Public-Employee Pensions and Retirements

Abstract

Around the time of their adoption, reformers argued pensions would help encourage public employees to retire rather than remain in their jobs indefinitely. However, the significance of specific pension policies for retirements within numerous plans is not well-known. Using new data on service retirements from 81 state-employee defined-benefit plans between 2001 and 2019, I examine pensions’ relationship with retirements, while also accounting for other variables that might matter. The results indicate that early retirement incentives, relaxed post-retirement work restrictions, and higher employee pension contributions are all associated with more retirements, while increased employer contributions and cost-of-living adjustments are associated with fewer. As governments continue to consider fiscal reforms to retirement plans, they also should keep their personnel needs and pensions’ relationships with them in mind.

Keywords

pensions states retirement personnel

Governments frequently insist that pensions aid in public management, facilitating the production of public goods and services. For example, Arkansas’ 2012 Comprehensive Annual Financial Report (CAFR) states: “The retirement system benefits the entire state and all Arkansans, not just public employees.” Pensions, which provide regular retirement income to vested employees, might offer one tool to help manage the supply of personnel via recruitment, retention, and retirement. Around the time of their adoption, reformers argued that pensions would help efficiently manage these behaviors in routinized ways (Clark, Craig, and Wilson 2003; Graebner 1980; Lazear 1979). Pensions also might be especially useful in the public sector, where there long has been concern talented employees either choose not to enter or leave for private-sector work (Gorina and Hoang 2020; Lewis 1991; Perry and Wise 1990). Here, I ask what relationship pensions have with state-employee retirements within plans and over time.

Pension formulas are products of employees’ years of service and average earnings. Plans typically stipulate normal and early retirement ages. Employees and employers make contributions, which are then invested. I focus on defined benefit (DB) pensions, which place risk with employers. While some governments have expanded defined contribution (DC) pensions in recent years, which place risk with employees, 89% of public employees still belonged to DB plans in 2014 (Ali and Frank 2019).¹ Aside from employees, politicians also benefit from pensions, which spread costs into the future. Complicated rules and payment automaticity also reduce visibility and insulate officials from blame (Anzia and Moe 2017; Weaver 1986).

Pensions incentivize labor by tying retirement income to tenure, but also subsidize exits by providing income to non-workers. Employees need to remain with the same employer or within the same system in order to maximize their payouts. Individual workers could react to these incentives in different ways, though: some might leave earlier, while others keep working. Pensions also are packages of policies, some of which could matter more than others. Plan complexity also could muddle workers’ awareness or understanding, undermining pensions’ behavioral incentives (Brown 2013; Clark, Morrill, and Vanderweide 2014; Mitchell 1988).

There is extensive variation in state and local pensions, as the national government plays little role in their governance. In light of rising costs, a number of states have made reforms in recent years, generally reducing generosity and increasing contributions (Anzia 2022; Novy-Marx and Rauh 2011; Peng and Wang 2017; Quinby and Sanzenbacher 2020). Prior research has not examined pensions and retirements across numerous plans and over time, alongside these types of reforms and other “large forces” that might matter (see Roberts 2013).

I argue that visible policies that make work less alluring will be associated with increased retirements. I outline several hypotheses, which I examine with original data from over 1,000 reports in 40 states between 2001 and 2019. Using a Poisson model, I regress service retirements in the following year and four-years-later on pension, economic, and political variables to consider these relationships in both the shorter and longer terms.

The findings show several policies are associated with increased retirements in the short term: early retirement incentives and relaxed post-retirement work restrictions. Additionally, retirements are greater as employee contributions are higher and employer contributions are lower, and especially so in plans with greater numbers of older workers. Enhancements in COLAs, in comparison, are associated with fewer retirements in both the short and long terms. Retiring pension-plan members, of course, cease making contributions and begin collecting their deferred income. Increased retirements due to generational change present challenges to plans’ fiscal health alongside national-level programs like Social Security and Medicare. Such reforms also potentially might detract from diversity in public-sector employment. As concerns linger about the attractiveness of public service, governments ought to appreciate these linkages between personnel and pensions, especially as they consider reforms.

Pensions and Public Personnel

Prior research shows that fringe benefits matter for shaping the public-sector workforce, both in terms of recruitment and retention (Asseburg and Homberg 2020; Condrey and Battaglio 2007; Lee and Wilkins 2011; Perry and Wise 1990). Pensions are an important such benefit, and can be thought of as kinds of contracts between employers and employees. Though recipients give up some short-term compensation, pension income helps them smooth spending over their life-cycles. Retirement plans also are a form of insurance for beneficiaries who cannot know how long they will live, and whether they will be healthy enough to work. Pensions might shape employees’ expectations over their careers, especially as they come to understand their benefits (Blinder 1981; Clark, Morrill, and Vanderweide 2014; Mitchell 1988).

Pensions could aid in public management if they help attract motivated and long-term workers (see Gailmard and Patty 2007; Ippolito 1987; Lazear and Moore 1988; Munnell, Haverstick, and Soto 2007). They might work alongside job protections to help recruit and retain employees, which is why their initial growth happened around the same time as the transition from the spoils to the merit system in the early 20th century (Asseburg and Homberg 2020; Clark, Craig, and Wilson 2003; Condrey and Battaglio 2007; Lee and Wilkins 2011; Quinby and Sanzenbacher 2020).

While pensions are not the only factors shaping workers’ careers, they could matter on the margins. Younger workers have reason to avoid termination and vest, at which point they become eligible for retirement payments. Vesting typically occurs when employers recover training costs. Employers may wish to retain employees longer, though, especially if onboarding costs are significant, thanks to factors like administrative rules or security clearances (Clark, Craig, and Wilson 2003). Annual pension income is smaller than salary pay for a reason: employers reward work over retirement. At the same time, governments do not want workers to remain forever. Retirements can make way for new personnel, which can help spur innovation and reform (Kellough and Osuna 1995; Meier and Hicklin 2008). Alternatively, governments may wish to downsize for fiscal or ideological reasons (Anzia 2022; Maciag 2017). Pensions, by design, help encourage retirement by levying implicit taxes on work: vested workers forego money they could receive by doing nothing. Assuming employees value leisure, working is no longer worth it (Clark, Craig, and Wilson 2003).

Recent research by Gorina and Hoang (2020) shows that state-level pension reforms in recent years significantly shape public-employees’ attitudes, thereby influencing recruitment and turnover. As the authors state: “the design of public retirement benefits will continue to influence the ability of the public sector to recruit and retain high-quality workforce.” Similarly, Quinby and Sanzenbacher (2020) find pensions play an important role recruiting workers into public-sector work, but the effect diminishes as employees contribute more into plans, suggesting costlier benefits have diminished personnel effects.

