In some situations, passive investing strategies should not be an automatic choice, even for advocates of indexing, says Morningstar's John Rekenthaler.
There’s no disputing the basic argument for indexing: Dull short-term results eventually lead to sparkling long-term figures. What’s less clear, however, is if and when there are exceptions to the rule. Are there fund categories in which indexing fails? Such discussions tend to be varied—which is not a good thing because many different answers do not mean many different realities. This means, instead, that most observers are wrong.
For example, consider the recent and common claim that domestic fund managers have been badly beaten by index funds during the past three years because U.S. stocks have become more highly correlated. This statement bundles three propositions: Active domestic funds are performing particularly poorly relative to indexes; U.S. stocks are trading more as a block; and a high correlation in stocks confounds the efforts of active managers.
The third proposition would be difficult to prove, but the claim fails on the initial premise. It is true that the average diversified U.S.-stock fund has lagged the no-cost Morningstar U.S. Stock Market Index by 67 basis points per year over the trailing three years (through Sept. 30, 2011), but it leads that same index by 77 basis points annually over the trailing decade. So, by that measure, yes, funds are faring worse than usual.
However, this analysis is inadequate because the index is capitalization-weighted, while the mutual fund average is equal-weighted. This is a problem as the mutual fund average incorporates a style effect, tilting toward smaller companies.
Adjust the study so that it compares fund averages for each of the nine Morningstar Style Box categories against their nine style indexes, and the story changes. Now, the funds lag by 118 basis points for three years and 49 basis points for 10 years—not a great difference. Given that survivorship bias boosts the funds over the longer time period, the claim disappears entirely.
In this article, I’ll address two common difficulties when analyzing the performance of active versus passive management—benchmark fallacy and benchmark choice. I’ll then suggest a better way of thinking about where indexing might fail.
Case Study: Vanguard Intermediate-Term
On Aug. 11, Morningstar’s Christine Benz noticed something quite peculiar—the single-best-performing intermediate-bond fund over the trailing month, out of 1,238 funds in the category, was Vanguard Intermediate- Term Bond Index VBIIX.
Not exactly The House That Jack Built. Certainly there are performance variations, and index funds wouldn’t be expected to place in the 45th percentile each and every month. But landing in top thousandth? After one month? What in the name of Bogle had happened?
What happened is a frequent problem for judging the attractiveness of passive versus active management: The index fund invests quite differently than do the other funds in the category. Vanguard Intermediate-Term Bond Index has one of the longest durations of any fund in the category (34th on the list) and one of the highest allocations to Treasuries (15th). With bond prices rallying amid a flight to quality, the fund rode the bull. It gained 4.6% for the month, triple the category average and 50 basis points ahead of any nonindex fund.
Most intermediate-term bond funds benchmark their performance relative to the Barclays Aggregate Bond Index. However, says Morningstar’s Eric Jacobson, bond portfolio managers have been unwilling to match the index fund’s current portfolio because of “philosophy, instincts, or client pressure.” By philosophy, bond managers don’t feel as tied to indexes as do stock managers. Managers by instinct have shied away from assuming the index’s level of risk as the bond rally has shrunk yields and raised durations, making bond funds more volatile. Finally, clients seek stability and yield from their bond funds, and neither are enhanced by the index fund’s Treasury-heavy strategy.
Index-fund wanderings occur more often than one might expect. For various reasons—ranging from the idiosyncratic design of an index to market architecture (that is, one security, one sector, or one country dominating the market) and fund-manager beliefs—an index fund frequently will not serve as a neutral position for its category. This can also occur because of external market preferences—for example, if foreign governments prefer long-dated Treasuries to shorter securities and to government agencies, then they will push up the prices of long Treasuries, thereby nudging the U.S. mutual funds toward the latter assets.
The Benchmark Fallacy
This gap between the structure of indexes and the structure of funds leads to major interpretive problems. If an index fund can differ from its peers so dramatically as to be the top fund in a huge category over a 30-day period, then it can also differ from its peers so dramatically as to invalidate active versus passive comparisons. In such instances, conclusions about the effectiveness of indexing a particular asset class are mistaken.
Consider what occurred with international stock funds when they almost universally trailed the Japan-laden MSCI EAFE index during the 1980s, then reversed and almost universally outperformed that same index the following decade. First, people argued that the managers were incompetent. Then, they flipped and argued that it’s easier to add value as an active manager running an international portfolio than a domestic portfolio. They couldn’t have been correct both times—but they could have been incorrect twice.
The benchmark fallacy is distressingly common. In the 1980s, it led to flawed arguments that active investment managers were particularly poor at investing internationally, followed in the 1990s by equally flawed arguments that managers had insights internationally but were dopey domestically. Similarly, recent evaluations of the ability of domestic-stock managers have primarily been driven by the relative success of giant-cap stocks, as the S&P 500 tends to own bigger companies than even most large-cap stock managers.
