New paper: What does it cost to guarantee returns?

The Center for Retirement Research at Boston College released a new issue brief, "What does it cost to guarantee returns?" by Alicia Munnell, Alex Golub-Sass, Richard Kopcke and Anthony Well, which analyzes the cost of protecting investors against market downturns such as the one recently experienced. This is an area I've had a great deal of interest in, so here's the introduction followed by some comments:

The financial crisis has dramatically demonstrated how a collapse in equity prices can decimate retirement accounts.  The crisis highlights the fragility of existing 401(k) plans as the only supplement to Social Security and has sparked proposals to reform the retirement income system.  One component of such a system could be a new tier of retirement accounts.  Given the declines in the share of earnings Social Security will replace, these accounts would bolster replacement rates for low-wage workers and increase the security of middle- and upper-wage workers who increasingly rely on their 401(k) plans to supplement Social Security.  However, these new accounts could face the same risk of collapse in value seen over the past year in 401(k)s.  So policymakers may find some form of guaranteed return or risk sharing desirable to prevent huge variations in outcomes.  This brief explores the feasible range and the cost of the first option – guarantees...

The paper makes a solid presentation of issues regarding guarantees against market risk, and a clear distinction between the fact that while such guarantees would have been cheap retrospectively, meaning that market returns have been sufficiently high that the guarantees would rarely have needed to be accessed, prospectively such guarantees would be very expensive to provide via financial markets. Guarantees are in essence put options, which allow you to sell your portfolio for a guaranteed minimum price even if the market price of that portfolio has fallen. Put options can easily be priced using the Black-Scholes formula, and indicate that market risk guarantees can be quite expensive. So far, so good – I agree with everything in the paper.

However, I'm a little wary of the discussion of the possibility that the government could provide such guarantees more cheaply than markets could, especially since analysts on both the left and the right have proposed such guarantees without, in my opinion, fully grappling with their costs. (For instance, the pension reform proposal from Teresa Ghilarducci would guarantee investors a 3 percent real return, while the Social Security proposal from Peter Ferrara would guarantee that personal accounts would be sufficient to fund current law scheduled benefits.)

In both cases, there's an implicit assumption that government can provide guarantees more cheaply than the market can, yet no strong argument why this is the case. The CRR issue brief argues that the government might be less risk averse than private markets, because it can borrow at riskless interest rates and can spread risk across generations, and therefore would "charge" a lower rate for such guarantees.

But here's how the Congressional Budget Office analyzed these issues:

The first argument--that the government can borrow at a risk-free rate--ignores the role of stakeholders in enhancing the government's credit quality. The Treasury can borrow at a relatively low rate (by creating nominally safe securities) in part because of its sovereign power to tax. However, the authority to draw on the resources of others to ensure repayment of debt obligations does not reduce the risk that the government assumes by extending risky loans and guarantees. Rather, it is the means by which such risk is shifted to taxpayers and beneficiaries of government programs, who are, in essence, equity holders in the government's financial activities.

For example, suppose the government borrows $1,000 through the sale of Treasury securities and makes a risky loan for $1,000. In balance-sheet terms, the government has acquired a risky asset that will pay $1,000 at most and a risk-free liability of $1,000. That transaction adversely affects stakeholders because they now bear more financial risk than before the loan was made: they are liable for repayment of the government securities, independent of the performance of the loan. If the loan returns only $900, stakeholders lose $100. In fact, financing a loan with a debt issue implies that stakeholders have the equivalent of a highly leveraged, and hence very risky, ownership position in the loan. The critical implication of that example is that the government's ability to create a risk-free liability results from its sovereign authority to draw on the people's resources. That authority does not protect stakeholders from market risk, nor does it increase the value of a loan above its market value.

The second argument--that the cost of risk is lower to the government because it can spread the risk more widely--is relevant to diversifiable risk but not to market risk. It is sometimes argued that the government can spread losses more widely over the population than, say, insurance companies can by compelling participation in the risk pool. However, as noted above, market risk cannot be eliminated by diversification because it results from an aggregate change in asset values. Even if the government eliminated the diversifiable risk inherent in its lending activities, the associated market risk would remain. At best, a government guarantee could shift the market risk from one group (lenders) to another (taxpayers and other government stakeholders).

A related argument is that the government's ability to borrow and repay that borrowing with future taxes allows it to reduce market risk by spreading the risk among generations. However, borrowing does not increase total resources; rather, it redistributes existing resources from lenders to borrowers. Moreover, risk is not reduced by the government's power to print money, because financing credit losses by creating money substitutes an inflation tax for a pecuniary tax. In the end, someone must bear the consequences of unpredictable financial returns, and markets determine a price for assuming that risk.

The short story is that the government isn't an entity unto itself, such that it can bear risk or have a risk aversion significantly different from that of the population that participates in markets. Rather, the government transfers risk from some stakeholders to others; a guarantee against risk for some people is exposure to risk to others.

In a paper on pricing Social Security personal account guarantees with Kent Smetters of Wharton and Clark Burdick of SSA, we summarized the research in this way:

The consensus in the academic literature is that it is unlikely that the government has much, if any, advantage in risk sharing relative to the private market. Moreover, even if the government did have an advantage, especially between generations, it is not obvious that the optimal direction of risk shifting is from older retirees to younger workers, as implicit in a benefit guarantee. As a result, policymakers arguably should not treat risk much differently than individual investors who consider both expected outcomes and risk.

We include some additional discussion of these government vs. market issues that may be of interest.

The point here isn't to beat up on the CRR paper, which is timely, useful and well written. However, given the direction of policy proposals from both left and right and the government's general inability to price risk in any context (see this AEI forum for copious examples), I wanted to discuss this issue in depth.