Central Banking Vol XVIII Number 1. , 1st August 2007
Nassim Nicholas Taleb, The Black Swan: The Impact of the Highly Improbable. Allen Lane, 2007.
The central thesis of this interesting book is our wish to ignore or tone down the importance and frequency of the unexpected. Nassim Nicholas Taleb writes that “Fat tails are the technical term for Black Swans” and proceeds to enlighten us about the ways in which the existence of these birds is often overlooked and, even when noticed, they are assumed to be more domesticated and less wild than they really are.
The blindness of observers often comes, first, from ignoring the fat tails in the distributions being studied and assuming that they are normal (or log normal). Secondly, fat tails (kurtosis) may not be revealed in small samples, but only become clear as more data are used. There are also the unknowable black swans, which cannot currently be seen by anyone, but will appear with experience and as new data become available.
The failures of recognition are most culpable when they belong to the first and second groups. Evidence must not be ignored because it is unwelcome or suppressed by considering only selected samples of data. This is aptly termed “data mining” and while, as I have written before, it is peculiarly the province of stockbrokers, it is by no means unknown among academic economists.* Even the third category is important. The unknowable should not be ignored.
The author rightly stresses that we cannot verify our hypotheses by tests; we can only succeed or fail to falsify them. Claims for certainty fail to recognise, as C.S. Peirce, an American philosopher remarked, that “The conclusions of science make no pretence to being more than probable.”** The prevalence of Black Swan blindness is encouraged by many incentives, which include memory, vanity and profit. We impose patterns on data to make them easier to recall and then assume that the patterns are more than just mnemonic devices. We prefer supporting to destructive evidence once we have decided,let alone publicly pronounced, our views. Wealth and income, which are often dependent on claims to past success, are threatened by doubt and readily produce fear and fury if the claims are questioned.
In a footnote of great current relevance the author writes about the problem presented by “… the time taken to compensate someone and the time one needs to be comfortable that he is not making a bet against the rare event.” The volatilities of market returns are not log normally distributed. About 90% of the time,volatility is below and returnsabove normal. The rest of the time volatility is above normal and returns are negative. It thus pays those rewarded for achieving above average returns to ignore the risks of bad times, preferably with the aid of leverage. They have a good chance of getting rich and are anyway likely to lose their jobs in bad times. As the ownership of debt has moved away from banks and the rewards for “performance” have increased, it seems likely that the exposure of financial markets to the next black swan has risen.
The book seems to me, however, to fall into some errors, particularly when discussing economics. For example, another footnote refers to “the mistake of confusing a model with the physical entity it means to describe.” This is an extremely important point. Models are not descriptions of the “real world” or even simplifications, they are conjectures. They should not therefore be judged by whether they are “true to life”, but whether they are testable and robust under testing. In two instances, however, the author seems to forget this. He attacks Milton Friedman for saying “models do not have to have realistic assumptions to be acceptable…” and, in an attack on rational expectations models: “…economists ignored the fact that people might prefer to do something other than maximise their economic interests.” Realism is not a necessary condition for testing or robustness.
The author ’s dislike of such economic models is widely shared and usually comes from a failure to appreciate that economic results are determined at the margin. The models do not claim that people are homogenously rational or driven by self-interest, but that these qualities are crucially important at the margin. Not only is a lack of realism no justification for discarding a testable and robust model, the accusations themselves are often due to misunderstanding. This is important as we naturally seek to understand the world and models which appear to be unrealistic seem unsatisfactory, even if robust. This observation does not, however, detract from the author ’s central thesis. While models based on rational expectations need not assume that everyone is rational, this does not make them correct. Even if they have proved robust, so far, under testing, they cannever be more than probable.
Notes
* See “There is little value in broker economics.” Andrew Smithers. Financial Times, January 4 2006.
** C.S. Peirce (together with Karl Popper) is, very properly in my view, one of the book’s heroes. The quotation is from the essay “Hypothesis and Imagination” from The Art of the Soluble by Peter Medawar. Methuen 1967.