Financial Economics
Applied Econometrics
Empirical Macroeconomics

Financial Economics
Investment Analysis
Econometrics
Macroeconomics
Microeconomics

October 2000, Financial Economics (Primary Oral)
May 2000, Econometrics (Secondary Oral)
May 1999, Macroeconomic and Microeconomic Theory (Written)

Essays on Two Financial Market Anomalies

Professor William C. Brainard
Professor Robert J. Shiller
Professor Zhiwu Chen

May 2004

Ph.D., Economics, Yale University, May 2004, expected
M.Phil., Economics, Yale University, May 2002
M.A., Economics, Yale University, May 2000
B.A., summa cum laude, Economics, Beijing University, July 1995

Yale University
     Summer Research Fellowship, 2002
     Dissertation Fellowship, 2002
     Yale University Fellowship, 1998-2002
Beijing University
     University Scholarship, 1991-1998

Teaching Assistant for Professor J. Geanakoplos: "Financial Theory", Fall 2000 and Fall 2001
Teaching Assistant for J. Hunt / P. Goldberg: "The Theory of Resource Allocation and Its Applications", Spring
   2001 and Spring 2002

Research Assistant for Professor Zhiwu Chen, Yale School of Management, Fall 2000

"Stock Price Interactions and Momentum: A Model with Borrowing Constraints", Yale University, January 2003 [job market paper]

"An Asset Allocation Sub Puzzle", Yale University, July 2003

"Within-Industry Momentum and Its Implications", Yale University, April 2001

"Informational Role of Trading Volume in Money Flow Index", Yale University, in progress

"Relative Strength Factor in Asset Pricing", Yale University, in progress

"International Trade: Theory and Applications", Master Thesis, Beijing University, July 1998

My dissertation documents new evidence related to two financial market anomalies and proposes models to explain them. The first anomaly relates to the apparent momentum in stock prices, i.e. stocks with higher past returns continue to outperform those with lower past returns in a medium-term time horizon; the second anomaly concerns violations of the mutual fund separation theorem.

Chapter I: Within-Industry Momentum and Its Implications

This chapter provides empirical evidence of within industry momentum, i.e., the zero-cost strategy to long past winners and short past losers within an industry can generate profits. Contrary to the conclusion in Moskowitz and Grinblatt (1999) that momentum should be negligible within industries, I find that momentum exists in most industries for most ranking/holding investment strategies, whether using the 20-industry classification of Moskowitz and Grinblatt (1999) or the 48-industry classification of Fama and French (1997). I also find that such within-industry momentum varies significantly across industries. Specially, a zero-cost within-industry momentum portfolio can generate a per dollar long monthly profit ranging from 0.03% to 0.85% in the 48-industry classification. To explain such differences, I examine the relationship between momentum profit and industry concentration. Using the Herfindahl index (which measures the sales concentration of an industry) as a proxy for industry concentration level, I find a positive relationship between these two factors, i.e. industries with higher concentration levels tend to generate higher momentum. The findings in this chapter provide new criteria for testing different momentum models and for identifying the source of momentum.

Chapter II: Stock Price Interactions and Momentum: A Model with Borrowing Constraints

This chapter provides a new explanation of momentum based on the following two observations. First, there are systematic interactions between prices of different stocks, which few asset pricing theories model. One of the main sources of such interactions is that many investors face borrowing constraints. In a two-stock setting, for example, when investors believe that one stock is more promising than the other, they buy the former and, under borrowing constraints, simultaneously sell the other. The returns of different stocks can be negatively correlated, although their underlying businesses may be unrelated. Secondly, momentum essentially reflects relative performances between stock returns. We need to study cross-correlations between stock prices in addition to auto-correlations of a single stock.

In this chapter, I develop a multi-asset Rational Expectations Equilibrium (REE) model in a noisy trading environment, where investors are heterogeneously informed and some investors have borrowing constraints. Momentum is generated by gradual diffusion of information and investors’ borrowing constraints. Cross-correlation as well as auto-correlation relation are derived. I further use this model to explain the within-industry momentum documented in Chapter I. The model suggests that when many investors use the same investment scope, such as industry and value/growth, they will buy better stocks and sell others under borrowing constraints. The aggregation effect will lead to momentum within such a category. Moreover, I test an empirical implication of the model, the existence of a positive correlation between momentum profits and macro economy. When economy is improving, investors tend to buy more stocks. There should be higher momentum profits since more investors face borrowing constraints. The result agrees with the model prediction.

This stock price interaction also has interesting implications for asset pricing. I propose a new factor, namely Relative Strength, in factor models. In this model, one stock’s Relative Strength may change with other stocks’ performance. Consequently, a stock’s return could increase or decrease as its relative strength changes, though its own factors such as beta, book to market value, and size remain the same.

Chapter III: An Asset Allocation Sub-Puzzle

The third chapter presents new facts related to the asset allocation puzzle of Canner, Mankiw, and Weil (1997) (CMW) by documenting a sub-puzzle in asset allocation: ratio of high-risk stocks to low-risk stocks changes with an investor’s risk attitude. CMW point out that empirically the ratio of bond to stock holdings of a typical investor depends upon her risk aversion: financial advisors generally recommend a lower ratio of bonds to stocks for more aggressive investors. This observation directly challenges the well-known mutual fund separation theorem of Tobin (1958) and Markowitz (1952). This chapter further examines the composition of the recommended stock portfolio. My result also contradicts the mutual fund separation theorem and its pattern is similar to that in CMW: a lower ratio of low-risk stocks to high-risk stocks for more aggressive investors. This finding is important because most current literature attempts to explain the asset allocation puzzle by studying optimal bonds to stocks ratio under different interest rate model settings. The new finding suggests that the puzzle involves more than just interest rates, and needs a more fundamental study from the perspective of an investor’s risk attitude. Drawing on the loss aversion literature, I propose a general model that can solve both the CMW puzzle and the sub-puzzle. The key is that investors are loss averse and care about their expected worst-case returns. When an expected worst-return constraint is added to investors’ utility maximization, the optimal ratio of low-risk to high-risk assets, be it bonds to stocks or low-risk stocks to high-risk stocks, varies with investors’ risk attitudes.