Traditional public sector defined benefit (DB) pension plans keep the participant’s benefit accrual (and any post-retirement inflation protection features) 100% fully protected and fixed, regardless of investment outcomes, with plans generally adjusting contribution rates when actuarial experience (positive or negative) occurs. This approach places all funding risk on the contribution rate paid by employers (read: taxpayers), shields participating members, and is more likely to cause bigger swings in contribution requirements. This contribution rate volatility makes it more difficult for participating employers to manage their budgets effectively.
Some governments have taken a different approach by using “flexible benefits” that allow future benefit promises to adjust or vary to help limit large fluctuations in the required contribution rates.
The South Dakota Retirement System (SDRS) and the Wisconsin Retirement System (WRS) offer two mature examples of how flexible-benefit designs can stabilize contribution requirements and support funded levels near 100 percent while providing rule-based, but not complete, protection of pension annuities against inflation.
How flex benefits fit within broader risk-sharing
Flex benefits are part of a broader set of variable benefit- and risk-sharing approaches. Each approach requires advance decisions about how investment, demographic, inflation, or funding actuarial experience will be shared among employers, employees, and retirees, rather than allowing all adverse experiences to automatically change employer contribution rates.
The main approaches include:
- Changing the accrual design: Some plans have moved from a traditional DB structure to hybrid DB/ defined contribution (DC) arrangements, cash balance, or pure DC designs. These alternative designs spread the funding risk between employers and participants by allowing the benefits to “flex” with market returns.
- Sharing changes in required contributions: Other plans retain a DB benefit structure but require employees to share in any increase or decrease in the actuarially determined or actuarially required contribution. This spreads contribution-rate volatility between employees and employers while leaving the benefit formula itself largely intact.
- Making post-retirement increases contingent: The flex-benefit approach varies post-retirement benefit increases/cost of living adjustments (COLAs) based on funding-level targets, investment-performance, or contribution-adequacy variables. These designs can help retirees maintain purchasing power but only when the specific contingencies are met. The contingencies limit future benefit growth when funding pressure would otherwise force higher required contributions.
Why SDRS and WRS are useful case studies
SDRS and WRS show two effective ways to make post-retirement increases contingent on a plan’s financial performance. SDRS limits retiree COLAs when providing them would exceed a fixed statutory rate. WRS uses post-retirement annuity adjustments only when investment performance targets are met. Both designs use predetermined rules to limit future retiree benefit increases and related liabilities when financial conditions warrant, reducing the likelihood of needing higher contributions from either employers or employees.
In both systems, the policy objective is not to eliminate risk but to set clear rules for how risk is managed between employees’ desire for contribution-rate stability and retirees’ need for post-retirement inflation protection. That makes them useful case studies for policymakers considering alternatives or complements to variable contribution approaches.
Let’s review how each system applies this concept and what their experiences suggest about contribution volatility, funded status, and protecting retiree purchasing power.
South Dakota: Contingent COLA tied to statutory-rate adequacy
South Dakota illustrates the most direct version of the flex-benefit concept. Rather than allowing funding pressure to flow primarily into higher employer contribution rates, SDRS maintains fixed statutory contribution rates—6 percent from employees and 6 percent from employers for the main class—and uses a contingent COLA as the primary mechanism for aligning future retiree benefit growth with what those fixed contributions can support.
Each year, the actuarial valuation tests whether the full statutory COLA range of 0 percent to 3.5 percent remains sustainable. If the full range is not affordable, the system calculates a restricted maximum COLA—the highest ongoing adjustment that still keeps SDRS fully funded under the fixed statutory contribution structure. The COLA increase is the lesser inflation, measured by CPI-W up to 3.5%, or the restricted maximum with a 0 percent floor.
Future COLAs in the actuarial valuation are then assumed at this lower restricted rate rather than the baseline assumption of 2.25%. This has two practical effects:
- It directly limits future benefit growth when pressure arises, lowering projected liabilities.
- It allows the valuation itself to adopt a lower long-term COLA assumption (currently 1.56%) instead of continuing to project the higher 2.5% baseline rate. That mechanical reduction in the assumed COLA helps keep the actuarially determined contribution at or below the fixed statutory rate.
This structure makes the balancing act explicit. Retirees receive some inflation protection when funding conditions permit, but the plan does not promise to provide 100 percent inflation protection every year. When funding or contribution-adequacy tests indicate pressure, future benefit increases are constrained before the system turns to higher contribution requirements.
