How bad is the state pension funding mess?

Dean Baker says not so bad; Kevin Drum, Paul Krugman, and others seem to take his side.  Josh Barro says it's bad.  I side with Barro.  Here is one Baker passage:

The total shortfall for the pension funds is less than 0.2 percent of projected gross state product over the next 30 years for most states. Even in the cases of the states with the largest shortfalls, the gap is less than 0.5 percent of projected state product.

Beware of the 30-year comparison I say.  A lot of sums look small compared to thirty years' worth of output.  I worry when I read sentences such as this:

The major reason that shortfalls exist at all was the downturn in the stock market following the collapse of the housing bubble, not inadequate contributions to pension funds.

In my house, that's what inadequate means.  I also see Baker relying on a dangerous version of an equity premium argument, when I'd rather see a probability distribution of scenarios.  I don't see Baker — not once — analyzing the public choice considerations of how state governments actually behave and treat their finances.  Or how about how state voters hate tax increases, reasonably or not, and think their governments should be forced to actually solve their mismanagement problems?  A crisis usually is an institutional crisis.

Here is a typical passage from the Barro piece:

New York taxpayers have learned about these dangers the hard way. There is a reason that the pension fixes enacted in 2009 were called “Tier V” and not “Tier II”: There had been three previous attempts to rein in the excessive cost of New York’s public-employee pensions by creating less generous pension “tiers” for newly hired employees. These reforms date back to the fiscal crisis of the 1970s, when unsustainably generous contracts with public-employee unions threatened to throw New York City into bankruptcy. Since then, though, New York’s public-worker unions have been highly successful in unwinding previously enacted pension reforms. The new Tier V is nearly identical to what Tier IV was at the time of its enactment in 1983–but Tier IV has been repeatedly, and retroactively, sweetened through increases in benefit formulas, cuts to employee contributions, and reductions in the retirement age. Similarly, by the time substantial numbers of workers actually start retiring under Tier V around 2040, this plan, too, will probably bear little resemblance to its current form.

Most of Barro's piece focuses on public choice considerations — of how state and local government institutions actually work — and thus it is the better analysis.  Here is a related piece by Eileen Norcross, closer to Barro than to Baker.


Re: rate of return assumption on pension funds, Jeremy Gold has been on top of this issue for over a decade. This paper (written in 2002) applies his insights to public financing plans, recognizing that public plans are different from private plans - but not because it's OK to use a "expected" rate of return on equities of 8%. Unfortunately the actuarial field has not moved enough to recognize his insights, so you get the kind of silliness that Baker is peddling.

I think that any arguments in the public pension debate should address these points. It recognizes that funding of benefits and arriving at the level of benefits are independent things.

The major reason that shortfalls exist at all was the downturn in the stock market following the collapse of the housing bubble, not inadequate contributions to pension funds.

In my house, that's what inadequate means.

To be fair, basically all of my clients (mostly municipal plans) were fully funded prior to the crash. I think the reasonable point here is that most public plans weren't being "sneaky" and deliberately keeping contributions lower than they thought were needed.

re: MVL, it has never made sense to me why that represents a remotely reasonable prediction about the performance of pension funds, given that the funds aren't actually invested in municipal debt.

(disclaimer: I'm not actually an actuary yet)

"fully funded prior to the crash"

At what assumed discount rate? Crashes are part of the overall return. Crashes and corrections happen in part because people over-estimate rates of return. Putting your retirement savings at the risk of (known optimistic) predictions of the future is oxymoronic.

One thirtieth of 30 years is about 1 year, just to pay pensions, based on projections. Projections involving pensions are often optimistic, based on unrealistic discount rates of return. Everyone I know near retirement has postponed retirement due to the stock market.

It seems certain people always come up with bizarre metrics to promote their point. It's not like I'm painting with a broad brush, I'm just looking at people that produce bizarre metrics and saying these people use bizarre metrics. I wonder if they use strange metrics in their high impact papers.

Over the long run, the good and bad years never balance out like they might hypothetically do so, because of public choice problems. Good years result in the increase in benefits, decreases in contributions, and all the other sweeteners that Barro conclusively documents. People don't object because the fund is over-funded at the time. Bad years result in taxpayers making up the difference.

