I wish someone had told me at the beginning of my career… Learn more math! I was good at math but didn’t appreciate how important it is to learn it when you’re young. Math is central to applied economics and could be used more in JDM psychology. The same is true for statistics— knowing a lot of tricks helps you get the most from data, win arguments, and figure out what you do and don’t believe in other people’s papers.
I most admire academically… (With apologies to many whom I’ve forgetfully excluded) Dick Thaler, for setting a good example by writing a small number of papers on important questions, and making each a gem. Danny Kahneman for being so wise, getting wiser every year (how does he do it?) and writing so beautifully. Amos Tversky, for a steeltrap mind and tenacity in digging on a topic until he had it figured out and expressed in a simple formalism. George Loewenstein for his gift of synthesizing lots of ideas and examples into an insight in a way that is very fruitful for others to then pursue. Gary Becker for seeing the interesting economic elements in so many kinds of choices (like having children, and crime). The economist Bob Shiller for being eclectic and for daring to write aggressively about the role of social forces in asset pricing (which everyone else thought was crazy and unmodellable but now is starting to gain traction). I also admire a lot of people JDMers may not have heard of in other fields. One is Joe Henrich, a cultural evolution anthropologist who did the first economics experiment in a small-scale society, which then led to an influential cross-society project. Three more are: Duncan Watts who knows a ton of things about social networks, Mike Kearns, a computer scientist who recently became interested in experiments on networks and problem solving, and Peter Dayan, a “dry” theoretical neuroscientist who is always coming up with remarkable bold ideas.
The best research project I have worked on during my career… If you’re doing it right, you almost always have the very genuine feeling that the paper you just finished is the best one (even though you had that same feeling N-1 times before). One of the best was our paper on taxicab driver labor supply (QJE 1997). It was a really simple insight and one of the earliest clear tests, outside of finance, between a behavioral alternative and a very standard economic idea—that labor supply curves slope upward (i.e., workers put in more hours when wages are higher). I was living in New York at the Russell Sage Foundation so off to the Taxi and Limo Commission I went. There sat a bored economist whose main job is to collect statistics so they can justify taxi fare increases every couple of years. It turned out they had done some studies asking drivers for information on the hours they drove on different days, so I left with a (free!) floppy disk full of data from them. We did not have any formal model in the paper, but others came along later, figured out the proper way to model it with reference-dependence, and replicated our basic finding.
With that paper, we also had a mixed editorial experience with a happy ending. We sent it to American Economic Review, where we ended up getting one silly short report basically saying “I don’t believe it” and mentioning measurement error, which we had addressed very squarely (with a good “instrumental variable”). A lot of economists were (mindlessly) hostile back then. We withdrew it and submitted it to a special issue of the QJE honoring Amos Tversky and got incredible help from the editor there (Larry Katz) who is an outstanding labor economist and told us exactly what to do.
The worst research project I have worked on during my career… The worst was the first experiment we did in Charlie Plott’s class in winter 1980 at Chicago GSB. Charlie was an incredibly patient and generous teacher, so he required us to actually run an experiment. We were interested in finance at the time so we created an experiment to test whether specialists in stock markets would smooth prices as they are supposed to do in theory, by buying during price drops and selling during price increases.
We made every possible mistake. First, there was only one specialist per session and a lot of live traders, but only the specialist’s behavior was interesting. So the design had incredibly fragile internal validity—a distracted or confused specialist would just produce terrible uninteresting results. The instructions were a mess. And of course we did not plan well so 10 minutes before the experiment we were in the library– a 5 minute walk from the lab– Xeroxing the instructions. Now I tell students that their first experiment will be their worst—hopefully!, since there is a learning curve—so they should just pick something and get started, rather than fret and ponder endlessly trying to make it perfect.
The most amazing or memorable experience when I was doing research… Probably the most memorable was a paper exploring whether you could create herd behavior in a horse race betting market. At the time, people in economics were just beginning to formally model “cascades”, in which you observe decisions other people make— like a crowd outside a new restaurant—and decide how to combine your own belief with what you infer from the crowd.
By mistake I once put in a ticket for a race that had not been held yet, and the terminal screen came up “Do you want to cancel your bet?” So I realized you could make bets and cancel them before the race. Then I got the idea to make large bets on a horse, $500 or $1000, and see if those bets influenced others to bet on the same horse (herding) or to stick with their own hunches and bet against me. Either result would be interesting.
It was fun to actually make the bets and see what happened. It was a matched-pair design in which races with two similar horses were picked, and I literally flipped a coin to decide which of the two to bet on (the other one was a within-race control). I had a little notebook and wrote down the betting totals every minute, it was fun being like a naturalist in the economic wild. It was also nerve-wracking because half the betting happens in the last three minutes, so there was always a chance I would get stuck in a slow line and not cancel the bet in time. Imagine having to explain to the university accountants why I needed to be reimbursed $1000 for a bet at the track!?
