Monday, September 14, 2009

How Seriously Wrong John Cochrane Is

I have stated here already that I have problems with Paul Krugman's analysis of what went wrong with macro, most notably his nearly total ignoring of the analysis made by many heterodox economists. However, the inflamed reply by John Cochrane has been a joke. He presents himself as speaking for "the economics profession" rather than just a narrow group of Chicago-based or inspired economists. He is the expert, and Krugman supposedly makes trivial technical errors such as not distinguishing CAPM from Black-Scholes, etc. So, I decided to see how wise Cochrane is by going back to look at his widely praised (including even by Robert Shiller: "impressive treatise of very high quality" and "can also serve as a textbook in an advanced finance course") 2001 book published by Princeton University Press, _Asset Pricing_. On what has happened in the last few years, it is utterly useless.

In the index I went to find the following words/phrases: "fat tails" "kurtosis" "leptokurtosis." They are not there, even though pretty much all practitioners and a large literature already existed discussing how to explain the ubiquity in practically all asset markets of exactly those phenomena. He has a big fat zero to say about them, nothing.

I did find "bubbles," which he discusses briefly near the end of the book (pp. 399-402). He seems to recognize that they might exist, but his discussion is strictly in terms of "rational bubbles" and cites Peter Garber ("Tulipmania" JPE, 1989) on how they are not econometrically identifiable (and later says they do not explain empirical variations of price/dividend ratios; oh, and Garber). The model of rational bubbles he discusses supposedly must go on forever, which he then sneers at because "Infinity is a long time...The solar system will end at some point." Somehow he ignores the Blanchard-Watson 1983 paper on stochastically crashing bubbles, which indeed crash in finite time after rising at an accelerating rate to provide a risk premium for the ever-rising probability of the inevitable crash. Ironically, this last model lies at the base of the forecasting model used by econophysicist Didier Sornette and his associates, whose not-too-bad recent forecasting record on crashes is ultimately based on estimating the moment that such a process (complicated by oscillations) blows to infinity.

Oh, and Peter Garber is also wrong, see "Complex bubble persistence in closed-end country funds," by Ehsan Ahmed, Roger Koppl, me, and Mark V. White, Journal of Economic Behavior and Organization, Jan. 1997, 32(1), 19-37, although the basic insight that one can identify the fundamental by the net asset value of a closed-end fund and thus can identify increases in price of the fund above its NAV as a bubble had been made by many others prior to us. So, it is simply ridiculous of Cochrane to have ignored this point as well (he does recognize on p. 408 that some such funds trade regularly at discounts but never mentions that some have shown massive increases in price above NAV or that this clearly shows the existence of a bubble).

17 comments:

Shag from Brookline said...

This quote of Johnny's

"Infinity is a long time...The solar system will end at some point."

perhaps reflects his physics education. But how does it apply to his economic financial models? It seems that for finance executives, their personal goals are short-term and long-term could be one-year to qualify for long term capital gain treatment.

seanm said...

I not in the business of defending Coam chrane and I do not have my copy of Asset Pricing to hand, but I do remember carefully reading his discussion of CAPM to check precisely what the constraints were on the model, and noting that it explicitly avoids any assumptions about higher moments on the distribution of asset prices, so your first point is unfair (and unecessary, since Cochrane has provided more than enough amunition anyway).

Anonymous said...

This post should be titled "How Seriously Infantile I Am".

If I wrote a book on the Civil War, would you accuse me of "ignoring" the Revolutionary War? Of course not. Similarly, Cochrane's book, while discussing a bit of the econometrics, is primarily a book about financial THEORY not financial ECONOMETRICS. Cochrane's book is now the standard text for Ph.D financial THEORY courses. Courses that focus more on the ECONOMETRICS primarily use Campbell, Lo, MacKinlay's "The Econometrics of Financial Markets".

If you were to look at the two financial theory books that Cochrane's book supercedes ("Foundations of Financial Economics" by Huang and Litzenberger, and "Theory of Financial Decision Making" by Ingersoll), you will also not find the words/phrases "fat tails", "kurtosis" or "leptokurtosis". Why? Because they are books on theory, not econometrics. BTW, the authors of these two books are from MIT, Wharton and Yale; i.e. not "freshwater" economists.

Marxelo said...

"Cochrane's book is now the standard text for Ph.D financial THEORY courses."

As long as people are indeed reading this crap and making decisions based on it, more financial crises will come.

