Three essays on asset pricing from the macroeconomic perspective
Student thesis: Doctoral Thesis
Related Research Unit(s)
The thesis consists of three essays, which investigate the topic of asset pricing from the macroeconomic perspective based on different kinds of econometric methodologies. In the first essay, we try to connect the financial market and macroeconomic data to various macro-based asset pricing model specifications and test the models' reliability and capability. Four groups of macro-based asset pricing models are developed and estimated by GMM. We compare their capacities in accounting for the asset market dynamics in Hong Kong, including Consumption-based models (CCAPM, Habit Formation Model and Recursive Utility Model), their Housing-augmented counterparts, Labor Income Model and Collateral Constraint Model. Two model comparison methods are used: the traditional RMSE/MAE to analyze the log-linear form of the first order conditions derived from the models and Hansen-Jagannathan Distance which preserves structural characteristics of Euler equations. The empirical results show that introducing housing into the Consumption-based models usually improves the models' performance; Recursive Utility and its housing-augmented version perform the best in forecasting asset prices in the reduced-form while the Labor Income Model produces the least Euler Equation Error; and Labor Income and Collateral Constraint are indeed relevant in explaining asset returns. Since the Recursive Utility Model is proved to be the best model form to predict asset prices in the first essay, the second essay extents the Recursive Utility Model by means of including the labor market search and matching into it. Thus this model tries to connect financial asset prices with labor market phenomenon and it naturally generates a new macroeconomic pricing factor which is "job market tightness" to predict the aggregate stock price. We use Bayesian estimation method which is among the first trying for this kind of model set up. Our posterior estimates show that: the Bayesian estimation method can generate the parameters which are generally consistent with the prior parameter values based on the consensus in the macro literature or data statistics; it also provides the estimates for the parameters which lack of literature consensus. Moreover, in order to assess the performance of this new pricing factor, we also compare this model with the Recursive Utility Model without labor market search and matching, by means of Bayesian posterior odds ratio analysis. The third essay investigates the asset prices from the perspective of international financial market transmissions, as global financial market integration makes the asset prices not only determined by domestic macro-fundamentals but also by international financial market shocks. It tries to analyze the pattern and the mechanism of international financial market transmissions between US and China. We focus on three markets: long-yield bond, stock and exchange rate market. We use Structural-VAR model to capture the contemporaneous spill-over effects based on a comparatively new identification scheme "sign restrictions" and utilize Bayesian estimation. The estimation results show that three spill-over patterns between these two countries' financial markets are consistent to our hypotheses such as "portfolio rebalancing", "investor's expectation", "forward premium puzzle" as well as "global financial cycle" theory.
- Capital assets pricing model