standard of living in later years of retirement once they begin receiving Social Security benefits. The annuity option choice is a one-time decision that will determine total lifetime wealth for pension participants. This paper examines the characteristics of a particular distribution option, Social Security Leveling, and the potential impact of this form of payout on the lifetime income of retirees. Social Security Leveling is an annuity option that allows individuals claiming retirement benefits before age 62 to receive a higher initial monthly pension benefits at the cost of lower benefits after one becomes eligible for Social Security. For individuals selecting this distribution option, the employer pension benefit received prior to age 62 is equal to the pension benefit plus estimated Social Security benefits after age 62. Some form of Social Security Leveling is used by one-third of state-managed plans where employees are also covered by Social Security. 2

3 Using administrative data on all North Carolina state and local government retirees from , this paper analyzes the decision of workers to select the Social Security Leveling annuity option and highlights the benefits and costs of doing so. Our discussion shows important roles for life expectancy and assumed discount rates in determining who might benefit most from the Social Security Leveling option. Using administrative data on individuals claiming retirement benefits between 2009 and 2014, we estimate a multivariate regression model. The estimates are consistent with theoretical predictions regarding who might benefit most from leveling. After presenting our empirical analysis of annuity choice and the choice of Social Security Leveling, we discuss the design of Social Security Leveling from a policy perspective. The crucial question is how an individual should finance consumption in the early years of retirement and how this impacts future retirement income security. Shoven and Slavov (2014a, b) highlight the value of postponing Social Security claiming and suggest options for borrowing against future pensions. However, because of the discount rate used in the North Carolina retirement plans to price annuity options, choosing Social Security Leveling is a relatively expensive way for retirees to finance consumption in the period prior to Social Security eligibility. I. Background on Social Security Leveling and Public Sector Annuity Options Prior research, focusing primarily on choices by retirees in private sector plans, has explored the tendency of individuals to under-annuitize pension wealth. 1 Less attention has been paid to the choice of annuity type, perhaps because in the private sector this decision is heavily 1 For example, see Benatzi, Previtero, and Thaler 2011; Brown et al. 2008; Brown 2001; Chalmers and Reuter 2012; and Butler and Teppa

4 influenced by the default option of a Joint and Survivor (J&S) benefit. 2 Public sector plans, which are not subject to ERISA, do not typically have a default to a J&S annuity. Thus, we can learn additional information about individuals preferences regarding annuity payout options by studying unconstrained choices within a public sector plan. In contrast to pension distributions chosen by private sector workers, retirees in the public sector rarely select lump sum distributions. Clark, Morrill, and Vanderweide (2014) show that because lump sum distributions in public plans tend to be based solely on employee contributions plus a credited interest rate, at or near retirement, workers present discounted value of a lifetime annuity almost always exceeds the value of a lump sum distribution. In comparison, lump sum distributions from ERISA-covered plans are calculated as the present value of the benefit annuity based on the benefit formula. In addition, public retirees who accept lump sum distributions often lose the option of remaining in the state retiree health plans. As a result, most benefit claimants with long careers who have attained retirement eligibility find that the present value of the annuity greatly exceeds the lump sum distribution amount (see Clark, Morrill, and Vanderweide, 2014). A. Is Social Security Leveling a Widely Offered Annuity Option? There is some limited evidence of Social Security Leveling being used by private plans toward the end of the twentieth century as reported in several articles in the Monthly Labor Review; however, we could find no systematic data indicating the incidence Social Security 2 In general, defined benefit plans have a benefit formula specifying the retirement benefit as an amount per year for the life of the retiree. The plan then offers the option of alternative annuities such as a J&S. The annual benefit of these annuities is determine so the present of lifetime benefits is equivalent to the single life annuity using an interest rate specified by federal guidelines. 4