Turning to empirical work on pensions and retirements, Morrill and Westall (2019) find that teachers who receive Social Security retire at higher rates than those who do not. Though Social Security is a universal rather than employment-based policy, the result shows retirement income alters older workers’ behavior. In looking at both public and private-sector pensions, French and Jones (2011) find reductions in benefits are associated with additional work. Within different types of plans, Nyce (2007) finds that both DC and DB pensions are associated with long careers, though DB employees work longer. That study also shows survey evidence workers highly value their pensions, which increase their commitments to their employers. Using panel survey data from the Health and Retirement Study, Papke (2019) finds that early retirement policies significantly increase the probability that state and local employees decide to retire.

Several studies also examine retirements in teacher plans, and generally find that they cluster around pensions’ stipulated retirement ages (Brown 2013; Costrell and McGee 2010; Friedberg and Turner 2010). These studies also note that these policies often are not well-matched to workers’ maximum productivities. However, Brown (2013) finds that while pensions can influence retirements, their role is small relative to underlying demographic factors. Though informative, these studies use data from just a handful of plans, and also only for teachers.

There is more to know about how pensions shape retirements across numerous plans. State and local pensions vary a great deal, and there has been no shortage of reforms adding to that heterogeneity in recent years (see Munnell 2012; Snell 2012; Thom 2017). While DB pensions require making promised payments, some changes do apply to current employees, like new early retirement policies, adjusted post-retirement work rules, shifting contributions and COLAs, and freezing or reorganizing plans. Any or all of these could shape retirements.

Theory

I expect pension policies shape retirements in different ways. A great deal of research shows that a dollar today is more valuable than a dollar tomorrow. A Pew survey of public employees found that a dollar of pension money is worth about 20 cents in salary (see Kahneman and Tversky 1979; Quinby and Sanzenbacher 2020). Reforms that trade off current for future money should incentivize retirement, while those that do not might push in the opposite direction, to the extent workers value deferred income. Sunsetting policies also should generate short rather than long-term impacts, since workers need to act in time to benefit.

Employees also may lack complete knowledge of their retirement plans, or make mistakes in their planning due to financial illiteracy (Clark, Morrill, and Vanderweide 2014; Mitchell 1988). Even so, while people may not always engage in optimal decision-making, they tend to act at their “most rational” when it comes to money, which facilitates quantifying self-interest (Chong, Citrin, and Conley 2001). Additionally, employees do not require perfect information for pensions to matter. Shifts in small but well-informed pluralities can induce “rational” aggregate effects.

Policies do need at least some visibility to matter, though. For example, employees might learn by listening to their co-workers or union representatives. Board elections and regular paychecks help remind employees about their pensions (see Lenz 2009). Chalmers, Johnson, and Reuter (2014) find significant peer effects, in which employees more strongly react to pensions as more of their coworkers belong to them. It also may well be the case that older workers nearer retirement should pay more attention to pensions (Almedia, Kenneally, and Madland 2009; Nyce 2007). Based on this discussion, I present four hypotheses regarding fairly visible pension policies that could plausibly affect current employees’ decision-making²:

H1: Consistent with the research discussed above, new early retirement incentives will increase retirements. These are often designed to reduce short-term payroll costs by increasing turnover.³ They can “increase the economic value of the standard retirement benefit, be a onetime payment that does not impact an ongoing defined benefit or defined contribution retirement benefit or provide other financial incentives to facilitate retirement before an employee’s otherwise planned retirement” (Government Finance Officers Association of the United States and Canada 2022). Unlike COLA increases, workers need to retire within a specified period to receive these benefits. Such policies usually are well-communicated to employees. Workers also may witness their colleagues considering retirement following these policies.

H2: Post-retirement work restrictions should decrease retirements, while easing such laws should increase them. These laws affect the rules surrounding returning to work following retirement. In many cases, they have been passed to prevent perceptions of abuse of the system by employees, who might retire to maximize their pension payouts, but then return to work for additional compensation. Many states passed such restrictions in the 1990s. However, states have since been more likely to relax these law to meet their short-term human capital needs, especially for teachers. Relaxing these laws reduces the cost or risks of retirement for current workers by expanding the opportunity to earn supplemental income.

H3: Cost-of-living adjustment (COLA) increases should be associated with fewer retirements, while decreases should be associated with more. COLAs are one way to adjust the amount of income retirees receive. They often account for inflation or shifts in local economies. Since COLA increases do not sunset like early retirement rules, I expect employees work longer in response to them. Workers also might take these as positive signals from employers.

H4: Higher employee contributions (as a share of total payroll) should be associated with more retirements, while more employer contributions should be associated with fewer. Employees’ contributions are regular and visible paycheck deductions. As workers funnel money into pensions, they have less income today. On average, they will be less excited about sticking around, particularly when they can separate and receive pension income. Similarly, Quinby and Sanzenbacher (2020) find poorer recruitment efforts as these contributions are greater.

Higher employer contributions, in comparison, mean governments pick up more of the tab. As costs have risen, employee and employer contributions have increased, as seen in Figure 1. However, between 2001 and 2019, the mean employees’ rate increased from 5.44% to 8.46% (a 55.5% increase), compared to the mean employer rate, which rose from 6.25% to 19.29% (a 209% increase). Clearly, states have been more reluctant to raise employee contributions. Workers also might take increases in their own contributions and decreases in employer contributions as negative signals. At the same time, I expect employer contributions, which do not appear in paychecks, will be less salient to workers than their own payments.

Figure 1.

Employee and employer contributions over time.

Data and Empirical Analysis

To examine these hypotheses, I exploit variation within pensions and states from 2001 to 2019. The unit of analysis is plan by year. Table 1 provides brief definitions and sources for each variable, as well as information about their distributions. The dependent variable is the number of service retirements, which I collect from CAFRs and actuarial variation reports in 81 plans across 40 states.⁴ The variable is right-skewed, as some pensions are especially large, and also as a small number of plans experience brief but large retirement spikes. Consistent with the prior discussion, Figure 2 shows that retirements have increased over time, reflecting both an aging public workforce and increased financial burdens for states.

Table 1.

Variable Summaries.