Over the period examined, this approach helped keep the actuarial determined contribution rate (ADC) in a narrow band of roughly 10.5 percent to 11.9 percent—consistently below the fixed 12 percent (6% employer plus 6% employee) statutory rate—while the system remained at or near full funding.
Research by the University at Albany’s Don Boyd, Gang Chen, and Yimeng Yin found that ordinary contingent COLAs (i.e., those that limit COLAs based on funding-level targets or exceeding investment-return assumptions) produce only moderate reductions in employer contribution volatility. The same papers note that more comprehensive designs “styled loosely after the South Dakota Retirement System”—those that adjust the COLA to achieve full funding within fixed resources—have much larger impacts. SDRS illustrates that stronger category.
Note: SDRS also uses a secondary flex-benefit mechanism that has not yet been used by the retirement board. SDRS benefits include the Variable Retirement Account (VRA)—a notional DC-type account that receives up to 1.5% covered compensation contribution and earns the same investment return (positive and negative) as the pension investments. The retirement board can reduce the 1.5% VRA contribution if needed to maintain the fixed statutory contribution rates. The 1.5% contribution to the VRA has remained at that level since inception in 2015, but it remains an available tool for maintaining the current statutory rate structure.
Wisconsin: Investment-performance adjustments with shared contributions
Wisconsin illustrates a different version of the same broader flex-benefit framework. WRS does reset contribution rates annually to the full ADC and shares those rates between employees and employers, but it also uses post-retirement annuity adjustments to share investment gains and losses with retirees. In that sense, WRS combines contribution-side risk-sharing with benefit-side flexibility.
WRS does not provide a traditional CPI-linked COLA. Instead, retiree annuities can increase or decrease in response to investment results through the two elements of the total annuity being paid:
- Core annuity portion: The WRS provides a base DB pension annuity to all participants. The core fund is invested to support these base DB annuity promises. Post-retirement adjustments are based on core fund performance relative to a 5 percent threshold. A core floor prevents this portion from falling below the initial amount set at retirement.
- Variable account portion: Employees may choose to invest a portion of their contribution in a separate all-equity variable trust fund. At retirement, the variable trust fund can be annuitized, and any post-retirement adjustments follow that fund’s performance with a higher threshold (±2 percent) and no floor; Importantly, the annuity portion from the variable account can rise or fall below the initial amount.
These post-retirement adjustments share investment gains and losses; they are not designed to track CPI inflation directly. That distinction is central to the case study: WRS can provide meaningful benefit growth in strong investment periods, but the inflation protection retirees experience depends on whether investment-driven adjustments keep pace with price growth.
The flex-benefit post-retirement adjustment mechanism used by WRS has helped keep the ADC in a tight range of approximately 13.6 percent to 14.4 percent. The effect of the variable trust fund on system-wide ADC volatility should be viewed as secondary because only about 12 percent of active members participate in that option.
Examples of less effective flex-benefit contingent COLA designs
Many state systems use contingent COLAs triggered by funded ratios or investment returns. Consistent with the Chen/Yin/Boyd research, these designs generally produce only moderate reductions in ADC volatility. Because their triggers are less tightly linked to contribution-rate adequacy, they provide less capacity to restrict COLAs when funding pressure rises, allowing greater contribution-rate volatility.
- Arizona Public Safety Personnel Retirement System (PSPRS) and Arizona State Retirement System (ASRS): COLAs are linked to CPI with caps that decline as the funded ratio falls (lower maximums below 90 percent, 80 percent, and 70 percent funded; COLA can be eliminated below 70 percent in some tiers). Pre-2013 ASRS tiers use a permanent benefit increase tied to excess investment returns. These provide some relief but have not prevented increases in contribution rates.
- Maryland State Retirement and Pension System: For post-2011 service, the COLA is CPI-based but capped at 2.5 percent if investment returns meet or exceed the assumed rate; otherwise, the cap falls to 1 percent. This pure return-based trigger moderates COLAs in weak years yet has allowed material contribution rate increases.
- Montana Public Employee Retirement System (PERS): Newer hires receive a COLA ranging from 0 percent to 1.5 percent depending on funded status. This funded-ratio ramp is directionally similar to Arizona’s approach and was modeled in the Chen/Yin/Boyd research as delivering only moderate volatility reduction.
- Rhode Island Employees’ Retirement System: The COLA is a hybrid of a five-year average investment return component and a CPI component. It has been suspended or heavily limited for extended periods, yet contribution rates still rose substantially, and repeated legislative intervention was required.
