William J. Bernstein
Selection Skill, Transactional Skill
The evidence that money managers cannot persistently earn excess returns is impressive. Burton Malkiel's famous monkey-throwing-darts-at-the-stock-page analogy resonates with anyone familiar with the data on manager performance persistence. Yes, in any given period a few simians will obtain superb results. But this is due to chance—last year's winners, in the aggregate, will have only average performance next year. (Actually, the empirical data show that the prior fund winners do ever-so-slightly better than average, due primarily to expense and momentum factors. But the margin is only a small fraction of the increased overall expense of active management.)
But what if I were to tell you that there is a manager who runs 13 funds, and that over the past 3 years every one of them, adjusting for expenses, has beaten its benchmark? The odds of flipping heads 13 times in succession is one in 8,192, and this in fact understates the odds of this occurring by chance, as frictional costs should lower the odds of a fund's gross return exceeding its benchmark to less than 50-50.
Most of you can guess whom I'm talking about: Gus Sauter, who runs Vanguard's index shop. The below figures (source: Morningstar Principia) are for the three-year period from May 1997 to April 2000, except for VISVX, VISGX, and VMCIX, which launched 6/1/98. VGSIX is a special casePrincipia does not contain its benchmark, the Morgan Stanley REIT Index, so I was stuck using the data in Vanguard's annual report, which covers 1/13/96 to 1/31/2000.
In order to make this table more clear, I've added each fund's expenses to its return to get the "Gross Return" in the fourth column, which is compared with the benchmark return in the next column to get the "Gross TE" (tracking error). The "Net TE" is the tracking error after expenses—i.e., the TE that the investor actually sees. Note that in all cases the "Gross TE" is positive. Especially amazing is that VFINX, the world's largest fund, comes within 2 basis points of making back its expenses.
Vanguard’s S&P 500 index funds provide another window onto the issue of transactional skill since there are a relatively large number of competing funds. Looking at the 34 funds with five-year track records from 1995 to 1999, the Vanguard Institutional and Index Trust 500 funds rank first and second, respectively. And this ranking is sorted on gross returns. In other words, the superiority of the Vanguard funds is due to trading strategy and not their expense advantage, which boosts their net returns relative to their peers even further.
How does he do it? Mr. Sauter must have the world's smallest ego, because he ain't telling. He’s not merely being contrary; Gus knows that revealing his strategy will spawn imitation, thus eliminating his advantage.
Dimensional Fund Advisors also has a positive TE in some of its funds, particularly its flagship U.S. 9-10 Small Company Fund, which has beaten its benchmark (the CRSP 9-10 Index) by 2% per annum since 1982. Their strategy is fairly well-known. Since they're the world's largest owner of equity in the microcap area, they have the capacity to execute market-clearing purchases of large sell-overhangs at below-market prices. But Vanguard is a bit player in all of its markets. Even VFINX owns only about 1% of the S&P 500, and Vanguard's small-cap funds do not have anywhere near the size and clout of DFA's.
As you can see below, there's a pretty good relationship between the TE and both market cap and turnover—the smaller the stocks and the higher the turnover, the more excess return is earned.
So, clearly, some sort of trading strategy is involved. One possibility is that this revolves around the "reconstitution" of the S&P indexes which occurs in January and July. But no, the annualized TEs occurring during the months of June, July, December, and January are actually less than during the rest of the year. Besides, when a stock is tossed from one index to the next, the obvious mechanism is simply to trade them in frictionless fashion by book-entry transactions from one fund to the other.
One thing is clear. Mr. Sauter has skill. Not selection skill, but transactional skill. It's a lot like figure-skating competition. Turn on the TV and you get the impression that it's all about the glamorous free-form event. But half of the performance is scored on the compulsory "school" maneuvers, which are duller than toast. Money management is no different; the glamour event is active selection, where there is no evidence of persistently superior performance. The compulsory maneuvers are what really matter, and here there does seem to be evidence of real persistence.
Undoubtedly there are active managers with transactional skill—John Montgomery and John Bogle Jr. come most easily to mind. But active security selection is such a noisy, random process that it completely obscures transactional performance, which has a much smaller scatter. So we'll never know. Only with index funds, which eliminate the noise of security selection, is it possible to detect transactional skill.
No doubt about it: there are skilled indexers. And just as clearly, as I pointed out with WEBSs in the Spring issue, there are also unskilled ones. A lot of excitement is being generated with the introduction of exchange traded funds (ETFs) by just about everybody. In a way, it is gratifying to see investors finally beginning to notice the difference between 18 bp and 8 bp of expense. But ironically, as the last bit of juice is being squeezed out of the expense lemon, skill—of a very particular and easily-measured variety—becomes for the first time critically important.
Copyright © 2000, William J. Bernstein. All rights reserved.