Variable	Source	Description	Min	Max	Mean	Count
Retirements	Plan Reports	Number of service retirements	31	35,486	3,938.84	1,501
Actives	Public Plans Database	Number of active employees	4,418	887,220	123,743.4	1,489
COLA Increase	NCSL & Brainard and Brown (2021)	Cost-of-living-adjustments increased	0	1	0.008	1,501
COLA Decrease	NCSL & Brainard and Brown (2021)	Cost-of-living-adjustments decreased	0	1	0.04	1,501
More Generous Formula	NCSL & Brainard and Brown (2018)	Increase in pension formula benefits	0	1	0.04	1,501
Less Generous Formula	NCSL & Brainard and Brown (2018)	Decrease in pension formula benefits	0	1	0.047	1,501
Undo Work Restriction	NCSL & Brainard and Brown (2018)	Law undoes or loosens retirees’ work restrictions	0	1	0.087	1,501
New Work Restriction	NCSL & Brainard and Brown (2018)	Law limits retired workers’ abilities to work	0	1	0.027	1,501
Early Retirement Incentive	NCSL & Brainard and Brown (2021)	Early retirement incentive or buyout	0	1	0.037	1,501
Merged In	NCSL & Brainard and Brown (2018)	Existing DB plan merged in	0	1	0.005	1,501
Introduce DC	NCSL & Brainard and Brown (2018)	Employees can opt to exit into new DC plan	0	1	0.028	1,501
New DB Plan	NCSL & Brainard and Brown (2018)	Employees can opt to exit into new DB plan	0	1	0.015	1,501
Transformed Employee Contributions	Public Plans Database	Inverse Hyperbolic Employee Contributions	0	5.01	1.97	1,470
Transformed Employer Contributions	Public Plans Database	Inverse Hyperbolic Employer Contributions	−0.00008	6.46	2.61	1,399
Average Salary	Public Plans Database	Average Active Employees’ Salary (Thousands)	17.06	109.32	46.45	1,487
Average Active Age	Public Plans Database and Plan Reports	Average Active Employees’ Age	39	52.52	45.34	1,281
Actives Over 55	Plan Reports	Number of Active Employees Over 55	596	213,027	27,847.22	1,181
Actives Over 60	Plan Reports	Number of Active Employees Over 60	178	107,294	12,568.65	1,200
Actives Over 65	Plan Reports	Number of Active Employees Over 65	35	39,601	3,957.59	1,167
Unemployment	U.S. Census	State Unemployment	2.4	13.5	5.68	1,501
% Republican	NCSL	% Lower Chamber Republican	0.1	0.87	0.51	1,402
Polarization	Shor and McCarty (2011)	Lower Chamber Polarization	0.49	3.04	1.64	1,355
Divided Government	NCSL	Unified = 0, Divided = 1	0	1	0.44	1,403

Note: The unit of analysis is plan × year.

Figure 2.

Retirements over time.

Amongst the independent variables, I include several binary indicators representing major legislative changes in pension policy, collected from the National Conference of State Legislatures (NCSL), as well as memos from the National Association of State Retirement Administrators (NASRA) (Brainard and Brown 2018, 2021).⁵ Each indicator variable takes on a value of 1 if the policy change affects active employees in a given year, and 0 otherwise. Legislatures sometimes change all their states’ pensions. In other cases, they single out particular plans, or specific plans’ boards of trustees enact changes. I code such cases appropriately.

There are 1,501 observations in which the dependent variable exists. Amongst these, there are 56 instances when new early retirement incentives appear. I include a variable representing if the plan undoes or loosens a work-in-retirement restriction, and another for when the state passes a new restriction. There are 131 such relaxations and 42 tightenings.

Additionally, I use four indicators representing both increases and decreases in COLAs and benefit formulas. Amongst these, there are 12 instances of increased COLAs, compared to 58 decreases. Similarly, changes in pension formulas can expand or contract the size of retirement income. There also are 59 increases in plans’ formulas, compared to 70 reductions. Clearly, decreases in both are more common than increases over this period.

A tradeoff with this approach is that in some cases, I overlook variation (e.g. I code 1% vs. a 2% increase in a pension’s benefit multiplier the same way). However, an advantage is that I can generalize major types of changes, while avoiding having numerous overly fine-grained independent variables. Doing so is useful to assess similar types of claims, and also appropriate since the analysis relies on variation within plans.

I also include several other controls from those two sources. One is a dummy representing whether an external plan has merged with the current one. In such cases, there could be increases in retirements since more employees suddenly belong to the plan, and potentially also if mergers affect employee decision-making.⁶ Additionally, I control for the introduction of new DB plans that might draw in current members: these might signal states wish to begin winding a plan down, or at least provide more options to employees. Similarly, I include a variable for the introduction of new DC plans that current DB active members could join.⁷

I merge the above data with the Public Plans Database (PPD), which is a joint effort between Boston College and NASRA. I include both employee and employer contributions as shares of total payroll to test the fourth hypothesis. To better comport with regression’s assumptions, I transform both these variables, which have right-skewed distributions.⁸ In addition, I control for average active member age, since plans with more older workers might experience more retirements.⁹ However, not all plans report average age information.¹⁰ As such, I include specifications with and without it, since its inclusion drops some observations. I also include original data on the number of active employees over age 55, 60, and 65, which allows me to take an alternative approach to account for the distribution of workers’ ages.¹¹

Likewise, I control for the average salary of active members, which could matter if well-paid employees retire earlier or later than their lower-paid counterparts, on average. I also control for annual state unemployment, which comes from the U.S. Census Bureau. This serves as a proxy for the labor market; it is possible there will be more retirements when unemployment is low, as employees have opportunities to earn more elsewhere.

Political forces also might play a role, for example if workers feel like the climate is more or less amenable to remaining in public service (see Bolton, de Figueiredo, and Lewis 2021; Dahlström and Holmgren 2019). To gain some insight here, I control for the partisan composition of the legislature, legislative polarization, divided government, and an interaction between the last two to account for gridlock (see Binder 2003).¹² I characterize legislative partisanship as the percent Republican control of the state’s lower house in each year. More conservative legislatures, for example, might pass legislation that is tough on employees and pensions. Alternatively, given that both parties have pursued similar pension strategies, at least until 2008, the variable might not matter (see Anzia and Moe 2017). Aside from that, polarized environments might encourage retirements by demoralizing workers. Alternatively, legislatures experiencing gridlock could provide employees with more discretion, encouraging them to remain. It also is possible employees do not pay a great deal of attention to politics when timing retirements. This data comes from the NCSL and research by Shor and McCarty (2011).¹³

Given the dependent variable’s count-nature and distribution, I use a Poisson approach to examine its relationship with the independent variables.¹⁴ In doing so, it is important to account for differences in the numbers of plan employees. For example, 5,000 employees retiring in New York is different than 5,000 in South Dakota. For that reason, I include models with two different exposure variables, which both account for the number of workers in plans. The first is the number of active employees from the prior year, given that retirees come from the population of prior-year employees. The second uses the count of all prior-year employees who are least 60, and therefore closer to retirement.¹⁵ The correlation coefficients between retirements and these two exposures are high and similar: 0.927 and 0.923, respectively.

