I have written on this blog before about the Government Accounting Standards Board (GASB) rules that allow public pension plans to discount their liabilities using the expected return on plan assets. This is no small technical matter … discounting in this way has the effect of dramatically understating the true economic value of the unfunded liabilities of public pension liabilities. This single accounting issue may understate the magnitude of the public pension funding problem by trillions of dollars!
GASB has issued proposed new standards and are in the process of accepting comments from the public. I submitted my comments last week, and thought I would post a few excerpts here.
“As members of the GASB Board may already know, I have been critical of existing GASB guidelines for the computation of public pension liabilities. I have a paper in the May 2009 American Economic Review (co-authored with David Wilcox of the Federal Reserve Board) in which we explain the basic economic rationale for the appropriate choice of a discount rate. Economic and financial theory is very clear that the choice of a discount rate depends on the risk of the cash flows being discounted.
This is true regardless of whether those cash flows are positive or negative, and whether they are being generated by public institutions, private institutions, or individuals. There is absolutely no economic or financial basis for discounting one set of cash flows based on the risk of a completely different set of cash flows. In other words, there is no logic whatsoever for discounting pension liabilities based on the risk of the pension plan assets.
A simple analogy illustrates this point. Suppose that at 9 a.m. on January 1, I borrow $100,000 for one year from a bank at an interest rate of 5 percent. I immediately invest the full amount in a diversified portfolio of risky assets (such as stocks and bonds) that has an “expected return” of 8 percent. By the time I have completed this transaction at 9:05 a.m. on January 1, how much do I owe the bank? Naturally, I still owe them $100,000 (ignoring the 5 minutes’ worth of interest). At this point, I have not changed by net worth at all. All I have changed is my risk exposure.
Suppose, however, that I follow GASB-like accounting rules to calculate my net worth. Because the expected return on my portfolio is 8 percent, I can use this to discount the $105,000 (principal plus 5 percent interest) that I will owe in one year (December 31). By this calculation, I would now only value the future liability as $97,222.22 (=$105,000 / 1.08), making it appear as if I have created $2,777.78 of wealth out of thin air.
Of course, if I tried to tell my lender at 9:05 a.m. on January 1 – just 5 minutes after closing the loan – that they should allow me to give them $97,222.22 and cancel my $100,000 debt, they would look at me with great puzzlement! And for good reason – I owe them $100,000 now, or $105,000 in one year, regardless of what I do with the funds in the interim! The transaction just described is absurd from the financial perspective. But as illogical as it seems, this approach closely mirrors the approach that has been taken by GASB in the past and that is still embedded in the proposed new public pension accounting guidelines.
The fundamental problem is that the use of “expected returns” as a discount rate is a largely a meaningless concept unless it is also accompanied by a discussion of, and accounting for, the accompanying risk.
The most straightforward way to do this is to discount pension liabilities based on the risk characteristics of the pension cash flows. For example, for the accumulated pension obligation (ABO) for a public pension in a state with strong constitutional guarantees against the impairment of retirement benefits, it would be appropriate to use a rate close to the risk-free rate. In other cases – such as states where pension benefits can be easily changed by the legislature and where the likelihood of change is correlated with broader economic activity – a higher discount rate should be used. In either case, what must determine the discount rate is the risk profile of the liabilities and NOT the expected return on plan assets.
In the exposure draft for the new rules, GASB seeks to replace the “expected return on plan assets” discount rate with a “blended rate.” This blended rate is a combination of the expected return on plan assets for the funded portion of the liability and a muni-bond index for the unfunded portion. Unfortunately, there is no theoretically coherent rationale for the proposed approach. Indeed, though this approach could be viewed by some as a “compromise,” the result produces an even less coherent outcome than existing policy.
There are several fatal flaws to the proposed approach, including:
1. There is no clear question to which the proposed measure is the right answer. Indeed, it is difficult to think of any outcome of interest that is meaningfully described by the output of a cash flow discounting exercise that uses the blended rate as described in the proposed rules.
2. Funded status is not a sufficient measure of the risk of pension liabilities. To be sure, it may be one such factor – at least insofar as one believes that participants in underfunded pensions are more likely to experience future benefit reductions – but it is far from a sufficient statistic. Therefore, it is an insufficient basis on which to evaluate the risk of the liabilities.
3. Even if funded status were a sufficient measure of risk (which it is not), the proposed GASB rules have blended the discount rates in the wrong proportions. If funded benefits are less risky than unfunded benefits, then the funded benefits should be discounted at a lower rate, and the unfunded ones that should be discounted at a higher rate. The proposed GASB rules turn this logic on its head, and the result is inconsistent with accepted procedures for risk adjustment.
4. Even if one wished to calculated a blended rate to account for differential risk based on funding status, the “expected return on plan assets” is not the right rate to use for the “risky” portion of benefits — unless the risk of the liabilities just so happens to correspond exactly to the risk of the asset portfolio, which is highly unlikely. For example, if liabilities are discounted using the expected return on a 60/40 equity/bond portfolio, this is equivalent to saying that the distribution of benefit payments to DB pension participants is just as risky as investing in a 60/40 portfolio. I suspect that few DB plan sponsors intend for their plans to be so risky, and fewer participants believe that their public pension is intended to be so uncertain.
5. The proposed rules provide an unattractive and dangerous incentive for plan sponsors to take on more investment risk than is optimal. If plan sponsors invest in a risker portfolio, they will then be able to “justify” a higher expected return under GASB rules. Indeed when they are permitted to use a higher expected return, they can show a larger share of their liabilities as being “funded.” This, in turn, also reduces the fraction of their liabilities that will then be discounted using the muni-bond index. In short, public pensions may be tempted to invest in a riskier asset portfolio in an attempt to shrink the reported size of their unfunded liabilities.
6. There is a good conceptual argument for using state or municipality’s bond returns as a discount rate for a public entity’s pension liabilities, at least to the extent that pension obligations and bond obligations bear comparable credit risk. However, this rate should reflect the risk of that particular state or municipality, rather than some aggregate index. More generally, different states and municipalities should use different discount rates when the risks of their pension obligations differ.
Because of these and other flaws, I believe it would be a mistake to adopt GASB’s the new proposed discounting rules. Instead, GASB should adopt standards that based discount rates on the risk of the liabilities.”