5 Level Income options. 3 Conversations with senior managers in the Bureau of Labor Statistics and the Office of Policy and Research of the Department of Labor confirm that there were no recent data on private sector defined benefit plans offering Social Security Leveling, and the general assessment was that it is rare to find this type of annuity option in a private plan. In contrast, as shown below, Social Security Level Income options are rather common in large state-managed public plans. Clark and Cowell (2017) reviewed the annuity options of 85 large state-managed public plans which cover teachers, state, and/or local employees and found that 20 of these plans offered a Social Security Leveling annuity option. 4 Employees and teachers in 17 of the plans are not covered by Social Security and obviously, none of these plans offer a leveling option. Thus, about 30 percent of the 68 plans in which participants are included in the Social Security system offer a Social Security Leveling option. Table 1 lists each public plan with this annuity option along with the type of employees covered by the plan, the number of active workers covered by the plan in 2015, and the age at which the pension benefit is reduced if Social Security Leveling is chosen. In total, these plans covered 3.3 million employees and beneficiaries in Fifteen of the twenty plans specify age 3 Wiatrowski (1990) reported that data from the 1988 Employee Benefits Survey indicated that one in eight defined benefit plan participants was in a plan that offered a transitional benefit to early retirees; however, these benefits were typically in the form of a uniform dollar amount for all plan participants regardless of salary or length of service. This type of early retirement incentive is not the same as the annuity option we are examining. Blostin (2003) has a brief statement that indicates that Social Security Leveling was used by some plans but provides no data on how frequent this option is offered and when offered it is selected. 4 These plans are described in bi-annual reports by the Wisconsin Legislative Council (2016). 5

6 62 as the age at which benefits are reduced while two plans set age 65, one age 66, one uses the full retirement age for Social Security benefits, and Virginia allows retirees to select any age between age 62 and the full retirement age for Social Security benefits. 5 These data indicate that Social Security Leveling is an important distribution option in many state retirement plans and thus merits additional analysis. [Table 1] B. North Carolina Retirement Plans and Annuity Options In order to examine the impact of Social Security Leveling on retirees choice of annuity, we examine the public retirement systems of North Carolina. The retirement plan for teachers and state employees and the state-managed pension plan for local employees in North Carolina are typical of state and local pension plans across the country. Teachers and state employees in North Carolina are covered by the Teachers and State Employees Retirement System (TSERS), while local government workers participate in the Local Governmental Employees Retirement System (LGERS). 6 Almost all participants in both plans are also covered by Social Security and TSERS pension annuitants are also allowed to remain in the state health plan. 5 In addition to these 20 plans, Georgia offers its retirees an Accelerated Benefit option that if chosen provides a monthly benefit equal to 135 percent of the single life benefit for the first five years of retirement after which time benefits are actuarially reduced. 6 The important characteristics of TSERS and LGERS are described in: and 6

7 The parameters of TSERS and LGERS are very similar. 7 In order to qualify for normal or unreduced benefits, the employee must have satisfied one of three criteria: reached age 65 with 5 years of membership service; reached age 60 with 25 years of service; or have attained 30 years of service at any age. Early retirement with reduced benefits are available to those who have reached age 50 and completed 20 years of creditable service and those who have reached age 60 and completed 5 years of service. Thus, for many public employees in North Carolina, these plans provide a strong economic incentive to retire in their 50 s, well before qualifying for Social Security at age 62. Upon termination and achieving the age and service requirements, retirees must request from the retirement system that their benefits begin and the annuity option they desire. This is a one-time option and no benefits are paid until the benefit request has been finalized. In other words, there is no default benefit, retirees must submit a request for benefits to be paid, and a choice of payout must be made. Both plans have the same six annuity options, which include a single life annuity, a 100 percent J&S, a 50 percent J&S, Social Security Leveling, and two additional J&S options with a pop-up provision if the retiree s beneficiary dies first. Plan 7 Both plans have five-year vesting, the same eligibility and retirement requirements, and are managed by the Department of State Treasurer. There is a slight difference in the generosity of the two plans in that the benefit formula for LGERS is 1.85 percent of final average salary per year of service while the TSERS formula is 1.82 percent of final average salary per year of service. Final average salary is determined by the four highest consecutive years of earnings. 7

8 actuaries set the terms of all annuity options so that they are considered present value neutral to the system. 8 The Social Security Leveling option was added to the two North Carolina plans in In the 1955 leveling option, the benefit assumed Social Security claiming at age 65, which at that time was the earliest age at which one could collect Social Security benefits. In 1959, the target age for Leveling was reduced to age 62 for women but age 65 was retained for men. Finally, the terms of the annuity option were amended in 1963 so that both men and women could select Social Security Leveling with the leveling age set at While we have found no record confirming the reason why the state modified the Social Security Leveling option to target age 62 instead of 65, one can speculate that the modification was influenced by the change in federal law allowing early claiming at age 62 and the subsequent surge in Social Security claiming at age 62. II. Calculating the Social Security Leveling Benefit The first step in determining the value of the leveling benefit is to calculate the single life annuity benefit, known as the Maximum Benefit Option. 10 The Maximum Benefit Option is derived directly from the benefit formula specified by the retirement system: B MAX = Early M YOS AFC 8 In this context, present value neutral to the retirement system means that the discounted value of the maximum benefit from retirement to death is expected to be equivalent to the discounted value of the two benefits from Social Security Leveling using the assumed rate of return to the plan assets. 9 These changes were the result of changes in federal legislation which first introduced early Social Security retirement benefits for women in 1956 and then for men in See Clark et al. (2017) for full details on the calculations described in this section. 8

