Historical Analysis

Unfunded Liabilities

The funded ratio is a percentage calculated by dividing the value of plan assets by the estimated liabilities of benefits promised to members. It is based on the actuarially smoothed value of assets or the market value of those assets. “Actuarially smoothed reporting of assets” refers to the practice of valuing assets by spreading investment gains and losses over a five-year period to prevent market volatilities from disturbing fiscal planning. The market value estimates are based on the most recent market price of assets (or declared market price).

The difference between plan assets and liabilities is the unfunded accrued liability (UAL), frequently referenced as “pension debt.” This measure is also referred to as Net Pension Liability (NPL).

Figure 1 shows the plans’ UAL in red and the funded ratio as the blue line. SERS is at a 50.4% funded ratio, as seen in Figure 1a. This indicates that SERS still has only about half of the assets it will need to fulfill pension promises made to public workers — short $20 billion. While still dire, SERS current position marks a significant improvement from its low of 36% funded in 2016.

Figure 1a. Unfunded Liability and Funded Ratio - Connecticut SERS

Figure 1b depicts the funded ratio for STRS, which is 59.8%. This is an improvement from the low of 56% in 2016. Despite this progress, STRS still has a $16.4 billion unfunded liability.

Figure 1b. Unfunded Liability and Funded Ratio - Connecticut STRS

The progress can be attributed to the fiscal guardrails implemented in 2017, which directed additional budget funding to stabilize the growth in unfunded liabilities (see Section I). This stabilization has been visible through slowing the growth and eventual reduction of unfunded liabilities past 2016, which is especially impressive considering the accounting assumption adjustments that have increased the present value of both plans’ liabilities (see Figure 3).

SERS’ funding position demonstrated considerably more improvement than STRS.’ This is due primarily to the State Treasurer’s decision to prioritize SERS for additional funding. Evidence of this prioritization: in September 2023, the State Treasurer allocated an additional payment of $1.05 billion to SERS to reduce its unfunded pension liability while only directing $272.8 million to STRS.

Both funds are recovering and forecasted to reach 100% funding by 2053, contingent on expectations being met precisely as projected. To protect the retirement security of a membership base that covers more than 5.6% of the state’s population, the state must remain committed to significant budgetary allocations for decades. This commitment will ensure the pension systems’ long-term viability, allowing them to overcome market volatility or unexpected deviations from accounting expectations.

Figure 2 shows the difference between the plans’ assets and liabilities, visualizing in gray the extent to which liabilities have grown faster than the fund’s assets.

Figure 2a. SERS Market Value of Assets vs Accrued Liability

Figure 2b. STRS Market Value of Assets vs Accrued Liability

Drivers of SERS Debt

What has driven SERS’ and STRS’ increase in unfunded liabilities from 2003/2004 to 2023?

Unfunded liability is the difference between a pension fund’s assets and its liabilities, which occurs because of persistent insufficient pension contributions. Pension liabilities can rise or fall for a multitude of reasons. When a rise in contributions does not immediately accompany a rise in liabilities, unfunded liabilities accumulate.

The $25.86 billion net increase in SERS and STRS pension debt between 2004 and 2024 can be attributed to several key factors (see Figure 3). Much of this increase stems from revisions in actuarial assumptions and methods — i.e., adjustments to investment and demographic expectations to better align with observed outcomes. A major contributor to this change was the necessary reduction in the pension liability discount rate to more accurately reflect investment returns. Pension discount rates are determined by the assumed rate of investment return (seeFigure 6).

Suppose a homeowner contracts a builder to construct a house for a fixed price of $600,000. If during construction the cost of lumber significantly exceeds forecasts, the contractor remains obligated to complete the project at the agreed price. The contractor absorbs the extra costs to fulfill the contract without charging the homeowner.

Suppose a homeowner contracts a builder to construct a house for a fixed price of $600,000. If during construction the cost of lumber significantly exceeds forecasts, the contractor remains obligated to complete the project at the agreed price. The contractor absorbs the extra costs to fulfill the contract without charging the homeowner.

Similarly, the state must fully cover the difference between expected and actual market outcomes to ensure that the promised benefits are fully paid. This situation places the financial burden of any underperformance in investment returns solely on the state, and by extension, the taxpayer.

Although lowering the assumed return rate increased the pension debt, it was a necessary and prudent decision to align contribution expectations with realistic market conditions, ultimately fostering more accurate and stable long-term financial planning.

After changes in actuarial assumptions, which added $7.09 billion to the UAL, the second most significant factor influencing SERS pension debt growth was the failure of its labor force demographic expectations to align with initial assumptions, thereby adding $4.1 billion to its unfunded liabilities from 2004 to 2023. This labor dynamic is explored further in Yankee Institute’s 2022 study, CT’s Growing Problem: Population Trends in the Constitution State.

