Arnold Kling  

EMH and Bubbles

Yuval Levin on Capitalism... See Us in Concert...

John Cassidy interviews John Cochrane. Cochrane says,

I think most people mean by a "bubble" just, "Prices were high and I wish I sold yesterday." The efficient markets (hypothesis) never told you that wasn't going to happen. What efficient markets says is that prices today contain the available information about the future. Why? Because there's competition. If you think it's going to go up tomorrow, you can put your money where your mouth is, and your doing it sends (the price) up today. Efficient markets are not clairvoyant markets. People say, "nobody foresaw saw the market crash." Well, that's exactly what an efficient market is--it's one in which nobody can tell you where it's going to go. Efficient markets doesn't say markets will never crash. It certainly doesn't say markets are clairvoyant. It just says that, at that moment, there are just as many people saying its undervalued as overvalued. That certainly seems to be the case.

Thanks to Mark Thoma for the pointer. Read the whole thing. I agree with much of what Cochrane has to say (Cassidy clearly does not).

It is tempting to say, "So-and-so called it a bubble in 2003, and it crashed in 2007. That proves that markets are not efficient." But the conclusion does not follow. I would point out that ex post markets almost never appear to be efficient. Yet, ex ante they are.

Recently, we discussed on this blog various people who were quoted as arguing that there was not a housing bubble. The efficient markets hypothesis would predict that you could find such quotes. If nobody believed that high house prices made sense, then they would have not have gotten so high. For a bubble to form, there has to be some plausible evidence that would persuade many reasonable people that there is a high probability that we are not in a bubble.

I quoted myself doubting the bubble hypothesis in 2004, and a commenter found a more recent post, in 2006, where I said that I thought that there was a 20 percent chance that we were in a bubble. Actually, 20 percent is extremely high. That is my version of a strong sell signal.

I would not trust anyone who would say "With probability 1, we are in a bubble." That person has way too much confidence in their own analysis. Markets can go wrong. But nobody can be 100 percent certain that they are smarter than the market. Nobody acts as if they are 100 percent certain that they are smarter than the market. Did your favorite prognisticator make a billion dollars betting on the Case-Shiller index? Why not? Because he was not as sure then as he is sure now that he was right.

Cochrane goes on to describe examples of what he sees as market inefficiencies. They tend to be relatively small. I also think there are market inefficiencies. Most recently, I have suggested that gold investors are betting in high inflation and bond investors are betting on low inflation. To me, that suggests that there is a profit to be made by some combination of going short gold, going short Treasuries, and going long indexed Treasuries. Will it work with probability 1? No. If it works, would that completely discredit the efficient markets hypothesis? No. It's a notion I have about some market prices that I think are out of alignment. If that notion turns out to be right, then good for me (although I am not doing that trade myself--there is too high a probability that I am wrong). If that notion turns out to be wrong, then once again the markets are smarter than I am. They don't always have to be smarter than I am to efficient. Just enough to keep me from getting infinitely rich.

Comments and Sharing

COMMENTS (12 to date)
Daniel Kuehn writes:

What bothers me most about that Cochrane statement is that he says: "what efficient markets says is that prices today contain the available information about the future". That's the primary problem. They don't. What they contain is "the available perception of what the future will look like", which is very, very different from "information about the future". If Cochrane thought about it in those terms, I think he'd realize that (1.) that's a more accurate version of EMH, but (2.) that's not a very original or encouraging insight. It is precisely because Cochrane confuses "information about the future" with "perceptions of the future" that he can just write off talk about a bubble.

Joshua Macy writes:

So what would an inefficient market look like? I suppose one where there was no trading going on...

eccdogg writes:

Well there really isn't any information about the future.

There is information about the present and the past and people have models(not necessarily computer models maybe just mental tools) that they use to create predictions about the future.

The market price is the price that balances the buyers and sellers based on these predictions.

Assuming that the market is in a bubble assumes that there is some "correct" model for predicting the future and that participants in the market are not using it. If this was the case a person able to find and adopt this model (and again I use the term loosely could be just intuition) should make outsized if they were confident in it. The problem is that no one knows what the "true" model is and most attempts to find winning strategies have on average have failed.