9 BMAX refers to the Maximum Benefit Option amount, YOS is the number of years of service at separation, and AFC is the average final compensation calculated using the highest four years of earnings. The pension multiplier, M, is for workers in TSERS and for workers in LGERS. Early is an early retirement reduction factor that is imposed for an individual claiming benefits prior to attaining the age and service requirements for unreduced benefits. The reduction factor is a function of claiming age and the number of years the retiree is short of qualifying for full or unreduced benefits. 11 Retirees selecting the Social Security Leveling option must submit an estimate of the anticipated Social Security benefit that they will be eligible to receive at age 62 provided by the Social Security Administration based on the assumption that they will not have any further earnings between their retirement and age The retirement system then calculates an initial retirement benefit that can be paid immediately to the retiree and a lower benefit that will be paid after age 62. The Leveling benefits in these two periods are priced relative to the maximum benefit. Before age 62 (period 1), the pension benefit amount is equal to the maximum benefit plus the expected Social Security payment (SS) times a leveling factor, F. After age 62 (period 2), the Social Security Leveling pension benefit is equal to the period 1 benefit minus the 11 For most employees, Early is one minus the lesser of 5 percent per year prior to 30 years of service and 3 percent per year between age 60 and 65 plus 5 percent per year prior to age The request for an estimate of the Social Security benefit beginning at age 62 is described in page 35 of If, at retirement, the member wants an estimate under Option 4, he or she must furnish the Retirement Systems Division with an estimate, obtained by the member from the Social Security Administration, of the Social Security benefit available to him or her at age 62. The member should request the age 62 Social Security estimate, in today s dollars, if he or she stops working at the age he or she will be on his or her effective date of retirement. This estimate should be obtained within 2 years prior to his or her effective date of retirement. 9

10 expected Social Security benefit, thus providing a level income throughout retirement of pension plus Social Security payments. 1 (1) B LEV = B MAX + SS F 2 1 (2) B LEV = B LEV SS The calculation of the Leveling benefit does not depend on whether the individual actually intends to claim Social Security at age 62. The system does not check to see if Social Security benefits are initiated at age 62 or whether the benefit received is equal to the estimated value used for leveling. The leveling factor is a function of the gender-specific mortality experience of the system, the ratio of male to female retirees, and the interest rate specified by the system. The objective is to determine the period 1 and period 2 benefit amounts so that the present value of the Social Security Leveling option is equal to the present value of the maximum benefit from the perspective of the retirement system. The benefit calculations use gender-specific mortality rates that are then combined using proportions of males and females that reflect the participants in the plan. This process yields a gender-neutral factor. To calculate the present value of the two benefit streams, the retirement system uses survival probabilities taken from the retirement system s experience study reports and the mortality tables referenced therein. Both of the North Carolina retirement plans assume that they will earn an annual return on investments of 7.25 percent so that benefit adjustments in the leveling option are made using the expected value of the return that the system could have earned 10

11 if they had not moved payments forward using the Leveling option. 13 Thus, theoretically the Social Security Leveling option is cost-neutral to the retirement system relative to the Maximum Benefit option, using the assumptions established by the Board of Trustees of the plans. A simple example illustrates the impact of Social Security Leveling on the amount and time pattern of the pension benefit. We consider a hypothetical retiree claiming benefits at age 57. We suppose the individual is eligible for a pension of $2,000 with the Maximum Benefit option. The retiring worker reports to the retirement system that her expected Social Security benefit at age 62 is $1,200. Using this information and the leveling factor based on the system s mortality experience and the assumed interest rate, the system would calculate both B1 and B2. In this example, the pension benefit would be $2,761 per month until age 62 and then $1,561 per month for the rest of the life of the retiree. Together with a Social Security benefit of $1,200, the post-62 total retirement income is level at $2,761. Figure 1 illustrates the values of the two options for this hypothetical retiree and the reduction in the monthly benefit at age 62. [Figure 1] Social Security Leveling and Maximum Benefit options are designed to be cost neutral to the system. Thus, the present value of the Maximum Benefit throughout the retiree s life is equal to the present value of Leveling (B1 from retirement until age 62 plus the present value of B2 from age 62 to death) with benefits being discounted at the assumed rate of return to the system s assets, currently 7.25 percent. Since the present value of the payout options are deemed by the retirement system to be equal, the system should not care how many retirees select this annuity 13 The interest rate was 7.5 percent prior to Most public defined benefit plans assume that they will earn an annual return of between 7.0 and 8.5 percent on their investments. These relatively high assumed returns have been criticized by many economists (Novy-Marx and Rauh, 2011). 11

