As an alternative to changing the decision-making process itself, employees and retirees could be required to share the risk associated with funding policy decisions. In Arizona, employees pay 50% of any unfunded liability amortization payment — which incentivizes lower assumed rates of return. In Wisconsin, retirees are promised a base pension benefit, and then can have that benefit increased when returns are strong, but decreased as low as the base benefit when returns underperform, meaning all parties share in the upside and downside of the investment allocation.
Other cost sharing models could be designed for SERS, so long as the paramount objective would be to incentivize better funding policy by linking decisions related to risk in the system with the liabilities created both by those decisions and by benefit design.
Understanding the Forecast
Baseline: The yellow line running across the chart above is the total employer contribution, combining the normal cost plus the unfunded liability amortization payment. This line represents the expected baseline forecast under the current plan assumptions – including the 6.9% assumed rate of return adopted under SEBAC in December 2016. The yellow line baseline representation will remain constant throughout the forecasting scenarios.
Normal Cost: The dark columns at the bottom are the employer’s share of normal cost. For the current fiscal year ending 2017, the employer share of normal cost for all tiers is 10.3%. Specific normal cost rates vary depending on the kind of employee, but to consider how any given set of changes would change expected contributions, it is best to look at the combined system as a whole. Note that normal cost is forecast to decline slightly over time, as the normal cost for Tier III (5.5%) is slightly less than the normal cost for legacy tiers.
Unfunded Liability Amortization Payment: The light columns at the top are the amortization payments, and are always paid by the employer. For the current fiscal year ending 2017, the unfunded liability amortization payment is 38.7%. Under the baseline that existed prior to the passage of the December 2016 SEBAC agreement, these were scheduled to end in 2033. With the passage of the agreement, the amortization payments will be stretched out past 2047.
Scenario: This scenario forecast assumes that the actual experience for SERS over the next 30 years is exactly what the actuarial assumptions expect, including actual annual returns of 6.9% and average COLA of 2.3% for Tier III.
Limitations: In order to create an apples-to-apples comparison, we have adopted all assumptions used by the plan (unless expressly indicated otherwise), but that does not mean we endorse those assumptions. The accuracy of these forecasts is only as strong as the reasonableness of the assumptions currently used by SERS. In that respect, we consider all of these forecasts to have underlying limitations in accuracy in relation to the assumptions being used.
Thus, the primary value of these forecasts is in comparing the difference between the scenarios and how a limited change will change the outlook, rather than in the specificity of a dollar amount forecasted 10 or 20 years from now. As previously stated, changes to the demographic assumptions of SERS are necessary to improve solvency, but a detailed analysis of how to apply such changes is necessarily outside the scope of this paper.
Any forecast becomes less reliable the longer out in time it goes, and that is no less true in our forecast than for forecasts by SERS itself.
Scenario 1: Lowering the Assumed Return to 5.5%
This forecast adopts a 5.5% assumed rate of return starting with FYE 2018, and then assumes the actual experience for SERS over the next 30 years aligns with actuarial assumptions, including actual annual returns of 5.5%, a 5.5% discount rate for valuing liabilities, and an average COLA of 2.3% for Tier III.
The fiscal effect of this change would be to increase gross normal cost for all tiers combined by 4.8% percentage points. We assume no change to the employee contribution rate in this scenario, so the employer would pay for the increase to a more accurately-priced normal cost and contribute 15.1% in fiscal year ending 2018 towards normal cost. The scenario would also change the discount rate to 5.5%, resulting in the recognition of more unfunded liabilities and therefore increase the amortization payment from 44.2% to 48.7% of payroll.
The solvency effect of this would be to reduce market risk exposure and contribution rate volatility while also improving the accuracy of normal cost pricing of benefits.
Scenario 1 Volatility Analysis:
The volatility effect of this change would be to reduce the range of probable increases in employer contribution rates because the asset allocation would change to include more stable investment vehicles, decreasing investment risk. The figures below compare volatility illustrations for the change in employer contributions rates given varying actual returns.
Scenario 2: Increasing Employee Contributions to 6%
This scenario forecast changes all employee contributions to 6% starting with FYE 2018, and then assumes that the actual experience for SERS over the next 30 years aligns with actuarial assumptions, including actual annual returns of 6.9% and an average COLA of 2.3% for Tier III.
The fiscal effect of this change would be to decrease the employer share of normal cost from 10.3% to 6.2%, producing taxpayer savings in the short-term and long- term.
The solvency effect of this change would be based on how the state utilized the savings from the change. If the savings were put back into the retirement system, then the unfunded liability would be reduced faster.
