It Takes (an Invisible) College

|
Posted in Uncategorized

Victory is claimed by all, defeat assigned to one alone, wrote the Roman historian Tacitus.  That’s certainly the case with the emerging field described as behavioral economics. Thus Richard Thaler, of the Booth School of Business of the University of Chicago, in his autobiography, Misbehaving: The Making of Behavior Economics, tells the story of the “Anomalies” column he wrote for many years, criticizing standard theory for the Journal of Economic Perspectives,

In The Undoing Project: A Friendship that Changed Our Minds, forthcoming in December (Norton), author Michael Lewis agrees with Thaler in assigning credit. Danny Kahneman and Amos Tversky, Israeli psychologists, who worked in tandem for nearly thirty years, “are more responsible than anybody for the powerful trend to mistrust human intuition and defer to algorithms,” the book’s advance material states. (Tversky died in 1996, at 59; Kahneman shared a Nobel Prize in Economic Sciences in 2002).

In fact, a good many seminal thinkers besides Kahneman and Tversky have contributed to the creation of behavioral economics, and what today is becoming part of economics’ engineering wing – perhaps as many as two or three dozen.  Some have died; more than a dozen have been recognized by the Swedes; several others may yet win a Nobel Prize.  A celebration last week of the sixty-fifth birthday of David Kreps, of Stanford University’s Graduate School of Business, marked the scholarly contributions of one of the leading figures in the integration of game theory into economics.  Somewhere near the beginning of the two-day Krepsfest, Joel Watson, of the University of California at San Diego, observed,

Writing books gives an author greater flexibility than afforded in journal articles.  Journals want simple ideas (so readers don’t have to work hard to get the main point) with fairly complicated analysis (to show muscle and credibly assert that others might not have been able to do it) that is very well connected to existing lines of thought in the literature (so that it makes others in the profession feel relevant and good about themselves).  With a book, the author can ruminate broadly (putting more ingredients into the picture) and speculate (sketch some ideas that may stimulate further research).

Hence Kreps’ reputation: he has done it all that

As a graduate student enrolled in Stanford’s operations research program, he shared a Palo Alto house apartment with future Nobel laureate Alvin Roth and Robert Leary. A third student in the program. He burst on the scene in the late 1970s with series of papers on financial markets, dynamic programming, and the theory of choice. When in 2010 the American Economic Association made Kreps a Distinguished Fellow, the citation listed at the very end his contributions to finance.  “Characterizing no-arbitrage restrictions in terms of martingale equivalent measure is the foundation of modern asset pricing theory,” the citation affirmed. Non-economists will recognize only the word “foundation.” At Krepsfest, finance came first on the program/ With Kreps’ student Chi-fu Huang, of DeepMacro.com, away on family business; Huang’s student, Ayman Hindy, of Capula Investment Management, filled in, delivering retrospective remarks. Both witnessed the spectacular rise and fall of Long Term Capital Management from the back of the bus.

(Kreps, who grew up near Montclair, N.J., and attended Wesleyan University and Dartmouth College, was a mathematician before he was an economist, thanks to summer math camps organized by the National Science Foundation. I asked him about that and he replied, “The summer I spent in 1966 at Ohio State was one of the transformative moments in my life.   I always ‘knew’ I wanted to do mathematics – at  least, past third grade – up to that point I thought astronomy was cool – but it wasn’t until that summer that I figured out what mathematics was. So Arnold E. Ross sits first among the teachers and mentors I’ve had. I don’t know how much your column has of biography, but despite what I’ve done in and around economics, one tiny accomplishment of which I am inordinately proud is that I have a tiny extension of the Hahn-Banach Theorem in mathematics with my name attached.”