12 option relative to the single life option. However, selection among claimants could alter this conclusion if those choosing Leveling have shorter life expectancy. III. The Relative Value of Social Security Leveling The retirement system determines the monthly benefits between the Maximum Benefit and Leveling based on system mortality experience and the assumed rate of return. If an individual has a personal discount rate of 7.25 percent, the present value of Leveling and the Maximum Benefit would be the same to the retiree. However, research indicates that discount rates vary across individuals with many economists arguing that a lower personal discount rate such as 2.9 percent would be a more reasonable representation of the rate used by the average retiree. 14 This lower rate is also more closely aligned with potential market yields to an individual investor. The assumed lower discount rate used by the retiree results in a higher present value of both annuity options. However, for individuals with lower personal discount rates, the present value of the Maximum Benefit would be higher than that of the Social Security Leveling option. Clearly the relative value to the retiree of the two options hinges on an individual s personal discount rate and the desire to smooth consumption throughout the retirement years. Table 2 illustrates the relative value of Social Security Leveling and the Maximum Benefit option for the hypothetical retiree described above. We calculate the present value of each annuity stream using a series of personal discount rates ranging from zero to 14.5 percent. These discount rates can be thought of as reflecting retiree preferences and/or market conditions for borrowing. Given that the formula used to determine the Maximum Benefit, B1, and B2 14 See Shoven and Slavov (2014a, 2014b) for a discussion of appropriate discount rates for recent retirees. 12

13 assumes a 7.25 percent discount rate, it should not be surprising that at a personal discount rate of 7.25 percent the Leveling and Maximum Benefit options are equivalent. [Table 2] Individuals with lower personal discount rates should prefer the maximum benefit option, which yields a higher value of lifetime pension payments. For example, an individual with a 2.9 percent discount rate, would have a 7.0 percent higher lifetime present value if she selected the Maximum Benefit instead of Social Security Leveling option. This difference is even larger with zero discount rate where Maximum Benefit is 11 percent higher. On the other hand, for very high personal discount rates, the present value of the lifetime benefit will be higher for Social Security Leveling. High personal discount rates could be due to time preferences or due to lack of liquidity in the pre-age 62 years. An alternative way to view the impact of discount rates on the lifetime value of pension benefits to retirees is to consider the values of B1 and B2 for the hypothetical retiree above if the monthly benefits had been calculated using an interest rate closer to the current market rate. If the retirement system had used a 2.9 percent discount rate to determine the leveling benefit, the monthly benefit would have been $2,892 until age 62 and $1,692 per month after age 62. Thus, using the lower discount rate of 2.9 percent would result in retirees receiving $131 per month higher monthly benefits before and after age 62. These values represent a 4.7 percent higher benefit in the early years and an 8.4 percent higher benefits after age 62 compared to the Leveling benefits based on a 7.25 percent discount rate. This comparison illustrates the importance of the choice of the interest or discount rate by the system and how this decision affects the value of the Social Security Leveling option relative to the maximum benefit. 13

14 We have three further observations that are important in determining the relative value of the benefit options. First, since the Leveling benefits are based on the average life expectancy of system participants, retirees who expect to live longer than average would have a higher lifetime present value if they select the Maximum Benefit, while those in bad health expecting fewer years in retirement would have higher total benefits from Leveling. This is easily seen if individuals believe that they expect to die prior to age 62. In such a case, benefits would be higher in every year between retirement and death for those with Social Security Leveling. Second, given that the retirement system prices the options based on a blended life expectancy between men and women, the Leveling option is relatively more favorably priced for men with average life expectancy than for women with average life expectancy. Gender may also proxy for other characteristics, however, so must be interpreted with caution. Finally, to the extent that the benefit level is correlated with household finances, we expect that those with higher benefits have alternative sources of liquidity. Social Security Leveling is more attractive to those facing a higher cost of borrowing or a greater need for liquidity. Moreover, higher earners have a longer life expectancy, on average (Galama, et al. 2018). It should be noted that there is no clear prediction regarding age at claiming. Empirically, we fix the age at claiming benefits and so are comparing only individuals at the time when they are making the claiming choice. Individuals that work during retirement or have access to other sources of income are more likely to prefer the Maximum Benefit. And, individuals that retire at younger ages might be more likely to plan to work in retirement. However, a longer time until retirement might mean more of a need to smooth consumption. Holding constant mortality expectations, personal discount rate, and the absolute level of 14

