Friday, September 01, 2006

Efficient Markets Hypothesis

I occasionally read The Big Picture, a Blog with interesting posts about the economy and financial markets. I recently came across an article on the efficient market hypothesis titled ``Random Walk and Outperforming Fund Managers." While I agree with some statements at the beginning, for example that one should use a variety of tools for stock selection, I find myself in much disagreement about the analysis of the Efficient Market Hypothesis, EMH, and its relation to mutual fund performance and the ``random walk" hypothesis. Before attacking the EMH one should at least make sure of what the EMH says and implies, and what it does not say or imply. This is what I intend to do in this Blog, plus I will try to clarify some other misconceptions that I have read in the mentioned article and related comments.

There are three forms of the EMH:
(i) Weak form: current prices incorporate the information content of past prices. That is, it is not possible to make abnormal returns (in excess of what the market expects from the stock) by predicting future stock prices based on current and past prices alone. This is the form that is related to the ``random walk" hypothesis.
(ii) Semi-Strong Form: current prices incorporate the information content of all publicly available information. This means that it is not possible for mutual funds to outperform an index strategy.
(iii) Strong-Form: current prices incorporate the information content of all, public plus private, available information. This means that even insiders cannot make abnormal returns.

The fact that there are 3 forms and that they can hold in special situations, it does not mean that economists, even proponents of the EMH believe the hypotheses are a good representation of actual financial markets. No one believes (iii), there is debate on (ii), and almost every academic economist believes (i). There are a lot of carefully executed analyses that show that mutual funds do not statistically outperform indexing, so (ii) is not rejected. But since valuing a stock is a very hard task, collecting and processing information is costly, and there is a big element of learning involved in the process, it's hard to believe that no investor can spot a value stock and profit from it. Some may be better than others and may earn superior returns. That is why I say (ii) is debated. But once you have paid the skilled manager (value finder), there is nothing left for the average investor, that is why indexing is just as good for most people.

1. EMH does not imply that prices necessarily follow a random walk process. It implies that you cannot make abnormal returns, on a risk adjusted basis, based on past prices alone. Prices may deviate from the random walk, but these deviation are not profitable, on a risk adjusted basis. Prices are very close to a random walk, and so (i) is not rejected in the data, that is what we mean when we say that the ``random walk" is a good representation of stock prices. The word martingale sometimes replaces ``random walk."

2. Malkiel is not the father of the EMH, Eugene Fama of the University of Chicago defined it as I have done above.

3. That some managers cosistently outperformed the S&P 500 does not prove EMH is rejected. One needs more than that. Just out of luck some one is going to do it. EMH does not say no one is going to do it.

4. The Weak version is not ``just a version that is less wrong". It has a precise definition, and there is overwhelming evidence that it is not rejected. This implies that technical analysis alone is useless. But I agree with the Big Picture that it can be part of a toolbox for stock selection.

5. Behavioral economics (BE) is not one of the reasons why EMH fails, or at least nobody has proven that. BE may explain the behavior of individuals under some circumstances. But to say that they act irrationally depends on what you mean by rational. BE does not say much about aggregation among individuals and the prices that would result after aggregation of these ``irrational" choices. This especially in a world of heterogeneous investors some of whom might not (I think many) be irrational. I also want to add that if you look closely at much of the evidence for ``irrationality" and behavioral biases, most of it comes from lab experiments with subjects making choices that most observers of financial markets would find far from reality (to say the least).

6. 2 Behavioral economists shared 1 Nobel prize. This is because of advancements in decision theory. But may or may not be a signal that BE explains most movements in stock prices.

7. The Big Picture rejects the weak form EMH by looking at the NASDAQ level in March 2000. Again, the weak form says that you cannot make a better prediction than the market, based on historical prices alone, about future stock prices. Whatever was embodied in the NASDAQ in March 2000 it could not be used to make abnormal returns. Even if you thought prices where not representing fundamentals, and therefore expected the bubble to burst at some point in the future, you could not use the fact that the NASDAQ was at 5100 to say when the crash was going to come. People had been predicting a crash since 1996. Even f the price is a ``bubble", EMH can still hold. Such bubbles are called ``rational bubbles." Even if you think that the price is above fundamental, a rational investor may still buy the stock, rationally expecting that with some probability the price will be even higher.

8. While many rational economic models that try to explain stock prices are rejected in the data, it does not mean that EMH alone is rejected. Tests of economic models are joint tests of the model plus EMH. It maybe that these models are rejected because they do not correctly capture risks faced by investors, not because there are inefficiencies. I haven not seen a careful study that rejects the weak form EMH alone.

9. EMH does not say that better track records than the S&P 500 cannot exist. It says that on a risk adjusted basis, there should not be a statistically significant number of investors that can do so. To reject the hypothesis, one should show that there are enough funds (a statistically significant amount) that can do that, after adjusting for risk. EMH is perfectly consistent with the fact that, by taking more risk, you can outperform the S&P. It is also perfectly consistent with the fact that some investors, even on a risk adjusted basis, can outperform the S&P. It says though, that there are relatively few of these, say no more than 5-10%.