Financial Economics
Empirical Macroeconomics
Applied Econometrics

Financial Economics
Investment Management
Fixed Income Security Analysis
International Finance
Econometrics

May 1999 (Oral) Financial Economics and Econometrics (with Distinction)
May 1998 (Written) Macroeconomics and Microeconomics Theory

Essays on Credit Risk, Interest Rate Risk and Macroeconomic Risk

Professor Robert J. Shiller
Professor Stefano G. Athanasoulis
Professor William C. Brainard
Professor Andrew M. Jeffrey

May 2003

Ph.D., Economics, Yale University (expected May 2003)
M.Phil., Economics, Yale University, 2000
M.A., Economics, Yale University, 1999
B.A., summa cum laude, School of Economics, Fudan University, China, 1996

Yale University
   Departmental Graduate Student Fellowship, 2002, 1997-1999
   Yale University Dissertation Fellowship, 2001
   Yale University Fellowship, 1997-2001
   Cowles Foundation Graduate Student Fellowship, 1999
Fudan University
   Special Merit Fellowship, 1994
   Dean’s scholarship (first class), 1992-1996

School of Management, Yale University
   Financial Economics (Doctoral Course), Teaching Assistant to Professor Hua He, 2000

Economics Department, Yale University
   Theory of Income Determination and Monetary and Fiscal Policy, Teaching Assistant to Professor
   George Hall, 2003
   Poverty under Post-Industrial Capitalism, Head Teaching Assistant to Professor Gerald Jaynes,
   2001, 2002
   Introductory Macroeconomics, Teaching Assistant to Professor William Nordhaus, 2000
   Introduction to Probability and Statistics, Teaching Assistant to Professor Donald Andrews, 1999

Geneva Executive Course, FAME, Switzerland
   Advanced Mathematics of Derivative Products, Teaching Assistant to Professor Salih Neftci, 2001

Executive course, ISMA Center, University of Reading, UK
   Credit Risk Management, Lecturer, summer 1999

Doctoral Research Fellow, International Center for Finance, Yale University, 2002
Research Assistant, Professor Salih Neftci, City University of New York, fall 2001, on the project, "The Notion of Correlation Across Credit Events: Some New Tools"
Research Assistant, Professor Robert Shiller, Yale University, summer 1999, on the book, "Irrational Exuberance"

"Integrating Market Risk and Credit Risk: A Dynamic Asset Allocation Perspective", 2002, manuscript, Yale University [job market paper]

"Hedging with Smile: An Empirical Analysis of the USD Swaption Markets", in progress

"Is Volatility Risk Priced in Swaption Markets?", in progress

"Optimal Investment with Default Risk" (with Xiangrong Jin), 2002, FAME research paper No. 46, Switzerland

"Testing the CAPM by a Synthetic Return on GDP as the Market Return", 2001, manuscript, Yale University

"Exchange Rate Determination: An Empirical Test", 1999, manuscript, Yale University

"A Note on Solving the Riskfree Rate Puzzle: Evidence from a Psychological Model", 1998, manuscript, Yale University

"An Analysis of the Dynamics of European Investment in China", 1996, Study of Multinational Economy, No. 8, China

American Finance Association Conference, Washington, D.C., 2003 (paper accepted)
Quantitative Methods in Finance Conference, Sydney, 2002 (paper accepted)
Inter-University Graduate Student Conference, College Park, Maryland, 2002
German Finance Association Conference, Cologne, 2002
The European Investment Review Annual Conference, London, 2002
Credit Risk Summit Europe, London, 2001

American Finance Association
Financial Management Association

This dissertation investigates credit risk, interest rate risk and macroeconomic risk in the context of asset pricing and asset allocation. The first chapter studies how the integration of credit risk and interest rate risk affects investor’s asset allocation. Credit risk, in this dissertation, mainly refers to the risk that an obligor fails to repay its debt. It is an important source of risk to investors in financial markets. Yet this issue has received little attention in the theoretical asset allocation literature. Based on recent theoretical results on credit-sensitive bond pricing, I study the investors’ behavior facing credit risk in a formal way. Investors in the model dynamically allocate their wealth across corporate bonds, Treasury bonds and equity in order to maximize utility. This chapter shows investors can gain sizable welfare improvement from investing in credit markets. The second chapter focuses on interest rate modeling from the perspective of hedging interest rate derivatives when implied volatilities across strike prices are non-flat (the so-called "volatility smiles"). Several recent extensions to the Libor Market Model, popular among market practitioners, have been advanced to cope with volatility smiles but their relative hedging performances are unknown. This chapter fills the gap by evaluating these models with a comprehensive European swaption dataset. The third chapter revisits the Capital Asset Pricing Model (CAPM) testing. Roll (1977) points out that the main difficulty in testing the CAPM is to find a suitable market proxy. This chapter attempts to answer the Roll’s critique by using a hypothetical aggregate portfolio that uses GDP flow as its dividend. The empirical results also contribute to the evaluation of US production efficiency in the mean-variance sense. The findings of this dissertation are shown to have important implications for portfolio choice, risk management and asset pricing.

