Abstract

The dissertation consists of three essays that address both the theoretical and empirical aspects of characteristics-based asset pricing models. In the first essay, we reformulate a characteristics-based model to demonstrate why firm characteristics explain cross-sectional expected returns. The model is based on an economic setting where the fully-rational group of investors adopts contrarian strategies against the quasi-rational group of investors. The key result is a parsimonious cross-sectional equation that is not only specified by the risk-return relationship, but is also determined by both market-wide and firm-specific adjustments. We offer consistent explanations for the behaviors of growth and value stocks, and also for the prominent cross-sectional patterns such as book-to-market, earnings-to-price, and size effects. In the second essay, we reformulate an asset pricing model where liquidity is an endogenous determinant of expected returns. The key result is that a firm's expected return can be explained by three components: an interest rate term that includes a market-average expected liquidity, a market risk term determined by a weighted average consumption beta, and a firm-specific term determined by a linear deviation of the firm's expected liquidity from that of the market portfolio. We test various empirical implications derived from the theory and find that the expected liquidity effect and the size effect are significant, but the risk-return relationship is flat in the Canadian market. In the third essay, we propose a characteristics-based asset-pricing model from an ex post perspective. We examine the widely used empirical procedure that groups stocks into portfolios by sorting firm characteristics, showing that the exhibited systematic patterns may be largely due to the way of forming portfolios. We design a new portfolio approach and perform robustness tests for the cross-sectional relationships between risk, liquidity, and returns using Canadian stock market data. We find a strong liquidity-return relationship and a significant risk-return relationship when conditioning on realized returns. Both the risk effect and the liquidity effect are highly robust across different portfolio formations.