behavioral finance

The Efficient Roulette Hypothesis

pnrj core principles, critique of neoclassical economics, financial advice, market failure, public policy , , , , , , ,

Nov 27 JDN 2459911

The efficient market hypothesis is often stated in several different ways, and these are often treated as equivalent. There are at least three very different definitions of it that people seem to use interchangeably:

  1. Market prices are optimal and efficient.
  2. Market prices aggregate and reflect all publicly-available relevant information.
  3. Market prices are difficult or impossible to predict.

The first reading, I will call the efficiency hypothesis, because, well, it is what we would expect a phrase like “efficient market hypothesis” to mean. The ordinary meaning of those words would imply that we are asserting that market prices are in some way optimal or near-optimal, that markets get prices “right” in some sense at least the vast majority of the time.

The second reading I’ll call the information hypothesis; it implies that market prices are an information aggregation mechanism which automatically incorporates all publicly-available information. This already seems quite different from efficiency, but it seems at least tangentially related, since information aggregation could be one useful function that markets serve.

The third reading I will call the unpredictability hypothesis; it says simply that market prices are very difficult to predict, and so you can’t reasonably expect to make money by anticipating market price changes far in advance of everyone else. But as I’ll get to in more detail shortly, that doesn’t have the slightest thing to do with efficiency.

The empirical data in favor of the unpredictability hypothesis is quite overwhelming. It’s exceedingly hard to beat the market, and for most people, most of the time, the smartest way to invest is just to buy a diversified portfolio and let it sit.

The empirical data in favor of the information hypothesis is mixed, but it’s at least plausible; most prices do seem to respond to public announcements of information in ways we would expect, and prediction markets can be surprisingly accurate at forecasting the future.

The empirical data in favor of the efficiency hypothesis, on the other hand, is basically nonexistent. On the one hand this is a difficult hypothesis to test directly, since it isn’t clear what sort of benchmark we should be comparing against—so it risks being not even wrong. But if you consider basically any plausible standard one could try to set for how an efficient market would run, our actual financial markets in no way resemble it. They are erratic, jumping up and down for stupid reasons or no reason at all. They are prone to bubbles, wildly overvaluing worthless assets. They have collapsed governments and ruined millions of lives without cause. They have resulted in the highest-paying people in the world doing jobs that accomplish basically nothing of genuine value. They are, in short, a paradigmatic example of what inefficiency looks like.

Yet, we still have economists who insist that “the efficient market hypothesis” is a proven fact, because the unpredictability hypothesis is clearly correct.

I do not think this is an accident. It’s not a mistake, or an awkwardly-chosen technical term that people are misinterpreting.

This is a motte and bailey doctrine.

Motte-and-bailey was a strategy in medieval warfare. Defending an entire region is very difficult, so instead what was often done was constructing a small, highly defensible fortification—the motte—while accepting that the land surrounding it—the bailey—would not be well-defended. Most of the time, the people stayed on the bailey, where the land was fertile and it was relatively pleasant to live. But should they be attacked, they could retreat to the motte and defend themselves until the danger was defeated.

A motte-and-bailey doctrine is an analogous strategy used in argumentation. You use the same words for two different versions of an idea: The motte is a narrow, defensible core of your idea that you can provide strong evidence for, but it isn’t very strong and may not even be interesting or controversial. The bailey is a broad, expansive version of your idea that is interesting and controversial and leads to lots of significant conclusions, but can’t be well-supported by evidence.

The bailey is the efficiency hypothesis: That market prices are optimal and we are fools to try to intervene or even regulate them because the almighty Invisible Hand is superior to us.

The motte is the unpredictability hypothesis: Market prices are very hard to predict, and most people who try to make money by beating the market fail.

By referring to both of these very different ideas as “the efficient market hypothesis”, economists can act as if they are defending the bailey, and prescribe policies that deregulate financial markets on the grounds that they are so optimal and efficient; but then when pressed for evidence to support their beliefs, they can pivot to the motte, and merely show that markets are unpredictable. As long as people don’t catch on and recognize that these are two very different meanings of “the efficient market hypothesis”, then they can use the evidence for unpredictability to support their goal of deregulation.