15 benefits, we find no clear prediction on the relationship between age at claiming and annuity option choice. IV. Regression Analysis of Selecting the Social Security Leveling Option We now examine the annuity choices of recent retirees in the two North Carolina retirement systems. Table 3 shows the distribution of annuity choices for public employees who claimed benefits between 2009 and During this time period, 71,968 public employees claimed benefits from either TSERS or LGERS with 56.0 percent of all retirees selecting the single life maximum benefit, 12.0 percent choosing Social Security Leveling, and 32.1 percent chose one of the four J&S options. However, half of these retirees claimed benefits after age 62 and thus, were not eligible choose the Leveling option. Restricting the sample of retirees to the 36,839 individuals who claimed benefits prior to age 62 (Column 2), we find that almost one quarter of retirees (23.4 percent) opted for Social Security Leveling. When the sample is further limited to only the 25,815 individuals who claimed benefits before age 62 and who selected a benefit based on a single life annuity (in other words, the Maximum Benefit or Social Security Leveling options), one third of these individuals selected Leveling. Thus, a significant proportion of North Carolina public sector retirees choose Social Security Leveling. This indicates that Social Security Leveling is not simply a curious anomaly made available to public employees but is an important annuity choice that should be a topic for policy review. [Table 3] To illustrate the pattern of annuity choice by age, Figure 2 plots the annuity type chosen by age at benefit claiming for all persons who retired and claimed benefits between 2009 and Here the annuity types are grouped in three categories: (1) Social Security Leveling, (2) Maximum Benefit, or (3) any Joint and Survivor Option. We can see that the ratio of single life 15

16 to J&S options is relatively constant. Figure 2 illustrates that about half of all individuals claiming a single life annuity prior to age 55 chose Social Security Leveling. This rate declines by age at claiming, with very few selecting leveling at age 61. This relatively high rate of selecting Social Security Leveling among younger retirees highlights the importance of an analysis of why retirees find this to be their preferred form of income in retirement. [Figure 2] V. Who Chooses Social Security Leveling? Almost one quarter of all public employees in North Carolina who retired prior to age 62 between 2009 and 2014 selected the Social Security Leveling annuity option. The previous discussion indicated that individuals with lower life expectancy, higher personal discount rates, and a higher cost of borrowing might benefit most from selecting Social Security Leveling. Using administrative data on all state and local government retirees, we explore whether the patterns of annuity choice and find evidence matching these predictions. Column 1, Table 4 presents the mean values for all individuals who retired prior to age 62 between 2009 and 2014, while the next two columns report the means for women and men separately. For this sample, 33.4 percent of retirees selected Social Security Leveling. The means by gender indicate that a higher proportion of men chose Leveling, 39.0 percent compared to only 31.5 percent for women. This finding is as expected as men have a shorter life expectancy and the Leveling benefits are based on a gender-neutral life table. The average age of claiming benefits is 56.9 years for the entire sample with about 40 percent of the sample retiring prior to achieving the age and service requirements for unreduced benefits. Men and women are quite similar in terms of age of claiming and years of service at retirement. Retirees are rather evenly distributed across the six years for which we have data. 16

17 To further explore the factors influencing the choice of Social Security Leveling, we conduct a multivariate regression analysis. Table 5 reports estimated average marginal effects from a Probit model where the dependent variable equals one if the retiree selected the Social Security Leveling annuity option and zero if the retiree selected the Maximum Benefit option. Holding other personal and economic factors constant, men are 3.6 percentage points more likely to selecting Social Security Leveling as their payout option. This sample ranges in age between 48 and 62 and the estimated marginal effect implies that each additional year of age at claiming lowers the probability of choosing leveling by 3.6 percentage points. Holding age constant, additional years of service are associated with a higher likelihood of selecting Leveling. There are several possible interpretations of the fact that individuals who retired with more years of service are more likely to choose Leveling. Those retiring with fewer years of service are more likely to have previous employment, perhaps in the private sector, and thus retirement income from a previous job. Further, those retiring with more years of service may be more likely to have continued in their public sector employment for additional years due to lower outside earnings potential. Both of these possible explanations suggest a reduced liquidity interpretation of the result, which is consistent with the above discussion. As predicted, those with higher benefit levels are less likely to choose the leveling options. Therefore, the estimated marginal effects of age, years of service, and initial benefit amounts are consistent with the prediction that retirees with larger benefits are less likely to choose Leveling. Finally, there does not seem to be a time trend in the use of Leveling, although slightly fewer retirees choose Leveling in 2011, 2013, and 2014 relative to

