Microeconomic Theory
Game Theory
Financial Economics

Microeconomic Theory
Game Theory
Financial Economics
Industrial Organization (undergraduate)

May 1999 (Oral) Microeconomic Theory (primary field, with distinction); Financial Economics (secondary field, with distinction)
May 1998 (Written) Microeconomics and Macroeconomics

Dynamic Coordination Games: Theory and Applications

Professor Stephen Morris
Professor Benjamin Polak
Professor Dirk Bergemann

December 2001

Ph.D., Economics, Yale University, 1997-2001 (to be awarded December 2001)
M.Phil., Economics, Yale University, 1998-2000
M.A., Economics, Yale University, 1997-1998
B.A., summa cum laude, Mathematics & Economics, Franklin and Marshall College, 1992-1996

Northwestern University
  Postdoctoral Fellowship, Center for Mathematical Studies in Economics and

   Management Science, 2001-02
Yale University
     Robert M. Leylan Dissertation Fellowship 2000-2001
     Cowles Foundation Prize 1999, 2000
     John F. Enders Fellowship 2000
     Yale University Fellowship 1997-2000
Franklin and Marshall College
     The Williamson Medal (valedictorian) 1996
     Phi Beta Kappa 1995
The American Mathematical Society
     Award for best undergraduate research presented at the AMS National Meetings, January 1996

Instructor
     Undergraduate Industrial Economics, Northwestern University, Spring 2002
     Graduate Review Course in Mathematics for Economists, Yale University, August 2000
Teaching Assistant
     Graduate Summer Course in Optimization Theory, Yale University, Summer 2000
     Graduate Advanced Microeconomic Theory, Yale University, Spring 2000
     Undergraduate Economic History, Yale University, Fall 1999

"Coordination, Learning, and Delay," Yale University, September 2001. (Job-market paper)

"Financial Contagion through Capital Connections: A Model of the Origin and Spread of Bank Panics," Yale University, August 2001.

"Does One Soros Make a Difference? A Theory of Currency Crises with Large and Small Traders," with Giancarlo Corsetti, Stephen Morris, and Hyun Shin, Yale University, August 2000. Revised and resubmitted, The Review of Economic Studies.

"Regionality Revisited: An Examination of the Direction and Spread of Currency Crises," Yale University, January 2000. Under revision for The Journal of International Money and Finance.

"Social Learning with Payoff Complementarities," Yale University, November 1999.

"Social Learning with Imperfect Observation and Payoff Complementarities," with Shachar Kariv (NYU), in progress.

"An Intersection Property of Sylow 2-Subgroups in Non-solvable Groups," Mathematical Proceedings of the Cambridge Philosophical Society, 1997:122, pp. 261-8, with Arnold Feldman

"A Model of a Kuiper Belt Small Grain Population and Resulting Far-Infrared Emission," The Astrophysical Journal, 1995:450, pp. L35-L38, with Dana Backman and Robert Stencel.

Visiting Scholar, Federal Reserve Bank of Minneapolis Research Department, February 2001
Visiting Scholar, The Econometric Institute, Erasmus University, Rotterdam, March 2001

J.P. Morgan & Company, Inc., New York, NY  1996-97
Financial Analysis, Emerging Markets Research and Fixed Income Research

Econometric Society Summer Meetings, College Park, Maryland, June 21, 2001
8^th World Congress of the Econometric Society, Seattle, Washington, August 15, 2000
Inter-university Graduate Student Conference, New York University, May 26, 2000

This dissertation extends the theory of dynamic coordination games with social learning. It then applies this theory to currency crises and bank panics: settings characterized by private information, strategic complementarities, and multiple periods with observable actions. Academic interest in such "self-fulfilling" phenomena dates back at least to Keynes. The modern literature has given rise to at least two distinct approaches to these problems. The first (e.g. Diamond and Dybvig 1983, Obstfeld 1986) focuses on coordination problems, but ignores the dynamics and social learning inherent in the original setting. The second (e.g. Banerjee 1992 and Bikhchandani, Hirshleifer, and Welch 1992) emphasizes dynamics and social learning, but ignores strategic elements by suppressing payoff externalities. In contrast, I model both the learning (backward-looking) and strategic (forward-looking) aspects of these problems together, and examine how they interact. To do this, I apply the new ‘global games’ techniques of Carlsson and van Damme (1993) and Morris and Shin (2000). My work extends their techniques to dynamic settings. Unlike previous approaches to equilibrium selection in dynamic coordination games (e.g. Frankel and Pauzner 2000, Burdzy, Frankel, and Pauzner 2001) my work incorporates Bayesian learning.

