We Need Accurate Predictions!

Brad Delong, a Berkeley economist whose work I generally respect, reviewed Judge Richard Posner’s book A Failure of Capitalism. The review is very critical of both the book and of the Chicago school of economics. The Chicago school is a branch of economics that has been quite influential over the past half century. The intellectual founder was the late Milton Friedman.

The Chicago school is deductive in its results; virtually all its theories rely on the assumption that consumers and producers are rational. To Chicago-school economists rationality means that consumers use all available information to make choices that maximize subjective well-being. While Chicago-school economists do not suppose that people are all knowing, they do assume that people are generally correct in their beliefs and assumptions about the future. Critics of the Chicago school typically overstate the assumptions of its adherents, claiming that such economists envision people as being omniscient.

Milton Friedman responded to some of these criticisms in an essay, "The Methodology of Positive Economics." According to Friedman, the way to test the validity of models is by analyzing the accuracy of their predictions, not by the accuracy of their assumptions. Thus, although people may not be as rational as the Chicago school supposes, Chicago-school theories are valid if they make accurate predictions. The reason that the Chicago school has been so successful is that its models have been confirmed by a wealth of data.

Furthermore, according to Friedman – and the principle’s embodied by Ockham’s razor – if multiple models make equally accurate predictions, the best model is the simplest. These views are not exclusive to economists: the great theoretical physicists Stephen Hawking makes similar claims in A Brief History of Time.

Anyway, back to Delong –

The review is based on an analogy comparing (a) Chicago-school adherents to 17th Century Jesuits and (b) enlightened economists (presumably behavioralists who acknowledge irrationality) to Copernicans. Jesuits believed that the sun revolved around the earth. Copernicus’s model claimed otherwise. Despite mounting evidence, the Jesuits clung to their beliefs by developing increasingly complicated models. Nonetheless, the Jesuit models couldn't predict planetary movements as successfully as Copernicus's simple but elegant model.

The Jesuit/Copernican analogy is the most ill-advised part of the review (other than perhaps DeLong’s failure to define rationality, as Posner points out in a response). The reason is that in economics, it is the behavioralists who are proposing increasingly complicated models and assumptions. Furthermore, with a few exceptions the behavioralist models are not able to predict behavior as effectively and accurately as rationality-based models. The assumptions of behavioralist models are often vague, ad hoc, and inconstant from model to model. Behavioral predictions are generally non-quantifiable and difficult to implement in non-laboratory settings. Until behavioral economics can make predictions that are more accurate than neoclassical predictions, its claims of irrationality will be meaningless.

This post may seem surprising, given the blog subtitle. I indeed identify as a behavioral economist - or rather as an economist who dabbles in behavioral research. In my view behavioralism can offer useful insights into economic decision-making. Nevertheless, behavioral economics will be little more than an interesting footnote until it can consistently make accurate predictions.

In a later post I'll identify some behavioral predictions that are more accurate than neoclassical predictions.