It's a ratchet.

The public pensions in the worst shape are in many cases the pensions in lieu of Social Security where the local governments save taxpayer money by exempting teachers and police and fire from the low return Social Security system and invest in Wall Street which always generates 0-12% returns compared to the low to negative returns on Social Security.

In 2000, this was the proof of the wisdom of replacing Social Security with private accounts invested in Wall Street.

And the high returns in Wall Street of these pension funds in the 90s allowed both cutting contributions while increasing benefits because Wall Street provides a free lunch of high benefits with all the security of Social Security.

Now that Wall Street has failed to deliver the promised free lunch, the view is the workers need to be saddled with all the risk while taxpayers get the benefit of low wages paid to public employees obtained by the promise of free lunch pension benefits.

Before the crash, state pensions were only fully funded under the ~8% discount rates. Robert Novy-Marx and I wrote the first draft of our paper using pre-crash data and found a $2 trillion unfunded liability ( Roughly one-third of the $3 trillion unfunded public pension liability is due to the financial crisis, the rest was there before.

Also, for those who think discounting at an expected return on assets makes sense, I offer the following. Consider how this method would work in the realm of personal finance. Imagine you personally have a $10,000 repayment of a loan due in 5 years. If you applied for a second loan, any lender would ask you to record your other debts on the loan application. Does the amount of debt you report depend on whether the savings you have is invested in stock or bonds? Obviously not. But in discounting pension liabilities using expected returns on assets, governments get to apply the returns they hope to achieve against their debts for reporting purposes.

Alex Godofsky said:

"re: MVL, it has never made sense to me why that represents a remotely reasonable prediction about the performance of pension funds, given that the funds aren't actually invested in municipal debt."

Only in the world of pension actuaries does it make any sense to discount pension liabilities at the "expected return" of the assets held by the plan. In that framework, $1 of equities is worth more than $1 of US Treasuries - the longer the liability, the larger the difference!

Who pays the price if the actual return on that $1 of equities falls short of the discount rate? If the answer isn't the pensioners, then the "expected" return of the funds' assets should have no bearing on the discount rate used to calculated the current liability. The pension liabilities are guaranteed (in many states ahead of most of their other liabilities) so they should be discounted at rates in line with assets with guaranteed returns.

I am a pension actuary, working with private plans, but the issues are similar. To determine the present value of pension liabilities, you need to discount the expected future cash flows of the pension plan. The cash flows are based on the plan provisions, the participant data and assumptions about future events (e.g., mortality, salary increases, retirement rates). Actuaries traditionally used the expected rate of return on the plan's assets (reflecting the asset mix). As noted above, people like Jeremy Gold have argued, based on financial economics, that the determination of the present value should not be based on expected returns, but on the discount rates for similar investments, such as high-quality corporate bonds. The IRS funding rules that apply to private plans and the corresponding accounting standards accept this view. Current discount rates on these bases are roughly 5.5%-6%, which if applied to public plans, would raise the liabilities significantly, and unfunded liabilities.

What is the right answer? Clearly, the expectation for those invested in equities is a return higher than those on corporate bonds. The price is increased volatility in contributions and funding status. As some have commented a perverse effect of good returns has been the granting of additional benefits to union members (that happened in NJ around the time of the 2000 crash). In any case, if the standards that apply to private plans were applied to public plans, the funding situation would be much worse than people like Baker are claiming. And that is without considering the post retirement health benefits that are also offered, which have large present values and which are not backed by assets - they are totally unfunded.

It's possible that it would be somewhat better to ask the question "how much money do I need to set aside now so that I have X% chance of being able to pay these benefits", for some X (this is what I assume Tyler is alluding to in his talk about scenarios), and we actually do run those sorts of tests, but that still doesn't suggest that fund returns should be projected at the cost of borrowing.

Now the agencies are waking up to the risks posed by pensions, with S&P's downgrade of NJ.

One thing that I find interesting in the debate over pension funds is the large mismatch between asset and liability risk characteristics and duration. For those of us with fixed income backgrounds it can be hard to understand the strategy behind asset allocations.