18 VI. Conclusions This study finds that the Social Security Leveling annuity, which provides a level income throughout retirement for those retiring prior to Social Security eligibility, is used by approximately one third of retirees in North Carolina who claim benefits prior to age 62. Men are more likely to find Social Security Leveling an appealing payout option, perhaps due to their shorter life expectancy than women. Those who retire at younger ages and those with lower pension benefits are also more likely to choose leveling. This is consistent with predictions that those with lower life expectancy, those who face a higher cost of borrowing, and those with higher personal discount rates have a higher present discounted value of the Social Security Leveling annuity option. An important policy concern is whether the retirees are potentially better off selecting Social Security Leveling over the Maximum Benefit. It is important to consider the positive and negative economic aspects of this annuity option. One advantage of Social Security Leveling is that it provides a way for retirees to smooth their annual income throughout their retirement years. Economic studies have often noted the favorable aspects of a smooth consumption path (Morduch, 1995). Leveling allows retirees to access more of their pension wealth at earlier ages and can be viewed as easing the transition from a career employer to complete retirement. In contrast, the Social Security Leveling annuity option may have several adverse effects on younger retirees. Retirees in their 50s who are healthy may have employment opportunities that would allow them to combine earnings with retirement income. The higher pension benefits available immediately upon leaving public employment may reduce the likelihood that the retiree will seek a bridge job or other types of employment after retirement. Second, the structure of the Social Security Leveling annuity results in the decline in pension benefits at age 62. This 18

19 provides a strong incentive for individuals to claim Social Security benefits at the earliest possible age. Recent studies have shown delaying claiming of Social Security is likely to increase the lifetime value of these benefits (Shoven and Slavov, 2014 a, b). Finally, the Social Security Leveling annuity results in retirees having lower income at older ages when they may have greater demands on income from health care expenditures. From a policy perspective, should retirement plans offer a Social Security Leveling annuity option like what is available in North Carolina? Policymakers have long focused on the role of annuity design in facilitating retirees ability to finance consumption throughout retirement. The design choices are particularly important when retirees often exit the labor force before becoming eligible to claim Social Security. For these retirees, financing consumption in the early years of retirement is particularly salient. Policymakers must therefore decide if the Social Security Leveling annuity option is a preferred vehicle for smoothing the transition from career employment to Social Security claiming. If so, one design change merits consideration. The North Carolina retirement plan assumes that the rate of growth on pension assets is 7.25 percent. Thus, when calculating the adjustment of benefits, the plan uses this same rate of 7.25 percent so that the annuity options are cost-neutral from the system s perspective. However, in essence, retirees choosing Social Security Leveling are taking a loan from the retirement system at a rate of 7.25 percent. As a result, Social Security Leveling constitutes a relatively expensive way for retirees to finance consumption in the period prior to Social Security eligibility. As shown earlier, if Leveling benefits were calculated using market interest rates, monthly benefits before and after age 62 would be significantly higher. This is similar to the finding by Shoven and Slavov (2014a, 2014b), Goda, et al. (2015), and Bronshtein, et al. (2016), who show that 19

20 claiming Social Security early is akin to borrowing from future benefits at a higher than market discount rate. Social Security Leveling can be a component of a well-designed set of annuity options within a retirement plan such as the North Carolina plans we study. Relative to the design in North Carolina, the Social Security Leveling option would represent a more attractive option for retirees if the retirement system used a lower discount rate. This option provides a way for individuals to smooth consumption throughout retirement. The evidence presented here suggests that this option is frequently chosen among North Carolina state and local government retirees. 20

