Premium
COULD DYNAMIC BETA MEASURES ENHANCE PERFORMANCE OF CAPITAL‐ASSET‐PRICING MODEL ON FITTING STOCK RETURNS? A REALITY TEST
Author(s) -
LI MINGYUAN LEON
Publication year - 2011
Publication title -
the manchester school
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.361
H-Index - 42
eISSN - 1467-9957
pISSN - 1463-6786
DOI - 10.1111/j.1467-9957.2009.02161.x
Subject(s) - econometrics , economics , capital asset pricing model , autoregressive model , autoregressive conditional heteroskedasticity , stock (firearms) , heteroscedasticity , beta (programming language) , markov chain , financial economics , mathematics , computer science , volatility (finance) , statistics , mechanical engineering , engineering , programming language
Certain dynamic beta measures based on the ARCH (autoregressive conditional heteroskedasticity)/GARCH (generalized ARCH) and Markov‐switching models are adopted and a comparative analysis derived from examining the performance of the capital asset pricing model on fitting stock returns among various dynamic beta models is introduced. Our empirical findings are consistent with the following notions. First, the dynamic beta measures encompassing both the time‐varying technique and state‐varying mechanism are significantly better than the constant beta and the pure time‐varying and state‐varying betas in matching the pattern of stock returns. Second, the misspecified models obtained with inappropriate beta settings could serve as one of the reasons for abnormal returns.