Scenario 3: Adopting a Max 2% COLA for All Tiers
This scenario forecast changes the COLA formula to be a maximum 2% benefit adjustment based on the change in CPI-W starting with FYE 2018, and then assumes the actual experience for SERS over the next 30 years aligns with actuarial assumptions, including actual annual returns of 6.9%. The average assumed COLA for all tiers in this scenario is 1.75% since the long-term average for inflation would likely be less than the 2% max.
The fiscal effect of changing the benefit formula would reduce outflows from plan assets and link COLAs with actual inflation instead of a percentage of change in inflation. The current formula has a minimum of 2% to 2.5% and maximum of 6% to 7.5% depending on hire date, and is based on a percentage of the change in CPI-W. However, for almost every year over the past two decades, inflation has been below the minimum COLA rate. From this perspective, COLAs are not benefit adjustments to keep up with inflation, they are simply a benefit increase.
Scenario 3 Liability Analysis: The liability effect of this change would be a reduction in the forecast of accrued liabilities, since the expected adjustment of benefits would be less under a system with a maximum 2% COLA compared to a minimum 2% COLA. As shown below, there would be a 7% change in liability growth over the next 30 years as a result of adopting a maximum 2% COLA.
The solvency effects would be lower benefit outflows from plan assets, allowing previously accrued contributions made in anticipation of higher COLA payments to be applied towards overall plan solvency, and lower growth in liabilities that are exposed to the aggressively optimistic actuarial assumptions of SERS.
Scenario 4: Adopting a $100,000 Pensionable Pay Cap for New Hires
This scenario forecast adopts a cap on pensionable compensation for new hires only at $100,000 starting with FYE 2018, and then assumes the actual experience for SERS over the next 30 years aligns with actuarial assumptions, including actual annual returns of 6.9% and an average COLA of 2.3% for Tier III. The scenario also assumes that new-hire employee contributions are based on only the first $100,000 of salary.
The fiscal effect of creating a Tier IV employee class with a pensionable compensation cap would depend on where the cap is placed and how many employees would become subject to the cap. The 2016 current cap from the IRS is $265,000, which applies to very few state employees. Lowering the pensionable compensation cap to a fixed $100,000 would mean paying out lower benefits, since the largest final average earnings figure would be the $100,000 limit.
The solvency effect would be based on lower growth in liabilities that are exposed to the aggressively optimistic actuarial assumptions of SERS.
The fiscal effect of creating a Tier IV employee class with a pensionable compensation cap would depend on where the cap is placed and how many employees would become subject to the cap. The 2016 current cap from the IRS is $215,000, which applies to very few state employees. Lowering the pensionable compensation cap to a fixed $100,000 would mean paying out lower benefits, since the largest final average earnings figure would be the $100,000 limit.
The solvency effect would be based on lower growth in liabilities that are exposed to the aggressively optimistic actuarial assumptions of SERS.
Scenario 4 Liability Analysis: The liability effect of adopting a lower pensionable compensation cap would be a reduction in the forecasted accrued liabilities, since the expected benefits for the new Tier IV would be less than the benefits expected for new hires into Tier III. As shown in the figure below, there would be a 24.6% change in liability growth over the next 30 years as a result of adopting $100,000 pensionable compensation cap.
Scenario 5: Adopt a DC Plan for New Hires, 7.7% Employer Rate
This scenario forecast adopts a Tier IV defined contribution plan only for new hires starting with FYE 2018, and then assumes the actual experience for SERS over the next 30 years aligns with actuarial assumptions, including actual annual returns of 6.9% and an average COLA of 2.3% for Tier III. The defined contribution plan modeled here would have an employer contribution of 7.7%, which is roughly equivalent to the employer’s share of normal cost for new hires into Tier III. The forecast assumes existing unfunded liabilities would be amortized over total payroll, with the same method and schedule as the status quo.
The fiscal effect of creating a defined contribution plan for new hires primarily depends on the contribution rate offered by the employer. If the DC employer rate exceeds expected (though probably underpriced) normal cost for new hires, then there will be a forecasted cost increase. If the DC employer rate is less than expected normal cost for new hires, then the forecast will expect savings.
The solvency effect created by switching to a defined contribution plan is also important for understanding the fiscal effects. Bringing all new hires into a plan with zero accrued liabilities means that, over time, the amount of pension promises exposed to the aggressive current actuarial assumptions of SERS will decrease rather than increase. Reducing the liabilities that will likely be underfunded by the current funding policy means that, over time, SERS will be better funded with a DC plan in place for new hires relative to the status quo.
Note: The same kind of solvency effects would be created by a cash balance plan or a DB-DC hybrid plan, though to a lesser extent.
