Executive Director’s Report | The Same Bet, A Thousand Times

Public pension plans are in better shape than they have been in a long time. Since the pandemic, funded ratios across the country have generally climbed — from a national average of 71.1% in 2020 to 81.2% at the end of 2025. And our analysis at Equable suggests the average funded ratio will rise to 85% by the end of 2026. The reasons for this improvement vary from state to state, but the overall factors are clear — financial markets have trended up, interest accumulation on pension debt is declining, and the growth rate of promised benefits has plateaued (largely a result of policy changes adopted in the years after the Global Financial Crisis). All of which raises the question — is this recovery real? 

The answer is yes, but it may not be sustainable.

The Recovery is Fragile, but Real

Reports from public plans aren’t just accounting gimmicks. Consider the following:

• There are still some retirement systems using unreasonably high investment assumptions that mask the true health of their pension plans, but the average assumed rate of return dropped to 6.9% by 2021 and has remained at that level since.
• State and local governments collectively are paying 100% of required contributions on average, with even New Jersey and Connecticut looking better than Illinois.
• Many plans are reaching a stage in their pension debt amortization schedules where contributions are winning the race against interest on unfunded liabilities. The share of unfunded liabilities that were due to interest on the debt peaked in 2020 and has declined $40 billion (or 13%), as of the most recent reports.
• And the growth rate in accrued liabilities has slowed down and leveled off. In the early 2000s, the average annual growth in promised benefits was 6% to 7%, but in the 2020s accrued liability growth has declined to around 3.6% to 4.2% annually. This is principally because state legislatures changed the rules after the Global Financial Crisis to reduce the value of benefits future employees could earn, and it took a few years for governments to hire enough new workers for this to make a meaningful change. 

These changes and trends have been welcome. And it is possible that all of these positive effects will persist and help drive the average public plan funded ratio over 90% and back toward 100%. Maybe.

But as nice as all of those changes have been, none of them would have mattered much to pension funded status if financial markets hadn’t recovered after the pandemic and then bent straight upward, propelled by government economic financing and trillions poured into A.I. infrastructure.

Between July 2020 and June 2026, the total assets of U.S. public plans jumped 34%, $4.3 trillion to $6.4 trillion. Fundamentally, that is why public plans have improved their funded status faster in the years since Covid-19 than they did for the decade-plus between the Global Financial Crisis and the start of the pandemic.

So, the question of whether or not the current public pension funded status is sustainable is really asking this: Are the financial market returns that state retirement systems have enjoyed sustainable?

And that is where the fragility is.

A Fragile Foundation

Public pension plans could have spent the last five to 15 years building a robust base of low-risk, modest-yield investments, long-duration bonds, and fixed-income-producing assets. But they’ve mostly done the opposite. Over the last 25 years, the average public plan has shifted from roughly a 1:2 split between defensive and return-seeking assets to closer to 1:3 — e.g., from about a third of every dollar oriented toward stability to just a quarter (shown in the figure below). The rest went searching for yield.

A shift toward private equity and hedge funds started earnestly in the wake of the Global Financial Crisis. Expansion into other alternative assets has grown steadily in the past few years, too, including a recent trend into private debt.

Throughout this period, there was a convergence of strategy, even as the underlying tactics diverged. 

Some pension funds leaned heavily into private equity, some into real estate, some into hedge funds. But these varied tactics converged on the same underlying risk posture: how much of the portfolio points at return-seeking growth versus defensive stability. Measured that way, plans have moved sharply together. 

Funds chose different routes to the same risk profile, which is why instrument dispersion and posture convergence can rise at the same time. The strategy has generally been to assemble a diverse set of growth bets, not to build a portfolio with robust countercyclical hedges that generate returns in any market environment.

What has been increasingly true over the last decade is that public plans have completely tied themselves to the overall whims and flows of financial markets. There is no substantive defensive buffer against geopolitical events or domestic economic performance that might influence private and public equity valuations or real estate.

The result: persistent fragility in returns. Market volatility contributes to pension debt paralysis. And pension funds survive but don’t thrive.

The Market Convergence

That brings us to today where public plans are essentially all making the same bets.

Theoretically, private capital markets do not have to follow the trends of public markets. The fundamentals of private equity investments can be different, and company life stages will vary. Theoretically, hedge funds that have “alternative” strategies principally focus on countercyclical strategies that generate returns no matter which direction financial markets trend. Theoretically, the valuations of real estate or private credit are based on company fundamentals rather than the interest rate environment.