The one story I always wanted to tell but never had a chance… In graduate school and my first two assistant professor jobs, I had a small independent record label. I always wanted to write a short casual paper on behavioral decision making and valuation under ambiguity in businesses based on my experience. It was fun and actually made a bit of money, which was a miracle.
A research project I wish I had done… A few years ago Dave Perrett came to Caltech and showed some beautiful work using facial morphing. After that a PhD student (I think it was Meghana Bhatt) suggested that maybe you could make people think about the future differently by showing them an aged version of their own face. We were lazy about actually doing it. Hershfeld et al. 2011 actually did this. The general method of facial morphing could be used in lots of other JDM research, too.
If I wasn’t doing this, I would be… I would be a photojournalist or a documentary filmmaker. My first job after college was working for a beach newspaper in Ocean City, MD. I loved the idea of taking pictures and had an excellent semipro photographer coaching me. (This was in the old days where serious photographers would develop the film in a darkroom, in a chemical bath—it was tedious but cool!) Sadly, my pictures were terrible. At the very end of the job we discovered there was a light leak in the camera (sadface) so my pictures weren’t so awful after all. Anyway, pictures of dramatic events, especially political events and war, can be so riveting and important (like Nick Ut’s famous picture of the napalmed Vietnamese child running down the street). Documentaries can make the same impact in a longer form. And they are actually surprisingly profitable as a whole because they are cheap to make and because of the long tail from the possible huge box office gross.
The biggest challenge for our field in the next 10 years… In my view, probably the biggest challenge and opportunity is to make use of the amazing change in accessibility of new field data (so-called “big data”). Economists have a head start on this because most of the data they work with are not experimental or survey data they produced, so they are well-equipped to find data and get answers out of it. Computer scientists are looking at these data too, and they have a huge edge in being able to get data (e.g. scraping websites etc.) If JDMers are stuck only in lab mode we will miss an opportunity to use both field and lab data to study robustness, whether interesting effects evident in short lab experiments persist over longer periods of time, and so on.
Keep your eyes open for where data are available. Lots of useful data are available from the web. In the US, the Freedom of Information Act (FOIA) requires governments to release any data they collected unless it’s classified. Many nonprofits and government agencies are interested in using behavioral science to make sense of what they do, and they are often eager to publish results (whereas companies may consider findings intellectual property and don’t want to publish it in order to keep it private). Tech companies like Google, Microsoft and Facebook have big research groups looking at their internal data and like having people spend time there as interns etc. A lot of people they hire are computer scientists who can be quite clueless about psychology and social science. JDM could add a lot.
Instead of thinking about what lab experiment to run, I hope some new researchers in JDM first think—what are the ideal field data to test my hypothesis?— then keep their eyes peeled for those data, including cold-calling companies asking for data. You can always run experiments as well if the field data are inconclusive about causality.
My advice for young researchers at the start of their career is… From a career point of view, it pays to specialize in a topic you find really interesting and explore it thoroughly using various tools. When you come up for tenure you want to be know as “Ms. Emotion and Risk” or “Mr. Overconfidence” or what have you. Don’t be shy about introducing yourself to senior researchers at conferences and sending them papers. Usually we won’t read the papers (or if so, not carefully enough to comment) but it gets your work into our memory.
Another important thing is to have a very clear understanding with your colleagues and department chair about what is expected of you to get tenure. Some places have very clear criteria, in terms of the number of papers and what journals count the most.
Another common mistake, in my opinion, is to invest too heavily in teaching pre-tenure. Teaching can be fun, you get positive feedback, and it’s deadline-driven. Research can be painful, frustrating, with negative feedback and no deadlines so that you can always procrastinate. To be very frank, as long as your teaching is adequate, research-oriented schools really do not care about teaching quality in making tenure decisions. If the colleagues who will be judging you say teaching does count a lot, get them to spell out what exactly that means and look carefully at the last 10 years or so of who actually did or didn’t get tenure. If star teachers with short vitas are getting fired that tells you what you need to do. When I was at Wharton business school there was a streak of people winning teaching awards then getting turned down for tenure just afterwards. It got so bad that people would start to worry if they won an award.
One more thing for women on the tenure track (and beyond): Many female colleagues complain that they get asked to do a disproportionate amount of service, such as serving on thesis committees, working on curriculum, recruiting, organizing speakers, and so on. Obviously these are activities that somebody has to do and you should feel obliged to do your share. The problem seems to be that women do too much. Maybe women feel more compelled to do it. Men seem to either not get asked as often or say No more often. It could also be that men do such a mediocre job that they get “punished” by not having to help out in the future. While your tenure clock is ticking, you need to guard your research time fiercely (or enlist a senior colleague who can help you do that).