In the spring issue of the Journal of Portfolio Management, we published some of our research results related to lifecycle or target-date funds. While the results have been startling to many and expected to some, it is the implication of our findings that has raised an uproar in some quarters. The main concern emanates from a perception that we have prescribed a most perverse asset allocation principle for retirement plan participants — they should increase their exposure to the stock market with age. This notion can be attributed to a misinterpretation of our findings.
First of all, let us clearly explain where we stand in the lifecycle-fund debate. We acknowledge that the autopilot feature of target-date funds can play a very important role, particularly for the vast majority of investors who, left on their own, would do precious little in managing their plan assets. It is the design of the autopilot with which we have issue. In its current avatar, the one and only variable in the target-date-fund asset allocation is the participant's age. Other important factors are completely ignored. For example, target-date funds do not care about market conditions, past or present. They choose to be oblivious to whether the plan investor's accumulation goals have been achieved or not. This is where, we think, the problem lies.
Now back to the JPM paper. In this particular study, we simulated pension fund outcomes for different variations of the conventional lifecycle strategy and compared them with corresponding contrarian strategies that reversed the direction of lifecycle switching, i.e., the investor initially invested in bonds and cash and gradually shifted to stocks. The simulated return paths for different asset classes (equities, bonds and cash) were created by drawing randomly from empirical distribution of past returns for those asset classes in the U.S. market. Our data spanned a period from 1900 to 2004 and included several good and bad market events. By conducting 10,000 simulation trials, we generated the distribution of wealth outcomes at retirement under both sets of strategies.
The simulation results showed that if one ignores the extremely poor (the worst 10%) outcomes, the contrarian results were far better than their lifecycle counterparts. Even when we compared the extremely poor outcomes, lifecycle strategies outperformed only by a very small margin. We argued that this downside protection was insignificant and offered little compensation for forgoing the upside potential of maintaining a relatively higher allocation to equities in the years leading up to retirement.
It is important to draw the reader's attention to an important methodological issue related to our study. Our simulations were based on the assumption that asset class returns are independently and identically distributed over time. Had we assumed returns are not independent and stock returns are mean reverting (and bond returns are mean averting) over long horizons — as argued by Jeremy Siegel, professor of finance at the Wharton School, University of Pennsylvania, and several others in the past — the case for switching from equities to fixed income over time would have been even weaker. The assumption of identical distribution of returns ensured that we did not arbitrarily impose any particular view about different return regimes (or regime changes) over the last 105 years as the investors would have little clue ex ante about what or how many regimes they would encounter during their career span.
Since results of our study did not show the lifecycle strategy in a favorable light, some commentators have been quick to assume that we are rooting for the contrarian strategy it was competing against. This is not correct. The paper states: “It is important to emphasize that we are clearly not suggesting that a retirement plan participant should follow any of the contrarian asset allocation strategies to allocate plan assets.” Actually, our study was not about picking a winning strategy but about finding whether portfolio size should be an important consideration in long-horizon allocation decisions. The pitting of two sets of strategies, lifecycle and contrarian, whose asset allocation over time exactly contrast each other, gave us important insight in answering this question.
Here is a summary of what we found. In the initial 20 to 25 years, the lifecycle and contrarian portfolios resulted in very similar wealth outcomes. But in the final 10 to 15 years to retirement, the contrarian portfolios really powered ahead of the lifecycle portfolios. This is due to what we call the portfolio size effect. In the early stages of their career, people generally have less income and therefore contributions going into the retirement account are typically very small. Whether their portfolio is tilted toward equities or fixed income at this stage does not make much of a difference on accumulated wealth. However, retirement portfolios grow in size in the later career stages because of the larger size of contributions resulting from higher earnings and also due to continuous accumulation and reinvestment of returns from past years. This growing size enormously escalates the impact of allocation strategies on the accumulated wealth. In other words, the importance of allocation increases manifold in the later years. This is the most important result of our paper but has remained lost to a few observers who are firmly focused on picking a winner between two incredibly naive strategies.
Because we found the terminal wealth in retirement accounts to be very sensitive to the asset allocation strategies pursued in the final 15 or 20 years, the question naturally arises whether there is an optimal strategy for this critical phase. The proponents of target-date funds apparently think there is one: moving away from equities to fixed income all the way. In view of the results obtained in our study, we are not so sure. The recent global financial crisis illustrates an important point of objection. Target-date fund investors who had the misfortune to have a birthday that automatically triggered the shift to a much more conservative investment portfolio after the market had fallen heavily had effectively sold out a large proportion of their equity holdings at precisely the wrong time.
By crystallizing the losses they were never allowed the opportunity to benefit from a subsequent market upturn because of the irreversible nature of the target-date fund glidepath.
Our most recent research — which was touched upon in our submission to the DOL and SEC joint hearing in June — puts forward an alternative dynamic model for target-date funds. This strategy acknowledges the importance of reducing portfolio volatility in the final years before retirement. But, unlike current target-date funds, it is not unconditionally tied to the glidepath. The strategy takes into account to what extent the investors' accumulation goals have been realized as they enter into the critical 10 to 20 years before retirement. If the investors are ahead of their accumulation target at any point of time during this phase, the portfolio is gradually tilted toward less volatile assets. However, if the portfolio has underperformed relative to the accumulation target, such asset switching is stalled and in some cases the portfolio is tilted toward equities. Therefore, unlike current target-date funds, the asset switching is neither automatic nor unidirectional.
Prospect theory, developed by 2002 Nobel economics laureate Daniel Kahneman and the late Amos Tversky, shows that decision-makers evaluate alternatives relative to a reference point. In our research, this reference point corresponds to a target return or accumulation and investors view risk in terms of underperforming that target (loss aversion). In this case, we find the dynamic strategy to be superior to conventional target-date funds as well as extremely aggressive strategies like 100% equities.
The strategy we propose makes intuitive sense as it aims to strike a balance between the investors' accumulation goals and the need to reduce portfolio volatility near retirement. It does not completely sacrifice one for the other. There is really no need to.
Anup K. Basu is senior lecturer and finance subject area coordinator, School of Economics and Finance, Queensland University of Technology. Michael Drew is professor of finance, discipline head of finance and financial planning, deputy head (finance) of the department of accounting, finance and economics, and head of research, sustainable enterprise research program, Griffith Business School. Both institutions are in Brisbane, Australia.