The following expresses my empirical approach: the i term represents the given plan, β₁–β₉ are vectors of the point estimates associated with the independent variables, t represents the year, and ε_it is the random error term.¹⁶ I also include year and plan fixed effects (FE), which are two vectors of dummy variables for all-but-one years and all-but-one plans. The independent variables are all lagged by a year to avoid post-treatment bias:

\begin{array}{l} % R e t i r e m e n t s_{i t} = β_{0} + β_{1} R e t i r e m e n t s_{i (t - 1)} \\ + β_{2} E a r l y R e t ._{i (t - 1)} + β_{3} W o r k R e s t ._{i (t - 1)} \\ + β_{4} C O L A s_{i (t - 1)} + β_{5} F o r m u l a G e n_{i (t - 1)} \\ + β_{6} C o n t . R a t e s_{i (t - 1)} + β_{7} P l a n & P e r s o n n e l_{i (t - 1)} \\ + β_{8} U n e m p l o y m e n t_{i (t - 1)} + β_{9} P o l i t i c s_{i (t - 1)} \\ + Y e a r F E + P l a n F E + ε_{i t} \end{array}

(1)

I control for a lagged version of the dependent variable to help focus on variation in retirements. Most plans have fairly stable retirements each year, meaning that failing to control for the prior year could present a form of omitted variable bias.¹⁷ Since it is difficult to know ex ante whether policy changes generate shorter or longer-term changes shocks in retirements, I examine both next-year effects and four-year-later effects. I also include models with and without the political controls, since it is useful to understand how pensions matter for retirements by themselves, but also potentially alongside political forces.

There are numerous characteristics for which I do not control. Some are observed, while others would be impossible to collect. For example, workers’ preferences for risk or job enjoyment could matter. Additionally, plans differ in how pension wealth grows over time (Brown 2013; Costrell and McGee 2010; Friedberg and Turner 2010). Plan FE examines variation within plans, differencing out these factors to focus on the relationships outlined above. In comparison, year FE account for broad forces that might influence various plans at once. A separate concern is collinearity amongst the variables. All of the models use robust standard errors, clustered at the state and plan levels to respectively account for inter-dependence between plans within states and consistencies in policies over time. Failing to make such adjustments could result in underestimating the standard errors (Cameron and Miller 2015; Wooldridge 2010).

Results

My fundamental question concerns how pensions shape aggregate state-employee retirements over time. To allow for more straight-forward interpretation, Tables 2 and 3 present the marginal effects of the log-linear coefficients, which are comparable to Ordinary Least Squares regressions.¹⁸ The coefficients are the number of retirements above or below the mean to expect in response to a one-unit increase in each independent variable. The even-numbered models include the political covariates. Additionally, models (1)−(4) employ the all prior-year active employee exposure, while (5)−(8) use the count of workers who are at least 60 years old.

Table 2.

Service Retirement Poissons.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Retirements	0.06* (0.02)	0.06* (0.02)	0.08* (0.04)	0.08* (0.03)	0.09* (0.02)	0.10* (0.02)	0.09* (0.03)	0.09* (0.03)
Early Ret.	569.53* (219.59)	595.66* (229.88)	424.94⁺ (251.56)	470.35⁺ (263.08)	434.79 (366.59)	513.22 (394.46)	605.86⁺ (330.76)	689.75⁺ (354.10)
Undo Work Rest.	358.71* (139.54)	375.69* (147.64)	266.97⁺ (139.30)	266.01⁺ (136.26)	218.66 (157.04)	214.77 (163.33)	245.63⁺ (133.54)	249.52⁺ (138.07)
Work Rest.	−153.29 (99.88)	−138.41 (108.48)	−103.59 (74.93)	−118.71⁺ (71.48)	−197.93⁺ (113.90)	−223.63⁺ (130.67)	−105.68 (75.07)	−141.98⁺ (83.64)
COLA Decrease	41.34 (110.82)	74.10 (108.95)	−8.36 (105.15)	2.82 (101.28)	−33.78 (111.95)	−38.07 (112.55)	−64.57 (107.36)	−68.20 (118.34)
COLA Increase	−624.80* (299.63)	−624.08* (303.05)	−459.41* (203.64)	−465.62* (218.54)	−621.31* (193.48)	−695.69* (218.17)	−518.91* (153.84)	−577.41* (182.43)
Less Gen.	−67.43 (139.75)	−94.90 (132.25)	107.84 (163.50)	113.12 (122.45)	90.10 (134.77)	90.56 (126.08)	169.38 (130.41)	169.91 (138.83)
More Gen.	−100.01 (122.12)	−70.45 (114.74)	−106.34 (103.63)	−131.06 (109.73)	70.67 (298.04)	26.50 (301.60)	−153.98 (181.81)	−236.84 (188.95)
Merged In	−9.89 (248.09)	50.22 (288.15)	173.11 (163.71)	239.05 (198.43)	−144.86 (327.99)	−144.79 (338.51)	31.93 (206.57)	6.11 (193.37)
Introduce DC	70.62 (353.70)	43.92 (380.91)	−85.01 (118.19)	−87.75 (124.24)	454.05 (420.57)	491.18 (443.83)	94.83 (120.62)	122.01 (123.24)
New DB Plan	164.82 (366.40)	279.99 (394.98)	220.32 (220.39)	279.44 (215.48)	290.49 (291.84)	490.10 (304.12)	360.96⁺ (198.27)	550.77* (165.61)
Trans. Employee Cont.	194.49* (83.91)	198.81* (85.32)	179.17* (60.50)	190.08* (63.44)	237.72* (58.75)	251.02* (64.95)	219.13* (37.89)	233.99* (47.51)
Trans. Employer Cont.	−77.41 (84.52)	−70.03 (93.78)	−83.79 (62.01)	−55.23 (59.56)	−147.12* (39.40)	−150.29* (43.10)	−216.76* (50.92)	−214.90* (47.06)
Avg. Salary	13.19 (10.51)	13.38 (10.72)	5.05 (7.02)	5.90 (6.84)	2.04 (4.87)	2.59 (4.64)	−0.71 (3.80)	0.09 (3.87)
Avg. Active Age			406.61* (64.50)	427.16* (70.10)			117.12 (81.45)	103.79 (81.81)
Unemployment	−1.65 (34.47)	−12.01 (32.39)	−28.28 (37.03)	−33.63 (37.31)	6.23 (38.07)	0.88 (38.15)	−36.99 (38.32)	−35.03 (40.71)
% Repub. Legis.		−83.78 (918.50)		−238.71 (772.45)		136.78 (1,153.56)		170.32 (807.02)
Legis. Polarization		176.16 (380.29)		−299.25 (371.64)		64.32 (437.31)		174.83 (428.49)
Div. Govt.		−221.72 (214.15)		−544.22* (182.87)		−309.73 (190.94)		−230.93 (155.09)
Pol. × Div. Govt.		203.55⁺ (115.19)		336.30* (113.25)		180.06⁺ (105.75)		100.37 (87.59)
Observations	1,337	1,273	1,132	1,069	1,068	1,005	981	918
Pseudo R²	0.970	0.971	0.973	0.974	0.976	0.977	0.976	0.977

