Susan Athey Awarded John Bates Clark Medal
by Joshua Gans
Susan Athey, a 1995 graduate of the Business School’s doctoral program in economics and currently a professor at Harvard, became the first woman to be honored with the very prestigious John Bates Clark Medal in April. The medal, which is one of economics’ oldest awards (60 years as compared to 40 years for the Nobel Memorial Prize), is given every two years to the best economist under 40 in the United States. It is one of economics’ most prestigious awards and a predictor of future Nobel glory. (At Stanford Ken Arrow, Michael Spence, and David Kreps are all past recipients.) While insiders already knew that Susan’s work was outstanding, the award tells outsiders (either to the field or to academia) of the news.
I’m an insider in this case, having been in graduate school with Susan in the early ’90s, and so have known her from before she made the contributions recognized this year. If I had been a betting person then, I happily would have taken a bet that she would win. Indeed, it would have been hard to find someone to take the other side in either the Economics Department, where I was located, or at the Business School, where Susan was a student.
Susan is an economic theorist, and in this respect, her award represents a “return to fundamentals” after a decade of applied economists receiving the award. For example, recent winner Steve Levitt, author of Freakonomics, has made a career of finding novel ways to explore economic forces in settings distinct from the drier subjects in the field, which has given his work notoriety rarely seen in economics. Susan’s work is on stuff far removed from Freakonomics, but to academic economists, it is no less clever and in many ways more brilliant.
In the book A Beautiful Mind about mathematician John Nash, Sylvia Nasar wrote that he saw the world differently from most people. In the same way, Susan operates at another level from most of her contemporaries. Some of us can see how to represent things mathematically; Susan sees mathematical structure and brings order where complexity has reigned. This is not an uncommon contribution of the best scientists, but precious few have this ability in economics. And it is that quality that will make her work stand the test of time.
The job of an economic theorist is to abstract enough from the complexity of the world to see through economic forces and ultimately make a prediction. To most, that means coming up with simple models, lots of assumptions, and hoping that the predictions are general. Take a simple thing like whether a fall in interest rates might stimulate investment. It is easy to theorize why that might be the case: The costs of finance are a big component in a firm’s investment decision—there can be lots more going on. After all, why have the interest rates fallen? It could be luck but it could also be because we are in a recession and the Fed is trying to stimulate the economy. In that case, while we might hope lower interest rates could stimulate investment, we can hardly be sure.
Susan’s contribution has been to give economists more confidence that their predictions are actually reasonable. She didn’t do this by analyzing any one thing—low interest rates lead to investment—but by considering what would improve predictions for all theoretical models.
Stanford in the early 1990s, when we were there, was an interesting place to think about such problems. The emerging internet revolution encouraged economists to really care about how the actions and strategies of multiple companies trying to make a profit from this new technology all fit together. In particular, when one company started to build a market, its profits would depend upon what others were doing; specifically, whether they were developing products that would enhance the value of their own products to customers. Susan’s (and my) thesis advisors, Paul Milgrom, PhD ’77, and John Roberts, had placed all this under the rubric of complementarities; that is, why does doing one thing better lead others to do other things better. This was especially an issue for organizations. Milgrom and Roberts argued that firms (at that time, U.S. firms) found it hard to imitate the efficient outcomes of other firms (at that time, Japanese ones) because the Japanese firms had constructed a system to complement their environment. If their imitators got one part wrong, they ended up with a poor outcome. The key issue was how do you predict what factors were important and what were not.
Milgrom and Roberts (with student Chris Shannon, PhD ’92) made significant advances in how to deal with this issue at precisely the time Susan and I were students. It came down to specifying when we can say if x changes, then y changes in a set direction. Work out what x is and you can predict what you should do with y.
This was all very well and good, but what if you had limited information? What if x was something you could not control? Then y was an uncertain outcome to be characterized in some way that could lead to a prediction. Milgrom, Roberts, and Shannon had tried but got nowhere.
Susan stepped into the breach. She showed that the same things we look for when, say, we assume that “for sure” Toyota is engaging in some practice also applied when we didn’t know for sure. For researchers, this meant that their results generalized and the confidence of their predictions went up accordingly. Indeed, it made prediction possible.
All this was in Susan’s Stanford dissertation. It started, in fact, with a more applied project including myself; Scott Stern, PhD ’96; and Scott Schaefer, PhD ’95, as co-authors. (Stern went on to win the Kaufmann Medal, which is like the Clark Medal but for entrepreneurial research.) We were trying to consider how organizations would change “who decides what” when their operating environment became more volatile and uncertain. Our paper had one little lemma to generate a comparison of options under uncertainty. Susan took that lemma and made 75 more in a few months. No sooner had she finished than she poured the proof statements back into a single fundamental theorem. It was simply amazing, and the reason why she had 25 job-interview flyouts and a write-up in the New York Times at age 25. Indeed, this generated a celebrity status. When we were in Japan a few months later, she was recognized on the street by someone who must have had an economics interest: “It’s you, it’s you.” (It is hard to forget stuff like that.)
Not content, Susan then did the same thing for what economists call “games with incomplete information” and auctions, where she started making predictions herself. She took the relationships at the heart of her work and with Stern showed how the techniques could be used to analyze the adoption of systems especially within organizations (for instance, how 911 operators could more effectively respond to calls). That work, although unpublished, is perhaps her most cited, completing the real task Milgrom and Roberts had set themselves a decade earlier.
I am a bigger fan, however, of her work in industrial organization. Again, using insights from monotone comparative statics, Susan (with Armin Schmutzler) looked at when investment by market leaders would reinforce their dominance, work that really nailed the dynamics of persistent market power. But, in more recent work with Kyle Bagwell, she has completely changed approaches to modeling dynamic competition. She makes the reasonable assumption that firms can observe things about other firms from their pricing choices, and this will soften or strengthen their competitive actions accordingly. This has all sorts of implications. One is that firms who want to collude will have trouble adjusting to change, so if you think oil companies collude on petrol prices, you should expect the price of oil to change less than it would if the prices were based on changes in underlying factors such as supply and demand. Another implication is that government regulators should not expect two competitive firms as sufficient to get real price competition.
Susan’s work is not as highly cited as others, but it is more fundamental and complete than any of her peers. When she tackles something, she does it all. The field is advanced 10 years in a single shot and, not surprisingly, the profession plays catch-up for a while. Indeed, it took years to digest and publish her papers. (Her 1995 dissertation work was published in 2002.)
To end, I am going to recount one story about Susan’s work habits. Last year, she was expecting her second child. I got a call that must have been late at night her time. She was trying to complete a few minor things on a project she and I were involved in on timber auctions in Australia. After a few minutes I asked, “So, you must be due soon; how is it going?” Well, apparently, it was going slowly. Susan was clearing her to-do list from the delivery room! We chatted for a while, and for the next hour emails continued to come in. The baby was born soon after.
Joshua Gans earned his PhD in economics from Stanford in 1995. He is now professor of management (information economics) at Melbourne Business School, University of Melbourne, and has a popular blog at http://economics.com.au.