The first edition of The Undercover Economist sported a pulp-fiction private investigator on the cover. I’d suggested the image because, well, why not? Little did I realise I was anticipating a trend: the economist as detective.
It seems an unlikely development, given the fondness of economists for convenient simplifying assumptions. (Had Sherlock Holmes been an economist, he would surely have assumed that the Hound of the Baskervilles was a perfectly spherical dog moving in a frictionless environment.)
But whether the development seems unlikely or not, “forensic economics” has arrived. A splendidly simple example was published in 1994. William Christie and Paul Schultz noticed that prices on the Nasdaq market, quoted in eighths of a dollar, actually varied by quarter-dollars. Quotes ending one-eighth, three-eighths, five-eighths and seven-eighths were rare. Christie and Schultz checked carefully before concluding that this was probably symptomatic of collusion between market-making investment banks, designed to keep bid-ask spreads plump. An investigation by the Securities and Exchange Commission, class-action lawsuits and out-of-court settlements all followed.
Other examples of forensic economics are more complex. A recent survey on forensic finance by Jay Ritter and a forthcoming one on forensic economics by Eric Zitzewitz jointly outline how forensic economics is evolving.
One approach is to use purely statistical analysis, the most famous example of which comes courtesy of Brian Jacob and Steve Levitt, who identified some improbable patterns in multiple-choice test results in some Chicago schools. These suspicious “answer strings” (with all students giving the same response to clusters of questions) were correlated with sharp, temporary improvements in average scores.
Another approach is to use deviations from what an economic theory would predict. Zitzewitz found evidence of “late trading” in mutual funds, which is illegal. He showed that trades were correlated with information the traders couldn’t legally have possessed and could not – according to the efficient markets hypothesis – have deduced. The suspicious patterns disappeared after an investigation was announced.
Field experiments are also increasingly popular as a way of revealing suspicious behaviour. Benjamin Olken studied corruption in road building in rural Indonesia, devising his own measures of how much construction material had been stolen and cross-checking them with an independent audit.
What emerges from forensic economics? Zitzewitz argues that a major theme is government failure: not just the failure of dysfunctional governments in poor countries, but the failure of police and regulatory authorities in the United States and other wealthy countries. Perhaps in these gloomy times that will not come as a huge shock.
It’s fair to ask why economists feel they are qualified to undertake this kind of analysis. They are a self-confident bunch and may not be inclined to ask such questions of themselves. But answers do exist.
Economics is unusual in requiring statistical sophistication plus the ability to think about man-made institutions and human motivations. And economists are naturally suspicious: rational economic man, after all, is smart and amoral, the kind of person you’d want to keep an eye on.
The forensic economics trend may be good for the economics profession, too. It requires economists not to spend too much time thinking about theory, but to pay close attention to data, and to the messy way in which markets actually work. These cannot be bad habits for economists to acquire.
Tim Harford’s book ‘Adapt’ (Little, Brown) is now available in paperback.