Forecasting Asymmetries in Stock Returns: Evidence from Higher Moments and Conditional Densities
Year: 2002
Type: preprint
Abstract: Chen, Hong, and Stein's (2001, JFE) time series test on Hong-Stein's (2002, RFS) Differences-of-Opinion model is not able to generate convincing results because of the lack of statistical power. This paper proposes a model that can be utilized to conduct this time series test with sufficient statistical power. Unlike Chen, Hong, and Stein, who use six-month sample skewness of daily returns as the estimate of the population skewness, this paper uses daily conditional skewness to estimate population skewness. An autoregressive conditional density model with an asymmetric Generalized t distribution is used to estimate the time-varying daily skewness. By modeling the distribution parameters as functions of prior return and turnover, we are able to estimate the time-varying effects of these two variables on the skewness. Different from the cross-sectional results in Chen, Hong, and Stein, in which the effect of prior turnover on skewness is found to be a negative constant and therefore supports the Hong-Stein model, this paper provides time series evidence that cannot be explained by the Hong-Stein model: If the prior return or the current conditional skewness is positive, the prior turnover cannot have a negative effect on skewness. Therefore, the implication of the Hong-Stein model that negative skewness is more pronounced if prior turnover is higher can possibly be true only if the prior return is negative and, conditional on this negative prior return, the conditional skewness is also negative.
Source: SSRN Electronic Journal
Authors: C. James Hueng, Robert Brooks
Institutions: Western Michigan University, University of Alabama
Language: English
Subfield: Finance
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W3136928573
