By Josh Barro
One of the many points of contention in Detroit’s bankruptcy is how underfunded the city’s pension systems are. Kevyn Orr, Detroit’s state-appointed emergency manager, says the pension funds are underfunded by $3.5 billion, out of $18 billion in total city liabilities. The funds’ managers say they are only short by $650 million, because they use more aggressive assumptions that lead to a higher estimate of fund assets and a lower estimate of liabilities.
Orr is closer to being right. But the dispute between the two sides shows a key reason that governments get into trouble with pensions: Their accounting is very complicated and highly subjective, meaning it’s easy to make promises that are larger than you understand, or larger than voters understand. Politicians make decisions about pension policy all the time without knowing how much their jurisdiction owes. In most places, these errors won't lead to insolvency. (Indeed, the pension liability is not the primary driver of Detroit’s insolvency.) But they do lead to taxpayers forking over more for pensions than they ever intended, and governments having a reduced ability to invest in infrastructure, provide public services, or cut taxes.
This post explains why Orr is right, and what’s at stake for Detroit in the calculation. I should note first a counterintuitive fact about the Detroit situation: a finding that the pension funds are deeply underfunded is likely in the interest of pensioners. Detroit is seeking, through bankruptcy, to reduce obligations to various creditors. In the case of the pensioners, Detroit’s obligation is the amount by which promised pensions exceed the asset balance in the pension funds.
If Detroit’s restructuring were based on the premise that the pensions were fully funded, pensioners would likely get nothing beyond the pension fund assets. The more underfunded the pensions are deemed to be, the stronger a claim the pensioners have to additional payments from Detroit. They still get the actual pension fund assets either way. Bondholders, who are competing with pensioners for Detroit’s limited dollars, therefore want the bankruptcy to proceed based on the idea that the pensions are as well-funded as possible.
The pension liability is hard to measure because both sides of the pension funds’ balance sheets are controversial: the value of their asset portfolios, and the cost of the promises they have made to retirees. Let’s take the asset side first, because it’s simpler. Pension funds invest in a mix of assets, typically principally equities (stocks) and fixed income (bonds). It’s easy to figure out their market value. But when pension funds report how well-funded they are, they don’t actually use the market value. Instead, they use what’s called an actuarial asset value, which relies on a smoothing of asset returns.
Let’s say a pension fund anticipates an annual return on assets of 8 percent. But there’s a bad year, and asset values decline by 12 percent. The pension fund will go ahead and book the 8 percent return anyway. Then, it will recognize the -20 percent deviation from expected returns over a period, usually five years. With five year smoothing, a pension fund would actually recognize a 4 percent gain in the year with a real 12 percent loss, and then phase in 4 percent losses over each of the next four years, to reach the 12 percent decline by the end of the fifth year. You might say “that’s B.S.” You’d be right. This practice leads to pension funds, including Detroit's, claiming to hold assets that don’t exist. Actuarial smoothing does serve a useful fiscal purpose—it prevents wild swings in pension funds’ reported funded status, and therefore prevents wild swings in the amount that actuaries are telling governments they should contribute to pension funds. That helps prevent the need for sudden tax increases or program cuts to accommodate pension funds.
What a smoothed asset value does not do is tell you, accurately, how well-funded a pension system is. Nor does the underlying premise of smoothing (that the sponsoring government can make up for missed payments now with added payments later) apply to Detroit, which is seeking to stop making payments into its pension systems. So, if you’re trying to figure out how large a gap there is in Detroit’s pension funds for the purpose of its bankruptcy, you shouldn’t use a smoothed asset value. Detroit is actually a bigger offender than usual on smoothing, using a seven year period instead of five. That means Detroit’s pension funds still haven’t fully recognized losses from the stock market declines of 2008-9. Orr hasn’t disclosed his smoothing assumption, but it’s likely he’s using true market value for assets, as he should be.
The other controversy regards the pension funds’ liabilities, which are a stream of payments to retirees due in the future. The key question here is, if you owe somebody $100 in ten years, what is your liability today? To calculate this, you apply a “discount rate,” which is like a reverse interest rate, to account for the fact that a payment due in the future is less burdensome than one due now. If you owed $105 in a year and applied a 5 percent discount rate, you would say that your present-value liability is $100. Public employee pension funds typically set their discount rates equal to their expected return on assets. And typically, they have used expected returns in the ballpark of 8 percent. That aligns with this rule of thumb about a pension fund's portfolio: 70 percent invested in equities returning 10 percent a year and 30 percent in fixed income securities returning 4 percent.
The use of such high discount rates even for healthy funds is controversial. Financial economists say that pension funds should actually use discount rates that align with the risk experienced by pensioners. That would mean a discount rate aligned with bond yields, probably in the ballpark of 4 percent or 5 percent. The lower the discount rate, the higher your reported liability, and the more money you need to set aside to cover your promises. When pension funds use a higher discount rate, they assume that equity returns are a free lunch, which they’re not—taxpayers provide valuable insurance of those returns, by agreeing to shore up pension funds which underperform when the economy does badly. The 8 percent discount rate is also controversial because, even if you do believe discount rates should be tied to expected returns, that figure is likely too high. Low inflation and low real interest rates mean that a 4 percent return on a fixed income portfolio is no longer realistic. In the last four years, many pension funds have cut their discount rates into the 7s because of this critique. But Detroit is still using 8 percent and Orr has said he thinks 7 percent would be more appropriate.
In Detroit’s case, any discount rate tied to expected asset returns is inappropriate. That’s because the use of an asset-linked discount rate assumes that the pension fund has an infinite time horizon, and can ride out any temporary market dips with the sponsoring government making additional contributions as necessary. Obviously, Detroit isn’t going to do that. Think of it this way. Let’s say that Detroit’s pension funds were 100 percent funded based on an 8 percent discount rate, and Detroit terminated any further obligations to pensioners, leaving them only with the assets in the fund. Would it be true that the pensioners lost nothing of value? Obviously not. Currently, they have an insurance policy: if the 8 percent return target isn’t met, Detroit taxpayers will backstop their pensions. Losing that insurance policy hurts pensioners.
The only way to make pensioners whole for the loss of the backing guarantee is to give them enough assets to cover all expected pension payments even if the assets are invested in low-risk bonds. To have that much money, you’d have to be 100 percent funded based on a low discount rate, perhaps below 4 percent. All of which is to say, Orr’s figures, if they are based on the 7 percent discount rate he has voiced support for, likely understate the funding gap in Detroit’s pension funds. They clearly don’t overstate it.
My qualified support for Orr's position does not necessarily mean that Detroit’s plan for bankruptcy restructuring allocates resources correctly between bondholders and pensioners, or that it ought to allocate more toward pensioners. Chapter 9 municipal bankruptcy is a much more flexible and arbitrary process than corporate or personal bankruptcies, and there might be good reasons for the city to prioritize some obligations over others. But we shouldn’t conclude that Detroit has made its proposal based on a trumped up claim that its pensions are underfunded. If anything, the city's assumptions are still too aggressive.