Bryan Caplan  

Past Performance and Probability

From Religion to Real Estate... Class Consciousness...

Half-Sigma has an interesting comment on my "Four Bad Role Models."

Merely looking at past performance of stocks is not a sufficient basis for saying that stocks will outperform Social Security. One of the basic mantras of investing is that past performance is no guarantee of future results.

In response, I'd say that, like most mantras, this one is pretty silly. After all, in the real world, nothing guarantees anything. All we've got is probability. And the historic performance of stocks and bonds makes it extremely probable that they will beat Social Security.

But how is this different from those lame mutual fund adds bragging about how well they did last year? The answer is that on average, such funds really don't do particularly well in subsequent years. So the right warning would not be "Past performance is no guarantee of future results," but "Past performance does not increase the expected value of future results."

In fact, the literature on mean reversion concludes that past performance negatively predicts future results. So when you see those lame ads, you arguably learn where not to invest!

Comments and Sharing

COMMENTS (12 to date)
jaimito writes:


Excellent note.

Now, taking the historical performance of stocks as an unit or a cycle, what is the probability that it will outperform (once more) Social Security?


John Ford writes:

Certainly the probabilistic phenomenon of regression to the mean could be in play in this setting. But could the notion that a rise in stock price is a negative predictor of future performance also reflect a higher P/E ratio?

Isn't this why, in speaking about the yin and yang of investing, greed and fear, Warren Buffett said "When the market gets scared I get greedy and when the market gets greedy, I get scared"

Keith Singer writes:

Given that so many companies go bankrupt routinely, are there are any studies that have computed the cumulative stock returns after allowing for the failure rate of joint-stock companies. Simply finding the stock returns on the basis of companies that have survived perenially would create an extreme bias in the calculation of common stock growth rates.

Bill Newman writes:

As far as I can see, everything written above here is as though the long-term outperformance of the stock market is just some historical pattern that people have happened to notice that might well go away in the future, like Dogs of the Dow or different returns on different exchanges or whatever. But the long-term outperformance of the stock market isn't just some pattern that people have happened to notice; there are good underlying reasons to expect it (all other things being equal, in a market where there's enough smart money, etc.) because stocks have higher volatility. The outperformance could still go away if those reasons turn out to be less important than people think they are, but the basic return/volatility relationship looks pretty natural and stable to me, and it has been holding up well for a long time since becoming common knowledge: >40 years since it was put in fairly sophisticated form in CAPM, e.g. (Or SS might catch up if it becomes politically acceptable both to put SS money into actual investments, and to allow those investments to be volatile, but I'm not gonna hold my breath.)

Dewey Munson writes:

Have any of you really done the numbers?

I lived through the entire period of SS. I have been retired for 18 years. Even correcting for the annual value of FICA I have been receiving a 25% return on "investment". Not only that, it is partially inflation protected,

Ridiculous but true. The error lies in the fact that although the benefit was initially valid there is no debiting provision for the fact that I am now consuming my deferred production credits.

Anyway, everything being written about SS seemingly ignores the fact that 50% of filers earn less than $25000 and benefits rise with earnings. Certainly not "progressive" .

Lawrance George Lux writes:

The Market for Stocks and Bonds has grown substantially since the start of the Reagan administration (actually, with the introduction of the 401k system in the 1970s). This growth has been both in invested Capital, and in Participants. It historically takes a deep Recession or Depression to clear these Markets of Part-Time Players, who use Speculation on Short-term Market changes, rather than Long-term growth trends.

The Stock Market is again going to take a long bath, ridding itself of Speculative players. It is not a question of 'If' but 'When'. The only question that remains is: How long will it take for Market forces to overturn the Government underwriting of the 401k system? Maybe years, maybe decades, maybe tomorrow; but it will come. lgl

Different River writes:

Bryan, I think you are slightly misunderstanding mean reversion. Mean reversion doesn't imply that "past performance negatively predicts future results," but rather that "outliers are not likely to be repeated."

In other words, when a mutual fund has extraordinary performance (good or bad) in one particular year, that year is likely to be an extreme value from that fund's own probability distribution. The mean of that distribution is likely to be where it always was, which is closer to the mean for all years than that extraordinary year is. The next year's performance is a still a draw from that same distribution, so it's likely to be closer to the mean than the extraordinary year's.

This is quite a bit different from "past performance negatively predicts future results." It does so only when past performance is conditioned to be extraordinary, which is a way of introducing selection bias.

Boonton writes:
In response, I'd say that, like most mantras, this one is pretty silly. After all, in the real world, nothing guarantees anything. All we've got is probability. And the historic performance of stocks and bonds makes it extremely probable that they will beat Social Security.

Which is meaningless. Certainly the winning NJ loto ticket will have a return that beats any stock, bond, Social Security, hedge fund, etc. The lottery's return, however, is whatever the jackpot is. If everyone buys the winning ticket the jackpot doesn't grow, the return will drop until it is slightly negative to pay for the administrative cost of the lottery machines as well as the state's take of the pot.

Knowing that a particular company's stock had a return of 20% for the last ten years, suppose I traveled back in time and convinced President Clinton to have all Social Security surplus money used to purchase that stock. In fact, suppose I convinced President Clinton to borrow $100,000 per US citizen and use it to create a 'private account' with that company's stock in it. Would the result be that every American would now enjoy a nice fat private account that had been growing at a rate of 20% for the last ten years?

No, the return on the stock would have immediately fall and have fallen dramatically. The total returns that everyone can enjoy are fixed by the overall growth of the economy. If a piece of the economy is growing faster then other pieces must grow smaller. In other words, the 20% return cannot hold unless somebody else is getting a 1% return or even a negative return.

Randy writes:


Re; "Anyway, everything being written about SS seemingly ignores the fact that 50% of filers earn less than $25000 and benefits rise with earnings. Certainly not "progressive"."

That it a very good point. So it seems that one unintended consequence of Social Security is to lock in the class structure in old age. Bush's proposal for progessive indexing would do much to remedy this.

Boonton writes:

How would that be?

Boonton writes:

I don't see how Social Security 'locks in' class structure. Yes benefits do rise with income but not in proportion so, roughly, a person who makes $80K per year every single year of his working life will get less than four times the benefit of someone who makes $20K per year for every year of his life.

Ike Coffman writes:

I know this is an old message and may not get much traffic, but I would to play devil's advocate for a bit. I suggest that the overall market does not really go up or down based on individual company performance, but because of money flows, which, as Mr. Lux mentions above, depends on demographics and governmental policies. In other words, the market goes up when money flows in, and goes down when money flows out.

But that is a trivial statement. I suggest that the market "requires" a positive flow just to maintain price levels, and net outflows cause a disproportionate decrease in market performance. This means that money invested in the market is worth less than percieved, because once money starts flowing out the market drops like a rock.

The danger in investing social security money in the stock market is not that investments will fail to gain value, but that investments will gain too much value. At some point people will cash out, either through necessity or greed. First movers make lots of money, everyone else gets screwed.

Heres my point: prove me wrong. In two out of the last four years the market went down even thought there were positive money flows in the domestic market. In 2002, when there was a net outflow, the outflow then wasn't very large. This is not evidence, certainly, but it is disconcerting to me, and should be to you also.

Comments for this entry have been closed
Return to top