Home Address:
  470 Prospect Street, Apt. 51
  New Haven, CT 06511

Telephone: (203) 376-5679 (cell)

Microeconomic Theory
Game Theory
Behavioral and Experimental Economics
Finance

Microeconomic Theory
Game Theory
Behavioral Economics
Finance

May 2002 (Written) Microeconomic (with distinction) and Macroeconomic Theory
May 2003 (Oral) Microeconomic Theory (with distinction) and Public Finance

Essays on Evolution, Reputation and Rationality

Professor Stephen Morris
Professor Benjamin Polak
Professor Shyam Sunder

May 2006

Ph.D., Economics, Yale University, expected May 2006
M.Phil., Economics, Yale University, May 2004
M.A., Economics, New Economic School, May 2001, diploma cum laude
B.Sc. & M.Sc., Mathematics, Moscow State University, May 1999, diploma cum laude

Yale University Dissertation Fellowship, 2005
Cowles Foundation Prize, 2004
Summer Graduate Fellowship, Summer 2002 and 2003
Four-year Fellowship at Yale Economic Growth Center Prize, 2001
Four-year Fellowship, Department of Economics, Yale University, 2001
Don Patinkin Award for the best Diploma thesis, New Economic School, Moscow, 2001
Prof. Alexandrov Fellowship for outstanding academic performance, Moscow State Univ., 1998

Instructor:
  Introduction to Game Theory, Yale University, Summer 2005
  Introduction to Game Theory, Yale University, Summer 2004
Teaching Assistant:
  Econometrics 2000-2001, New Economic School
  Game Theory, Fall 2003 and Fall 2004, Yale University
  Mathematical Economics, Spring 2004, Yale University
  Financial Markets, Spring 2006 (expected), Yale University
Math and Science Tutor:
  Yale College, 2005-2006, Tutor all undergraduate Economics and Math courses

Energy Economics

General programming: Matlab, Stata
Experiment design: zTree, PHP, HTML

Professor Stephen Morris
Center for Advanced Studies
  in the Behavioral Sciences
75 Alta Road
Stanford, CA 94305
Tel: (650) 321-2052
Fax: (650) 321-2052
Email: smorris@princeton.edu

Professor Nicholas C. Barberis
Professor of Finance
Yale School of Management
137 Prospect Street
PO Box 208200
New Haven, CT 06520-8200
Tel: (203) 436-0777
Fax: (203) 432-6970
Email: nick.barberis@yale.edu

The first chapter of my dissertation analyzes the stock market dynamic under the presence of agents with heterogeneous beliefs and connects it to the recently observed decline of equity premium and assets’ overvaluation. While I use a behavioral approach by considering agents with non-rational expectations, I do not exploit such asymmetric assumptions as optimistic bias or short-sale constraints to obtain my results. The second chapter experimentally analyzes the importance of non-monetary and strategic considerations in subjects’ reasoning. The results show that such recent theories as fairness, altruism, reciprocation, etc. have a rather modest effect and explain less than half of the subjects’ deviation from payoff-maximizing behavior. The third chapter describes the interactions between a borrower (country) and creditors (investors), where the borrower has private information, and the loan sizes are determined endogenously by the competitive credit market. I show that without behaviorally honest types all equilibria but one are inefficient and result in the country’s eventual default. The result helps to explain why developing countries often fail to establish a good reputation in order to attract potential investors.

1. Markets with Heterogeneous Expectations and the Equity Premium Decline

As early as the 1950s, Milton Friedman argued that, since rational agents will consistently make more money than agents with incorrect expectations, the latter will eventually be driven out of the market and thus can be ignored. This served as a justification for considering models where all agents have rational expectations. Recently, however, many authors have argued that noise traders should not be disregarded because it might take a long time before their wealth will become insignificant. Moreover, even with negligible wealth, noise traders can have considerable impact on prices. Following this literature many authors suggested asset market models where agents have heterogeneous non-rational expectations about future returns. Even the simplest deterministic models of this sort were capable of generating a complicated dynamic that resembled the actual stock market and could explain some financial puzzles.

In the first chapter of my dissertation, I also analyze a model in which different types of agents have different non-rational expectations about the future price and cash flows of a risky asset. The distribution of beliefs evolves in such a way that types with higher profit increase their market weight. We could think that either the wealth of successful agents is increasing or that other agents adopt the beliefs of successful types. My model is different from the literature (for example, Brock and Hommes, Econometrica 1997) in that evolution is modeled at the level of the aggregate distribution of beliefs about returns, not at the level of individual learning rules. At this aggregate level I can allow for a very general class of evolution dynamics. Effectively agents can use any individual learning techniques as long as, at the aggregate level, the change in distribution of beliefs satisfies some natural properties such as monotonicity. This change in modeling assumptions produces very different results.

