Government Policy on Distribution Methods for Assets in Individual Accounts for Retirees

As the baby boom generation begins to retire, fewer and fewer private-sector workers have traditional defined benefit pension plans, which usually pay lifetime annuity benefits. Instead, they have accumulated considerable assets in 401(k) plans and individual retirement accounts (IRAs) that have no particular method of payout. Federal government policy, which has regulated defined benefit plans heavily and mandated plan designs for distributions, has tread more lightly on defined contribution plans because of their historical secondary nature. Questions are now arising about whether this stance needs to change and, if so, how. At the same time, discussions have heated up in state governments about replacing underfunded pension plans for government workers with defined contribution plans and about providing uncovered private-sector workers with individual account plans. Hence, the policy and design issue of how most retired households can get orderly lifetime distributions from their current or potential future accounts must be faced soon and squarely.

This paper examines the two basic methods put forth in professional literature and practice as opposing solutions to this problem facing retired households. The first is a simple insurance contract that has been around for centuries: the immediate life annuity. In exchange for a premium paid to the issuing insurer, it makes payments periodically (sometimes annually, usually monthly) at a level fixed at issuance for insured’s lifetime. The contract has no liquidity and little or no bequest value. The second is a simple, easy-to-explain, and popular rule that many financial advisors employ and financial companies recommend. A fixed percentage, often 4 percent, of an invested portfolio is distributed initially, and its dollar value is adjusted every year by the actual inflation rate. This approach gives the retiree complete liquidity and allows the use of the account as a bequest, but also comes with the possibility of running out of spending resources if returns are poor or the participant has a long life. Moreover, the inflation-adjusted income flows, at least initially and for several years thereafter, will likely be lower than those of a life annuity because of the latter’s “mortality credit,” the sharing of mortality gains among survivors in the annuity purchase pool. There are theoretically more complex strategies and products on both sides of this debate, but both practically and for policy purposes, an empirical investigation into the basic choices is an excellent starting point for discussion.

I calculate and compare the income levels and risks these two methods produce over retirees’ lifetimes using a historical simulation of asset returns, interest rates, and inflation. I also consult more recent data on the pricing of immediate life annuities by several insurance companies. The empirical investigation of risk and return properties of single-premium immediate annuities modeled using historical data is a unique and original contribution here. This comparison examines the characteristics of the methods and serves as an empirical basis for discussions about retirement-income policy issues, such as projected-income illustration requirements, default annuitization in 401(k) plans, minimum distribution rules for individual accounts, and retirement-plan design by sponsors. This analysis will also be useful, in future work, as a basis for understanding, explaining, and designing more sophisticated distribution methods that plan sponsors and financial organizations can use to assist retired households, as well as the government policies that affect those products and strategies.

The life annuity is indeed an effective instrument for distributing retirement assets to produce lifetime income; it functions generally somewhat better than the withdrawal rules in widespread use. Because the life annuity is subject to inflation risk and is illiquid, and because household needs and preferences are so diverse and critical, the governmental stance toward this issue should be one of mild encouragement of and education about life annuities. Policies discouraging creative strategy and plan design using annuities should be abandoned. When designing retirement plans for government employees, if defined benefit plans are to be replaced by defined contribution plans and individual accounts, the position in favor of the individual income annuities should be stronger still—as a default selection. These considerations would also be relevant in designing a system of individual accounts in a reformed Social Security program.

Literature Review

There are large professional literatures on retirement income, in particular on withdrawal rules of thumb by financial planners and on life income annuities by economists. Those literatures, because of their diverse authorships, do not often talk to one another, but the summary review here undertakes such a conversation, examining implicit assumptions and key premises and results.

Withdrawal Rules

William P. Bengen, a solo practitioner of financial planning and investment management, published an influential paper in his profession’s primary journal (Bengen (1994)) giving a solution to the retirement-income problem. In a simple empirical analysis, he showed how a 4 percent initial annual withdrawal rate from the portfolio, subsequently increased by the rate of inflation (or decreased by the rate of deflation), could be sustained for more than 30 years from an investment portfolio evenly and consistently allocated between stocks and bonds (50/50). Thirty years was considered the most relevant horizon because the life expectancy of most retiring households, even with some years added for conservatism, did not extend past this horizon unless a particularly early retirement age was chosen. (In fact, at age 62, the probability of a single man surviving to age 92 is 12 percent, and the probability of a single woman surviving to age 92 is 23 percent; for a mixed-gender couple, both age 62, the probability of at least one surviving to age 92 is 33 percent. At age 65, the comparable probabilities—the chances of surviving 30 years to age 95—are 6 percent, 13 percent, and 18 percent, respectively. At age 70, the comparable probabilities of survival to age 100 are 1 percent, 4 percent, and 5 percent, respectively. So the focus on a 30-year horizon, although standard, bears some risk for most retirees, especially younger ones, and researchers should examine longer horizons when possible.)

Continue reading