Investment returns falling short of expectations added $3.3 billion, highlighting the gap between assumed and actual returns. Net amortization contributed $1.5 billion, as debt payments were insufficient to offset high-interest accruals. (Massive interest costs have prevented significant debt reduction.) Pay increases deviating from actuarial expectations and the rising costs for cost-of-living adjustments (COLA) added $1.4 billion and $1.2 billion, respectively.

Figure 3a. Origins of SERS Pension Debt 2004-2023

Source: Pension Integrity Project analysis of SERS valuation and reports.

Like SERS, between 2003 and 2023, the most significant contributor to the growth of STRS pension debt was changes to actuarial assumptions/methods, which added $6.9 billion to unfunded liabilities. Investment performance falling short of expectations was the second most significant factor, adding $6.1 billion. Net amortization contributed $3.5 billion, as debt payments were insufficient to cover high-interest accruals. Faulty demographic assumptions to align with initial assumptions added $1.02 billion, while uncategorized liability experience further increased the debt by $839 million.

Certain factors reduced STRS’s unfunded liabilities during the period. Reductions in COLA expectations led to a $442 million decrease and pay increases deviating from actuarial expectations reduced liabilities by $1.5 billion. Additionally, revenue from general obligation bonds reduced the unfunded liabilities by $2 billion, and receivable contributions further decreased the debt by $2.6 billion. Despite these estimated decreases and additional funds, the net change resulted in an increase of $11.9 billion in unfunded liabilities over the period.

Figure 3b. Origins of STRS Pension Debt 2003-2023

Source: Pension Integrity Project analysis of STRS valuation and reports.

The receivable transfers observed in green for both SERS and STRS, which reduced unfunded liabilities by $4.2 billion, represent the additional General Fund surplus transfers arising from the mechanisms in Connecticut’s fiscal guardrails.

Figure 4 illustrates the fund’s challenges in reducing debt despite steadily increasing contributions, highlighting the different elements driving interest costs for the pension debt of SERS and STRS from 2004 to 2023. This shows that interest accrued on unfunded liabilities for both plans often exceeded the contributions meant to reduce them.

This negative amortization position indicates that the system’s debt often grew faster than it was being paid down, posing a challenge to reducing the overall pension debt burden.

For SERS (Figure 4a), the trend shows that interest on debt (orange bars) consistently exceeded amortization contributions (blue bars), leading to net amortization (light blue bar) that was often negative — until 2017, when the impact of the additional contributions visibly reverted the trend for most of years since then.

Figure 4a. SERS Interest on Debt Exceeding Amortization 2004-2023

SERS has been in a negative amortization position since at least 1996. However, as the graph demonstrates, it appears to have reverted from this negative amortization, maintaining positive net amortization for more consecutive years than ever since 2004.

Similarly, STRS (Figure 4b) displays a comparable pattern, with interest costs frequently outpacing contributions and a more persistent trend of negative net amortization given the funding priority given to SERS (see section 1, unfunded liabilities).

Figure 4b. STRS Interest on Debt Exceeding Amortization 2003-2023

Figure 5a and 5b display the trend of SERS and STRS unfunded liability interest costs contributing to their pension system’s. The orange section represents unfunded liabilities from other sources, while the shaded gray section indicates the unpaid interest on these unfunded liabilities. This visually illustrates the impact of accrued interest, compounding interest costs, on the overall debt.

Figure 5a. SERS Interest Costs Adding to Debt 2011-2023

Figure 5b. STRS Interest Costs Adding to Debt 2003-2023

Investment Performance and Asset Allocation

Figure 6a illustrates the investment returns on SERS fund assets from 2001 to 2023. The yellow line represents the assumed return rate, which was initially set at 8.3%. It was first lowered to 8% in 2013 and then prudently lowered to 6.9% in 2017 to more accurately represent the fund’s investment returns. The blue line shows actual market returns, while the black and orange lines indicate the 5-year and 10-year rolling average returns at different periods.

Figure 6a. Investment Returns on SERS Fund Assets 2010-2023

Figure 6b illustrates the investment returns on STRS fund assets from 2001 to 2023. The yellow line represents the assumed return rate — which, like SERS, was initially set at 8.5%, then reduced to 8% in 2017 and further to 6.9% in 2018 to more accurately represent the fund’s investment returns. The blue line shows actual market returns, while the black and orange lines indicate the 5-year and 10-year rolling average returns at different periods.