It was in 1982 that a little paper with Paul Milgrom, John Roberts and Robert Wilson, appeared in the Journal of Economic Theory.  “Rational Cooperation in the Finitely Repeated Prisoners’ Dilemma,” showed how, with a little uncertainty about the each other’s personality, cooperation could emerge among players whom standard theory predicted would inevitably choose to betray one another at the end of the day – a phenomenon ubiquitous in daily life but, to that point, about which economics had little or nothing to say, “intractable” as it seemed.  The authors were quickly dubbed the “gang of four”; the joke being that, by joining forces, the authors (and the editor, Karl Shell) assured that whatever path might lead one or more to an eventual  Nobel Prize, would be pursued separately, since the Swedes never cited more than three persons in one award.  Kreps and Wilson’s “Reputation and Imperfect Information” and Milgrom and Roberts’s “Predation, Reputation, and Entry Deterrence,” appeared separately in the same issue. “Sequential Equilibrium,” by Kreps and Wilson, subsequently elaborated the tool.   With Drew Fudenberg, Kreps wrote papers modeling aspects of learning that started skeins of influential work; with In-koo Cho, he showed how players dealt with new facts they learned about each other.  In 1989, Kreps was recognized with the John Bates Clark Medal as having made the most significant contributions among his economist cohort before the age of forty.

The next year came A Course in Microeconomic Theory, a textbook for first-year graduate students  – the “Black Book,” so called for its lightly-decorated cover. It was here that many of the best students of the next generation discovered that ideas previously taught in advanced “topic” courses in the second year had become basic  for talented undergraduates: moral hazard and adverse selection, signaling and screening behavior, entry deterrence, implicit collusion, and reputation effects.  Ahead lay incentive alignment and mechanism design. The format was pure Kreps: larger type for essential matters, technical material confined to in smaller type, designed to be read after the gist was mastered; interspersed with  occasional philosophical digressions and, every once in a while, a sly joke. “It was like having a highly interesting and patient mentor in the room, talking with you and explaining things,” is the way political scientist Jonathan Bendor, of GSB Stanford, recalled.  Here is the way Kreps introduced the notions of behavioral type and reputation formation at the heart of sequential equilibrium:

Imagine two individuals, whom we will call (without much imagination) player 1 and player 2. Player 2 is unhappily something of a bully. He likes to pick fights with cowards.  But he is equally averse to picking one with someone with courage; he will pick a fight with someone only if he assesses probability .5 or more that this individual is a coward. (He is indifferent between picking and not picking a fight at assessment.5.) He considers picking a fight with player 1, but he is uncertain whether player 1 is a coward or is brave.  He initially assesses probability .8 that player 1 is brave and .2 that player 1 is a coward (these being the only possibilities), and so it seems that he will not pick a fight with this individual.

Before deciding whether to pick a fight with player 1, player 2 gets a signal about player 1’s disposition that may be of value, namely what player 1 will have for breakfast.  If player 1 is brave, then she prefers a breakfast of beer to a breakfast of quiche. If player 2 is a coward she prefers quiche to beer.  So, it might seem, player 2 should observe what player 1 has for breakfast and conclude: if player 1 has quiche, she is certainly a coward; pick a fight.  But if she has beer, she is certainly brave; avoid a fight. But complicating this is that whether brave or a coward, player 1 prefers player 2 not pick a fight with her, and cares more about this than she does about the type of breakfast that she has. And player 1 realizes that player 2 is lurking at the street corner, watching carefully what player 1 has for breakfast.

The eighties were a time of high excitement in microeconomics.  In an essay for a special issue of the quarterly Daedalus, “American Academic Culture in Transformation: Fifty Years, Four Disciplines,” Kreps explicated the transformation of microeconomics in the last quarter of the twentieth century. Twenty years later, Kreps’s account remains the clearest I know of why so much has changed. He describes the effects of the mathematization of economics in terms of translation of one language to another. The powers of formal deductive models and statistical techniques were great, he wrote; as was the status that their use conferred. Armed with mathematical methods, he says, economists tackled the easiest problems first: the price theory that had emerged from natural language over two hundred years.