p < .10. *p < .05. **p < .01. ***p < .001

All models include plan & year fixed effects. Exposure is all prior-year active employees in (1)−(4), and all actives ≥60 in (5)−(8).

All independent variables lagged by one year. Two-way robust cluster standard errors in parentheses.

Table 3.

Service Retirement Poissons, Four Years Later.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
Retirements	0.02 (0.02)	0.02 (0.02)	0.00 (0.01)	0.01 (0.01)	0.04* (0.02)	0.04* (0.01)	0.00 (0.02)	0.01 (0.02)
Early Ret.	−202.14 (126.66)	−184.88 (130.19)	−80.08 (70.92)	−68.58 (71.64)	−371.69* (188.10)	−365.36* (182.21)	−101.52 (115.89)	−111.78 (112.70)
Undo Work Rest.	101.51 (104.55)	69.45 (114.74)	−70.51 (77.36)	−86.60 (74.47)	45.40 (117.11)	8.27 (115.17)	−87.63 (99.56)	−99.54 (91.75)
Work Rest.	128.62 (151.91)	111.40 (154.87)	−84.98 (63.33)	−68.44 (63.39)	−110.29 (72.41)	−101.56 (85.89)	−104.20 (70.34)	−78.77 (83.42)
COLA Decrease	−15.45 (114.95)	−17.15 (110.12)	−14.17 (87.56)	−12.78 (95.16)	−64.66 (130.55)	−50.19 (128.44)	−38.62 (110.09)	−35.21 (114.42)
COLA Increase	−675.40* (162.73)	−692.20* (180.58)	−491.54* (130.63)	−463.69* (98.54)	−576.62* (256.22)	−464.13* (214.93)	−513.37* (237.04)	−464.61* (214.15)
Less Gen.	−165.04 (106.81)	−165.02⁺ (98.28)	−87.59 (78.58)	−68.37 (94.35)	13.29 (137.99)	25.86 (112.67)	24.44 (151.29)	76.64 (119.22)
More Gen.	98.84 (126.97)	104.69 (132.64)	−5.77 (86.62)	5.02 (86.80)	147.03 (202.85)	157.03 (225.61)	29.27 (140.40)	79.33 (139.59)
Merged In	140.70 (527.44)	138.42 (524.48)	168.25 (491.05)	195.00 (477.52)	28.63 (456.35)	16.49 (458.94)	83.56 (447.56)	80.15 (448.48)
Introduce DC	−6.24 (145.46)	59.74 (158.90)	25.00 (105.67)	35.29 (114.66)	306.21⁺ (172.99)	349.06⁺ (189.06)	178.44 (146.11)	189.41 (168.31)
New DB Plan	−52.91 (253.05)	−121.06 (243.64)	33.53 (158.25)	−66.80 (133.75)	−12.49 (308.50)	−147.59 (288.95)	21.10 (243.81)	−105.59 (162.99)
Trans. Employee Cont.	89.64 (113.91)	101.96 (112.72)	215.18* (69.31)	203.72* (74.41)	264.76* (81.29)	267.53* (70.14)	248.52* (64.08)	239.41* (57.03)
Trans. Employer Cont.	−83.29 (97.81)	−113.91 (105.47)	−4.13 (70.91)	12.67 (88.94)	−88.16 (90.98)	−136.45 (83.63)	−106.68⁺ (63.01)	−132.95⁺ (73.77)
Avg. Salary	7.46 (8.16)	7.56 (8.63)	−0.80 (4.43)	−0.27 (4.41)	−1.44 (3.74)	−0.88 (4.65)	−7.22 (4.80)	−5.97⁺ (3.51)
Unemployment	−85.99 (56.80)	−90.09⁺ (54.69)	−94.75⁺ (51.46)	−103.30* (50.24)	−5.51 (43.36)	−9.97 (38.23)	−29.55 (58.18)	−35.48 (62.79)
Avg. Active Age			548.69* (99.75)	561.78* (109.22)			261.12⁺ (135.13)	218.50 (148.64)
% Repub. Legis.		−1812.56* (869.95)		−1126.97* (574.04)		−1108.65 (910.20)		−1007.42 (694.82)
Legis. Polarization		745.40 (710.09)		550.81 (584.89)		1,236.41⁺ (633.48)		1,165.76* (507.23)
Div. Govt.		−321.03 (209.68)		−225.56* (68.07)		−134.88 (157.04)		−66.62 (137.70)
Pol. × Div. Govt.		173.10 (107.68)		165.48* (45.26)		76.73 (73.02)		40.98 (76.17)
Observations	1,100	1,046	921	868	862	809	787	734
Pseudo R²	0.975	0.976	0.978	0.979	0.977	0.977	0.976	0.977

Note: All models include plan & year fixed effects. Exposure is all prior-year active employees in (1)−(4), and all actives ≥60 in (5)−(8).

All independent variables lagged by one year. Two-way robust cluster standard errors in parentheses.

p < .10. *p < .05. **p < .01. ***p < .001.

The results suggest several pension policies contribute to variation in retirements. The early retirement incentive variable is associated with an increase of about 450–700 retirements above the mean in the following year in Table 2. In other words, the mean plan would expect an increase of from about 3,795 to between 4,245 to 4,495 retirements. In Table 3, the direction of the relationship reverses, though it is only significant in two of the eight models. This could reflect that following the initial boost in retirements from this policy, fewer workers who could retire remain. The difference also could reflect a selection effect in which workers less likely to respond to an early retirement offer remain longer in their jobs.