I work in a trading environment for pretty illiquid commodities that should be the much less efficient than stocks or mortgage backed securities. Yet even in this realm finding strategies that outperform the market is VERY difficult. Everyone who believes in market enefficiency should try trading their veiws. Even if you make money it is a very humbling experience and gives great insight into how hard it is to create and informational advantage on the market.

bjk writes:

There's a difference between "bubble" and "this is not going higher." Bubbles can get bubblier. And a prediction that "a is going to be lower in the future" is not tradable if you don't know a) when or b) how it's going to get there. That's why shorting is not symmetrical with going long. It wasn't until very late in the bubble that it was possible to short housing.

mark writes:

In some respects you have a communication problem. A) Genuine miscomprehension or difference of view of how much EMH is purporting to explain. B) Deliberate exaggeration of EMH for ideological reasons by both proponents and opponents of market orientation. C) Whenever one talks about intangibles, it is extraordinarily difficult to come up with a comprehensive, bright line definition of what they are and what they are not.

But a small criticism of Cochrane's statements: to what extent is EMH prescriptive and not merely descriptive, if, as discussed, it simply means: the prices of recent transactions in X signal the views of those who transacted or considered transacting about the value of X. It is hard for me as I read his statements to find a lot of prescriptive power in that thesis.

eccdogg writes:

Wow, I really should have proofread what I wrote better.

Matt C writes:

What Joshua Macy said. What use is the EMH if it is a near-tautology?

> Did your favorite prognisticator make a billion dollars betting on the Case-Shiller index?

Many people did make money betting against the bubble, or got out of the market early and preserved their wealth. I recall reading about one guy who made billions.

pushmedia1 writes:

Matt, think of the EMH in finance as the assumption in physics that physical laws are immutable in space and time. These are prior assumptions you need to be able to test other theories. Neither assumption can be tested directly.

In finance, if you assume markets are efficient then you can test theories about asset valuation. If your test fails you can say the valuation theory was wrong or EMH is wrong.

In my opinion, deciding the EMH is false rather than the valuation theory is like finding an "anomaly" in some theory of physics and attributing it to the fact that physical laws are in fact different in different times and place. To me, its more fruitful to keep the prior assumptions intact and look at the theory instead. In other words, what looks like a market bubble is really a fault in your explicit or implicit valuation model.

Rimfax writes:

As you or another economist pointed out a year or so ago, even those who foresaw the bubble bursting could not predict when it would burst with enough accuracy to profit off of it. Several investment scenarios were described in which a bubble bear went short on a foreseen bubble and still yielded worse performance than a bull who bailed out after the bust had started. The break even point for the bubble bear appeared to be somewhere within 12 months of the beginning of the bust, which would have required a level of telepathic omniscience that has not been demonstrated to date.

Even with the bursting of the bubble, the market appears more efficient and growth producing than the risk averse approach. The memory of the crash may make it more efficient, but it may actually encourage more bubbles rather than discouraging them.

Matt C writes:

> If your test fails you can say the valuation theory was wrong or EMH is wrong.

So those people who looked at factors other than market price in 1999 and claimed that was not really worth a billion dollars, those people were actually wrong at the time, despite what happened in 2000?

> still yielded worse performance than a bull who bailed out after the bust had started

What is the strong EMH justification for pulling out of the stock market "after the bust has started"?

pushmedia1 writes:

Matt, the EMH is an assumption about markets, not individuals.

random variable writes:

The EMH teaches us a few things. First, most professional money managers don't add much value once you account for their fees. Second, if you are getting high promised returns, you must be taking some risk. There is a lot of evidence for both those predictions. Of course they are not always 100% true, but OK first approximations. Though the value adding one is looking better and better to me right now.

Third, the EMH typically requires competitive securities markets, and small transactions costs, with easy short selling and so on. It is exactly in situations that violate such conditions that we find the most empirical evidence against the EMH. So the theory doesn't only tell you things are perfect in a magic way, it also points to where to look when the predictions screw up; something useful empirically, and helpful at thinking about benchmarks.

The EMH is theory that provides useful intellectual discipline for looking at the world. No more, no less.

Anyone with any economics training also understands the big empirical problem with the EMH-testing is requires an additional hypothesis that you have the 'correct' model for required returns. Gene Fama pointed that out in the 1960s or so. It is also ironic that Fama is probably the most prolific and clever empirical person who has worked on this. Fama has documented and empirically organized most of the robust patterns in financial data over the last 40 or so years.

Notice that all the anti-EMH people are arguing that the government should be more in charge of investments. (And that logically follows). After all, what is the stimulus but the government deciding what a large chunk of money should be invested in. That's why they are fighting so hard to discredit the EMH.

If people are really as irrational as the anti-EMH types think, I wonder we they want to remove even more market discipline, and have decisions made primarily by people who are not interested in returns, but instead in getting re-elected. Is the idea that the politicians are more rational than the investors somehow? Is it that having big decisions made by people mainly worried about getting re-elected leads to better outcomes than decentralized markets making those decisions? Because that's what it must come down to.

Comments for this entry have been closed
Return to top