The first chapter, Coordination, Learning, and Delay, studies how the introduction of social learning with costs to delay affects the equilibrium set of coordination games with incomplete information. A continuum of agents chooses whether and when to invest in a risky project. The project succeeds if a sufficiently large proportion of agents participate, given the state of the world. All agents receive private signals about the state. In addition, those who invest late can noisily observe the proportion of early investors. When the project succeeds, it pays a higher return to those who invest earlier. Hence, there is a cost to delay. I show that this game has a unique monotone equilibrium. This is true regardless of whether the order of actions is exogenously specified or endogenously chosen. One can, therefore, compare the endogenous order game with the exogenous order game and with the limiting static games. Welfare is highest in the endogenous order game. In this game, agents with high private signals invest early, and those with intermediate signals invest late. Those with very low signals do not invest at all. Endogenous ordering leads to strict improvement in the efficiency with which agents can coordinate relative to any game where the order of actions is predetermined. Further, coordination is most efficient at intermediate costs to delay. These results have implications for the initial public offering of debt, and for the adoption of new technology under incomplete information.

The second chapter, Social Learning with Payoff Complementarities, studies a related problem. Unlike chapter 1, players are discrete. They enter in exogenous order to play a stag-hunt game with private information and observable actions. The use of discrete players enables me to incorporate a strategic element missing from chapter 1. Discrete agents, knowing that their actions will be observed by successors, take this into account in making their decisions: they signal. This set-up includes the canonical herding model as a special case when the payoff externality is eliminated. I demonstrate the conditions under which herds and cascades may arise, and characterize the informational requirements for coordinated risk-taking in games with finite but unboundedly large numbers of players.

The third chapter, Does One Soros Make a Difference? A Theory of Currency Crises with Large and Small Traders (with Giancarlo Corsetti, Stephen Morris, and Hyun Shin) addressed the following question: To what extent do large investors increase the vulnerability of a country to speculative attacks in foreign exchange markets? We build a model of currency crises where a single large investor and a continuum of small investors decide whether to attack a currency based on their private information about fundamentals. We examine a static version of the model in which all traders act together, and a dynamic version in which the large trader can signal to small traders using her visible short position in the currency. Even without signaling, the presence of the large investor makes all other traders more aggressive in their selling. The difference, however, depends on the relative precision of the small and large investors’ information. Adding signaling makes the influence of the large trader on small traders’ behavior stronger. In particular when the large trader is better informed than the small traders, she can completely solve the coordination problem in the market. Small investors exhibit herd behavior: they attack if and only if the large trader attacks.

The fourth chapter, Financial Contagion through Capital Connections: A Model of the Origin and Spread of Bank Panics, examines the way in which a crisis in one financial institution may affect another. Leading examples of such financial contagion were the widespread bank panics in the U.S. in the late 19^th and early 20^th centuries. I model financial contagion as an equilibrium phenomenon in a noisy dynamic coordination game with multiple banks. The probability of bank failure is endogenously determined. Bank cross-hold deposits due to the imperfect correlation of regional liquidity shocks. It is these cross-holdings that then help bank failures spread. I show that contagion occurs in the unique monotone equilibrium of the economy. In addition, I identify a direction for contagion: from debtors to creditors. This fits the pattern of bank panics in the U.S. under the National Banking System (Sprague 1910, Wicker 2000): panics typically spread from debtor banks in New York to creditor banks in the interior of the country. Simulations identify the optimal level of inter-bank deposit holdings. They suggest that when the probability of bank failure is low, high levels of inter-bank holdings are optimal. When the cross-holding of deposits is complete, the intensity of contagion increases in the size of regional liquidity shocks.

In Regionality Revisited: An Examination of the Direction of Spread of Currency Crises, I empirically analyze the apparent geographical clustering of financial panics. My study extends Glick and Rose’s (1999) analysis of this question. Does geographical proximity, trade competition, or macroeconomic similarity, drive the spread of financial panics? I use recent techniques in the Bayesian analysis of binary data introduced by Albert and Chibb (1993) to examine this question. These techniques allow me to overcome small-sample biases and to meaningfully collate lessons across waves of crises. My findings indicate that trade competition and macroeconomic similarity may drive the spread of currency crises. Geographical proximity does not appear to be important.