Investing is not supposed to be gambling, but when you give your money to a fund manager you're exposing yourself to some combination of skill and luck. That's true even for passive index funds, to say nothing of actively managed funds. But how do you know which—skill or luck—will most determine the outcome? And is there a way to ensure that skill plays at least as large a role as luck? If not, is asset management just an elaborate lottery?
It's an age-old question that Legg Mason analyst Michael Mauboussin tackles in his fourth book, The Success Equation: Untangling Skill and Luck in Business, Sports and Investing. He starts with a simple test for discerning where skill matters more than luck: If it's possible to lose at something on purpose, that something involves skill. If it's not possible, the something is a game of luck.
Since your fund manager can't normally lose on purpose, you might be tempted to conclude that he's effectively gambling with your retirement. Indeed, on the luck-versus-skill continuum, Mauboussin places investing closer to roulette than to chess. But Warren Buffett is pretty good evidence that investing outcomes are not purely functions of luck, even if Buffett can't lose entirely on purpose.
How does an investor go about finding fund managers who can be reasonably credited with skill? Two ways, says Mauboussin. You can look at the manager's track record, but reliable track records take years to amass. Better, says Mauboussin, to look at his process. If it's a sound process—picking undervalued stocks, for instance—it should succeed in the long run whatever the short-term vicissitudes of the market.
Of course, the typical shareholder is wholly unequipped to size up a fund manager's process, which is why Mauboussin and many others recommend index funds for most retail investors. For investors who don't mind wading into some technical weeds, though, Mauboussin suggests some concrete tools.
One is to look at "active share" and "tracking error" together. Active share is the portion of a portfolio that departs from the benchmark, expressed as a percentage (zero represents exact replication of the benchmark and 100 no resemblance at all). Tracking error is the precision with which the portfolio mimics a benchmark index.
There is wide variation in performance among funds with high active share (say, above 60 percent), but those with both a high active share and moderate tracking error tend to generate higher risk-adjusted returns than those with high active share and high tracking error, notes Mauboussin.
Even the view from the weeds doesn't tell you everything, of course. There are lots of ways to look at the interplay of skill and randomness, a subject that has produced a minor publishing industry in recent years exemplified by such popular books as Nassim Taleb's The Black Swan and James Surowieki's The Wisdom of Crowds.
We chatted with Mauboussin recently about some of the concepts central to those books and his—mean reversion, the "paradox of skill" (how a rise in collective skill levels makes luck more, not less, relevant), what makes for a useful statistic, and other questions. An edited transcript:
Even if we agree there is such a thing as investment skill, a typical investor would be hard-pressed to find it. So does it make sense for an average investor to be anywhere other than index funds?
If an investor is not interested in putting some effort into trying to figure out which managers potentially have differential skill, index funds make an enormous amount of sense for most people. Two things I would add, though: First, there is a test in economics called the macro-consistency test, which basically asks the question, "What would happen if everybody does something at once?" Unfortunately, index or passive investing generally fails that test. In other words, not everyone can do it at the same time.
Investing sits toward the luck side of the continuum not because investors are not skillful but because, in effect, their skills offset one another through efficient prices. The classic random-walk argument goes roughly as follows: All relevant information about stocks is reflected in the price today, therefore only new information should change stock prices, and new information, by definition, is random. That's not perfectly accurate in the real world, but as a rough sketch of what's going on, it's not a bad place to start. There has to be some active management in order to ensure that information is reflected in prices.
The second thing is, I think the evidence shows that while there's a lot of randomness in investing, there is indeed differential skill. And there are ways to think about trying to identify a priori that differential skill. For people who are not interested in getting into that, indexing makes a lot of sense; for those that are motivated, there may be some paths to trying to do that intelligently.
Can a portfolio manager move close enough to the skill side that shareholders can be confident there's at least as much skill as luck at play?
It's a super-interesting question. I think that someone like Buffett you could say quite confidently is very skillful. I think he's enjoyed a bit of good luck along the way, but that's fine. This gets to an essential question, which is, "How might you come to that conclusion?" There are two fundamental approaches to this. This first is to look at outcomes. With someone like Buffett, the track record tells you all you need to know. But for most portfolio managers, their track records are simply too short.
So the second approach, which I advocate for, is to focus on their process. That, to me, is the key to illuminating skill. For instance, I might give you $1,000 to see if you know how to play blackjack. You go off and play for the evening, then return and give me whatever you have made. We know the results are going to be largely influenced by luck. The alternative is to give you $1,000, then look over your shoulder and watch your decision-making to see if you're playing basic strategy. If you are, I know that you're going to do as well as you can over time because your process is a good process.
Do you suggest a minimum tenure needed to discern skill from luck?
I've always tended to balk at this kind of approach because it assumes a purely random process. I wouldn't give a number on that. There can be investors who have a poor process but a pretty good track record, and I wouldn't want to incrementally give them any of my dollars. And there can be people with relatively short track records who do seem to have a good process, and I'd be more comfortable giving them money.
If you think fees are too high, would performance fees be the answer? Could the fund industry run on that basis?
I think it's less true for retail, but in the institutional money-management world, there is already a fair bit of that—performance fees. What you really want to reward are people who are operating with a proper and thoughtful process every single day. Over the long haul the best processes win, but in the short term they may not. Whenever compensation gets too linked to luck, good or bad, it makes me a little bit leery.