Yet when you look closely at this argument, it collapses. Unpredictability is not evidence of efficiency; if anything, it’s the opposite. Since the world doesn’t really change on a minute-by-minute basis, an efficient system should actually be relatively predictable in the short term. If prices reflected the real value of companies, they would change only very gradually, as the fortunes of the company change as a result of real-world events. An earthquake or a discovery of a new mine would change stock prices in relevant industries; but most of the time, they’d be basically flat. The occurrence of minute-by-minute or even second-by-second changes in prices basically proves that we are not tracking any genuine changes in value.

Roulette wheels are extremely unpredictable by design—by law, even—and yet no one would accuse them of being an efficient way of allocating resources. If you bet on roulette wheels and try to beat the house, you will almost surely fail, just as you would if you try to beat the stock market—and dare I say, for much the same reasons?

So if we’re going to insist that “efficiency” just means unpredictability, rather than actual, you know, efficiency, then we should all speak of the Efficient Roulette Hypothesis. Anything we can’t predict is now automatically “efficient” and should therefore be left unregulated.

Small deviations can have large consequences.

pnrj core principles, critique of neoclassical economics, market failure , , , , , ,

Jun 26 JDN 2459787

A common rejoinder that behavioral economists get from neoclassical economists is that most people are mostly rational most of the time, so what’s the big deal? If humans are 90% rational, why worry so much about the other 10%?

Well, it turns out that small deviations from rationality can have surprisingly large consequences. Let’s consider an example.

Suppose we have a market for some asset. Without even trying to veil my ulterior motive, let’s make that asset Bitcoin. Its fundamental value is of course $0; it’s not backed by anything (not even taxes or a central bank), it has no particular uses that aren’t already better served by existing methods, and it’s not even scalable.

Now, suppose that 99% of the population rationally recognizes that the fundamental value of the asset is indeed $0. But 1% of the population doesn’t; they irrationally believe that the asset is worth $20,000. What will the price of that asset be, in equilibrium?

If you assume that the majority will prevail, it should be $0. If you did some kind of weighted average, you’d think maybe its price will be something positive but relatively small, like $200. But is this actually the price it will take on?

Consider someone who currently owns 1 unit of the asset, and recognizes that it is fundamentally worthless. What should they do? Well, if they also know that there are people out there who believe it is worth $20,000, the answer is obvious: They should sell it to those people. Indeed, they should sell it for something quite close to $20,000 if they can.

Now, suppose they don’t already own the asset, but are considering whether or not to buy it. They know it’s worthless, but they also know that there are people who will buy it for close to $20,000. Here’s the kicker: This is a reason for them to buy it at anything meaningfully less than $20,000.

Suppose, for instance, they could buy it for $10,000. Spending $10,000 to buy something you know is worthless seems like a terribly irrational thing to do. But it isn’t irrational, if you also know that somewhere out there is someone who will pay $20,000 for that same asset and you have a reasonable chance of finding that person and selling it to them.

The equilibrium outcome, then, is that the price of the asset will be almost $20,000! Even though 99% of the population recognizes that this asset is worthless, the fact that 1% of people believe it’s worth as much as a car will result in it selling at that price. Thus, even a slight deviation from a perfectly-rational population can result in a market that is radically at odds with reality.

And it gets worse! Suppose that in fact everyone knows that the asset is worthless, but most people think that there is some small portion of the population who believes the asset has value. Then, it will still be priced at that value in equilibrium, as people trade it back and forth searching in vain for the person who really wants it! (This is called the Greater Fool theory.)

That is, the price of an asset in a free market—even in a market where most people are mostly rational most of the time—will in fact be determined by the highest price anyone believes that anyone else thinks it has. And this is true of essentially any asset market—any market where people are buying something, not to use it, but to sell it to someone else.

Of course, beliefs—and particularly beliefs about beliefs—can very easily change, so that equilibrium price could move in any direction basically without warning.

Suddenly, the cycle of bubble and crash, boom and bust, doesn’t seem so surprising does it? The wonder is that prices ever become stable at all.


Then again, do they? Last I checked, the only prices that were remotely stable were for goods like apples and cars and televisions, goods that are bought and sold to be consumed. (Or national currencies managed by competent central banks, whose entire job involves doing whatever it takes to keep those prices stable.) For pretty much everything else—and certainly any purely financial asset that isn’t a national currency—prices are indeed precisely as wildly unpredictable and utterly irrational as this model would predict.

So much for the Efficient Market Hypothesis? Sadly I doubt that the people who still believe this nonsense will be convinced.