Scenario 5 Liability Analysis: The liability effect of adopting a defined contribution plan for new hires would be a directional change in the forecast of accrued liabilities, since new hires into Tier IV would produce no liabilities. As shown in the below figure, liabilities will grow slightly in the first few years following the adoption of a DC plan for new hires, because members already in the defined benefit tiers of SERS would continue to accrue pension benefits until they retire. After about 10 years, the liabilities begin to decline and eventually fall to zero.
SERS is clearly a troubled pension plan, with $21.7 billion to $25 billion in unfunded liabilities (depending on how they are valued). Over the past few decades, investment returns have consistently underperformed expectations by a wide margin, while the asset allocation has been shifting toward riskier investments in an effort to compensate for these shortfalls and chase higher yields. Over the past 15 years, the share of relatively safer, fixed income products has been reduced from about one-third of plan assets to only a one- fifth of plan assets.
Given SERS’s current actuarial assumptions and funding policies, there is a high degree of volatility in prospective future employer contribution rates, creating budgeting challenges down the road. The amortization methods used for paying down unfunded liabilities over the past few decades have been focused just on keeping near-term payments low, rather than actually reducing or eliminating pension liabilities. And even when the state has paid 100% of the actuarially determined contributions — a practice that has been anything but consistent — they haven’t been enough to fund the plan properly because the discount rate used to value liabilities has been too high. Collectively, the net effect of these problems has been spiraling pension payments, which crowd out spending on other government services and require higher taxes.
Those with power in the decision-making process — including members of the State Employees Retirement Commission, labor leadership associated with SEBAC, and prior state governments — have failed to adequately ensure the long-term solvency of SERS. Prior collective bargaining agreements ignored the need to adjust actuarial assumptions to account for demographic and market changes, while explicitly allowing the underfunding of actuarially determined contributions.
Solving these problems requires all interested parties in Connecticut to focus on ensuring the long-term solvency of SERS; provide retirement security for its members; stabilize contribution rates; reduce taxpayer exposure to financial risk; reduce long-term costs; ensure the ability to recruit 21st century employees; and improve the incentive structures within the current governance of the plan.
The most substantive action taken recently with respect to addressing SERS’s problems was the December 2016 SEBAC agreement. However, the plan as presented to the General Assembly took only one limited step toward improving the actuarial assumptions of the plan — lowering the assumed return from 8% to 6.9% — while taking several steps backwards in once again extending the schedule for paying of unfunded liabilities. This “solution” — adding more years to the timeline for paying o the debt in order to reduce payments in the near term — just repeats failed policies of the past that contributed to the problem today. The net outcome of the agreement adds $8 billion to $9 billion in additional interest payments on the unfunded liabilities for taxpayers in the future, just to make budgeting in the next decade easier.
The next set of solutions should start with careful consideration of the menu of meaningful reform options set forth in this paper, including:
- Lowering the assumed rate of return to a level that would allow a less risky asset allocation and more accurately priced normal cost;
- Lowering the discount rate to a level consistent with the market value of liabilities;
- Increasing employee contributions;
- Changing the formula for cost-of-living adjustments;
- Adopting a cap on pensionable compensation for new hires;
- O ering new hires a more appropriately priced and governed de ned bene t plan;
- O ering new hires a de ned contribution plan, cash balance plan, “DB-DC” hybrid plan;
- O ering new hires an optional de ned contribution only plan;
- Re-organizing the governing process for SERS such that the parties with the most liabilities have the greatest degree of control over funding policies.
Addressing pension challenges is no easy task. It is a complicated, multifaceted problem with a wide range of competing, powerful, and often-vocal political interests. But what’s even more obvious is that inaction would be catastrophic.
There are sensible and sustainable options on the table. Now it’s time for the people’s representatives to summon the statesmanship and courage necessary to keep past promises by protecting existing pension benefits, and securing our state’s future by ensuring that future state worker retirement bene ts do not undermine the financial condition of the state and the taxpayers they have been elected to serve.
About the Authors
Anthony Randazzo is managing director for the Pension Integrity Project at Reason Foundation and the author of numerous studies on the solvency challenges for public sector retirement systems. He holds an M.A. in behavioral political economy from New York University.
Daniel Takash is a policy analyst for the Pension Integrity Project at Reason Foundation. He holds a B.S. in Applied Mathematics & Statistics and Political Science from Johns Hopkins University.
Adam L. Rich is a Fellow of the Casualty Actuarial Society and holds a B.S. in Mathematics from Brigham Young University. He lives in South Windsor, Connecticut.