21 REFERENCES Benartzi, Shlomo, Alessandro Previtero, and Richard Thaler Annuitization puzzles, Journal of Economic Perspectives, 25(4): Blostin, Allan Distribution of retirement income benefits, Monthly Labor Review, April, pp Bronshtein, Gila, Jason Scott, John Shoven, and Sita Slavov Leaving big money on the table: Arbitrage opportunities in delaying Social Security NBER Working Paper #22853, November Brown, Jeffrey Private pensions, mortality risk, and the decision to annuitize, Journal of Public Economics, 82(1): Brown, Jeffrey R., Jeffrey R. Kling, Sendhil Mullainathan, and Marian V. Wrobel Why Don t People Insure Late-Life Consumption? A Framing Explanation of the Under- Annuitization Puzzle, American Economic Review: Papers & Proceedings, 98:2, Bütler, M., and Teppa, F The choice between an annuity and a lump sum: Results from Swiss pension funds. Journal of Public Economics, 91(10), Chalmers, J., and Reuter, J How do retirees value life annuities? Evidence from public employees, The Review of Financial Studies, 25(8): Clark, Robert and Janet Cowell, Annuity Options in Public Pension Plans, in Financial Decision Making and Retirement Security in an Aging World, Olivia Mitchell, Brett Hammond, and Stephen Utkus (eds.), Oxford, UK: Oxford University Pres 2017, pp Clark, R. L., Hammond, R. G., Morrill, M. S., & Vanderweide, D Annuity Options in Public Pension Plans: The Curious Case of Social Security Leveling (No ). National Bureau of Economic Research. Clark, Robert, Melinda Morrill, and David Vanderweide Defined Benefit Pension Plan Distribution Decisions by Public Sector Employees, Journal of Public Economics, 116:

22 Galama, T. J., Lleras-Muney, A., & van Kippersluis, H. (2018). The Effect of Education on Health and Mortality: A Review of Experimental and Quasi-Experimental Evidence (No. w24225). National Bureau of Economic Research. Goda, Gopi, Shanthi Ramnath, John Shoven, and Sita Slavov The Financial Feasibility of Delaying Social Security: Evidence from Administrative Tax Data, NBER Working Paper #21544, September Local Government Employees Retirement System. Your Retirement Benefits. Morduch, J., Income smoothing and consumption smoothing. Journal of Economic Perspectives, 9(3), pp Novy-Marx, Robert and Joshua Rauh Policy options for state pension systems and their impact on plan liabilities, Journal of Pension Economics and Finance, 10(2): Shoven, John and Sita Slavov. 2014a. Does It Pay to Delay Social Security? Journal of Pension Economics and Finance, 13(2): Shoven, John and Sita Slavov. 2014b. Recent Changes in the Gains from Delaying Social Security, Journal of Financial Planning, 27(3): Teachers and State Employees Retirement System. Your Retirement Benefits Wiatrowski, William. Supplementing retirement until Social Security begins, Monthly Labor Review, February 1990 pp Wisconsin Legislative Council Comparative Study of Major Public Employee Retirement Systems, December df 22

23 Figure 1. Illustration of Single Life Annuity Options for a Hypothetical Retiree* Pre-62 Leveling Pension Benefit = $2,761 Monthly Income Max Benefit = $2,000 Monthly Post-62 Leveling Pension Benefit = $1,561 Age 62 Current Age *The numbers assume a retiree claims benefits at age 57 and is eligible for a maximum single life benefit of $2,000 per month. We assume the retiree is eligible for a reduced Social Security benefit of $1,200 at age 62 (this implies a PIA of $1,600). The Social Security Leveling benefit would then be $2,761 prior to age 62 and $1,561 after age 62 ($1,561 + $1,200 yield a level income of $2,761). Thus, an individual selecting the Leveling option would have retirement income of $2,761 per month before age 62 (all employer pension) and after 62 (employer pension plus Social Security benefit. 23

24 Figure 2. Age and Time Pattern of Social Security Leveling* Notes: The sample is all benefit claimants from 2009 to 2014 as in Table 3, Column (1). Individuals claiming before age 48 or after age 70 are omitted for clarity. 24

