|Contributed by: martingale |
Will a mutual fund that has been a winner for the last five years
more likely be a winner next year? Will the stock of a company that
has been rocketing up for the past few weeks continue its ascent?
Will returns in an S&P-500 index ETF over the next twenty years be roughly
what it has been on average over the last hundred? Absolutely not. The
past performance of a fund, stock, or even stock market as a whole is
no indication of the future return.
It's not just some lawyer's disclaimer that you can ignore,"Past
performance is no indication of future returns" is a fundamental
principle of investment. In fact, good past performance is more often
associated with poor future returns, for reasons that go to the very
heart of the theory of investment. In this article we'll explore what
you can learn from looking at historical data, and what you can't. Let
the buyer beware.
Here's a scam: Start up ten or twenty investment newsletters, and in each
of them publish a different prediction about where the market is headed:
"The bulls will return!", "The bear draws near!", "Market to flat-line!". At
the end of the year discontinue the newsletters that got it wrong but keep publishing the winners--with new random predictions. Start up a fresh batch to replace the losers. After a few
years you'll always have a newsletter lying around that you can point to as proof of your financial genius--the one that, by dumb luck, got it right five years running. If only investors had
subscribed they'd all be rich by now! If anybody mentions the losers just tell them you are skilled enough to know when to cut your losses, then show them a pretty chart on glossy paper depicting the fortunes they'd have made if only they'd followed your advice.
This scam depends on "survivor bias", since only the successful newsletters are still around you seem to have a good track record. Only if someone looks back at the failed publications will they realize you operate on sheer dumb luck. Welcome to the world of financial publishing, investment advice--and mutual fund management.
Mutual funds and survivor bias
Mutual fund companies always have a few funds on hand that have posted spectacular results over the past five years--returning 20-25%
even during bear markets. They tell you this is because they've hired simply the best management money can buy. What's really going on is that these returns are based on the same kind of dumb luck survivor bias as in the newsletter scam. If a fund produces a crushing loss for investors customers flee and the unprofitable, unpopular fund is wound up (usually rolled into some other fund.) Since the worst funds cease to exist after a few years the survivors all look pretty good--and a few look spectacular just as with the newsletter scam. But it's just dumb luck--study after study shows that the average fund performs worse than the market average.
The fund companies have no scruples about marketing their dumb luck as if it were insight and skill. They will draw up pretty graphs on glossy paper showing the spectacular returns you would have had if only you'd had the dumb luck to invest in their dumb luck funds at just the right moment. The unstated implication is that if you buy last year's winner you might get the same return next year. Somewhere underneath in the fine print, though, it will say "past performance is no guarantee of future returns."
Total market return and survivor bias
Many of you were aware of all this already. But were you aware that the
same thing applies to the stock market as a whole? Have you been fooled into thinking an TSX index funds will have a 10% return, because that's what the TSX has had in the past? It's the same flawed
logic. People who say this point to decades of market data, and show that in the long run it returned about 10%. But you know what? It's still just survivor bias!
The markets which are still around today are those that survived two world wars, a great depression, the revolutions of the early 1800's, and the communist revolutions of the 1930's. Many of the stock markets that existed 200 years ago simply don't exist now--those that managed to survive did well, especially those that managed,
like Canada and America, to avoid the devastation of the world wars. But how could anyone have known back then which would survive? Even many
nations that did survive, like France, met with dismal returns over a 200 year period due to the hardships encountered.
The long-term return of the Canadian and American stock market is survivorship bias. The average stock market over that time period did not post similar returns primarily because they experienced hardships not experienced in Canada or the United States. There is no way to know, going forward, which of the worlds markets--if any--will post the same kinds of returns. The past performance of the American stock market is no indication of future returns.
Does this mean historical data is useless? No. But it isn't as
simple as looking at the past performance. If you look at the past
volatility, how much the stock or fund bounced around, you
can gain some appreciation as to the risk you are taking.
In general, the more volatile a fund or stock, the higher the
expected return should be, because you get paid to take on risk.
Good past returns may mean poor future returns
There's something even more fundamental here, too: A company that
has had a major run up is likely now a safe bet. It was a startup,
but now it's a proven business; or it was a basket-case, but it's
been turned around. You get paid to take these kinds of risks--those
who invest at the darkest hour are well rewarded on the turnaround.
Those who invest once the company is safe and sound get much more
modest returns. What this means is that frequently excellent past
performance indicates poor future returns. If the company
or fund or market is not now over-valued, at they very least it has become a safe bet.
Similarly, a hundred years ago Canada and the United States were still developing economies fraught with risk and peril for investors. Who knew they'd grow into stable, mature world leaders? The effect of that transition was impressive stock market returns--but we are now mature economies, we cannot transition from developing to developed status again.Once again, past returns are no indication of the future.
The efficient market hypothesis
It gets worse. With so many stock analysts fed by so many sources of information poring over the value of every fund and stock there is, as soon as something becomes known about a security--or even suspected--people immediately act on it. If you knew the price was going to go up tommorow, you would start buying like crazy--and your very buying would drive the price up today. Everything that everyone knows or even suspects or guesses at has already been included in the current price of the stock. This is called the "efficient market hypothesis" -- the idea that information is rapidly included in a stock price. The best estimate of tommorow's price is today's price (plus interest).
What does all that mean when you're looking at the historical prices of a stock or a market? What it means is that anything that you could have learned by looking at the past performance of the stock has already been included in the price. If the past were some proof that the stock would go up in value then it already has gone up in value, that effect is "used up".
So what should you do?
What should you do then? Here's the standard advice that goes with the above analysis: Determine your risk level. Invest in a diversified basket of securities that matches that risk level--some split of stocks and short-term bonds. Either ignore the long-run historical data, or bet against it. In theory it's possible to increase your return by preferring those stocks and markets that have been falling in value or doing worse than average, the opposite of the conventional advice. But realize when you do this that this means picking out companies and markets that are in trouble--you get the higher return only by increasing the risk you take.