Economists Need to Learn a Messier Kind of Math
People aren’t rational calculating machines. So why do models of the economy -- models that inform policy in some of the world’s largest and wealthiest countries -- still assume that they are? It might be no more than a matter of taste.
Economists love the rigor and tidiness of mathematical equations. But building a model with them requires making a lot of simplifying assumptions -- for example, that people make optimal decisions based on a perfect understanding of all constraints and possible outcomes, and that economies tend toward equilibrium. Unfortunately, those assumptions are so flawed that the models have failed to recognize the possibility of crucial events, such as the last global recession.
Some economists are trying to make the models a bit more realistic. Xavier Gabaix of New York University, for example, has been experimenting with one that seeks to account for human limitations in foreseeing the future and adjusting their behavior. One important insight: Because people don’t actually smooth their consumption perfectly over the life cycle, fiscal stimulus or “helicopter drops of money” really can get them to spend more and help pull an economy out of recession. The model, however, still requires decisions to be close to optimal, so it’s a small tweak.
A bolder and more natural approach would be to take psychology seriously and accept that real people rarely optimize. Instead, they make decisions using a wide range of heuristics, or simple rules of thumb. We copy others, or seek out experts and take their advice. We learn by trial and error, or ignore information that we judge to be irrelevant (rightly or wrongly). And we listen to our emotions. Informal observations suggest that corporate executives base roughly half of all their decisions on gut feelings, only afterwards seeking to justify them with logical reasoning (or long reports by consultants).
Flawed as heuristics may be, some psychologists argue that they’re actually superior to calculation when problems are complex or information is lacking, as often happens in a highly uncertain world. So why, then, don’t more economists try to build models using such heuristics? One convincing explanation -- offered in a new book, “The End of Theory,” by financial economist Rick Bookstaber -- is that it would require using a kind of math in which economists haven’t been trained, and which they are therefore reluctant to use.
Consider a financial market. Bankers and investors make decisions based on society-wide moods, details of their day-to-day interactions and perceptions of what others may know or not know. Let millions of such agents loose in a model, and you’ll get something akin to the chaos of real markets. No pretty set of equations determines what happens. Events unfold in an algorithmic way, with one triggering another. The outcome can’t be predicted, and may well be something that no one could have even imagined beforehand.
Economists don’t like to have this kind of uncertainty in their models. They know it exists, but the math of optimization works only if they assume that people know -- and agree on -- the full spectrum of things that can happen, as well as the probabilities. So there can be no true surprises, even though it is precisely such surprises that drive many of the most important events in the real world.
The only way forward, Bookstaber argues, is to accept uncertainty, fixating much less on optimization and equilibrium. This would involve developing models populated with large numbers of agents, all acting on their own heuristics. Such models can include as much of the actual “plumbing” of markets or economies as necessary -- including laws, imperfections and details of structure -- rather than trying in vain to reflect some beautiful theory in the abstract.
Some economists and other scientists are already working on such agent-based models. We need more, so that one day we might have an economics with fewer impressive equations and a lot more relevance in the real world.
This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.
To contact the author of this story:
Mark Buchanan at firstname.lastname@example.org
To contact the editor responsible for this story:
Mark Whitehouse at email@example.com