JOB MARKET PAPER
Stock Prices, Risk-free Rate and Learning
The co-movement between stock and short-term bond markets in US data appears weak in terms of the correlation between stock price-dividend ratio and risk-free rate and the variance decomposition of stock excess returns. It is essential to market participants and policy makers to understand the lack of empirical relationship, especially in light of the fact that several rational expectation asset pricing models that match stock market volatility actually imply a much stronger relationship between stock and short-term bond markets than empirically observed. To explain this apparent inconsistency, this paper presents a small open economy model with "Internally Rational" agents, who optimally update their subjective beliefs on stock prices given their own model. Compared with risk-free rate's variation, agents' subjective beliefs are essential in generating stock price volatility. When testing our model using the method of simulated moments, quantitatively it can simultaneously match moments of the stock and bond markets as well as the weak co-movement between two markets.
Key Words: stock price, risk-free rate, learning, correlation, variance decomposition
JEL Class. No.: G12, E44, D84
Understanding AH Premium in China Stock Markets (with Renbin Zhang)
There are 88 companies (AH share) dual-listed in both China mainland stock markets (A share) and Hong Kong stock market (H share) accounted for 20% of total A share. The ‘Shanghai-Hong Kong Stock Connect’ program starting at November, 2014 makes previously two segmented markets--Shanghai and Hong Kong stock markets--connected. The prices difference of AH share in Shanghai and Hong Kong stock markets, measured by Hang Seng AH Premium Index, instead of converging persistently divergences, and even reaches 50% higher in Shanghai market. We have shown that asset pricing models with heterogeneity agents with different risk aversions or diverse beliefs in the complete market and incomplete markets cannot generate any AH premium. Transaction cost and different dividend taxes between Shanghai and Hong Kong markets also fails to explain such high and volatile AH premium. We, hence, propose an ‘Internal Rationality’ learning model, in which agents don’t know the pricing function from fundamentals to the stock prices and have different subjective beliefs about tomorrow’s capital gains in Shanghai and Hong Kong markets. Our learning model can successfully generate data-like weekly AH premium. We also show that convergence traders with strategy short in Shanghai and long in Hong Kong will lose money with 33% probability.
Key Words: AH Premium, Shanghai-Hong Kong Stock Connect, Heterogeneous Agents Asset Pricing Model
JEL Class. No.: G12, G14
Working in Progress
Exchange Rate Puzzles and Learning (with Jian Wang and Jianfeng Yu)