But none of that has been the reality post-Global Financial Crisis. The problem for pension funds is that asset classes themselves have become heavily correlated. For example:

• The same economic forces that drive public markets also drive private equity values — often just with high management fees and debt. As companies have stayed private longer, all of the money invested in mid- to large-sized private companies has become exposed to returns based on the same consumer demand, supply chain, and economic cycle forces that influence publicly traded companies.
•  Within public markets, returns have become heavily concentrated in a small set of mega-cap “Magnificent 7” companies. Because public pensions heavily index-manage and cap-weight their equity portfolios, that concentration passes directly into public plan returns — most funds own the same handful of company stock in roughly the same proportions, regardless of their broader allocation differences.

The net result is that plans have converged on the same risk posture, despite varying tactics — and a system that has taken on the same exposure tends to earn the same way. The convergence of strategy shows up as a convergence of outcomes. The left figure below shows the range of returns for each fiscal year, where the smaller the box the more similar the reported returns were. The right figure shows that same data but aligns the boxes to put each year’s median return in the center — an approach that makes it even easier to see the narrowed variance of investment performance.

These figures show that the distance between performance at the 25^th and 75^th quartiles has declined from an average of 6.6 points between 2001 and 2020 to 4.3 points from 2021 to 2025.  

Future Markets

Pension funds vary on their portfolio mix of private capital, real estate, hedge funds, and other alternatives, but those are increasingly different routes to the same place. And that place is being entirely exposed to market ebbs and flows.

So pension funds can stay along for the ride or try to change something.

Ray Dalio, founder of hedge fund Bridgewater Associates, argued earlier this year that investors, including institutional funds like public pensions, must emphasize a portfolio of uncorrelated bets — even as hard as that might be in our current financial landscape.^[1] He contends it is a “mathematical certainty” that over time a countercyclically diversified portfolio will deliver structurally better returns for risk versus a concentrated portfolio.

Still, where there is the administrative and political will to shift investment strategies to something that could provide long-term stability, it is by no means obvious what reallocation to the right set of countercyclical bets should be. The market today is not following traditional signals, many of which are loudly disagreeing with each other. Aggregate economic data generally look good. Market sentiment does not.

Consider three independent surveys from 2026 — the General Social Survey, Gallup’s happiness data, and the University of Michigan’s consumer sentiment panel.^[2] All of these have put American self-reported financial sentiment at or near record lows. 

Political and social commentators offer a range of theories about why financial sentiment is low, but broadly everyone agrees that a key factor is prices — the effects of inflation on everyday expenses, the cost burdens of healthcare and childcare, and (perhaps most of all) housing, where the cost burden of buying a new home is around a multi-decade high. People are consuming, but they do not feel they are getting ahead. The economy can be doing well while people are not feeling like they are winning. Both things are true at once.

At the same time, even investing where there is growth could be buying into an overhyped market. The U.S. stock-market capitalization is roughly 220% of GDP going into the summer of 2026 — significantly higher than during either the peak of the Dot-Com Bubble (around 135% in Q1 2000) or Housing Bubble (around 110% in Q3 2007).^[3] The total S&P 500 investment return of 22.5% over the period July 1, 2025 to June 30, 2026 has been largely driven by seven mega-cap technology firms that account for roughly half of all market returns.^[4]

While no one debates that there is a lot of concentrated risk in financial markets today, there is plenty of debate around the degree of the risk, how much economic sentiment matters to market performance anymore, whether A.I.-driven gains can outpace potential A.I.-created losses, and what exogenous factors (like health outcomes) could be meaningful economic and investment return influencers.

An optimistic case for the future comes from the same place most of the recent returns did: artificial intelligence.

The argument is straightforward: If A.I. delivers the productivity gains its proponents expect, then today’s valuations are not a bubble but an early estimate of enormous future earnings, and the concentration in a handful of firms simply reflects that those firms got there first. On this view, the market is not overpricing assets so much as it’s ahead of the evidence, and the gap closes as the productivity shows up in the data. Such an outcome would vindicate pension portfolios who ride that wave.

This isn’t an unreasonable expectation. There are ways it could fall short — human beings as bottlenecks might slow adoption and lead to a more muted A.I. growth picture, or efforts to prevent a rogue A.I. (nominally called “alignment”) may fail. But otherwise, it’s not hard to see all of the ways that A.I. could positively affect people’s lives.

But notice what it does to the risk picture. If A.I. is what justifies today’s valuations, and all of the valuations in a highly correlated market move together, then A.I. is also what the entire system is now leaning on — which means the thousands of bets that public pensions have made in a collective growth-focused strategy are really just a single bet that A.I.-driven productivity will save public pension portfolios. An optimistic case built entirely on one source of growth is not a hedge against concentration. It is just hope.

A.I. Is Three Risks, Not One

Treating A.I. only as a market story understates it. For public pensions specifically, A.I. influences funded status through three separate doors, and they do not have to open in the same direction.

The first risk is assets. A.I.-valuations concentrate market returns in a few mega-cap firms, and because pensions cap-weight and heavily index-manage their equity portfolios, that concentration passes straight through to plan balance sheets. 