Figure 6b. Investment Returns on STRS Fund Assets 2001-2023

Connecticut’s SERS and STRS have had similar investment performance over the past 20 years, though STRS appears more volatile. Both funds initially set high assumed return rates (SERS at 8.3% and STRS at 8.5%) that did not accurately represent their achieved returns, which were later reduced to 6.9%. The 5-year and 10-year rolling average returns for both plans are aligned with the new assumed return rate of 6.9%, indicating a more accurate reflection of long-term investment performance under the adjusted return assumptions. Past performance, however, does not guarantee similar results for the plans’ future, as shown in the subsequent Investment Return Probability section.

Figure 7 shows the allocation of SERS and STRS investment assets from 2000 to 2021, highlighting the shifts in asset allocation over time across various asset classes.

Connecticut’s SERS and STRS have exhibited similar asset allocation trends over the past two decades. Both funds have significantly reduced their bond holdings from approximately 25% to about 10% of the total portfolio. Equities have consistently remained the dominant asset class, making up around 60-70% of total assets throughout the period. Real estate and commodities have maintained relatively stable, smaller portions of the allocation.

Both plans have moderately increased their allocations to alternative assets, expanding their real estate, hedge fund and commodity allocations, while keeping the share of their portfolio allocated to private equity stable over the past 20 years.

This trend extends beyond SERS and STRS, as pension systems nationwide have increasingly turned to alternative assets, seeking higher returns through riskier investments. Rather than lowering their assumed return rates to reflect the actual rates achievable with a safer portfolio, some pensions have opted to take on more risk to meet these higher assumptions. While this strategy can boost returns, it also introduces greater volatility and significant downside risks, potentially increasing debt and costs for taxpayers.

Figure 7a. SERS Allocation of Investment Assets 2001-2023

Figure 7b. STRS Allocation of Investment Assets 2001-2023

Investment Return Probability

The assumed rate of return (ARR) represents the expected rate of return on the pension fund’s investments, which guides how much Connecticut’s government must contribute annually to meet pension obligations. This rate also serves as the discount rate for calculating the present value of future liabilities. A higher discount rate lowers the present value of liabilities, while a lower rate increases it.

Pension policymakers set the ARR based on investment advisors’ projections. Pension boards first establish the asset allocation — bonds, stocks, alternative investments, etc. — and then determine the ARR from a weighted average of the expected returns in each asset class, aiming for a 50% probability of achievement.

A higher ARR reduces the contributions required from employees and their employers (in this case, the state government), while a lower ARR increases the necessary contributions to fulfill promised benefits. That is why overestimating market returns can ultimately create unfunded liabilities, as the pension fund ends up with fewer funds than anticipated.

SERS and STRS have lowered their ARR to 6.9% over the last few years. Based on market forecasts, what are the chances of these plans meeting its ARR in the next 10 or 20 years?

The probability of the plans meeting their ARR is assessed by applying the current asset allocation to market forecasts from various research and investment firms. Figure 8 illustrates these probabilities based on different sources and historical data.

Figure 8a. Connecticut SERS Unfunded Liability and Funded Ratio

For each of the market forecasts given SERS asset allocations, 10,000 Monte Carlo simulations of investment returns were run. Each of the dots represents a few dozen of these simulations. Green indicates the ARR will be achieved, red indicates it will not. The simulations run using JP Morgan and Horizon’s forecasts indicate that there is a 57-62% chance that SERS will achieve the ARR of 6.9%. The BNY Mellon and Research Associates’ forecasts indicate that there is only a 39-49% chance that SERS will achieve an ARR of 6.9%. The STRS forecasts were slightly more volatile, but largely mirrored the SERS forecast. This mixed outlook highlights the uncertainty and the challenges the funds face.

Basing the probability on the plan’s historical performance suggests only a 25% and 26% chance of SERS and STRS meeting their ARR, respectively.

The Yale School of Management has shown how Connecticut’s historical investment performance has frequently ranked at the bottom of pension investment return rankings. Over the last decade, a generic no-overhead 80/20 investment portfolio, without active management and simply tracking the S&P 500, would have outperformed Connecticut’s portfolio by 3% annually.

If realized, the plan’s current outlook of a 6.9% long-term return rate would improve the fund’s performance compared to the last few decades. Policymakers should be aware that assuming a result that hasn’t occurred in recent years carries significant risk. Although the current assumption aligns with some market forecasts, it is not a guaranteed outcome, and prudent planning should account for variability.

Figure 8b. Connecticut STRS Unfunded Liability and Funded Ratio

If realized, the plan’s current outlook of a 6.9% long-term return rate would improve the fund’s performance compared to the last few decades. Policymakers should be aware that assuming a result that hasn’t panned out in recent years carries significant risk. While the current assumption aligns with some market forecasts, it is not a guaranteed outcome, and prudent planning should account for variability.