Behavioural Science is a rapidly expanding field and everyday new research is being developed in academia, tested and implemented by practitioners in financial organisation, development agencies, government ‘nudge’ units and more. This interview is part of a series interviewing prominent people in the field. And in today's interview the answers are provided by Colin Camerer. Colin is an American Behavioral Financier and a Robert Kirby Professor of Behavioral Finance and Economics at the California Institute of Technology (Caltech). According to Wikipedia, he is a former child prodigy who received his received a B.A. in quantitative studies from Johns Hopkins University at age 17, followed by an M.B.A. in finance from the University of Chicago and a PhD in behavioral decision theory at age 21 (I'm not kidding, look it up). Since then his academic career has been world-renowned (he literally doesn't need this introduction). It's a miracle he found the time to do this interview (and edited it!) especially as it turned out to be an 11 minute read. So let's dive into the world of Colin Camerer:
Who or what got you into Behavioural Economics?
Horses and people.
In my hometown there was a racetrack called Timonium Fairgrounds which held a one-month horse racing meeting which overlapped with a country fair, in August. The track was casual about who could come and go, so I would go there starting as a teenager. It was amazing to me that the physical horses walking up to the starting post, and jockeys, all looked the same. It looked like any of them could win (not like in human sports such as football or basketball where there are big physical differences). But the programs and Daily Racing Form had a ton of statistical data about how each horse had performed in the past. The crowd somehow dimension-reduced all this data into consensus odds about which horse would win. And the odds were fairly well-calibrated: Supppose you took a big group of horses who were all at 10-1 odds. That means if you bet $1 you would win $10 (and get your dollar back) if your horse won. The implied probability of winning is 1/(1+10)=.09. If you did studies about what happened to that group of horses about 9% of them did actually win (except for a pronounced bias to overbet the longshots with high odds and low probabilities). How did the crowd synthesize all that information? And could you beat the market if there was some pattern the crowd missed?
After Timonium, my Dad and I went to bigger nearby racetracks with a colleague of his who owned a race horse, and who had become a stockbroker after working with my dad. He was my finance tutor. I became interested in the stock market which was very similar to the race track in terms of mysterious information aggregation. Except the stock market was even more interesting because you current prices depended on expectations about future prices, and the future did not end twenty minutes later when the race was run. So you could have bubbles in which companies were overvalued but overvaluation persisted for a long time. (I later studied this in graduate school but it began at Timonium in 1971).
In college I studied a lot of different subjects. Abstract math was too hard (damn you, number theory). Physics was elegant but didn’t have people. Psychology was interesting but didn’t have math. Economics had a good mix of both. I ended up at the GSB (now called Booth) at the University of Chicago studying for a Ph.D, intending to specialize in finance. At Chicago you took a general exam and a field exam. My economics background was weak— the only microeconomics course I had taken was at Towson State College one summer-- so finance was challenging for me. It also became clear that the core researchable topics were, at that time (1977-81), what seemed to me a boring, small set of questions: Capital asset pricing (with taxes! Exciting!), studies of stock market reactions to events, beginning of option pricing. It was several years before behavioral finance would get traction, and Chicago finance— at that time—was extremely hostile to any models or even intuitions about limits on rationality. In any case, I passed the exams in behavioral science (general) and finance (field). Nobody had passed that combination before, because the typical finance students all took the general economics exam.
Happily for me, Chicago had an unusual behavioral science faculty. In many business schools, the psychologists were studying organizations and taught courses like leadership, organization design, etc. Chicago had hired a remarkable “industrial psychologist”, with a Ph.D. from Wayne State, named Hillel (Hilly) Einhorn. He had a wonderful student at the GSB, Robin Hogarth, who finished in three years, worked in Europe, then returned to the GSB as a faculty member. They were my thesis co-advisors.
They thrived at Chicago for two reasons. First, they both loved to argue and were smart. At Chicago data either win arguments, or at least keep the debate alive. The economics/finance faculty that dominated the school “tolerated” other fields like marketing and behavioral science as long as you could see technical merit and empirical value in what those areas were talking about. They both had mathematical psychology-type training so they knew how to criticize Bayesian updating, expected utility theory, etc. in a sophisticated, even surgical way. Second, Hilly and Robin had developed essentially a Managerial Judgment course for MBAs. Chicago practiced what it preached about consumer sovereignty so there was a very, very lean “core” of required MBA courses. Behavioral science wasn’t required, nor was finance. The theory was that by having few requirements areas had to teach well to attract students. It mostly worked. And students really liked behavioral science because it was interesting, and it was an antidote to the people-are-rational ideas they would get in so many other courses.
Another important event in my drift toward behavioral economics was that Charlie Plott, an experimental economist from Caltech, visited at Chicago in the winter of 1981. He had a tiny class because none of the serious economics or finance students cared about experiments at that time. He loves to push students immediately into projects and was very generous with his time (and even money to pay subjects). I did an experiment on specialists in asset markets that was a total disaster. We made every amateur mistake: It was overdesigned, highly unrobust (the whole market depended on what one subject did), had long opaque instructions, which we printed out at the library 15 minutes before the experiment began three buildings away. It was a very educational disaster.
