In a recent blog on the Detroit bankruptcy, I noted that Detroit’s Emergency Manager, Kevyn Orr, may have inflated pension liabilities by lowering the assumed return on pension fund assets. Because most of the cost of pension benefits comes from investment earnings, this can double or even triple the cost of these benefits (Orr quintupled the cost, which suggests other factors are involved).

Public pension fund actuaries use an expected rate of return rooted in historical experience—in practice, usually slightly lower than realized returns over the long run. Some financial economists are critical of this practice, and argue that, although expected returns on risky assets take into account the possibility of losses due to default risk and the like, they don’t factor in risk aversion—the fact that most investors prefer guaranteed returns over volatile ones even if the average expected return is the same in both cases. In the critics’ view, this means contributions to fund future pension benefits should be based on yields on “risk-free” government bonds in order to avoid shifting risk from current to future generations of taxpayers. (A variation of this argument, which seems contradicted by present circumstances, is that pension benefits are guaranteed and should therefore be discounted using a guaranteed rate of return.)

Though this seems to be an issue that pits actuaries against economists based on a different understanding of risk, this is an oversimplification. Both sides agree that the expected return on diversified pension fund assets is higher than the risk-free rate, and that the difference represents a risk premium that compensates investors for taking on risk. The issue, rather, is when is it appropriate to use the risk-free rate versus expected returns in calculations? Many pension fund critics, who claim to be basing their arguments on financial economics, engage in a bait-and-switch tactic, using the risk-free rate in contexts in which the expected rate is understood.

While actuarial projections don’t take into account taxpayers’ risk aversion, they are unbiased, at least in theory. If the projections are off, they’re as likely to be high as to be low. If actuaries instead used the risk-free rate for the purpose of projecting budget outlays necessary to fund future pension benefits, the projections would be biased because the assumed rate of return would be lower than the expected return on pension fund assets. As a result, current contributions would double or triple while future contributions would be much lower than projected.

Critics of current pension fund practices rarely acknowledge the practical implications of basing contributions on an assumed rate of return that is lower than the expected return on pension fund assets. Clearly, they hope most people will interpret their high cost estimates as unbiased projections of budget outlays, as opposed to estimates of the contributions that would be necessary if, counterfactually, pension funds invested only in government bonds.

If public pension fund critics are concerned about risk, why don’t they propose that pension funds actually invest only in government bonds? A few do, but most are proponents of 401(k)-style plans. Most experts agree that 401(k) participants should be more risk averse than pension funds, which have long investment horizons, are professionally managed, and are designed to pool risk. Among those caught in this bind is Robert Pozen, who argued for using a lower rate of return assumption to estimate Detroit’s pension obligations as a guest on WBUR’s On Point. As Ross Eisenbrey, Dean Baker and I have pointed out, Pozen and other members and staff of President Bush’s Commission to Strengthen Social Security assumed higher returns on investments when arguing for Social Security privatization (for a more detailed discussion of this issue, see this EPI briefing paper).