How Do Lifecycle Investment Strategies Affect the Distribution of Retirement Wealth?
The typical employer provided pension has changed dramatically over the past twenty years. The fraction of private sector employers offering defined benefit (DB) plans, in which employees receive a fixed benefit determined by a formula, has declined, while there has been a sharp increase in the fraction offering defined contribution (DC) plans such as 401(k)s, in which the employer's contribution is specified but benefits depend on asset returns. Among employees with pensions, the share with DC plans rose from 40 percent in 1983 to 79 percent in 1998, while the share with DB plans fell from 87 percent in 1983 to 44 percent in 1998.
This change has shifted responsibility for managing retirement assets to employees, who must now decide how to allocate assets across broad asset classes and many different financial products. Many policy analysts and others have raised questions about whether DC plan participants are sufficiently well informed to make these decisions, which have very important consequences for retirement wealth accumulation.
In response to these concerns, some plan sponsors have begun to offer investment options that simplify investment decision-making. One such option is the "lifecycle fund," which automatically adjusts the portfolio allocation depending on the participant's age or years until retirement, typically shifting assets from stocks to bonds and cash as participants age. These funds have grown rapidly in the past decade, with $47 Billion held in such funds in 2005 and nearly 40 percent of all 401(k) plans offering them.
In "Lifecycle Asset Allocation Strategies and the Distribution of 401(k) Retirement Wealth" (NBER Working Paper 11974), James Poterba, Joshua Rauh, Steven Venti, and David Wise examine how different asset allocation strategies affect the distribution of retirement wealth. In their analysis, the authors contrast lifecycle strategies with investment rules that allocate the portfolio across assets in fixed percentages and do not vary with age.
To conduct this analysis, the authors must model the path of plan contributions over an individual's working life and combine this with information on asset returns. They start with real earnings histories for 1,400 households. They assume that households contribute nine percent of their earnings to the plan each year, starting at age 28 until retirement at age 63. Next, they randomly assign return histories to these contribution paths, based on the empirical distribution of returns from 1926 to 2002. Each household is run through this simulation 200,000 times, to generate a distribution of possible retirement wealth outcomes.
Over this period, large cap U.S. stocks had an average annual real return of 9.0 percent and a standard deviation of 20.7 percent, while long-term U.S. government bonds had an average real return on 3.2 percent and a standard deviation of 10.0 percent. Treasury Inflation Protected Securities (TIPS), which were only introduced recently, are assumed to provide a 2 percent real riskless return.
The authors examine a number of asset allocation strategies, including investing solely in one assets (TIPS, government bonds, or stock), investing in stocks and bonds in fixed proportions, and the "No Lose" strategy proposed by Martin Feldstein, which calls for making sufficient investments in TIPS to guarantee assets of at least the amount originally contributed then investing the remainder in stock. They also consider several lifecycle options, either investing a fixed percentage (110 minus age of household head) of assets in stock and the rest in TIPS or bonds or picking the stock share to mimic typical lifecycle funds on the market.
The results of this analysis are shown on Table 1; these results are for households with a high school degree but no college degree, but the relative ranking of different strategies is similar for other education groups.
Among those strategies that focus exclusively on one asset, investing all assets in stocks yields a much higher average return, $812,000 vs. $162,600 for TIPS or $192,700 for bonds. However, one percent of the time, the all stock portfolio returns just $12,800 or less, while the TIPS strategy is guaranteed to return $162,600. In the four lifecycle funds, average returns fall between the all stock and the all bonds or all TIPS values, varying between $303,600 and $438,200. The first percentile values are much higher than in the all stock case, varying between $37,300 and $64,300. The lifecycle funds achieve very similar results to the fixed stocks and bonds portfolio, both in average and 1st percentile return. Finally, the "no lose" plan offers a similar average return to the lifecycle funds but much higher 1st percentile return.
Noting that some analysts question the assumption that future stock returns will be as high as they have been in the past, the authors repeat their analysis lowering the average stock return by 3 percent. Naturally, the average return on the all stock portfolio falls, to $404,800, and the 1st percentile return also falls to $7,300. As the authors note, "this emphasizes the risk associated with holding stocks: a very small chance of a very poor outcome." The average return for all strategies holding stocks fall, but the "no lose" strategy falls more than the lifecycle funds because it has relatively more stock exposure.
Finally, the authors incorporate these results into an expected utility model, to see which portfolio is preferred by households when you incorporate their aversion to risk. They find that at modest levels of risk aversion, the all stock portfolio is preferred. However, the "no lose" and lifecycle plans become more attractive for investors with high levels of risk aversion and when the expected return on stocks is reduced relative to the expected return on bonds.
The authors gratefully acknowledge support from the National Institute on Aging (grants P01 AG05842 and P30 AG12810)