Tuesday, November 27, 2012

The EMH Isn't Testable and That's OK - Part I

After reading this feisty exchange between Stephen Williamson and John Quiggin, I wanted to address an argument Quiggin raises against the Efficient Markets Hypothesis (EMH). I read the EMH chapter in his book, Zombie Economics, first, just to make sure I was really seeing the strongest form of his point and not some watered down blog version. While there is much in the book I disagree with, I think Quiggin attempts to grapple with his opponents’ strongest arguments. And he’s a first rate economist (top 1% according to this ranking).

What is the Efficient Markets Hypothesis?

Quiggin gives this definition, “financial markets are the best possible guide to the value of economic assets and therefore to decisions about investment and production. This requires not only that financial markets make the most efficient possible use of information, but that they are sufficiently well-developed to encompass all economically relevant sources of risk.” I honestly don’t know what that means. What is an economic asset? Are there non-economic assets? “Best possible guide for who”? Encompassing all risk sounds like a notion of market completeness, but that isn’t a requirement of market efficiency. Market completeness is just the notion that any risk I have can be insured. This is obviously not true; for instance, I think I'll get a PhD, but I might not. I could flunk out. I'd love to buy insurance against that possibility, but I can't. Does that mean IBM’s stock isn’t fairly priced?

I’m going to use a simple definition given by Eugene Fama in this podcast and seminal paper, and since Fama is one of the major foils in the book, that seems especially appropriate. EMH says, “prices reflect all available information.” This obviously leads to the question, what is the “available information”? Fama broke the EMH into three forms, in order to try to encompass the types of tests people were doing at the time (he regrets this, btw). The forms are weak – the relevant information is past prices, semi-strong – the relevant information is all publically available data (financial statements, earnings announcements…), or strong all public and private data. Quiggin is mostly talking about the semi-strong form, which is standard.

We just have to cover the “prices reflect” part and this is really the crux of the issue. Quiggin argues that EMH implies that prices generated by markets are “right” (scare quotes in the original). The only distinction Fama would make is that they are the best guess, but not necessarily correct. Otherwise they would agree on this statement by Quiggin, “the value of an asset is determined by the flow of income it generates over the period for which it is held and its disposal value. This stream of payments can be converted into a current value by a discounting procedure: [at the] “right” discount rate.” I like this statement. To know what the price should be, we need to know its future cash flows and we need to know what rate to discount the flows.

Here is the crux of the issue. In order to know if the market prices are right we need some theory to tell us what the cash flows will be and more importantly, how to discount them. But how do we know if our theory is correct? In order to test our theory, we have to assume markets are efficient and see if our theory matches market prices. This is called the joint hypothesis problem and has been taught in finance courses for a long time.

This problem really should have been central to the chapter (as it is in Finance courses), because every one of the reasons Quiggin gives in refutation of the EMH is subject to the joint hypothesis problems. Stock prices are too volatile. How volatile should prices be? We need some theory. Is the theory wrong or EMH? Quiggin places great weight on the recent financial crisis, and after every market crash there is always lots of clamoring against EMH. But what theory says markets can’t crash?

Quiggin argues that EMH has been redefined in response to criticism to make it unfalsifiable. As such it isn’t science, but a new sham hoisted on you by the finance community. In Part II, I’ll argue at the very least, it’s an old sham :-). And that maybe there is some hope.


  1. You make a good point with "How volatile should prices be?" Is there a better example of "how volatile prices are" than... how volatile prices ARE? Does he propose a replacement for the public ownership of corporations through stock? Is there, ostensibly, a better solution in his opinion?

    I really want to read this material now. Hope I have some extra time in the next few days.

  2. Thanks for the comment.

    Definitely, check out the book. He's a smart guy, and it's a good critique. But alas, I always end up underwhelmed, when it comes to what should we do instead? He definitely doesn't try to offer a replacement or a solution, but he does address at the end of the chapter how we should move forward.

    He offers policy recommendations. They aren't so radical as no stock. Mostly classic, social democrat type stuff. He thinks ideas like the EMH implicitly have political statements, which is a premise I don't bye. But for example, he says the gov't should try to funnel money out of bubbles (ie. limit housing credit when prices are "too high").

    He offers some ideas for better theory. He likes behavioral finance. He wants more study on bubbles and the limits to arbitrage. He quotes Robert Shiller approvingly (who is well known for bubble research), so presumably more of that.

    Maybe it is too much to ask for a solution. I think he's trying to give his best guess for where the answers are.

  3. " I think I’d love to buy insurance against the possibility that I don’t [something missing here, I think, and I'll add for concreteness "keep my job"] Does that mean IBM’s stock isn’t fairly priced?"

    In this case, the answer is yes as long as equity returns are correlated with uninsurable employment risk. It doesn't mean that IBM is unfairly priced with respect to other similar stocks, but that the rate of return to equity is too high - this is the equity premium puzzle.

  4. Hi John,

    Thanks for your comment! This will surely lead to a post.

    Respectfully, I disagree. The equity premium puzzle is a macro puzzle. It says the RBC model makes predictions about what the equity premium should be and these predictions don't match the historical premiums that have actually been observed. If markets are inefficient, there is no puzzle. Perhaps, its the market that's wrong. Who can say otherwise? This is the joint hypothesis problem.

    Here is Fama on the equity premium puzzle:


    "Has the equity premium puzzle gone away?

    EFF: There never was one. The "puzzle" comes out of a simplified economic model that says the average spread of the equity market return over the t-bill return has been too high, given the risk of equities. It is easy to show that this argument is silly. Thus, the returns from equity investing are quite risky. As a result, if the high average stock return of the past is the true long-term expected return, the high volatility of stock returns nevertheless means that getting a positive equity premium (of any size) is highly likely only for holding periods of 35 years (an investment lifetime) or more. Given this result, the historical equity premium does not seem too high.

    The simplified model that produces the equity premium puzzle, says that the premium (the difference between the expected returns on stocks and bills) should be about 1% per year (or even less). If the premium were this small, the required holding period to be relatively sure of getting a positive premium would be about 1600 years. Who would be willing to hold equity on these terms?"

  5. I should have said GE instead of RBC, since Mehra and Prescott (1985) formulate the approach with Lucas (1978) and not an RBC model.

    Here's the seminal pape:

    "Historically the average return on equity has far exceeded the average return on short-term virtually default-free debt. Over the ninety-year period 1889-1978 the average real annual yield on the Standard and Poor 500 Index was seven percent, while the average yield on short-term debt was less than one percent.The question addressed in this paper is whether this large differential in average yields can be accounted for by models that abstract from transactions costs, liquidity constraints and other frictions absent in the Arrow-Debreu set-up. Our finding is that it cannot be, at least not for the class of economies considered. Our conclusion is that most likely some equilibrium model with a friction will be the one that successfully accounts for the large average equity premium."

    Translated: we have this model we like and it makes predictions about the equity premium. Those predictions fail to match the historical equity premium. We need a better model, probably one with incomplete markets.

    They could have said, "our model must be right. People are overly pessimistic and don't invest enough in stocks. The EMH must be wrong."

  6. P.S. Thanks for pointing out the error in editing (and miraculously reading through what I was trying to say).

    I've fixed it now to say:

    "This is obviously not true; for instance, I think I'll get a PhD, but I might not. I could flunk out. I'd love to buy insurance against that possibility, but I can't. Does that mean IBM’s stock isn’t fairly priced?"

  7. Here's me and Simon Grant on equity premium