The Rise Of Alternative Designs For Public Plans

Introduction

State and local government workers have traditionally had defined benefit (DB) pensions that insulate them from the stock market and ensure that they do not outlive their savings. Funding these plans, however, has become burdensome for many governments, with costs rising dramatically in the immediate wake of the 2008 financial crisis and continuing to grow since then. In response, some governments have introduced alternative plan designs that shift investment and/or longevity risk onto workers.

Given the rise in alternative designs, this brief answers two questions: 1) how prevalent were alternative designs before the financial crisis? and 2) how have they evolved since then? The discussion proceeds as follows. The first section introduces the various ways that state and local employers can shift risk onto their workers. The second section describes recent trends in plan design. The third and fourth sections identify the types of plans with alternative designs and explore what factors lead them to adopt these features.

The data show that the share of plans with alternative designs has grown continuously since 2008, so that plans with some risk sharing now cover roughly half of the state and local workers. A decline in funded status and Republican leadership often predict the shift towards alternative designs, while local plans and those covering public safety workers are more likely to stay traditional DBs. The final section concludes that risk sharing in the public sector is likely here to stay, but that workers still have significant protections.

How Do States and Localities Shift Risk to Workers?

In a traditional DB plan, workers earn benefits that are paid as an annuity throughout their retirement. To help fund these benefits, public sector plans typically require employees to contribute a percentage of their salary to the pension’s trust fund, but employers are responsible for any shortfall between the assets accumulated in the fund and promised benefit payments. This arrangement imposes two types of risk on employers: 1) investment risk, if assets in the trust fund underperform target returns; and 2) longevity risk, if retirees live longer than expected. The 2008 financial crisis was a stark example of investment risk: the stock market crash reduced the aggregate funded ratio of state and local plans from 86 percent in 2007 to a low of 72 percent in 2013.¹

To mitigate such risks, many employers have shifted at least some of it onto employees through alternative plan designs. In practice, these designs often follow one of the following models:
Stand-alone defined contribution (DC) plan. Like 401(k)s in the private sector, public sector DCs rely on individual accounts to which workers and employers each contribute a set percentage of the employee’s salary. Workers decide how to invest their assets and draw down the funds in retirement. Hence, workers bear all the investment risk during their working years as well as the risk of outliving their savings once they retire.²

Stand-alone cash balance (CB) plan. These plans also rely on individual accounts, but the employer determines how to invest the contributions and guarantees a minimum investment return. Account balances are automatically annuitized at retirement, which protects retirees from outliving their savings while placing longevity risk on the employer.

Hybrid plan. Some states and localities pair a smaller traditional DB with a DC or CB plan.³ The notion is that the DB provides a modest base of core income support, while the DC or CB component insulates employers from bearing all the risk.

These three plan designs represent a fundamental shift away from traditional DBs. Yet, plan sponsors can also act within the DB structure to insulate themselves from rising cost in the following ways:

Variable employee contribution rate. Public sector DBs often set the level of the worker’s contribution in state statute. Employers are then on the hook for any increase in the actuarially required contribution (ARC). To reduce this exposure, some employers instead set the employee contribution rate as a percentage of the ARC or explicitly set conditions under which employee contributions will increase. Effectively, workers’ take-home pay is cut when the plan does poorly, and increased when the plan does well.

Variable cost-of-living adjustment (COLA). Public DB plans can also share risk by making COLAs contingent on the plan’s financial condition in two ways. First, it has been long-standing practice for some plans to fund their COLAs solely from “excess return” accounts where funds are deposited whenever a plan’s investment performance exceeds actuarial targets (or some other threshold).⁴ This approach implicitly links the availability and size of COLA payments to the plan’s investment performance. More recently, some plans have explicitly linked annual COLA payments to either the plan’s funded ratio, recent investment returns, or both.

Clearly, these incremental risk-sharing features within traditional DBs have much less impact on employees than shifting away from the DB structure. However, all these designs transfer some degree of risk to the worker compared to the basic DB model. The question is, to what extent have these alternative designs gained traction in recent years?

