There are many theories about why America has over a trillion dollars in public pension debt. And there is certainly more than one reason why, leading to a tangled web of factors causing unfunded pension liabilities. But until now there wasn’t clarity on to what degree any specific factor had contributed to the national funding shortfall for state retirement systems.

We recently collected a large amount of data from statewide retirement systems that has helped to answer some critical questions: what are the specific financial reasons that the defined benefit plans say has caused this underfunding phenomenon? Have benefit increases been a major contributor to the growth in unfunded liabilities? How much has the failure of state legislatures to pay their pension bills contributed to the funding shortfall? What share of the overall unfunded liability amount is because of changes to actuarial assumptions?

The answers to these questions and more are part of a new working paper from Equable Institute that presents actuarial gain/loss data as gathered from statewide retirement systems. We are publishing this in preliminary form to solicit feedback from others about what the implications of this data might be and for comment on how we have classified the data. We encourage those interested in this analysis to , as we are happy to share the information and discuss the findings.

Download the Working Paper

Preliminary Findings

America’s state defined benefit plans have not always been underfunded. At the start of the 2000s, statewide retirement systems were collectively around $40 billion overfunded and nearly all plans were 100% funded (or close). However, by the end of fiscal 2019 statewide defined benefit plans held a collective $1.39 trillion in unfunded liabilities, on an actuarial basis. There is no shortage of academic literature analyzing the general political, institutional, and economic reasons why this has happened. But what are the specific financial reasons that the defined benefit plans say has caused this underfunding phenomenon?

State-administrated plans publish data each year showing what factors led their funded status to increase or decrease, typically known as actuarial gain/loss analysis. At the start of the 2000s statewide retirement systems were collectively around $40 billion overfunded and nearly all plans were 100% funded (or close). However, by the end of fiscal 2019 statewide defined benefit plans held a collective $1.39 trillion in unfunded liabilities, on an actuarial basis. Actuarial gain/loss data can help us understand the scope of each possible source of growth in state pension debt.

Based on our analyses of these data, the primary factors that contributed to the growth in state defined benefit plan unfunded liabilities over the past two decades were: (1) underperforming investment experience, (2) changes to actuarial assumptions, and (3) interest on unfunded liabilities growing faster than contributions.

Specifically, the data indicate that 40.85% of the change in unfunded liabilities can be attributed to “Investment Experience.” That is, some $540 billion of the growth in unfunded liabilities since 2000 is due to plans’ actual investment returns underperforming their assumed rates of return, plus related asset experience.

Much of the remaining change in unfunded liabilities can be attributed to “Assumption Changes” (27.5%) and “Contribution Experience & Interest on the Debt” (25.9%). Both of these are categories that we’ve developed based on the terminology use in actuarial valuations, and they are discussed at length in the working paper.

Notably, factors like individuals living longer than expected and retirement patterns varying from expectations were not significant contributors to unfunded liability growth. Moreover, while benefit design improvements and unfunded COLAs did contribute to an increase in unfunded liabilities overall, COLA reductions in the past few years reduced unfunded liabilities by a greater amount, such that benefit experience was also not a significant contributor to unfunded liability growth.

This report is the first comprehensive assessment of detailed actuarial gain/loss data across all major state-administrated retirement systems. It is designed to provide specific attribution to different factors that have contributed to the growth in state pension debt over the past two decades. While each of these sources have been long known to be contributing factors to the growth in pension debt, until now there has not been an estimate of the specific share that each has provided to the expansion of unfunded liabilities.

Why this Project is Important

A frequent challenge for any state wrestling with how to manage its public pension plans or other retirement systems is a lack of agreement over what problems they are really trying to solve. What this project aims to show is the specific dollar amounts behind each of these causes and others, providing a clear picture of how much any one of these factors contributed to the total level of unfunded liabilities. The data we have compiled for this project can help offer clarity to those debates. Consider the following common arguments related to pension policy:

“States have unfunded pension liabilities because governments provide overly generous benefits and increased those benefits without paying for the changes.” — This argument is often associated with Republicans, conservatives, or others who advocate in opposition to public labor unions.

“States have unfunded pension liabilities because governments have not paid their required contributions, leaving them shortchanged.” — This argument is often associated with Democrats, progressives, or labor union leaders.

Irrespective of whether either of these arguments are made in good-faith, or are just part of broader political debates, we can directly address them both in this project. We can look at what share of the $1.39 trillion shortfall in 2019 is because of benefit increases and/or because of contribution rate deficiencies.

We can also dive deeper into these categories by looking at what share of benefit experience is related to changes to benefit formulas and what is related to cost-of-living adjustments. We can even assess whether many of the benefit value reductions adopted over the past decade have offset the benefit enhancements seen at the beginning of the decade. We can even differentiate between contribution shortfalls due to legislatures explicitly not paying required contribution rates, and implicit funding shortfalls where the interest on unfunded liabilities is growing at a faster rate than contribution inflows— even when legislatures fully fund actuarially determined rates. Another reason why this project is important is that it can help with policy questions related to pension funding in the future. Consider these two questions:

Between FYE 2009 and 2019 (e.g., after accounting for the losses of the financial crisis), the reported level of unfunded liabilities for state defined benefit plans increased by roughly $260 billion, despite financial markets experiencing a historic bull run. At the same time, the average assumed rate of return used by those same plans fell from 8% to 7.2%. What share of that $260 billion increase in unfunded liabilities is associated with changes to assumptions (which were generally made to adopt more realistic expectations about the future and reflect responsible policy decisions) as opposed to underperforming investment returns? If we control for changes to assumptions, did state defined benefit funding status actually improve over the decade after the financial crisis?

At the end of FYE 2001 only a few state pension plans had unfunded liabilities, and many were overfunded or around 100% funded. For states that entered this century with those unfunded liabilities, what share of their funding shortfalls today are actually a legacy of 20th century funding policy decisions as opposed to being related to policies and experiences over the past two decades?

Understanding answers to these questions could help inform how funding policies should be designed going forward, such as whether or not “legacy” unfunded liabilities should be paid off using a different time frame than contemporarily developed funding shortfalls.

Download the Working Paper