When the Wells Fargo scandal broke last year, revealing that the bank’s employees had opened millions of fake accounts in customers’ names, the general consensus was that the company’s aggressive, unrealistic sales targets drove the workers to such rampant fraud. Instead of fostering a culture in which employees know they’ll be rewarded for doing the right thing for customers, Wells Fargo had put in place a system that created perverse incentives to cheat the very people the bank should have been serving. Once word got out to the public, Wells Fargo’s reputation took a hit that may ultimately cost the company far more than the nearly $200 million in fines that they’ve already paid.
It’s easy to see what went wrong at Wells Fargo. It’s harder to see what the bank might have done to motivate the mal-motivated employees in ways that would be good for the employees, the bank, and its customers. This issue clearly involves ideas from organizational behavior. But also it raises economic questions, questions that traditional economics seem ill-suited to address. Over the past four decades or so, however, economists have widened the scope of their discipline to address organizational culture, reputation, and motivation, all key issues in this story. One of the leaders in these developments is David Kreps, Adams Distinguished Professor of Management at Stanford Graduate School of Business.
Organizational economics is only one of several fields to which Kreps has made seminal contributions; his prolific career also included path-breaking work in choice theory, finance, and game theory.
The contributions Kreps made before he was 40 earned him the John Bates Clark Medal in 1989 (then given only once every two years); that and subsequent work led to honors as a Member of the National Academy of Sciences, a Fellow of the American Academy of Arts and Sciences, and the Econometric Society, and a Distinguished Fellow of the American Economic Association. And, while doing this, he has had a significant impact on education and administration: he was senior associate dean for Academic Affairs for nearly a decade, and he is author and coauthor of several seminal textbooks. Jonathan Levin, current dean of the GSB (and himself a John Bates Clark medalist), wrote in an open letter to Kreps that when he was a first-year grad student in economics considering switching to a more exciting field, encountering one of Kreps’ textbooks “was like stumbling into a conversation with someone who knew all the secrets of economic theory” — a process that lured Levin into that conversation.
Back in September, dozens of colleagues and former PhD students gathered to honor Kreps for his powerful, wide-ranging influence. The common denominator for all this disparate work, says GSB economist Andy Skrzypacz, is that Kreps is “the master of formal modeling of economic phenomena” — with an unparalleled knack at taking informal, messy notions such as reputation and preference for flexibility, and conceptualizing their essential elements through the precise language of mathematics in ways that lead to new insights.
Kreps received his PhD from the Department of Operations Research in the School of Engineering at Stanford; his advisor was Evan Porteus, Sanwa Bank, Limited, Professor of Management Science, Emeritus. His dissertation involved the theory of dynamic programming, and this work led him to ask: What is it about dynamic choices made by individuals that is different from the standard models of static or one-time choices?
As Princeton economist Faruk Gul explained, two papers represent the most important work Kreps did in answering this question. One is his 1979 paper on the preference for flexibility, the desire to keep options open, just in case. The key to this paper was finding the right context in which to model preference for flexibility: Instead of thinking of the choice of objects, Kreps asked how individuals choose opportunity sets from which they will choose an object at some later time. This formulation of the problem makes it simple to identify a preference for flexibility and allowed Kreps to show how preference for flexibility is precisely (and only) “as if” the person choosing is unsure what she will want at the later date.
The second paper, coauthored with Evan Porteus, concerns preference for early or later resolution of uncertainty. Imagine that a coin will be flipped, and you will either win $100,000 or lose $50,000. Imagine that money will only change hands in three months. This opens up the possibility that you can be told the outcome today, or you can be told the outcome in three months. Does it matter? Of course it does, and most people prefer early resolution of uncertainty. But, for other gambles, people prefer to delay hearing the outcome. Again, the crucial innovation was to find the right way to depict lotteries that distinguish between learning the outcome today versus in three months. Once the right model was identified, expressing preferences that distinguished between different times of resolution was straightforward.
Kreps has a PhD in operations research. Most people think of him as an economist. His chair title is Adams Distinguished Professor of Management. And yet his most cited paper, from 1979 and coauthored with Michael Harrison, Adams Distinguished Professor of Management, Emeritus (and an emeritus member of the OIT group), is in finance. This highly technical paper addresses issues raised by the famous Black–Scholes Option Pricing formula: In what sense are options priced by arbitrage? What else is priced by arbitrage? (In the Black–Scholes world, pretty much everything.) How do these notions generalize to worlds with, for instance, uncertain future interest rates? This paper was the origin of the term equivalent martingale measure, now a standard term of art in the practice of rocket-science finance.