We track key pension data with our software- you can see a sample at our blog: We also track actuarial assumptions, which vary widely by fund.

I stopped reading Barro's article at the first sentence of the second paragraph which is factually inaccurate. If he cannot spend thirty seconds on fact-checking, I will not read his article.

He said:

But then, in the early months of 2009, the stock market went into free fall tells me the S&P 500 (the first thing I looked up) dropped from over 900 to less than 700 at the start of 2009.

What's the problem?

I agree with Norcross that the states should use MVL discounts but from what I've gathered states weren't even funding pension plans at the current discount rate assumptions. I also believe that the media coverage has been very off base with their coverage of this topic. This is not an argument about what is better, DB or DC, it is an argument to get states to develop a process to fully fund their obligations. When people talk about the relative savings of DC they don't address the fact that most of that savings comes from the fact that employees don't fund their 401k's at a level that will adequately meet their retirement expectations. The savings has nothing to do with reduced expenses (quite the opposite actually), and everything to do with poor individual retirement planning.

Philo: congestion taxes?

But the stock market is back up, so shouldn't current figures on pension fund balances show improvements? Or is the problem that yields on bonds are still low? So maybe we need a good dose of inflation to solve the problem. Are state and local pensions typically indexed for inflation, or is that just Social Security?

Pensions are problematic because they hide the risk from the beneficiary, and there is no good solution to this since binding the future isn't really possible. Like it or not, your pension might be cut in the future if the contributions you made don't have the returns you or the administrator expected (or the fund might even end up empty before you retire). And if you fall back onto having some other party pay the shortfall you didn't anticipate, that party could well tell you to go pound sand up your ass (taxpayers and future contributors). For once, I find myself in complete agreement with Bill- 401K or IRAs are far more transparent, and put the risk explicitly where it always was anyway.

Another factor that has not been discussed is that the funding level (at least for WI and probably for other state funds) is based on actuarial or smoothed assets. The smoothed assets phase in investment losses over time with the result that they were higher than market value at the dates described. IIRC, if the MV were used instead of smoothing, then the underfunding for WI increased by roughly $9 billion which was about 10-11% of the liablities. Even with this underfunding, WI was in better shape than most states, which isn't saying much.

Fear not -- in under 2 years, we'll have a GOP President, and then Krugman will suddenly discover that Government pensions are a disaster, and he will scream about them non-stop.

I forgot to mention The PEW Center. GO TO: States / UTAH (on map) / Trillion Dollar Gap / Utah fact sheet. DUH?

My first comment, which did not post, was that Barro is more of anadvocate for terminating public defined benefit pension plans than he is an analyst of the causes, severity and cures of under_funded public pension plans.

You might as well just say "terminating defined benefit pension plans", since they're already practically extinct in the private sector. The idea that someone other than the retiree might contract to bear some risk, any risk, associated with retirement *and then be held to that contract* is just anathema to some people.

The idea that someone other than the retiree might contract to bear some risk, any risk, associated with retirement *and then be held to that contract* is just anathema to some people.

The last part of that is the illusion that one can be sure to hold them to that contract, and it is made even more uncertain in that, being long-term contracts, you are contracting with people who may not even be making the decisions in the future- or paying the costs. For example- the public plans are attempts to bind future taxpayers. Those future taxpayers will either pay to cover the shortfalls, or not, and if they don't, it is not immoral or unethical for them to not do so- they were not the ones who made the contracts. Employees need to stop taking promises from people who won't be paying them off. Show me the money, don't show me a promise of money paid by someone else 20 years from now.

I keep looking for meaningful numbers in all these pension op-ed, and not seeing them. I don't want to know whether pensions are 3 trillion dollars in aggregate, or .3% of state GDP; I'd like them to be compared with something meaningful-- say, with the state budget. Please, op-ed writers, make your numbers meaningful.

Baker's arguments on taking advantage of the equity premium sounded like the Social Security privatization crew's!

It makes sense that state pension plans operate under less strict funding rules than do corporations. But the particular rules seem to have encouraged more risky investment choices. And I do not understand why the state plan actuary associations stuck to the letter of the law in declaring plans adequately funded. I imagine they are under some heat at the moment.

Comments for this post are closed