25 Table 1. State Annuity Options Social Security Leveling State Plan Age Covered Participants Plan Information Alaska PERS 65 State and local employees (49,805) Left SS in 1986, DC in Leveling only for DB. Idaho PERS SS FRA State, local, and teachers (109,665) Illinois SRS 66 State employees (131,227) Illinois MRF 62 Local employees (290,860) Indiana PERF 62 State and local employees (217,857) Hybrid plan annuity from DB component. Indiana TRF 62 Teachers (56,560) Hybrid plan annuity from DB component. Kentucky KERS 62 State employees (81,325) Kentucky CERS 62 Local employees (133,503) Michigan SERS 65 State employees (71,019) DB plan frozen 1997, New hires in DC plan. Michigan PSERS 62 Teachers (402,608) North Carolina TSERS North Carolina LGERS North Dakota TRF 62 or SS FRA 62 State employees and teachers (501,920) 62 Local employees (183,592) Teachers (18,539) Rhode Island ERS 62 State employees and teachers (46,409) South Dakota SDRS 62 State, local, and teachers (65,039) Tennessee CRS 62 State, local and teachers (10,926) Vermont SRS 62 State employees (11,669) Vermont TRS 62 Teachers (15,779) Virginia SRS 62 - SS FRA State, local, and teachers (521,158) Wisconsin WRS 62 State, local, and teachers (413,254) Similar Annuity Option Georgia ERS Retiree chooses age between 62 and SS FRA Accelerated Benefit. A monthly benefit equal to 135% of the Maximum Plan Benefit, payable for the first five continuous years of your retirement. After 5 years, monthly benefit is actuarially reduced. Notes: This information is based on a review of retirement system websites. The numbers of active employees covered are provided in the 2015 report by the Wisconsin Legislative Council, 25

26 Table 2. Hypothetical Relative Value of Benefits by Personal Discount Rates* Present Value of Pension Benefits Claimed at Age 57, Social Security Benefit at 62 = $1,200, Maximum Benefit Pension = $2,000 Assumed Personal Discount Rate 0% 2.9% 7.25% 14.5% SS Leveling $569,834 $401,727 $272,972 $179,203 Maximum Benefit $639,017 $429,908 $273,024 $163,598 Percent Difference: (Leveling Maximum)/Maximum -11% -7% 0% 10% *Present values are calculated using the mortality assumptions from the retirement system and the assumed personal discount rate as indicated in column headings. See Clark et al. (2017) for full details on these calculations. 26

27 Table 3. Means of Individuals Claiming Retirement Benefits between 2009 and 2014 All Retirees Claiming Before 62 Claiming Before 62 and Selecting a Single Life Annuity (1) (2) (3) Number of Benefit Accounts 71,968 36,839 25,815 Annuity Type: Social Security Leveling 12.0% 23.4% 33.4% Maximum Benefit Option 56.0% 46.7% 66.6% J&S Annuity Options 32.1% 21.8% Notes: The data include all North Carolina state and local government retirees who claimed benefits between 2009 and

28 Table 4. Sample Means for Retirees Less Than 62 at Claiming All Retirees Male Retirees Female Retirees Number of Observations 25,815 6,359 19,456 Leveling Male Age at claiming (3.603) (3.860) (3.503) Early benefit Years of Service Less than or more Initial Benefit Amount (1K) (1.281) (1.426) (1.228) Agency of Employment State Government Community College Local Government Schools University Year of Claiming Notes: The data are from administrative records and include all North Carolina state and local government retirees who claimed benefits between 2009 and 2014, who were younger than 62 at claiming, and who selected a single-life annuity pay-out option. Sample proportions are shown for dichotomous variables. Sample means are shown for continuous variables, with standard deviations in parentheses. 28

29 Table 5. Choice of Social Security Leveling Among Single Life Annuitants All Retirees Male Retirees Female Retirees (1) (2) (3) Male 0.036*** (0.007) Age at Claiming *** *** *** (0.001) (0.002) (0.001) Years of Service Less than *** *** *** (0.012) (0.025) (0.014) *** *** *** (0.010) (0.022) (0.012) *** * *** (0.008) (0.016) (0.009) Maximum Initial *** *** *** Benefit Amount (1K) (0.003) (0.006) (0.004) Agency of Employment Community College (0.017) (0.032) (0.020) Local Government *** *** *** (0.009) (0.015) (0.011) Schools *** *** *** (0.007) (0.014) (0.009) Universities *** *** *** (0.013) (0.024) (0.015) Year of Claiming (0.010) (0.020) (0.011) *** * ** (0.009) (0.020) (0.011) (0.010) (0.020) (0.011) ** * (0.010) (0.020) (0.011) *** * *** (0.010) (0.020) (0.011) Number of observations 25,815 6,359 19,456 Notes: The data are from administrative records and include all North Carolina state and local government retirees who claimed benefits between 2009 and 2014, who were younger than 62 at claiming, and who selected a single-life annuity pay-out option. The dependent variable is selecting the Social Security Leveling option. Average marginal effects from a Probit model are presented. Standard errors are in parentheses. *** p<0.01, ** p<0.05, * p<