RE: A Challenge to Michael Emmett Brady
Updated 31 August 2013 to address a misreading of what I wrote. Revision is in boldface type.
In the course of researching Keynes, probability theory, and other matters pertaining to economic theory, I've encountered a lot of book reviews by a person named Michael Emmett Brady. A link to his profile on Amazon is here; and here is his page on the Social Sciences Research Network (SSRN). He wrote over 700 reviews, mostly in very dense and hastily-typed bursts that are often difficult to follow, mostly of books on economics—in fact, usually on Keynesian economics or attacks on Keynesian economics. He has been published in the History of Economics Review, History of Political Economy, The British Journal for the Philosophy of Science, International Studies in the Philosophy of Science, Psychological Reports, and Synthese.. His book reviews are usually mild, but he has sparred with a school of economics called Post-Keynesianism, whose most famous adherent is probably Paul Davidson.
This is not the sort of thing that attracts screaming headlines or anything, but I could never help wondering if there was something to Mr. Brady's allegations that the Post-Keynesians failed to understand Keynes's Treatise on Probability, and this affected their understanding of The General Theory.
Then someone named Philip Pilkington wrote a general rebuttal to Michael Emmett Brady here. The essence of the rebuttal is that Mr. Brady is a mysterious person who is saying everyone of repute is wrong about what John Maynard Keynes said, particularly with regard to decision theory. Or, as paraphrased by the blogger "Lord Keynes" (a pseudonym),
Post Keynesians have not understood Keynes’s Treatise on Probability properly, and in particular his idea that of relegating frequency interpretation of probability to a highly limited domain and that non-objective probabilities (in the sense of non a priori probabilities or non relative frequency probabilities) can nevertheless be mathematically represented with imprecise interval estimates (apparently called“approximation” by Keynes). This leads into a Keynesian decision making theory with some mathematics behind it superior to the Ramsey/Savage/neoclassical EUT and consistent with Daniel Ellsberg’s decision making theory.Mr. Pilkington argues that, if Mr. Brady is so confident of his judgment about decision theory, he ought to put his theories to the test at a hedge fund.
The problem with this is that, of course, Mr. Brady is not a full-time trader by profession. A common argument by some economists is that if a theory is valid, it ought to survive a test in a real time exchange (although this almost never actually happens). But in this case, Mr. Brady's gloss of Keynes pertains to the nature of inferences about the future by people with the data they have. That's the whole point of the first three chapters of Keynes's Treatise on Probability. Which means, if Mr. Brady knows he doesn't understand financial markets, he knows his foray would be a disaster.
In other words, Mr. Brady's field of expertise is on J.M. Keynes as a philosophically-minded mathematician that later turned to economics, not the trading of stocks. Any choice he would make in the milieu of the trading floor would be less valuable than it would have been if his life's work was as a market-maker or seasoned day-trader. It will no doubt be objected that this is special pleading: Mr. Brady ought to be able to explain his theory to a hedge fund manager, who could then apply the theory and use it to beat the market (according to Mr. Pilkington).
One well-known problem is, of course, that efforts to exploit different mathematical models and techniques—all questions about their logical foundations and accuracy being put aside—have long ago been stretched to the breaking point. As soon as an improved model is developed, financial experts use it to harvest rents as fast as possible. The rents never disappear, but neither do massive failures—major busts and panics. The reason is that the technique of weighted hedging of risk becomes itself an investment vehicle, and hence another data point. Data processing, mathematics, and leverage are the limiting factors, while practical application of weighted hedging is the driving force.
I don't pretend to speak for Mr. Brady, but I think this explains why his arguments on Keynes's probability theory are unrelated to the business of hedge funds. In fact, Mr. Brady has spared no opportunity to denounce speculation on the basis of virtue ethicsanyway. Saying that one's opponent's entire Weltanschauung rises or falls based on some application of one's own choosing is an argument that is unlikely to change anyone's mind. Another point of concern for Mr. Pilkington is this:
Brady appears to be making the claim that buried within Keynes’ Treatise on Probability is some sort of theory of microeconomic behavior. This has always seemed to me a bizarre assertion. Surely for someone to make decisions in line with Keynes’ work on probability they would have to have first read and understood this work. Given that such an argument is coming from a scholar that complains that not enough people read the Treatise the irony is, of course, enormous. Indeed, in many ways Brady’s argument is structured in the same way as a rational expectations argument in that the economist assumes that people act in line with his model of them — so, it implicitly assumes that such a model exists “in their heads”,which seems unlikely unless they have deeply studied and adopted said model.Something has to be said about this. One is that any theory about human behavior could be, to the same degree, accused of tacitly assuming that everyone it purports to describe "knows" the theory. By this reasoning, Freudian psychology, likewise, "assumes people have first read and understood Freud"—for surely, for someone to make decisions in line with Freud's work on psychology they would have to have first read and understood this work. This does not work as a rebuttal.
Another is that the Rational Expectations Hypothesis differs in so far as it uses the "stylized" or "schematic" assumption that people's estimates of the future will not be consistently wrong. For the record, defending rational expectations is one of the last things I would ever do, but claiming that it is so obviously wrong that no sensible person would embrace it is probably not a good idea. I would definitely agree that it is wrong, but I would also agree that debunking it is not easy.
Mr. Pilkington writes,
I don’t believe in mathematical decision-making theories, [no] matter what type of probability theory they’re based on. A useful decision-making theory has to say something about the real world. Since the most relevant place where such a theory can be applied is the financial markets then this, for me, is the ultimate test of any theory. Indeed, this is what the Efficient Market Hypothesis does; it posits a theory of decision-making and then makes actual market predictions (no one can beat the market etc.).Belief in mathematical decision-making theory probably ought to depend at least in part on what the mathematics purports to explain. If the mathematics purports to explain what people are compelled to do based on available evidence and logical necessity, then a case-by-case assessment is in order. If you really take the position that no decision theory can have any pertinence to real decision making, then I'd say you've repudiated the field of economics because you've rejected any logical inference about economic behavior.
"Since the most relevant place where such a theory can be applied is the financial markets then this, for me, is the ultimate test of any theory."The problem with this point of view is explained above: if a theory's inferences are orthogonal to questions like the frequency distribution of outliers, then no, that's an unreasonable opinion. Believe it or not, there are wiser economists who know better than claiming to predict stock market behavior off-the-cuff.
"[T]his is what the Efficient Market Hypothesis does; it posits a theory of decision-making and then makes actual market predictions"—no, the EMH posits a theory of what markets are capable of, viz., capturing all relevant information immediately after it becomes available. The EMH has been discredited, but doing so was not easy. If the EMH was a decision theory (which it is not, although it does have behavioral descriptions and assumptions), it would not appeal to common sense, because it is not reasonable to assume people are always making correct decisions about the future of volatile prices, but it is reasonable to assume the market, in the aggregate, is extremely difficult to outperform and requires something like inside information to do so. Again, for clarity, it's reasonable but mistaken.
One can finally ask, So James, in what sense is Michael Emmett Brady circulating a useful decision theory? In what sense does it have something to say about the real world? And I would have to say that Mr. Brady has always discussed its value in the realm of public policy, and I would also say that if his reading of Keynes is correct—something I am not qualified to arbitrate—then its value lies there.
Sources and Additional Reading
The blogger known as "Lord Keynes" has addressed Mr.Brady's criticisms more rigorously here: "Post Keynesians and Degrees of Uncertainty."
Ammons Scientific - Psychological Reports (for future reference)
The British Journal for the Philosophy of Science
International Studies in the Philosophy of Science
Synthese (an International Journal for Epistemology, Methodology and Philosophy of Science)