Trends in Alternative Plan Design

Although the 2008 financial crisis was a watershed moment for alternative plan designs, a handful of public retirement systems – 34 of the 250 plans in the Public Plans Database (PPD) – had previously adopted some type of risk sharing. Prior to 2001, most of these arrangements involved risk sharing within traditional DBs – variable employee contributions or investment-linked COLAs (see Figure 1). But 9 plans already had DC, CB, or hybrid structures.⁵

Between 2001 and 2007, another handful of plans switched to DC, CB, and hybrid designs.⁶ Unlike the reforms to follow, these transitions were not necessarily perceived as detrimental to workers as they took advantage of a strong stock market during the period.

The 2008 financial crisis changed the picture completely. The number of state and local plans shifting investment risk onto workers more than doubled in the seven years following the crash, from 34 plans in 2007 to 79 in 2014. Importantly, legislators in this period had little appetite for moving workers into a stand-alone DC plan, where the employee bears all the risk. Instead, many reforms involved shifting new employees into less risky (for the worker) CB and hybrid plans, as well as introducing some risk-sharing within existing traditional DB plans.

In the decade since, alternative plan designs have continued to proliferate, with 107 state and local plans currently having some form of risk sharing (see Appendix Table A1).⁷ To put these numbers into perspective, Figure 2 shows the percentage of all state and local plans with alternative plan designs along with the total active members in those systems. The first cluster bar shows the share with a traditional DB, while the shorter bars to the right break the remainder down into the five different types of risk sharing. Among plans with DC elements, the hybrid model has clearly gained the most traction – affecting plans that cover 15 percent of all active retirement system members. For risk sharing within the DB, COLA-based risk sharing is the most prevalent.⁸

It is important to keep in mind, however, that this figure may overstate the share of active members currently impacted by alternative designs as many plans only applied the new features to employees hired after the reform. That said, it gives a sense of the widespread nature of risk sharing and highlights that the approach to alternative plan designs has remained protective of workers – very few plans have introduced stand-alone DC plans.

Are States or Localities More Likely to Adopt Alternative Designs?

While the trend indicates a shift away from traditional DBs, somewhat less than half of plans have done so. Which plans are making these changes? The first step here is to consider whether the changes are affecting large state-administered systems or smaller local plans. Figure 3 shows the percentage of state and locally administered plans in the PPD that currently have alternative designs. Clearly, most activity has been concentrated among state plans – more than half of which now have some alternativ design. In contrast, less than a third of local plans have adopted alternative designs.

The difference in adoption rates is perhaps surprising given that local plans for a long while had consistently lower funded ratios (see Figure 4) and higher costs than state plans. Compounding these financial challenges, local government revenue is less diversified than state revenue due to reliance on property taxes – a weakness that hit localities hard during the 2008 financial crisis.⁹ In addition, relative to a typical DC plan, self-administering a DB plan requires significant staff resources and investment expertise. Larger state governments arguably have more capacity to manage these plans, making the persistence of traditional DBs run by local governments even more surprising.

One reason for this somewhat counterintuitive trend could be the political environment in which state and local plans operate. State plans are governed by legislators who often represent a broader swath of stakeholders. Local politicians, on the other hand, generally deal with a narrower constituency and might be more easily affected by the views of employee groups. Another reason could be the type of workers covered by state and local plans. While roughly half of both state and local plans are either teacher or public safety plans, local plans are much more likely to cover police and firefighters – two groups who particularly value traditional DB benefits, are heavily unionized, and are often active in local politics (see Figure 5).¹⁰

To further examine the potential reasons behind the adoption of alternative plan designs, the next section turns to a regression analysis.

Why Did Some Plans Adopt Alternative Designs?

The sharp rise in alternative plan designs after the financial crisis suggests a defensive motivation: to avoid the costs associated with large unfunded liabilities and to unload some of the investment and longevity risk associated with traditional DB plans. But the difference in uptake among states and localities also suggests political forces at play. To check whether this story is supported by the data, we used regression analysis to find the factors associated with the probability that a plan sponsor would switch to an alternative design. The analysis includes data on each plan in the PPD from 2001-2024.¹¹ The dependent variable is set equal to zero if no action was taken in the year and 1 if the government introduced a mandatory (or default) alternative design. The independent variables include:

• Change in the plan’s funded ratio since 2001: plans experiencing a steady deterioration in funding may be more likely to switch to an alternative design.
• Republican control: jurisdictions with Republican leadership may be more ideologically motivated to adopt alternative designs. For state plans, we identify Republican leadership when the governor is Republican and all legislative bodies are majority Republican. For local plans, we classify Republican leadership as when the mayor is Republican and the city council (or other major governing body) is majority Republican.
• Social Security coverage: around one-quarter of state and local workers are not covered by Social Security because their employer has agreed to provide comparable benefits.¹² These non-covered plans may be less likely to adopt alternative designs to ensure that their members receive the annuity that they otherwise would have from Social Security.
• State plan: in general, states have greater fiscal and managerial capacity to manage a DB plan. At the same time, state legislators may consider the preferences of a broader constituency than local governing bodies.¹³
• Teacher plan: teachers are more likely to spend their whole career in the public sector and so may be more likely to value traditional DB benefits.¹⁴ They are also better represented in state and local politics, suggesting that plans catering to these workers are more likely to remain traditional DBs.
• Public safety plan: similarly, police and firefighters are also more likely to value traditional DB benefits and are often strongly represented in local politics through their unions, suggesting that plans for these workers are more likely to remain traditional DBs.

The results are shown in Figure 5 (with more details in Appendix Table A2).¹⁵ The bars show the correlation between each factor and the probability of introducing an alternative plan design in any given year. As expected, a plan’s deteriorating funded status is predictive of the switch. For ease of interpretation, the figure scales this result to reflect a 20-percentage-point drop in the funded ratio, as experienced by public plans on average between 2001 and 2009. It shows that this drop is associated with a 0.34-percentage-point increase each year in the probability of adopting some form of risk sharing.

Similarly, Republican control is associated with a 0.92-percentage-point increase in the annual probability of switching plan design, while being state-administered is associated with a 1.03-percentage-point increase. Conversely, as expected, plans covering police officers and/or firefighters are 0.87-percentage-points less likely to switch in any given year. While the coefficient on teacher plan is also negative, as expected, it is not statistically significant. And interestingly, we find no impact of Social Security coverage.¹⁶

While a 1-percentage-point change in annual likelihood might seem small, it becomes meaningful when compounded over our 25-year analysis period. For example, although 42 percent of plans currently have alternative designs, the average likelihood of adopting one in any given year was less than 2 percent. So, a 1-percentage-point increase in that baseline likelihood is actually substantial.

Conclusion

Alternative plan designs – where investment and longevity risk are shared between employees and employers – are now well established in the public sector. Indeed, plans that cover roughly half of state and local workers currently have some form of risk sharing. The shift away from traditional pensions has been driven by sharply declining funded ratios after the financial crisis, with state and local politics also playing a role. Nevertheless, most reforms to date have remained protective of workers, eschewing stand-alone 401(k)-style plans in favor of other designs that alleviate some employer burden while still providing core annuity benefits to retirees.

References

Aubry, Jean-Pierre, Siyan Liu, Alicia H. Munnell, Laura D. Quinby, and Glenn R. Springstead. 2022. “State and Local Government Employees Without Social Security Coverage: What Percentage Will Earn Pension Benefits That Fall Short of Social Security Equivalence?” Social Security Bulletin 82(3): 1-20.

Brainard, Keith and Alex Brown. 2018. “In-depth: Risk Sharing Retirement Plans.” Lexington, KY: National Association of State Retirement Administrators.

National Association of State Retirement Administrators. 2025. State Hybrid Retirement Plans: A Comprehensive Overview of State Hybrid Plans. Issue Brief. Lexington, KY.

Pew Charitable Trusts. 2017. Cost-Sharing Features of State Defined Benefit Pension Plans. Report. Washington, DC.

Public Plans Database. 2001-2024. Center for Retirement Research at Boston College, Center for State and Local Government Excellence, and National Association of State Retirement Administrators.

Quinby, Laura D., Jean-Pierre Aubry, and Alicia H. Munnell. 2020. “Pensions for State and Local Government Workers Not Covered by Social Security: Do Benefits Meet Federal Standards?” Social Security Bulletin 80(3): 1-29.

Quinby, Laura D. and Geoffrey T. Sanzenbacher. 2020. “Do State and Local Government Employees Save Outside of Their Defined Benefit Plans When They Need To?” Working Paper 2020-17. Chestnut Hill, MA: Center for Retirement Research at Boston College.