In Table 2, there is evidence in models (1)−(4) to support rejecting the null for the undoing work-restriction hypothesis. The policy is associated with about 250–300 more retirements above the mean, depending on the specification. In the longer-term models, the variable is not significant. In comparison, new restrictions have coefficients in the expected direction, but are not significant. In Table 3, the relationship’s direction varies and is not significant.

The COLA decrease variable is not significant in either table. However, increases have large and significant effects in all specifications in both: increases are associated with about 450–700 fewer retirements in the mean plan in both the short and long terms. Though the relationship is robust, I might signal some caution, since just 12 such increases occur in the data. Neither increases nor decreases in formula generosity are significant, in comparison.

Employee contributions are significant and in the expected direction in Table 2. In models (1)−(4), which consider retirements with respect to all active employees, increases have a marginal effect of around 190 retirements, implying that a one percent increase in employee contribution rates is associated with 1.9 more retirements above the mean.¹⁹ Larger increases, of course, should be associated with additional retirements, while reductions are associated with fewer. In models (5)−(8), the effect is larger, averaging about 235 retirements above the mean. In comparison, employer contribution rates are in the hypothesized direction in models (5)−(8) in Table 2. The coefficients are smaller in magnitude, though. The variable is not significant in most of the models in Table 3, though. In Figure 3, I plot the contribution coefficients for the full specification across each exposure variable. As plans have more “older” employees, larger contributions have effects that are greater in magnitude in both the short and long terms. Again, employee contributions seem to matter more than employer ones.

Figure 3.

Marginal effects of contributions on retirements by age exposure.

The merger, new DB, and new DC plan variables all are not significant across the various specifications. The salary variable also is not significant, which might reflect that some employees wish to keep earning more as their salaries are higher, while others opt to retire. As expected, the coefficient for average age is positive and significant in both specifications (3) and (4), in both the short and long term. Less surprisingly, it is not significant in (7)−(8), which already takes a measure of age into account in the exposure variable. There are about 415 more retirements in the following year as the mean plan’s employees are a year older. Four years later, there are about 555 additional retirements, reflecting that plans with older employees experience even more retirements down the line.

There is mixed or little evidence that the political and economic variables matter. The coefficients for legislative partisanship, polarization, and divided government have mixed significance across the different specifications.²⁰ While political forces could play a role in retirements in some cases, they are less meaningful than demographics and pension policies.

Discussion

The results provide insight into the connections between policy and personnel within numerous plans and during an era of reform and an aging workforce. They also show that pensions are composed of different moving parts. Similar to the finding in Quinby and Sanzenbacher (2020) that greater employee contributions dampen recruitment, I show that they also are associated with additional retirements. Many states have raised contributions in recent years to enhance pensions’ funded status. The finding also speaks to how workers view retirement plans: contributions are arguably pensions’ most visible policy, and the evidence suggests increases in these contributions leave workers with less pay today, making retirement more alluring.

Early retirement incentives also should be relatively easy for employees to learn about, especially as governments usually broadcast their efforts. The policies also sunset, unlike COLA adjustments: if workers do not retire, they miss out. Early retirement also makes deferred income more immediate, increasing its value. Lastly, relaxed work-retirement rules reduce the cost or risk of retiring, giving workers more flexibility as they ease into retirement. Workers may become more aware of these rules as they see former colleagues return to the office. Loosening restrictions could be useful for governments facing personnel shortages due to difficulties recruiting new workers, but also could end up encouraging new retirements.

In comparison, other pension variables matter less. Adjustments to formula generosity and decreases in COLAs, as well as the introduction of new DC or DB plans have no discernible effects. It is possible some of these variables are less well-known to workers, especially if they do not pay close attention to pensions. Given many employees do not know whether their own plans are DB or DC, it is probable that only some employees will notice these types of reforms, much less connect them to their retirement decision-making. It also is possible that these policies might encourage some workers to retire, while others prefer to stick around.

Implicit in the results is that “non-work” or leisure plays an important role. Long-tenured employees may wish to “cash in” or select leisure if they think additional work will do little to enhance their pensions. Alternatively, they might look for new work elsewhere. Plans with more older employees also experience more retirements. Such plans also are more sensitive to contribution levels, which also speaks to how older pensioned workers view the costs and benefits of continued work. That said, at least some retirees are willing to return to their old jobs in a part-time capacity, given relaxations of work-in-retirement policies.

The results also are relevant in light of Anzia’s (2022) finding that governments more often opt to reduce their workforces rather than raise revenue to fund pensions. Of course, governments also may wish to downsize for other fiscal or ideological reasons. Though state employees may retire for a variety of reasons, pension reform offers one potential means to encouraging it: greater employee contribution rates, new early retirement packages, and relaxed work restrictions all reduce the number of employees, on average. While these changes may prevent retirees’ pensions from growing larger, they still present longer-term costs by reducing the pool of active contributors into funds, requiring additional payments to newly eligible retirees, and diminishing institutional knowledge. Workers also might take such changes as negative signals from employers, especially if they are publicized as workforce-reduction measures.

While the results highlight the importance of recent reforms, a limitation is that in grouping similar changes together, I might overlook some variation. More fundamentally, I rely on observational data: it is impossible to rerun history or vary policies for any individual plan to generate true counterfactuals. It also is difficult to control for all variables that might matter, such as managerial competence or job satisfaction. However, substantial and sudden aggregate shifts in these unobserved variables within plans would be required to bias estimates. Controlling for different types of observed variables plausibly connected to retirements and looking at variation within plans both help mitigate this potential threat. Even so, further research should explore the connections between pensions and these factors.

Conclusion

In order to provide public goods and services, governments need to be able to manage their personnel. This includes both attracting and retaining talent in public service. Pensions might offer one tool to aid in public management, thanks to the incentives they create for workers. I focus on the relationship between pensions and retirements across a number of plans and over time. Retirements are a common way for workers to leave work. While they can create opportunities to bring in new talent, they also entail a loss of institutional knowledge. In other words, sudden retirement waves represent significant losses of human capital. Of course, other factors besides pensions also shape retirements, making it also important to consider variables like age, salary, and the number of retirements in the prior year. Building on earlier work, I examine retirements in light of several plausibly visible pension policies. I argue that policies that make continued labor appear more expensive should be associated with more retirements. The findings largely support this theory. Attempts to increase retirements by incentivizing early retirement have their intended effect. Relaxed post-exit work restrictions also are associated with more retirements. Additionally, higher employee contribution rates are associated with more retirements, suggesting that pensions might lose their personnel power if employees perceive them, alongside continued work, as too expensive (see Quinby and Sanzenbacher 2020). Other policies, however, like reductions in COLAs and changes in formula generosity fail to achieve statistical significance. There also is evidence that there are more retirements as plans’ active members are older. In the longer term, pension policies seem to have more muted effects, with the important exceptions of contributions and COLA increases.