- Colorado Public Employee Retirement Association (PERA): Starting in 2018, Colorado combined a contingent retiree COLA with automatic employer and participant contribution-rate increases to keep the plan on a 30-year schedule for addressing its unfunded liabilities; at the time, the funded ratio across all divisions was about 60 percent. Eligible retirees receive a COLA of up to 2.0 percent. If funding progress falls behind schedule, the cap can be reduced by up to 0.25 percentage points per year, but not below 0.5 percent; the current cap is 1.0 percent. Because the COLA cannot fall below 0.5 percent, adverse actuarial experience can still require higher employer and employee contributions. For example, both rates increased by 0.5 percentage points in 2020. With an already-low COLA and a statutory floor, PERA has less capacity than SDRS or WRS to absorb investment-market shocks through benefit adjustments.
In each case, the contingency is real but less automatic, less tightly calibrated to contribution adequacy, and less able to force all residual pressure onto the benefit side. As a result, these designs have proven less effective at stabilizing the ADC than the SDRS and WRS structures.
Flex-benefit policies protect retirees against most, but not all, lost purchasing power
From 2015 through 2025, annual CPI inflation averaged about 3.0 percent using CPI-U data published by the U.S. Bureau of Labor Statistics. That average masks two very different periods: relatively modest inflation before 2021 and substantially higher inflation during 2021–2023.
Against that benchmark, the SDRS and WRS adjustments protected most, but not all, of retirees’ lost purchasing power. Using a simple comparison of average annual benefit adjustments to average annual CPI inflation, SDRS COLAs averaging about 2.15 percent protected roughly 72 percent of the CPI inflation experienced over the period. WRS Core annuity adjustments averaging about 2.61 percent protected roughly 87 percent. The WRS Variable option averaged about 7.45 percent, or about 248 percent of average CPI inflation, but that figure reflects much greater investment-driven volatility and does not mean every retiree in this much smaller group of participants was consistently protected in each year.
Comparing the two states, WRS core retirees received somewhat stronger average protection over the full period, but through an investment-based adjustment rather than a CPI-linked formula. Because core adjustments are driven by investment performance and subject to a floor at the initial annuity amount, they can protect purchasing power in favorable markets but may still lag inflation when prices rise faster than recognized investment gains.
The WRS variable option provided the highest average adjustment over the period, but it also exposed retirees to the greatest year-to-year variation. For the limited group of retirees with variable exposure, purchasing-power protection depends heavily on market timing and the sequence of returns; upside potential is greater, but so is the risk of temporary or lasting real-benefit erosion.
Overall, the results underscore the central policy trade-off. Providing 100 percent CPI protection every year would better preserve retiree purchasing power, but it would also leave less room for the benefit-adjustment mechanism to absorb adverse experiences and would increase pressure on ADC contribution rates. SDRS and WRS instead demonstrate different ways to provide partial, rule-based protection while preserving contribution-rate stability.
Policy implications for other states
States that rely primarily on variable ADC rates, or that retain more rigid post-retirement benefit structures, can draw practical lessons without necessarily adopting fixed statutory rates.
First, stronger, more automatic flex-benefit features—particularly contingent COLAs tightly linked to contribution adequacy or clear funded-status targets—can materially reduce ADC volatility. Ordinary contingent COLAs deliver only moderate effects; designs closer to the SDRS model produce larger stabilizing results.
Second, policymakers should evaluate the full package—the interaction between benefit flexibility, contribution policy, and actuarial assumptions. A contingent COLA that also permits a lower long-term actuarial assumption when restricted is more powerful than a stand-alone COLA trigger.
Third, transparency and credibility matter. Both SDRS and WRS apply their rules consistently and communicate them clearly. Members understand that future increases (or, in the variable case, current payments) can vary. That clarity supports long-term acceptance. In contrast, retirees face much higher uncertainty about COLAs when states provide COLAs only when funding levels are met or on an ad hoc basis, when and if the legislature approves an increase. This uncertainty and unpredictability do not meet transparency and credibility objectives.
Finally, flex benefits are not a complete substitute for sound funding policy. They work best when paired with disciplined assumption setting, realistic asset allocation, and a clear funding target. States considering reforms can treat automatic benefit flexibility as one tool among others—potentially allowing more stable contribution patterns while still sharing risk with members.
The experiences of South Dakota and Wisconsin, set against the more limited results of typical contingent COLA designs in other states, demonstrate that well-designed flexible benefit structures can deliver contribution stability and facilitate healthy funding of pension benefits. For policymakers seeking alternatives or complements to purely variable ADC approaches, these models offer concrete, tested options worth careful examination.