Quinby, Laura D. and Gal Wettstein. 2021. “Do Deferred Benefit Cuts for Current Employees Increase Separation?” Labour Economics 73(102081): 1-14.

U.S. Census Bureau. 2001-2023. State and Local Government Finances. Washington, DC. https://www.census.gov/programs-surveys/gov-finances/data/datasets.html.

Appendix

Click here to view appendix tables.

1. Public Plans Database (2001-2024). Fluctuations in
   investment returns take some time to fully affect a
   plan’s funded ratio because most plans smooth gains
   and losses over a period (typically 5 years) to prevent
   sharp changes in required contributions. ︎
2. Of course, the flip side is that workers also enjoy
   all the investment gains and have the option of leaving
   a bequest. ︎
3. Many states and localities also offer employees in
   traditional DB plans the option of participating in
   supplemental DC plans – 457 or 403(b) plans. These
   optional accounts are not included in this brief. See
   Quinby and Sanzenbacher (2020) for more information. ︎
4. For example, Connecticut Teachers and many of
   the state-administered Louisiana plans have used
   these types of accounts since the 1990s as the mechanism
   to provide for (or abstain from) the payment of
   COLAs. ︎
5. Texas County and District and Texas Municipal
   were opened as cash balance plans. Indiana Teachers
   and PERF were opened as hybrid DB-DC plans. Various
   plans within the Washington Retirement Systems
   began enrolling employees in a hybrid DB-DC plan
   in 1996. Finally, Michigan SERS began enrolling new
   employees in a DC-only plan in 1996. ︎
6. In 2003, Nebraska State and County plans transitioned
   from stand-alone DC plans to CB plans for
   new members. In 2004, Oregon PERS shifted from a
   traditional DB plan to a hybrid plan for new members
   that consists of a DB plan funded solely by employer
   contributions and a DC-type plan funded solely by
   employee contributions. In 2006, Alaska PERS and
   TRS shifted from a traditional DB plan to a DC plan
   for new members. ︎
7. In total, 115 plans in the PPD have had an alternative
   plan design at some point, but 8 of them shifted
   back to a traditional DB by 2025. ︎
8. Included in the variable employee approach are
   plans with fixed statutory employee and employer
   contributions defined explicitly as either a share of the
   total cost, or in direct relation to each other. While the contribution rates for these plans do not float explicitly
   with plan finances, the plans have shown a history
   of moving both the employee and employer contributions
   in lock-step whenever changes to statutory rates
   are made. ︎
9. U.S. Census Bureau (2001-2023). ︎
10. Police and firefighters, who often spend most of
    their career in the public sector, benefit less from the
    enhanced portability of DC plans. Police and firefighters
    also tend to retire earlier than those in other occupations
    due to the physically intense nature of their
    jobs, so the annuity feature of DB plans is particularly
    appealing to them (Aubry et al. 2022). ︎
11. Due to data limitations, the analysis period is
    from 2001 to 2024. The focus is on initial shifts
    away from the traditional DB from 2001 to 2024. As
    a result, plans are removed from the sample once
    their initial action is taken and plans with alternative
    designs prior to 2001 are excluded from the analysis
    completely. ︎
12. Quinby, Aubry, and Munnell (2020). ︎
13. In results not shown, we also included a variable
    for the change in the sponsor’s own-source revenue
    since 2001. The coefficient was negative – suggesting
    that plans sponsored by governments with slower
    revenue growth were more likely to take action.
    However, the value was economically and statistically
    insignificant – so it was omitted for parsimony. ︎
14. Aubry et al. (2022) and Quinby and Wettstein
    (2021). ︎
15. Robust standard errors are clustered at the plan
    level. ︎
16. This finding aligns with what has transpired for
    plans in Colorado, Ohio, Maine, and Alaska – all
    states with a very high proportion of non-covered
    workers. Although we surmised that non-covered
    plans would want to avoid alternative designs, Colorado
    PERA, Ohio Teachers, and Maine PERS all have
    implemented COLA risk-sharing and allow for variable
    employee contributions. In Alaska, despite the
    fact that nearly three-quarters of public employees
    are not covered by Social Security, all new hires are
    required to join a DC plan. ︎