David said...

I don't see the contradiction here. Cochrane's book may indeed be only about theory. Then the problem is that the real-world issues experienced by financial markets over the past few years have been all about fat tails and bubbles, about which Cochrane's book has (apparently) little or nothing to say.

The original point of the post wasn't that the book was bad (at least not by my understanding), but that it didn't treat the relevant issues. And considering that one of the key, key questions these days is whether rational markets preclude bubbles, I'd say his slapdash coverage of the topic is some cause for real concern.

Michael Rand said...

David -

John Cochrane's point about Krugman was that he made trivial mistakes distinguishing between various financial economics models. Whether or not Cochrane's book deals with the relevant issues is beside the point. Does the following logical analysis look right to you?

Premise: Cochrane claims that Krugman made a basic mistake in confusing CAPM and Black Scholes.

Premise: Cochrane's book does not include advanced topics like leptokurtosis.

Conclusion: Cochrane cannot be trusted to identify simple mistakes like the one Krugman made.

rosserjb@jmu.edu said...

Ah, some defenders of Cochrane have shown up!

Seanm,

That he says nothing about them does not mean he allows for them. Just the opposite. His entire discussion of CAPM boils down to the mean-variance frontier, which is not satisfactory outside of a normal distribution. (BTW, a normal distribution has no skewness, but does have kurtosis equal to 3, so it is kurtosis above that which becomes the problem, "excess kurtosis," or, of course, skewness).

Anonymous,

I love it when "Anonymous" people engage in name calling, especially when their arguments are weak.

This is not a matter of theory versus econometrics. There is indeed a considerable theoretical literature, much of it quite old as about Levy distributions that date back to the 1920s (and it was an economist, Pareto, who first proposed power law distributions that many current econophysicists love and which accommodate excess kurtosis in the 1890s, this stuff ain't new). Nevertheless, you are right that Campbell (a student of Shiller's), Lo, and MacKinlay do mention these matters (and Lo has recently gone on a semi-unorthodox path pursuing "adaptive" models of finance outside of any of these textbooks).

However, even Campbell et al is instructive in how little attention gets paid to these phenomena. In their 611 pages, Campbell et all deal with the matter on a whopping six, even though in their initial (and longest discussion) in the first chapter they state (p. 16), "At short horizons, historical returns show weak evidence of skewness and strong evidence of excess kurtosis." In the following two pages they do at least define both skewness and kurtosis, and have a footnote on p. 17 that mentions Levy stable and Paretian distributions (in a footnote!).

The only other mentions are on p. 81 where students are asked to calculate skewness and kurtosis for a given data series, no further discussion there.

Then on p. 480 it is noted that time-varying variance will imply excess kurtosis and a way of measuring excessive fat-tailedness is given. This is in connection with the discussion of GARCH techniques, and on p. 488 there is a single sentence noting that there are excess kurtosis in residuals off a GARCH process, the following, "Unfortunately there is excess kurtosis in the standard residuals of GARCH models, albeit less than in the raw returns." And that is it, still ridiculously insufficient, if more than the big fat zero in Cochrane.

Two further points. You decide to lump me with saltwater as in them versus freshwater. However, in my many comments in various locations (including here), I have dismissed that entire discussion as a waste of time, which even Krugman kind of led up to. His problem was that, aside from Shiller, he never identified the people who might have been helpful in explaining all this mess, such as Minsky. I am outside this fresh vs saltwater hullabaloo. I am more in with Minsky and that gang.

Finally, the really nauseating aspect of Cochrane is that in his post he claims that Chicago "knows about fat tails," with his piece of evidence being this ridiculous account of Fama's thesis. Well, I am sure they do know about fat tails, which makes it all the more outrageous that they never talk about them in their most important writings and spout on about models ignoring them as if they tell us about reality, except of course in theses more than 40 years old. All in all, the performance of Cochrane, both in his widely used book, and more recently in his blog outbursts, has been thoroughly discreditable and indefensible.

To the others, thanks.

rosserjb@jmu.edu said...

Michael,

Your message came in while I was writing my last comment.

Let me be clear that I think Cochrane scored some points on Krugman. PK does understand the difference between the two, but did smudge it in his piece. However, was that smudging of any significance to his general argument other than to show that Krugman is sloppy? (And, in case you do not know, I have been highly critical of Krugman on a number of occasions for a number of things).

