Behavioural economics needs a unifying theory4 min read . Updated: 09 Jan 2018, 02:02 AM IST
Economics models are mathematical in nature, and representing the full complexity of human behaviour with math would make models unwieldy
Behavioural economics has always met with a bit more resistance than it deserved. This is true even though a number of behavioural researchers have won the Nobel—Daniel Kahneman, Robert Shiller and Richard Thaler just last year. Despite this and other forms of official recognition at the highest levels, there continue to be some economists who have an almost instinctive aversion to behavioural ideas themselves.
Meanwhile, though research in behavioural economics itself continues, behavioural ideas haven’t spread—in most fields of econ, the vast majority of models continue to be based on an idealized, perfectly rational homo economicus.
What’s the reason for the continued resistance? Part of it surely has to do with the cultural residue of the 1970s and 1980s. In those days, many economists saw their purpose as defending free markets (a few doubtless still do), and the idea that human beings are irrational seemed to threaten that ideal.
But ideology isn’t the only reason for the reluctance. Behavioural ideas are difficult to put into economic models. For one thing, many psychological biases and irrationalities involve people behaving in very complex, situation-dependent ways. Economics models are mathematical in nature, and representing the full complexity of human behaviour with math would make models unwieldy.
A second problem is that drawing too many different insights can lead to a problem called overfitting. Overfitting is when scientists make models that are so flexible that they seem to explain everything but are able to predict nothing. It’s possible to overfit by making a single model that’s too malleable, but the same result can be achieved by cooking up a different explanation for every different piece of data that you want to explain.
Economics already has plenty of the latter kind of overfitting. Each new question tends to be answered with a new model. Most of the models are mutually contradictory, but economists tend not to lose much sleep over this fact. Given that this practice is already widespread, many economists feel that they don’t need behavioural insights in addition—if they can make a new model that relies only on assumptions about technology, or information, cost structures, or other things that would concern a purely rational actor, why bother to add ideas from psychology? Meanwhile, those economists who do worry about the overfitting problem fret that bringing psychology into the mix will just make things worse.
So behavioural economics probably won’t become widespread without a more unified theory to tie the paradigm together. Although this hurdle is unfair, the reality is that behaviourism can’t afford to remain as a collection of disparate effects and anomalies.
Enter Xavier Gabaix. The French-born Harvard professor has been on somewhat of a mission to incorporate behavioural economics into the mainstream. In a recent paper, he made behavioural modifications to a standard macroeconomic model, and showed how this could resolve a number of puzzling questions all at once. Now, he’s setting his sights higher, with an attempt to create a unified theory of behavioural economics.
Gabaix’s unified theory is based on limited human attention. Standard economic theory requires that consumers and businesspeople pay close attention to a vast array of prices, quantities and other information. Because keeping track of this flood of information is very costly, it makes sense to think that people economize on what they pay attention to. That kind of economizing is called bounded rationality, a term coined by polymathic social scientist Herbert Simon, himself a recipient of the Nobel.
Gabaix puts Simon’s idea into elegant yet powerful math. When people in Gabaix’s model receive new information, they only partially incorporate it into their thinking. He applies this idea to a dazzling array of economic questions, and finds that it tends to explain lots of the apparent anomalies and questions that have plagued researchers. If taxes change, the impact may be reduced because people don’t immediately pay attention to the ways that their incentives are different. People may temporarily ignore changes in inflation and other macroeconomic variables, throwing a wrench into the economy and creating more of a role for activist monetary and fiscal policy. Lack of attention to the future can make consumers short-sighted. Overconfidence and other psychological biases can also be represented as forms of limited attention. Gabaix even shows how limited attention can lead to both overreaction and underreaction in asset markets.
Will Gabaix’s unified theory of limited attention catch on? It will be an uphill battle. Adding limited attention means one more variable for economic theorists to deal with when solving their equations, and one more parameter for empiricists to estimate. But the idea is a deep and powerful one, and Gabaix himself is the perfect salesman. He’s no crusader—he uses purely rational models as freely as behavioural ones, and has the respect of even the staunchest defenders of the old paradigm. In trying to create a new unified theory of economic behaviour, he’s following in the footsteps of another French economist: the great Gerard Debreu, whose purely rational model of “general equilibrium" now forms the foundation of much of economics. The odds seem long for Gabaix’s model to fully supplant Debreu’s as the standard, but it could herald a new era of increased influence for the behavioural approach. Bloomberg View
Noah Smith is a Bloomberg View columnist.
Comments are welcome firstname.lastname@example.org