The second risk is government revenues. The money that funds pension contributions comes out of state and local budgets, and those budgets rest on tax bases that A.I. is beginning to reshape. Over the past year, Meta Platforms, Microsoft, Block, and Oracle have all cut staff — in several cases while reporting strong earnings and explicitly crediting A.I. efficiency rather than financial distress. Those reductions hit high earners disproportionately, who disproportionately concentrate in the same states (California, New York, New Jersey, Illinois) that carry the largest pension liabilities. The chain from a displaced six-figure salary to reduced state income-tax receipts to strained contribution capacity is short and direct. Whether A.I. ultimately destroys more jobs than it creates remains unsettled — but a pension system does not need mass unemployment to feel this. It only needs the high end of the wage distribution to soften in the states that depend on it most.

The third risk is liabilities. Advances in health technology that extend lives also extend the obligations pensions owe. A retiree who lives longer collects benefits longer, and a force that improves population health — better treatments, earlier intervention, whatever the next decade brings — raises the present value of every promise already made. This is an unambiguous good for the people living those longer lives and a complication for the plans paying for them. Actuaries do not currently calibrate most longevity assumptions for this factor.

The unifying point is unavoidable: A single force is now wired into the asset side, the revenue side, and the liability side at once, and it can push them in different directions simultaneously. For example, a scenario where A.I. disappoints in markets, softens the tax base, and still lengthens liabilities through unrelated health gains is not the base case — but it is a coherent possibility.

None of this requires believing A.I. is going to be a catastrophe to still be concerned. A.I. adoption and influences could simply fall into normal business cycle flows with headwinds, tailwinds, shocks, and troughs — all creating uneven economic and social changes. And since ordinary business cycles typically eliminate incumbents, a coherent story also exists in which companies adopt A.I. in a generally positive way but future start-ups wipe out the early leaders like OpenAI and NVIDIA, leading to losses for investors in those companies even where the market thrives. You do not need an apocalypse to justify diversification. Routine uncertainty is enough. That is the whole point of not putting every chip on one square.

What to Watch, and Where This Lands

Since today’s funded status fragility is structural, it is useful to monitor indicators that tell us risk exposure is turning into actual damage. Below are a few leading indicators we’re watching going into the rest of 2026 and 2027.

• State and local revenue: We will see whether budgets hold up as tariffs and geopolitical cost pressure feed through to tax receipts and spending adjustments. At least half of states have already revised their 2026 revenue expectations down, and because governments contract much of their procurement annually, some of that pressure arrives on a lag — meaning the visible hit may trail the underlying cause by a fiscal year.
• Contribution discipline under stress: Whether states keep paying close to 100% of required contributions is the single most helpful pattern of the past decade — and the one most exposed to a revenue downturn. Discipline that holds in a good year is easy. The test of discipline is a series of bad years.
• Allocation changes to address A.I. concentration: We will track whether funds actually shift away from the crowded consensus — adding genuinely countercyclical positions, rebuilding fixed income, or trimming mega-cap concentration — rather than simply rotate between growth strategies that carry the same underlying risk. Movement here would be the clearest sign that plans recognize the consolidation and are willing to trade some upside for a real hedge. Its absence would confirm the convergence is holding.

None of this guarantees the future. A.I. companies delivering on their optimistic promise, tax revenue holding, funding discipline persisting, actuaries appropriately updating liability measurements to demographic and health trends — these are all uncertainties about the future, and they increasingly correlate with one another. 

Public pension systems have improved. But they have arranged themselves so that their fate now rises and falls with the same crowded bet as everyone else’s.

Notes

1. Source: Ray Dalio. Note: To be clear, pension funds do live with constraints that individual investors don’t have to deal with — political and governance influences on investment strategy, cash flow and liquidity management needs to pay benefits, asset-liability matching strategies to consider. The majority of state retirement systems have not drifted into the crowded consensus out of carelessness. There are structural considerations driving them. But the destination is the same regardless of how they arrived: a landscape of plans that look diverse on paper and behave alike in a drawdown.
2. Sources: General Social Survey, NORC at the University of Chicago (gss.norc.org); World Happiness Report, powered by the Gallup World Poll (worldhappiness.report); and the University of Michigan Surveys of Consumers (sca.isr.umich.edu).
3. Source: Federal Reserve Economic Data. Measures total market capitalization (Nonfinancial Corporate Business; Corporate Equities; Liability, Level; sourced from Federal Reserve Z.1 financial accounts) against Gross Domestic Product (sourced from U.S. Bureau of Economic Analysis).
4. Source: Goldman Sachs and Fidelity Investments analysis of the performance of Alphabet, Amazon, Apple, Meta Platforms, Microsoft, NVIDIA, and Tesla.