Richard Thaler v Adam Smith: Behavioral Economics v the Economics of Behavior

 

 

Awarding this year’s (2017) Nobel Prize in Economics to Richard Thaler has naturally excited behavioral economists. This is in fact the second time that a Nobel Prize has been awarded to pioneers of behavioral economics. I wish to explain why regardless of this euphoria many economists remain skeptical of the scientific contribution of behavioral economics to their discipline. 

Ever since Adam Smith, classical economics has been axiomatic. The twin axioms of the pursuit of self-interest and rationality have produced a rich corpus of economic theory, which purports to explain the behavior of individuals, families, firms, groups and countries. In this, economics has much in common with other axiomatic sciences, such as physics and chemistry. Axiomatic science should not be judged by its axioms, but by its ability based on these axioms to explain and predict the world. In his landmark essay on “The Methodology of Positive Economics ” (1953) Milton Friedman remarked that the success of biology in predicting patterns of vegetation, as if plants can solve complex differential equations, is justified axiomatically. The axiom that plants understand optimization theory may be unrealistic, but predictions based on this axiom can be tested. Utility maximization is an axiom to consumers just as the optimization of sunlight is an axiom to vegetation.  Science is just as good (or bad) as the axioms on which it is based. Karl Popper argued that while all science is axiomatic, better science uses fewer axioms to explain the same phenomena, or it uses new axioms to explain phenomena that incumbent science failed to explain. Einstein’s axioms have supplanted Newton’s.  

Economics is the only social science with a Nobel Prize. This status stems from its unique axiomatic character among the social sciences. Moreover, economists invented econometrics to test their hypotheses, which is increasingly being taught and used by sociologists and political scientists.  By contrast, psychology, sociology, political science and other social sciences have shunned the adoption of basic axioms. For example, there is no unified theory of psychology, which may be functional, structural, behavioral, cognitive, psycho-analytical, social, evolutionary, and neurological.  Students of psychology have to contend with a bewildering array of unrelated ideas. By contrast, students of economics, like students of physics, study a unified body of theory. (Or, at least they did before behavioral economics featured in curricula.)

Behavioral economists have sought to attack the axioms of classical economics by reporting empirical “anomalies”, which appear to conflict with these axioms. Typically, these anomalies are informed by small non-representative samples, such as students, who agreed to participate in hypothetical experiments. These experiments often entail virtual or symbolic pay-offs, and are very different to real life experiences. Indeed, how people behave in the rarified atmosphere of a social science laboratory may be very different to how they would behave in real life. Psychologists refer to this phenomenon as “Hawthorne effects”, which are usually ignored by behavioral economists. This reminds me of an interview on Israeli television of Daniel Kahneman after he was awarded his Nobel Prize in 2002. Laboratory experiments showed that people value objects that they possess above their market value. This phenomenon, known as the endowment effect, had been previously discovered by Thaler as an “anomaly”.  On entering the studio, Kahneman presented a mug with a Princeton University logo to a cameraman, who accepted it with some embarrassment. Kahneman reassured him that it only cost $10. During the interview, Kahneman asked the cameraman to return the mug. Before he managed to add that he meant to pay for it, the cameraman returned the mug. Kahneman insisted that he should pay, but the embarrassed cameraman refused. According to endowment theory, the cameraman was supposed to agree to sell the mug back for much more than $10, but he was happy to return it for nothing. The interviewer reassured the flustered Kahneman that perhaps what happens in real life may be different to what happens in laboratories. 

So called “anomalies” arise in all sciences, but they do not necessarily constitute refutations of theory or paradigms. For example, a medical procedure might be successful for most patients, but not for some. The latter “anomalies” obviously do not imply that something is wrong with medical theory. During the 1930s econometricians and statisticians developed the concept of probabilistic hypothesis testing. Good theory does not have to explain everything. Rather the null hypothesis cannot be rejected empirically with some probability. If this probability is 1, there can be no anomalies by definition. In regression models the absence of anomalies would require R2 = 1, which is an unreasonable criterion for hypothesis testing. Most investigators wish to be 95% sure that the null hypothesis cannot be rejected, which implies the existence of anomalies. Behavioral economists seem to have forgotten this.

Whereas rationality implies uniqueness, there is an infinite variety of irrational or behavioral outcomes. To take an extreme example, rationality implies 1 + 1 = 2. Behavioral arithmetic would imply any answer other than 2. Behavioral psychologists such as Kahneman and Tversky have tried to discover behavioral patterns in irrational behavior. For example, in their “prospect theory” they claimed that individuals do not distinguish between different high probabilities and different low probabilities. In his “mental accounting theory”, Thaler claimed that consumers indulge when they happen to be spending more for other reasons. Hyperbolic discounting theory implies that savers are short-sighted, and under-invest in their pensions. Endowment theory implies that possession matters for valuation. Do these claimed behavioral patterns constitute new axioms that might replace or complement the classical axioms of rationality and efficiency?

Ever since David Hume, epistemoligists such as Karl Popper have argued that good science is deductive rather than inductive. Induction is based on the false premise that nature eventually gives up its secrets, and empirical observation leads to the discovery of truth ex post. By contrast, deduction begins with a priori reasoning based on axioms. Hypothesis testing is probabilistic as explained. Axioms are not chosen by eye-balling the data as they are with induction. Behavioral economists search for empirical regularities in irrational behavior is essentially inductive. Instead of looking for a new holy grail, they need to axiomatise irrational behavior. This is a difficult if not impossible challenge because there is an infinite variety of irrational behaviors. Behavioral economics is a methodological dead-end.

So why has behavioral economics grown in popularity? One reason is that it is entertaining. There is always a good story line behind every “anomaly”. Another is that it is less rigorous than mainstream economics. Behavioral economists don’t have to invest so much in mathematics, statistics and econometrics. Third, whereas mainstream economics is “modernist”, behavioral economics is a post-modern phenomenon, which is more suited to our times. In many respects, behavioral economics is to economics what behavioral genetics is to genetics. The study of twins and adoptees is no substitute for DNA based genetic theory. Behavioral economics is no substitute for the economics of behavior.

Michael Beenstock