Of course, large forces ultimately drive total public-sector employment (Roberts 2013). State and local revenue, which helps fund workers’ salaries and pensions, is highly contingent on the economy. Both the Great Recession and the COVID-19 pandemic led to reductions in the size of the workforce (Maciag 2017). Likewise, employment increased between these events as the economy improved. Still, specific aspects of pensions seem to influence retirements on the margins. Moreover, pensions have at least partially contributed to governments’ fiscal challenges in the first place, given their significance as financial obligations. The COVID- 19 pandemic has only made these concerns more acute, while also augmenting employees’ concerns about their jobs and finances: “As economies reopen and COVID-19 cases potentially spike in many states, it is imperative that state and local governments have a resilient workforce that can respond to these unprecedented and lingering challenges” (Liss-Levinson 2020).

A separate concern relates to the potential impact of pension reform on the public sector’s diversity. Prior research shows that less educated workers tend to prefer DB to DC or no pension plans (Ali and Frank 2019; Frank, Gianakis, and Neshkova 2012). Education is also tied to financial literacy, and such workers tend to have lower retirement savings outside of their pensions and Social Security, provided that they are eligible. Reforms that shrink DB benefits should have more severe effects on these employees. Historically, people of color and women make up greater proportions of the workforce in positions requiring less education. The results imply that changes like requiring greater employee contributions or passing early retirement policies should encourage more retirements amongst such workers, diminishing both their retirement income and the public sector’s diversity.

Going forward, states might do more to synthesize pensions and personnel, given these concerns. Reforms could incorporate more employee feedback. Adjusting boards of trustees to include more employees, for example, might assist in this end (see Brooks 2019). Future research should continue explore the relationship between pensions and public management. Continual employee and employer surveys would provide updated insights. It would be useful to have more insight into how pensions matter for different occupations. Textual analysis of legislation and plan reports also could highlight important new connections. In summary, pensions matter for state-employee retirements, though in multi-faceted ways. Governments should be aware of their personnel and the impact pensions and reforms can have on them, especially in the context of an aging workforce and strained costs.

Supplemental Material

sj-docx-1-slg-10.1177_0160323X221127387 – Supplemental material for Public-Employee Pensions and Retirements

Supplemental material, sj-docx-1-slg-10.1177_0160323X221127387 for Public-Employee Pensions and Retirements by John Brooks in State and Local Government Review

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

John Brooks

Supplemental Material

Supplemental material for this article is available online.

Notes

Author Biography

John Brooks is an assistant professor of public administration in the Department of Government and Justice Studies at Appalachian State University. His research concerns the nexus between public administration, budgeting, and politics at all levels of government. His dissertation, from which this article originates, concerns the politics and governance of public-employee pensions. His research appears in Journal of Public Policy, Journal of Politics, and Electoral Studies.

References

Ali

S. B.

Frank

H. A.

2019. “Retirement Planning Decisions: Choices Between Defined Benefit and Defined Contribution Plans.” The American Review of Public Administration 49 (2): 218–35.

Almedia

Kenneally

Madland

2009. “The New Intersection on the Road to Retirement: Public Pensions, Economics, Perceptions, Politics, and Interest Groups.” In The Future of Public Employee Retirement Systems, edited by Mitchell

O. S.

Anderson

New York, NY: Oxford University Press. 294–326.

Anzia

S. F.

2022. “Pensions in the Trenches: How Pension Spending is Affecting U.S. Local Government.” Urban Affairs Review 58 (1): 3–32.

Anzia

S. F.

Moe

T. M.

2017. “Polarization and Policy: The Politics of Public-Sector Pensions.” Legislative Studies Quarterly 42 (1): 33–62.

Asseburg

Homberg

2020. “Public Service Motivation or Sector Rewards? Two Studies on the Determinants of Sector Attraction.” Review of Public Personnel Administration 40 (1): 82–111.

Binder

S. A.

2003. Stalemate: Causes and Consequences of Legislative Gridlock. Washington, DC: Brookings Institution Press.

Blinder

A. S.

1981. “Private Pensions and Public Pensions: Theory and Fact. W.S.” Woytinski Lecture No. 5. University of Michigan Institute of Public Policy Studies.

Bolton

de Figueiredo

J. M.

Lewis

D. E.

2021. “Elections, Ideology, and Turnover in the US Federal Government.” Journal of Public Administration Research and Theory 31:451–66.

Brainard

Brown

2018. Spotlight on: Significant Reforms to State Retirement Systems. 1–97. Lexington, KY: National Association of State Retirement Administrators.

10.

Brainard

Brown

2021. NASRA Issue Brief: Cost-of-Living Adjustments. 1–16. Lexington, KY: National Association of State Retirement Administrators. https://arxiv.org/abs/1903.01633

11.

Brooks

2019. “Board on the Job: Public-Pension Governance in the United States (US) States.” Journal of Public Policy 39 (1): 1–34.

12.

Brown

K. M.

2013. “The Link Between Pensions and Retirement Timing: Lessons From California Teachers.” Journal of Public Economics 98:1–14.

13.

Cameron

Miller

D. L.

2015. “A Practitioner’s Guide to Cluster-Robust Inference.” Journal of Human Resources 50 (2): 317–72.

14.

Chalmers

Johnson

W. T.

Reuter

2014. “The Effect of Pension Design on Employer Costs and Employee Retirement Choices: Evidence From Oregon.” Journal of Public Economics 116:17–34.

15.

Chong

Citrin

Conley

2001. “When Self-Interest Matters.” Political Psychology 22 (3): 541–70.

16.

Clark

R. L.

Craig

L. A.

Wilson

J. W.

2003. A History of Public Sector Pensions in the United States. Philadelphia, PA: University of Pennsylvania Press.

17.

Clark

R. L.

Morrill

M. S.

Vanderweide

2014. “Defined Benefit Pension Plan Distribution Decisions by Public Sector Employees.” Journal of Public Economics 116:73–88.

18.

Condrey

S. E.

Battaglio

R. P.

2007. “A Return to Spoils? Revisiting Radical Civil Service Reform in the United States.” Public Administration Review 67 (3): 425–36.

19.

Correia

Guimarães

Zylkin

2019. “Verifying the Existence of Maximum Likelihood Estimates for Generalized Linear Models.”

20.