In the case of Cochrane's reply, he is way off. He goes on about a model that is supposed to represent "economics" that is a joke, that is totally irrelevant to discussing some of the most important aspects of what has happened. To make things worse, he makes these claims that the stuff the models do not explain are "known of" in Chicago, citing a more than 40 year old thesis as showing this, even though this 40 year old thesis would provide a basis for ignoring fat tails on insufficient grounds. His self-righteousness here is absurd.

rosserjb@jmu.edu said...

Michael,

Just to be clear, I did not say that Cochrane was wrong about the CAPM vs Black-Scholes claims by Cochrane. I said nothing about them. The question for you, if you want to stress this matter, is, so what?

PQuincy said...

Pro forma, IANAE: but I can ask whether a book about financial THEORY that does not seriously investigate the nature, structure, and, well, THEORY of financial bubbles can be described as a good book about such theory?

It seems unlikely that anyone who has reviewed any history (or been alive for the last 15 years) could argue that there are never any bubbles in financial assets, or that bubbles in financial assets are always based on rational expectations. Indeed, Cochrane himself, in his publication Contra Krugmanum, separates "bubbles, distinguishing them from rationally low risk premiums," so he must believe they do exist.

Frankly, some of Krugman's arguments, and almost all of Cochrane's, remind me of nothing so much as late Scholastic theology, anyway.

Michael Rand said...

The start of the post, as written, seems to imply the following:

a) Cochrane's claim about Krugman's being wrong about CAPM vs. BS is wrong because Cochrane's book stinks, therefore

b) The rest of Cochrane's argument is gibberish.

I agree that Cochrane's nitpick on Krugman's "confusion" is a "so what" moment. I'm just arguing that you don't elevate the debate by performing exactly the same "so what" analysis on Cochrane's textbook.

There's enough other material in Cochrane's response to attack.

message_in_a_bottle said...

Can anybody enlighten me in which sense economic agents can be said to act "rationally"?

Certainly not in the sense of utility maximization.
How many agents do you know that have a utility function?

Cochrane's example that sticks in mind is the following:

Governemnt spending cannot increase
demand. If government increases spending economic agents know that future taxes will be higher and will thus save money accordingly.

Does anybody believe this?

I submit that most householders do not think in these terms at all and that moreover the basic premise is false also:

Government could also print money in which case your savings will be confiscated by inflation.

message_in_a_bottle said...

I sincerely hope that Cochrane's text is not the standard PhD. level text in American Ivy League colleges on financial theory.

The mere thought of that amuses me greatly.

Barkley Rosser said...

Michael,

I do not believe that I said that Cochrane's claim about Krugman's goof was incorrect because of the uselessness of Cochrane's book in explaining anything that has happened. I basically said that Krugman's error was trivial, a so what matter. Cochrane made the remark to show that he somehow knows more than Krugman, who is not as up on what is what as Cochrane and his pals at Chicago. This is what drove me to look at what Cochrane had covered in his famous and widely used book, only to find it so useless.

gpp,

Certainly there are questions about rationality, and behaioral finance attempts to deal with those. However, the problems in Cochrane's book goes way beyond this assumption, which is made by all kinds of economists who make more sense than Cochrane.

Shag from Brookline said...

Leiter Reports: A Philosophy Blog at

http://leiterreports.typepad.com/

features "Alex Rosenberg on Cochrane and Economics" that may be of interest on this thread. Comments will be permitted, but apparently the mechanism in not yet in place. I note that Rosenberg uses "satsificers" which I can't find in my dictionary. If it is a misspelling, I can't figure out of what.

Barkley Rosser said...

Shag,

"Satisficing" is a term invented by the late Herbert Simon, father of the idea of bounded rationality. It is to be contrasted with "maximizing," as in do firms maximize profits as assumed in standard economic theory. Simon said, "no," partly because they do not know how to do so. Therefore, managers seek a satisfactory level of profits that will keep the owners happy. When they are following such an approach, they are satisficing. The term does not get used as much now as it did back in the 1960s, and I am not surprised it is not in some dictionaries, being somewhat of an economics jargon term that has become obscure.

Jack said...

Satisficing, a portmanteau is a decision-making strategy that attempts to meet criteria for adequacy, rather than to identify an optimal solution. satisficing is a behavior which attempts to achieve at least some minimum level of a particular variable, but which does not necessarily maximize its value.
antique jewelry