The first chapter investigates how credit risk and market risk (interest rate risk in this chapter) affect investors’ portfolio choice and how the two types of risk interact in this context. Credit markets develop at a rapid pace during the past decade and provide a distinct risk-return profile to investors. This chapter formally incorporates credit risk in the asset allocation framework. The instrument that bears credit risk in this model is assumed to be a defaultable zero-coupon bond. Should default occur, the amount recovered from the defaulted bond is assumed to be proportional to its market value prior to default. The credit spread, which is defined as the products of the loss rate and the hazard rate, is assumed to follow a mean-reverting process. Investors maximize utility by dynamically allocating their wealth across corporate bonds, Treasury bonds, equity and a money market account. I obtain a closed-form solution to this investment problem, which enables me to analyze the impact on investors’ decisions of various risk parameters. One interesting insight is that a non-zero recovery rate of the credit-risky bond affects investors' decision in a fundamental way. This is manifested in a dividend-like adjustment term in the drift of the stochastic differential equation (SDE) of the defaultable zero-coupon bond’s return process. Based on the valuable information in credit spread, investors in this model attempt to time the market conditions in their decision making process. The optimal asset allocation involves the "separation effect" and "integration effect". The separation effect means that the optimal holding of each asset category is intimately linked to the main risk inherent in that asset. The integration effect refers to the finding that the summation of the optimal demand for the Treasury bond and that for the defaultable bond contains no hedging term against credit risk, even though the individual optimal demands do. As a result, the relation between myopic demands for bonds and market prices of risk becomes complicated compared with that in the traditional setup. In addition, I show the cross-markets correlation is a potentially important factor in the asset allocation decision. In particular, it affects investors’ ability to hedge against or speculate on the stochastic risk premium of the defaultable bond. Numerical examples show that the inclusion of credit markets significantly enhances investors' welfare. The gains measured by the annualized rate of return in certainty equivalent wealth with credit markets in the investment opportunity set against that without credit markets is between 1.6% and 2% for reasonable parameter values. Risk premium and market correlation are found to be the main sources for the welfare improvement.

The second chapter (in progress) focuses on interest rate modeling in the context of interest rate derivatives pricing and hedging. The Libor Market Model (LMM) justifies the market practice of using Black (1976) formula in pricing interest rate options. Due to its theoretical merits and easy implementation, the LMM has become very popular recently. Yet this model in its original form neglects the volatility smiles commonly observed in the interest rate derivatives markets. The existence of volatility smiles means that the distribution of interest rates implied in the derivatives markets has fatter tails than the lognormal distribution assumed in the LMM. This has profound implications for risk management and asset pricing (for example, pricing of exotic derivatives). Several recent extensions to the LMM have been designed to cope with the existence of volatility smiles in the interest rate derivatives markets, but their relative performances are still unknown. I evaluate the hedging performances of these models with a European swaption dataset. Moreover, given the growing influence of the stochastic volatility models, I compare a stochastic LMM with deterministic extensions such as the constant elasticity volatility (CEV) model and the mixture of lognormal model. This research is one of the first to examine the popular market models of Libor rates with swaption smiles data.

The third chapter revisits the testing of the CAPM model. Testing the CAPM is susceptible to many difficulties. This is highlighted in the Roll’s critique. In particular, because the true market portfolio is unobservable, testing the CAPM is difficult. The third chapter tackles this problem by carrying out the CAPM testing by using a hypothetical aggregate portfolio that uses GDP as its dividend. Since GDP reflects a nation’s production capacity (including tangible and non-tangible components), this aggregate portfolio constitutes a broad, legitimate market portfolio proxy. Though this aggregate portfolio is not observable, GDP flows are. Using the constant expected rate of return model, I can work back from the GDP data and test the CAPM without the need to explicitly calculate the market portfolio. As a result, I show in this chapter that the CAPM is indeed testable, thus addressing Roll’s critique. The linearity hypothesis, the Black-version CAPM hypothesis and the mean-variance efficiency of the aggregate portfolio hypothesis are tested with US data. The testing results show that the linearity hypothesis cannot be rejected, that the Black-version CAPM is econometrically viable, and that the hypothetical aggregate portfolio is mean-variance efficient most of the time within the sample period.