Correia

Guimarães

Zylkin

2020. “Fast Poisson Estimation With High-Dimensional Fixed Effects.” The Stata Journal: Promoting communications on statistics and Stata 20:95–115.

21.

Costrell

R. M.

McGee

J. B.

2010. “Teacher Pension Incentives, Retirement Behavior, and Potential for Reform in Arkansas.” Education Finance and Policy 5 (4): 492–518.

22.

Dahlström

Holmgren

2019. “The Political Dynamics of Bureaucratic Turnover.” British Journal of Political Science 49 (3): 823–36.

23.

Frank

Gianakis

Neshkova

M. I.

2012. “Critical Questions for the Transition to Defined Contribution Pension Systems in the Public Sector.” The American Review of Public Administration 42 (4): 375–99.

24.

French

Jones

J. B.

2011. “The Effects of Health Insurance and Self-Insurance on Retirement Behavior.” Econometrica 79 (3): 693–732.

25.

Friedberg

Turner

2010. “Labor Market Effects of Pensions and Implications for Teachers.” Education Finance and Policy 5 (4): 463–91.

26.

Gailmard

Patty

J. W.

2007. “Slackers and Zealots: Civil Service, Policy Discretion, and Bureaucratic Expertise.” American Journal of Political Science 51 (4): 873–89.

27.

Gorina

Hoang

2020. “Pension Reforms and Public Sector Turnover.” Journal of Public Administration Research and Theory 30 (1): 96–112.

28.

Government Finance Officers Association of the United States and Canada. 2022. The Use of Early Retirement Incentives: State and Local Governments Should Not use ERIs for a Variety of Reasons. https://www.gfoa.org/materials/the-use-of-early-retirement-incentives. Accessed: 2022-06-13.

29.

Graebner

1980. A History of Retirement. New Haven, CT: Yale University Press.

30.

Ilmanen

Kabiller

D. G.

Siegel

L. B.

Sullivan

R. N.

2017. “Defined Contribution Retirement Plans Should Look and Feel More Like Defined Benefit Plans.” The Journal of Portfolio Management 43 (2): 61–76.

31.

Ippolito

R. A.

1987. “Why Federal Workers Don’t Quit.” Journal of Human Resources 22 (2): 281–99.

32.

Kahneman

Tversky

1979. “Prospect Theory: An Analysis of Decision Under Risk.” Econometrica 47 (2): 263–92.

33.

Kellough

J. E.

Osuna

1995. “Cross-Agency Comparisons of Quit Rates in the Federal Service: Another Look at the Evidence.” Review of Public Personnel Administration 15 (4): 58–68.

34.

Lazear

Moore

1988. “Pensions and Turnover.” In Pensions in the U.S. Economy, edited by Bodie

Wise

Chicago, IL; Chicago: University of Chicago Press. 163–188.

35.

Lazear

E. P.

1979. “Why is There Mandatory Retirement?” Journal of Political Economy 87 (6): 1261–84.

36.

Lee

Y.-J.

Wilkins

V. M.

2011. “More Similarities or More Differences? Comparing Public and Nonprofit Managers’ Job Motivations.” Public Administration Review 71 (1): 45–56.

37.

Lenz

G. S.

2009. “Learning and Opinion Change, Not Priming: Reconsidering the Priming Hypothesis.” American Journal of Political Science 53 (4): 821–37.

38.

Lewis

G. B.

1991. “Turnover and the Quiet Crisis in the Federal Civil Service.” Public Administration Review 51 (2): 145–55.

39.

Liss-Levinson

2020. Survey Results: Public Sector Employee Views on Finances and Employment Outlook due to COVID-19. Washington, DC: The Center for State and Local Government Excellence.

40.

Maciag

2017. A Downsized Public Workforce May Be a Permanent Consequence of the Recession: State and Local Governments Still Haven’t Regained Many of the Jobs They Cut, and They’re Unlikely to Anytime Soon. Governing: The Future of States and Localities.

41.

Meier

K. J.

Hicklin

2008. “Employee Turnover and Organizational Performance: Testing a Hypothesis From Classical Public Administration.” Journal of Public Administration Research and Theory 18 (4): 573–90.

42.

Mitchell

O. S.

1988. “Worker Knowledge of Pension Provisions.” Journal of Labor Economics 6 (1): 21–39.

43.

Morrill

M. S.

Westall

2019. “Social Security and Retirement Timing: Evidence From a National Sample of Teachers.” Journal of Pension Economics and Finance 18 (4): 549–64.

44.

Munnell

A. H.

2012. State and Local Pensions: What Now? Washington, DC; Washington: Brookings Institution Press.

45.

Munnell

A. H.

Haverstick

Soto

2007. Why Have Defined Benefit Plans Survived in the Public Sector? 2. Boston, MA: Center for Retirement Research at Boston College.

46.

Novy-Marx

Rauh

2011. “Public Pension Promises: How Big Are They and What Are They Worth?” The Journal of Finance 66 (4): 1211–49.

47.

Nyce

S. A.

2007. “Behavioral Effects of Employer-Sponsored Retirement Plans.” Journal of Pensions Economics and Finance 6 (3): 251–85.

48.

Papke

L. E.

2019. “Retirement Choices by State and Local Public Sector Employees: The Role of Eligibility and Financial Incentives.” Journal of Pensions Economics and Finance 18 (04): 515–28.

49.

Peng

Wang

2017. “Affordability of Public Pension Benefit: A Historical and Empirical Analysis of US State and Local Government Pension Contributions.” Journal of Pensions Economics and Finance 16 (1): 21–42.

50.

Perry

J. L.

Wise

L. R.

1990. “The Motivational Bases of Public Service.” Public Administration Review 50 (3): 367–73.

51.

Quinby

L. D.

Sanzenbacher

G. T.

2020. “Do Pensions Matter for Recruiting State and Local Workers?” State and Local Government Review 52 (1): 6–17.

52.

Roberts

2013. Large Forces: What’s Missing in Public Administration.

53.

Shor

McCarty

2011. “The Ideological Mapping of American Legislatures.” American Political Science Review 105 (3): 530–51.

54.

Snell

2012. “State Pension Reform, 2009-2011.” 1-6. Washington, DC: National Conference of State Legislatures.

55.

Thom

2017. “The Drivers of Public Sector Pension Reform Across the U.S. States.” The American Review of Public Administration 47 (4): 431–42.

56.

Weaver

R. K.

1986. “The Politics of Blame Avoidance.” Journal of Public Policy 6 (4): 371–98.

57.

Wooldridge

J. M.

2010. Econometric Analysis of Cross Section and Panel Data. Cambridge: MIT Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.09 MB