In that course, it gradually dawned on me that economics experiments—markets, and games—could be used to test many claims about whether behavioral effects would or would not affect market prices. Plott himself was not at all a believer in behavioral economics, but he was fascinated by design and getting people excited about what you can learn from elegant economics experiments (which was a tiny part of the economics profession at the time).
All of this got me interested in using psychology to make economics more realistic, while showing causality in experiments and thinking outside the lab. At that time, a lot of behavioral economics was incubated in business schools rather than in economics departments. Thaler was at Cornell, Shefrin and Statman were doing early work on behavioral finance in Santa Clara, Wharton and Carnegie-Mellon SDS (a very interdisciplinary department) had a good group. So the GSB was a good environment to learn even though there was enormous hostility in the famous Department of Economics next door (which was literally linked to the GSB by an enclosed walkway because the winter was so bad).
What is the accomplishment you are proudest of as a behavioural economist?
In a two-year period 1997-99, I published five papers with different colleagues on different topics with different methods: Taxicab driver labor supply field data (QJE); an early field experiment on race track betting (JPE), a general model of learning in games (Econometrica) and two experimental papers on cognitive hierarchy and overconfidence (AER). Of course, it was partly a coincidence these all got published in a tight cluster but I was very proud of them all, and some of them were also an editorial struggle. For example, we only got one review from AER about the cabs paper, saying all our results were probably a mistake because of measurement error (that was wrong and really insulting because we addressed the question as clearly as we could in the paper). Then QJE announced a special issue in honor of Amos Tversky so we sent the cabs paper to them instead. The two AER papers were “shorter papers” because the journal seemed to treat lab experiments as minor papers that did not need to be long.
And the MacArthur Foundation awarded me a fellowship in 2013. So I’ve got that going for me.
If you weren’t a behavioural economist, what would you be doing?
My fantasy plan B jobs are investigative journalist or documentary filmmaker. I worked as a semi-pro journalist after college and one summer during PhD (meaning, not well-paid and with small audience). It was really fun, especially after college. I learned a ton about human nature, politics, human vanity, institutions, culture. I also learned how to write on deadline. Whatever came out of my manual typewriter at 2pm on Monday would get printed in the newspaper. Everything after that was too late.
How do you apply behavioural economics in your personal life?
Mostly I use behavioral economics as a tool to understand why me and my family face judgment and choice challenges, and to implement normative choices. Mostly it is elementary economics, like recognizing opportunity cost and hiring a lot of people to help us do things we do not like (e.g. dishwashing, lawn-mowing, not driving your brother to the airport which takes 90 minutes and is stressful but can be delegated to rideshare for $50). My teenage son has also been cured of sunk cost fallacy. He also knows that if an online restaurant menu has no prices, the prices are probably high. Happy about that. I’ve lost tussles about optimal family portfolio diversification, however. Not happy about that.
With all your experience, what skills would you say are needed to be a behavioural economist? Are there any recommendations you would make?
I’m going to answer as if the question was asked by somebody applying to graduate school in economics or in closely applied fields (e.g., policy or business schools). First, you need to know the “rules” of economics—the basic canon and methods—very well. (That was a big advantage for me at Chicago in graduate school, it is a crucible for learning to “think like an economist”.) To break the rules you need to know the rules.
Second, in my opinion, if you want to succeed in behavioral economics it is a big help to be very fluent in an adjacent social science. A lot of behavioral economics is in the business of importing ideas and translating them, redesigning and “selling” them inside economics. So you need to become bilingual and know what psychology, or neuroscience, media studies, or whatever, is solid, and has a long good empirical pedigree. Figuring that out can be difficult.
Third, nowadays you really should be able to do lab (and online) experiments, know about quasi-experimental designs (IV, diff-in-diff, regression discontinuity) and know some machine learning. It is often said that most of the methods you will use in your long research career are those you learned in graduate school. It is like packing for a long, long trip to a place where there are no stores in case you forgot to pack anything. Fill that backpack with methods.
Behavioral economics has been slow to embrace machine learning (for reasons discussed in the next section— it got on the BEAM reading list very late), which is unfortunate. As a result, a lot of the exciting work in behavioral science is being done in computational social science by sociologists, cognitive science, cultural anthropologists, etc.
How do you think behavioural economics will develop (in the next 10 years)?
I am going to mix some thoughts on what will happen and what (in my opinion) should happen. Frankly, I think behavioral economics right now is a little stuck. If you look at the papers presented at focused conferences like BEAM (which is invitation-only) and SITE, there is a faddish quality about concepts that drive research for a 5-10 year cycle. Behavioral economists get excited about something like “correlation neglect”, which is the idea that people neglect how one information source can influence different people and get mistakenly double-counted. It is a reduced-form concept trying to capture something much more basic, which is why and when representations of information structure are simplified. To be clear, I think this kind of error can occur, but I don't think it is a fundamental major construct that is going to account for a lot of different effects. Twenty years from now people will look back and reminisce about it, as they reminisce about fashion fads.