The study of economic institutions – a big portion of mainstream economics before World War II – dwindled in importance. The newly dominant dialect of mathematical modeling lacked some topically important vocabulary; rather than speak in an unfashionable dialect, some things were just not discussed….

From 1975 to the present, what we have seen in economics is 1) the development of further techniques that play by the basic rules of established (mathematical) economics but that broaden its scope; and 2) the application of those techniques to specific contextual questions, with the result that the field seems to be returning to something like the breadth of the discipline before World War II.

Borrowing a conceit from economist Paul Romer, then a GSB Stanford professor, Kreps compared the process to the image of an hourglass:  broad traditional concerns in natural language narrowing dramatically as the initial process of translation takes place, then broadening out again as novel forms of the new language are devised to analyze familiar features of the economic landscape.

In 2000, Kreps became senior associate dean for academic affairs at GSB Stanford, and, over the next nine years worked with Dean Robert Joss to build a large new campus, improve relations with the rest of the university, and continue to press the annual contest with Harvard Business School to enroll the students who have a choice. Stanford’s share of home-grown senior faculty, its preferred method of transmitting and preserving the school’s culture, rose steadily to the desired 50 percent. Plans were laid for an extensive new campus. Cooperation with the economics department has catapulted Stanford into a newly prominent position in the profession, vying with the combination of Harvard and MIT in the East. Five of the last ten Clark medal winners are on campus – Yuliy Sannikov, Matthew Gentzkow, Raj Chetty, Jonathan Levin, and Susan Athey. Levin, for several years chair of the economics department, recently agreed to put aside his research career to become GSB dean.  Meanwhile, Stanford has replaced Princeton as the university that Harvard views as it principal institutional competitor.

Against the odds, Kreps returned to teaching and research at 60. He began a new series of graduate texts:  Microeconomic Foundations I (the math of pre-game theory economics), with volumes II (covering information economics and game theor(y and III (behavior and institutions) not yet in sight. Meanwhile, The Other Big Mo:  Motivation on the Job, a Norton trade book, is expected to appear next year. Its point of departure (after review of the literature):  “Pay for performance, in the form of piece-rate pay or year-end bonuses or, more generally, any rewards for performance, can be a powerful motivational tool.  That is clearly shown in the case of Safelite Glass. But, as the other anecdotes from the last chapter make clear, for jobs more complex than that of Safelite technician, pay for performance can be difficult to get right; and it may even be dysfunctional.”

At dinner, Robert Gibbons, of MIT”s Sloan School of Management, one of the organizers of the conference, summed up:

In my view, for about the decade of the 1980s, Stanford GSB was  more than a great business school: it gave  the world one of the two greatest epochs in business-school history (the other being Carnegie, for a decade around 1960) .It’s tough  to prove that Kreps’s arrival in 1975 caused this  spectacular epoch, but he clearly played a huge role – not just by his celebrated contributions to economics, but also by the bridge he helped build from economics to finance, accounting, and other management fields, as well as by the tone he modeled for the school’s teaching and research. Given the enormous impact of David’s research on economics, it surprises me to realize that, at this institutional level, his contributions to the school’s tone may have been even more important than his academic work.

So much, at least for today, of some of the victories of behavioral economics. What about its most recent embarrassing defeat?  I mean, of course, the persistent nudge, codified in corporate culture at Wells Fargo Bank, which led more than 5,000 people to fake those new accounts – a spectacular example of organizational engineering gone wrong.  Some practitioner or strategist or architect came up with the design. Others elaborated it, studied it and signed off. Eventually case studies will be written, after the shouting dies down; some are doubtless taking shape already, and there will be Kreps’s book, which of course isn’t about Wells Fargo per se.  Many accidents happen naturally in the early days of new technologies.  It’s how engineers have always learned. Behavioral and institutional economics, along with their engineering wings, market and organization design, will continue to improve, if that’s the word.