Bedtime for market efficiency
Dan Davies

“An Infinite Regress. Of Dumb”: More Reflections on “Market Efficiency”

tl;dr: Richard Thaler is a genius. His arguments for the importance of behavioral finance are unanswerable.

The last time I saw Richard Thaler speak, he talked about the “Beauty Contest” game. In the “Beauty Contest” game, a bunch of people each pick a number between 0 and 100. The winner is the person whose number is closest to 2/3 of the average. And he talked about the person who, when he proposed to try the game on a group of alumni at a development function, advised him not to: it would be boring because everyone would choose zero.

The reasoning? People will think: “If people pick at random the average will be 50, and 2/3 of that is 33. But if people on average pick 33, I should pick 22. And if people on average pick 22, I should pick 15…” The unique Nash equilibrium of the game is for everybody to pick zero.

And then Richard Thaler said:

If you pick zero in the ‘Beauty Contest’ game…
You are smart enough to have solved the game for the Nash equilibrium…
You are dumb enough to think everyone else can and has solved for the Nash equilibrium…
You are dumb enough to think that nobody else will consider that some others will not have solved for the Nash equilibrium…
You are dumb enough to think that nobody else will consider that some others will consider that some others will not have solved for the Nash equilibrium…
You can see where this is going.
An infinite regress.
Of dumb.

And then he concluded:

The belief that asset prices are always rational…
I give that a gentleman’s C-. In a world where nobody gets less than a B.

(Exact Richard Thaler wording not guaranteed.)

I’ve done the Beauty Contest game a lot of times. In a population of people with MBAs, the average number is usually in the 22–33 range, with the winning number in the 15–22 range.

This is a situation where we have the fundamental value nailed, and in which everyone in the market has an MBA. Yet to trade based on your rock-solid knowledge of the fundamental efficient-markets value — 0 — is to lose. And the fundamental efficient-markets value does not tell you what the price will be.

And so our very own Dan Davies wrote:

Even with respect to some of the most liquid markets in the world (S&P500 equities) and with respect to the weakest possible version of the hypothesis (so-called “weak form” — the hypothesis that there is no information in the past history of share prices which can be used to predict the future), it doesn’t work…. Momentum effects exist and can’t be rationalised away as statistical noise; fund companies like AQR have been offering funds based on them, and generally outperforming, for ages. And when you get to anything stronger than the very-weak form versions, the performance is really quite embarrassing….
Market prices are… in some sense a weighted average of the views of a large group of well-resourced and intelligent people with an incentive to get things right. But nobody would build a theory of politics around the infallibility of opinion polls — business schools certainly wouldn’t let any other department get away with claiming that it’s definitionally impossible to improve on the current state of affairs. All that’s really left of market efficiency is a sort of woolly idea that “it’s difficult to make money in the stock market”. Which it is, but it’s pretty difficult to make money in any other way too…. Efficient markets theories have the advantage for economics academics that they provide a useful response to an often-asked question by class smart-alecs (“If you’re so smart, why aren’t you rich?”), and… provide… sensible advice for personal finance (tracker funds). But this really isn’t a reason to keep them on the syllabus.

In Dan’s view, the law of one price for financial assets is much like the law of one price for goods and services: You aren’t going to make a lot of money off of it unless you have, somehow, durable barriers to entry protecting you from entering competitors doing the same thing. This has the implication that future price changes will not be clearly and substantially predictable — but that is a very, very different thing from saying that market prices are “efficient” in being co-state variables associated with some maximization problem that properly use either all easily-known or all publicly-available or all information. What you should say, instead, is that the only market inefficiencies that you can rule out are those that could be detected and then ironed away by arbitrage at sufficient scale.

The interesting question in the sociology and history of economics is why anybody should ever have thought otherwise. The puzzle is particularly sharp in the case of Michael Jensen, whose three big messages during his career were:

  1. Markets are efficient.
  2. Investors pay a fortune to money managers who make losing trades relative to any reasonable diversified-index benchmark.
  3. CEOs are underpaid.

Ignore (3), which is part of a different debate. Consider just (1) and (2). (2) necessarily entails that investors are making huge and expensive mistakes in who they trust with their money — and thus also, presumably, making huge and expensive mistakes when they trade on their own. (2) necessarily entails that money managers are also making huge and expensive trading mistakes. For that and for (1) to be true simultaneously requires that these trading and advisor-picking mistakes somehow exert no pressure on prices.

One would have imagined that such an intellectual puzzle would have attracted a great deal of intellectual energy, and that there would be a large literature trying to square (2) and (1). But there is very little, other than Milton Friedman’s long-ago claim that investors and managers who make trading mistakes will tend to buy when prices are above “fundamentals”, sell when prices are below “fundamentals”, lose their money, and vanish from the marketplace. There are holes in this argument as a matter of theory. And I am aware of no demonstrations that this argument has much empirical traction.

One would think that Grossman and Stiglitz’s now long-ago demonstration that the claim that markets are “informationally efficient” is wrong in every possible world would have had more impact. It was mathy. It was insightful. It led to the interesting conclusions like: *

  • there is smart money in the market,
  • the smart money can and is outperforming right now,
  • but that smart money isn’t YOU,
  • so sit down and index.

And everyone agreed that Grossman and Stiglitz were logically and technically right. But then the financial-markets-are-efficient posse went on doing exactly what they had been doing beforehand. At most, they have added one extra wrinkle — that of saying: the extra wrinkle of also saying “Oh. And the representative agent suddenly became risk-neutral over 1998–1999. And then the representative agent then became strongly averse to equity risk between 2000–2003…”

Thaler also quoted the much-deserved Nobel Prize-winning Robert Shiller:

The inference from the unpredictability to the rationality of stock prices is the most remarkable error in the history of economic thought…

1230 words