The Craft (Not Science) of Asset Allocation

On Second Thought Wednesday's headline on Ignites (a paywalled trade-industry site, no link) read, "Target-Date Funds Can't Beat 60/40 Funds: Study." I assumed the study compared the total returns of target-date funds to that of balanced funds--or possibly a customized benchmark--and found that target-date funds had trailed. Before reading further, I began to prepare my rebuttal.

Such research is commonly cited, and it's usually flawed. The performance gaps tend to be accidental. Inevitably, the two fund categories have different asset allocations. One has more stocks, or more foreign securities, or something else that fares relatively well, and it leads for the trailing five-year period. Then the markets reverse, and the other type shines. Its managers suddenly got smart!

(The classic case of such an analysis involved not mutual funds, but 401(k) plans. In 1999, a study circulated showing that the 401(k) accounts of employees who worked at several leading financial-services companies sharply lagged the naive asset allocation of 60% S&P 500/40% long-term bonds. True ... but that allocation gained 21% annually during the study period! The next several years, the S&P 500 sold off sharply, most mutual funds held up better, and the study was discontinued.)

However, reading the underlying paper, I am pleased to report that my initial impression was false. The study, authored Peter Forsyth and Kenneth Vetzal of the University of Waterloo, and Graham Westmacott of PWL Capital, did not compare the returns of target-date funds to balanced funds--or indeed any version of mutual fund to any other version. Instead, "Target Wealth: The Evolution of Target Date Funds" compared theoretical performances of asset allocations, using simulations generated from historical market returns.

Simulating Reality Well, that's better. There's something more than slightly suspect about implicitly assuming that past market returns, achieved at times during the Great Depression, under the gold standard, and during great inflationary bouts, will serve as an accurate guide to the future. On the other hand, inventing a model is even more suspect. To paraphrase Winston Churchill, bootstrapping historical returns is the worst form of investment research--except for all the other forms that have been tried.

(Naturally, investment topics never being as simple as they first seem, the issue is more complicated than I have let on. There are many ways to simulate historical returns. The authors try a couple of them, claiming that the second method--one that selects longer chunks of time than the customary one-month periods--is better. I can't speak to that, but they arrive at similar conclusions with either approach, so no worries.)

Thus, I'm willing to believe the results, as much as they can be believed. One reason that asset allocation is a craft is that although the outputs can be measured precisely through basic math, the inputs cannot. They must be invented, meaning that they are never fully correct, they might prove woefully wrong, and they carry unknown degrees of error. Will all that said, we must proceed in some fashion--and the authors' grounding seems as firm as any.

Not Better, Not Worse Their finding matches the Ignites headline, but more literally than I had expected. It is true that the study shows that, for an investor with a multidecade time horizon, target-date glide paths don't beat holding a simple 60/40 bond mix. At a given level of risk, they don't create higher average wealth, lower probabilities of shortfall, or an improved chance of becoming very wealthy, should the financial markets be unusually strong. However, the optimal glide path doesn't trail, either. It pretty much accomplishes the same thing; effectively, the contest is a draw.

The authors argue that because most actual target-date funds stray from the optimal glide path (as they have defined it) and carry expenses, target-date fund investors will fare worse than if they held the static 60/40 mix. That may be, but as the authors don't estimate the size of the straying effect, and funds (as opposed to simulations) come with expenses regardless of whether they follow glide paths or have fixed allocations, those points are minor. It's still a draw.

The question then becomes, if glide paths do not create better investment outcomes than static allocations, why does every target-date fund use the former? The companies that sponsor such funds, of course, will provide white papers explaining their investment rationales. But such articles typically assume the answer: They start with the premise that target-date funds follow glide paths, then justify why their firm's glide path is better than the competition's.

A Rose by Any Other Name The authors identify the answer: perception. They present the following choice:

A) An investment strategy that has a high return potential early on and a move to safety in later years.

B) An investment strategy that has the same expected return throughout.

Pretty clearly, those describe A) target-date funds and B) balanced funds, and equally clearly, A) is better. Who wouldn't wish to make the big money early, then protect those gains (with "safety" no less) once they have been achieved?

Or, investors could have this choice:

C) An investment strategy that has the greatest potential for losses early on, and a limited potential for gain in later years.

D) An investment strategy that has the same expected return throughout.

Once again, those are dressed-up portrayals of target-date and balanced funds, but this time the balanced funds seem superior. Who would want to lose heavily coming out the gate, then be hobbled as they try to recover?

Target-date funds control the debate. They have taught the entire marketplace plan sponsors, investors, the media--to regard them as option A) and not option C). Target-date sponsors could be selling their gasoline while levying a credit card surcharge, or by offering a cash discount. (Once again, the same thing, described differently.) They have convinced onlookers that it is a cash discount.

Relativity I would add a further argument. While simulations don't pause to reflect as each year passes, adjusting their expectations as the markets rise and fall, people most certainly do. An investor who profited during his fifties by holding a 60/40 fund through a stock bull market, when many target-date funds were more cautiously positioned, won't be happy if stocks subsequently crater. His ending wealth might be the same as if he had owned a target-date fund, but he will feel regret. His reference point is not his beginning investment age of 25, nor the amount of his 40-year gains, but rather his investment peak at age 60. He now feels like a loser.

There is more to be said about relative loss aversion--and next Friday, I will discuss the authors' main recommendation, that target-date funds follow a "Target Wealth" strategy. Suffice it to say that, between the hazards of estimating future market performances and the necessity to convince investors that they have made sound decisions, asset allocation is far from a hard science.

John Rekenthaler has been researching the fund industry since 1988. He is now a columnist for Morningstar.com and a member of Morningstar's investment research department. John is quick to point out that while Morningstar typically agrees with the views of the Rekenthaler Report, his views are his own.