I first show that the equilibrium price converges but that, even in the long-run, heterogeneous beliefs survive. The main result of the paper is that, unlike in earlier papers, the long-run equilibrium price is the risk-neutral fundamental price although all agents are risk-averse. Even when I allow the evolution to depend on the realization of random dividend shocks, the equilibrium price is absorbed with probability one by a symmetric interval around the risk-neutral price. An immediate consequence of this result is that the asset is going to be overvalued from the classical point of view, and its excess return will be low. This result might explain the observed decline of the equity premium in the U.S. stock market and the overvaluation of the assets. In the literature, there are many behavioral theories of asset overvaluation. Unlike most of these, my model provides an explanation of overvaluation that does not depend on ‘asymmetry’ assumptions such as optimistic bias and short-sale constraints.

2. Separating Non-monetary and Strategic Motives in Public Good Games

A well-established finding in experimental economics is that people tend to be considerably more cooperative than individual payoff-maximization would suggest. Recently, behavioral theories have addressed this issue by stressing different factors such as fairness, altruism, reciprocity, etc. In my second essay I try to understand the importance of these factors in people’s reasoning.I divide the theories suggested in the literature into three groups: utility interdependence or UI (concern about other people’s payoffs, e.g., fairness, altruism); action interdependence or AI (subjects want to influence future opponents' decisions or to reciprocate past actions, e.g., reciprocation, encouragement, reputation); and learning. I compare subjects' behavior in three different treatments: a benchmark treatment; a phantom treatment when UI and AI are not applicable; and a two-type treatment when only UI is not applicable.

The benchmark treatment is a standard public-good game, where contributing zero is a dominant strategy. In the phantom treatment, subjects are randomly matched with the decisions that were made in the (separate) benchmark treatment. Since the opponents do not get any payoff and their actions cannot be influenced, it removes both UI and AI considerations from the subjects’ behavior. In the two-type treatments, one type of subjects gets a fixed payment regardless of the outcome, and the second type has a standard payoff function, with only the opponent's type being known. Thus, when subjects are matched with a person who gets fixed payoff they do not have UI considerations. The advantage of the suggested treatments is that they do not change main aspects of the game such as strategic uncertainty, information and payoff structures. The main result is that non-monetary and strategic considerations have a rather modest effect. They explain less than half of the over-contribution. Specifically, the removal of both AI and UI factors decreases the over-contribution by 40% and the removal of UI considerations decreases the over-contribution by 25% as compared to the benchmark. Something else, perhaps a lack of understanding of the optimal strategy, must explain most of the over-contribution.

3. Reputation dynamics in Credit Market

In the third essay, I use a game-theoretical approach to investigate why countries, governments or local authorities often fail to create a good reputation in order to attract potential investors. I model the interactions between investors and a country as an infinite partnership game where investors are uncertain about the country's patience and the level of investment is set endogenously based on the investors’ information and confidence about the country. The motivation of the players is standard: the country wants to maximize its discounted wealth by choosing the best time (if ever) to default. The investors are competitive and risk-neutral. In the equilibrium they correctly estimate the probability of confiscation, and provide such an amount of investments that their expected rate of return is equal to the riskless interest rate.

I show that there are three types of equilibria depending on the initial confidence level. There is a unique efficient equilibrium when the initial confidence is high. In that case the investment risk and the effect of incomplete information are asymptotically eliminated, and more patient types will enjoy the efficient level of investments. For the continuum of intermediate values of initial confidence, the level of investments grows at first but after some point it starts to decline and all types will eventually default. For low initial values, the investments decline from the beginning and again all types will eventually default. The possibility of inefficient outcomes arises from the fact that all types in the model are opportunistically rational. As the game progresses, it is revealed that the country’s type is more and more patient. However, this knowledge per se does not decrease the risk of investment because, even though the country is patient, it still might prefer to default depending on the path of future payoffs. Consequently, investors might choose low investments despite their information about the country’s patience. The model might explain, for example, the Russian case, in which confidence and foreign investment levels first rose and then fell leading eventually, to default (in 1998) and to alleged asset confiscation (in 2003).