Tests of the Relations Among Marketwide Factors, Firm-specific Variables, and Stock Returns Using a Conditional Asset Pricing Model

Abstract

In this paper we generalize Harvey’s (1989) empirical specification of conditional asset pricing models to allow for both time-varying covariances between stock returns and marketwide factors and time-varying reward-to-covariabilities. The model is then applied to examine the effects of firm size and book-to-market equity ratios. We find that the traditional asset pricing model with commonly used factors can only explain a small portion of the stock returns predicted by firm size and book-to-market equity ratios. The results indicate that allowing time-varying covariances and time-varying reward-to-covariabilities does little to salvage the traditional asset pricing models. Tests of the Relations Among Marketwide Factors, Firm-specific Variables, and Stock Returns Using a Conditional Asset Pricing Model Asset pricing theory posits that the expected excess return on a financial asset is the (summed) product of the conditional covariance (or beta) of the asset return with marketwide factors and the reward-to-covariabilities (or factor premiums). Models based on this basic relationship are widely used in financial decisions. However, the unconditional versions of the models, especially the Sharpe (1964)-Lintner (1965) Capital Asset Pricing Model (CAPM), are frequently rejected by data and are known to leave some anomalies. Fama and French (1992) conclude that in explaining the cross-section of asset returns, betas are overwhelmed by two firm specific variables: the market value of a firm’s equity (ME), and the book-to-market ratio of equity (BM). The Fama-French result has stimulated many subsequent studies to explain the ME and BM anomaly. In this paper, we investigate whether the cross-sectional explanatory power of ME and BM is consistent with a conditional multi-factor asset pricing model. At the most general level, the asset pricing theory does not specify the functional form of the relation between conditional covariances and conditioning information, nor the functional form of the reward-to-covariability. This relation is empirically determined. Since the ME and BM effects are cross-sectional relations, they are more relevant to the covariance terms in the asset pricing model. Many previous studies (including Fama and French (1992)) assume that covariances, or betas, are constant over a fixed period of time, while others adopt some functional forms. The failure of such constant specifications to explain the ME and BM effects may well be due to the misspecification of the functional form of the conditional covariances. To avoid the consequence of misspecifying conditional covariances, we adopt in this paper an empirical multi-factor model based on Harvey (1989). The main advantage of Harvey’s specification is that it admits a general structure for conditional covariances between stock returns and marketwide

3 Figures and Tables

Cite this paper

@inproceedings{He1996TestsOT, title={Tests of the Relations Among Marketwide Factors, Firm-specific Variables, and Stock Returns Using a Conditional Asset Pricing Model}, author={Jia He and Raymond Kan and Lilian Ng and Chu Zhang and Mark M. Carhart and Kar C. Chan and Kent D. Daniel and Bernard Dumas and Wayne E. Ferson and Campbell R. Harvey and Ravi Jagannathan and George Kirikos and Angelo Melino and Madeline Stead}, year={1996} }