In a second and very different paper with Harrison, the authors model speculative trade in financial markets. In their model, speculator A buys a security not because he thinks it has good prospects for paying dividends but because he anticipates circumstances in which speculator/investor B will decide the security is valuable. This phenomenon allows “speculative bubbles” to occur, but these are bubbles that never need pop, because they are based on enduring differences of opinion. Nobel Laureate Thomas Sargent describes this as a formal treatment of J. M. Keynes’ discussion of “Beauty Contents” where the idea is not to select from a photo gallery the 10 pictures you think are most beautiful, but the 10 that you think most other people will think that most other people will think … are beautiful. The simple model in this paper led to a large literature on investor overconfidence and speculative bubbles and is often listed as one of the seminal papers in behavioral finance.
When Kreps began his career in the early 1970s, game theory was something of a sideshow within economics. Back then most economists were more interested in anonymous markets than in the strategic interactions between individuals or among firms. The rise of information economics in the 1970s changed that, and economists found that the language and ideas of non-cooperative game theory were just what was needed to “do” information economics formally. Through the late 1970s and into the 1980s, game theory allied to information economics revolutionized economics, and Kreps was one of that revolution’s leaders.
As economists began to speak in the language of game theory, two modeling concepts were central. The first, called games with incomplete information, was a way to model and study situations in which some participants (or players) knew things about which others were uncertain. The second, called perfection, was crucial to the analysis of dynamic games in which one player took an action to which others would respond. The issue here is non-credible threats: If I threaten to blow up a bomb that will hurt both of us if you don’t give me your money, and if you believe me, you’ll give me your money. But the threat is not credible. If you don’t give me your money, will I really set off the bomb? Perfection is how formally to rule out non-credible threats.
When it came to bringing together the two notions of incomplete information and perfection, though, technical difficulties arose. There was a machine for resolving these difficulties — called trembling-hand perfection — but it was impractical to apply in all but the simplest of situations.
This is where Kreps, working with Bob Wilson (yet another Adams Distinguished Professor, Emeritus), enters the story. Kreps and Wilson, in 1982, devised a simpler way to discuss perfection in complex, dynamic games. Previously, when analyzing games, the discussion was (solely) about the strategies chosen by the players. Kreps and Wilson suggested that one should discuss both the players’ strategies and their beliefs about what had happened in the game so far. By working simultaneously with strategies and beliefs, checking for incredible threats was easy. And, to illustrate these ideas, Kreps and Wilson, along with John Roberts (John H. Scully Professor of Economics, Strategic Management, and International Business, Emeritus) and Paul Milgrom (GSB PhD ‘79, and today Shirley R. and Leonard W. Ely Jr. Professor of Humanities and Sciences in the Stanford Department of Economics), contributed the famous Gang of Four paper, which showed how repeated play combined with a little incomplete information could dramatically change how people behave.
But while this gave a machine for analyzing non-credible threats, a new problem arose: non-credible beliefs. For example, my threat to set off the bomb is not credible because it will harm me, but what if you carry a gun and I think there is a small chance that you are the sort of person who might shoot me with it? If I demand your money and I expect you to give it to me, and then you do not, I might reassess as high the probability that you will shoot me if I don’t set off the bomb, which makes my threat to do so entirely credible. This story may seem fanciful, but it describes the strategic situation between certain rival firms.
With In-koo Cho of the University of Illinois, Kreps took on the project of finding criteria for what people “ought to” believe when faced with unexpected circumstances. This spawned a whole literature on belief-based refinements of perfect equilibria.
What is corporate culture in economic terms, and how does a corporate culture help a firm run more efficiently? Kreps was the first economist to tackle these questions.
He pointed out that in any company, situations inevitably come up that can’t possibly be foreseen and that therefore can’t be written into a contract between the company and its employees. Therefore, Kreps argued, only something like a culture—a set of general principles that get passed on throughout the workforce—can guide employees and managers on how to respond appropriately to such situations. As Kreps wrote in his famous paper on this issue, “Corporate Culture and Economic Theory,” corporate culture “says how things are done, and how they are meant to be done in an organization.” Over time, a company develops a reputation for its culture, and individual employees become invested in maintaining both their company’s reputation and their own by adhering to the culture’s principles.
This line of thinking led to a large body of economic research on “relational contracts,” or unwritten codes of conduct that help ensure trustworthy behavior in long-term dealings.
More broadly, Kreps, in collaboration with former GSB colleague James Baron, tackled the question: How should you conceptualize employment relationships on which bear both economic and social-psychological forces? In their book, Strategic Human Resources, written at the level of an MBA textbook, they construct a model of the employment relationship that very naturally weaves together the economic and social character of employment, giving us grist for answering the question: What would have been a better motivational scheme for Wells Fargo?
In all these fields, and in the textbooks he writes, what sets Kreps apart has been his ability to give insight into complex economic phenomena with formal models that are remarkably transparent. In his hands, formal (mathematical) models are not a fog that prevents us from seeing the forest, but instead a tool for seeing through the fog of complexity.
By Marina Krakovsky and Helen Chang