This problem extends into other areas besides fund evaluations. For example, in the 1990s, a consultant published a widely cited paper showing that 401(k) plans had dramatically lagged a composite index that was made up of 60% of the S&P 500 Index and 40% of the BarCap Aggregate Bond Index. His interpretation that 401(k) investors were hopelessly incompetent was accepted without reservation by the financial media.
This conclusion was at best preliminary, at worst wrong. The consultant’s index was made up of two assets that had recently soared: blue-chip U.S. stocks and high-quality U.S. bonds. Meanwhile, most of the large 401(k) plans that made up the study were diversified per Modern Portfolio Theory and held multiple alternate asset classes—all of which had trailed the two assets that made up the consultant’s composite index.
You know the sequel. Over the next 10 years, U.S. blue-chip stocks were awful, so poor that even good performance by high-quality U.S. bonds could barely get the composite index in the black. Meanwhile, Modern Portfolio Theory paid the rent. Foreign blue chips weren’t good either, but small-company, real estate, commodity, and emerging-markets stocks were excellent. For the decade, three fourths of mutual fund categories outgained the consultant’s composite index. (Yes, these calculations account for survivorship bias among funds.) The study faded into obscurity; there was no follow-up.
The Benchmark Choice
Exacerbating the difficulty is that there are several reasonable benchmarks for most mutual fund categories. As the benchmarks can differ substantially in performance, so choosing one versus another necessarily colors the results. Over the 10 years through Dec. 31, 2010, adjusted for survivorship bias, intermediate-bond mutual fund managers have, on an asset-weighted basis, outlegged their prospectus benchmark: the BarCap Aggregate. Hey, terrific! But they have lagged the benchmark used by Morningstar to measure their category, the BarCap U.S. Govt/Credit 5–10 Year Index, by 60 basis points per year. Oh, bad! Similarly, over the same time period and using the same methodology, large-blend funds went toe-to-toe with the costless S&P 500—an ironic showing, given that the Oughts were the decade in which the alleged “failure” of active management demonstrated to many onlookers the superiority of the passive approach. However, those same funds trailed both the Russell 1000 and Wilshire 5000 indexes. Again, hit or miss?
With these examples, the benchmarks that outgained the funds (the BarCap U.S. Govt/ Credit 5–10 Year, Russell 1000, and Wilshire 5000 indexes) are probably better fits for the two categories than are the more-famous benchmarks that lagged. Nonetheless, it’s tough to argue that actively managed funds got beaten. After all, nearly everyone entering the 21st century would have agreed that the familiar S&P 500 and Lehman Aggregate indexes would have been the appropriate indexes for those categories—particularly because those indexes were the ones that had been converted into large mutual funds. Thus, it’s a fair argument to state that for the decade, the active funds won, and the passive funds lost.
The Better Path
Given the frequent mismatch of benchmark and fund attributes and the availability of multiple benchmarks, speculating about where portfolio managers might be able to add value and where they cannot looks to be a mug’s game. Yes, people engage in such exercises, but simply examining a different time period or selecting an alternate index would likely change their conclusions.
It makes more sense to avoid the question altogether. The reality is that the market conditions largely drive the conclusions. As a general rule, indexes of booming asset classes outgain funds of that category, and indexes of losing asset classes get beaten by the funds. This phenomenon is frequently attributed to managers holding cash, but the principle that passive management fares best with bull-market assets is broader than that. Few active mutual funds have as pure an exposure to a given asset class as does an index, so by definition a fund will struggle to keep pace with the index if the asset class finishes first over the time period. Conversely, it takes a specially incompetent or unfortunate fund not to beat the index of the last-place asset class.
An example may be seen in Exhibit 1. The x axis shows the relative performance of the nine sections of the Morningstar Style Box, expressed in terms of annualized performance for the trailing 10-year period through Sept. 30, 2011. Thus, a figure of 3.00 would indicate that the style was highly successful, beating the U.S. stock averages by 300 basis points per year. The y axis shows how the style index fared versus funds. For the seven categories that had positive numbers, the index outgained the average fund. With the remaining two categories, the funds were the better performers.
Three areas in which indexing might fail are in markets that have bubbles, are narrow and specialized, or are illiquid. An example of the first would be the aforementioned market-cap global indexes in the 1980s, which were distorted by the extraordinarily high prices assigned to Japanese stocks. Many country indexes fall into the second category, such as the Finnish market, long dominated by the single stock of Nokia NOK. Finally, because of infrequent trading of the underlying assets, indexes of lower-quality muni bonds tend to be theoretical rather than practical constructions.
That indexing might fail in such markets doesn’t mean that it will. There will always be market conditions and time periods that favor passive strategies, even in the areas that seem least suited to be indexed. However, in these markets indexing is no longer an automatic tactic; even avid passive-management advocates will need to pause, evaluate the specific situation, and make certain that the available index works for them.