Fundamental Analysis of Chinese Stock Market

Abstract

I generate a fundamental signal library with more than 8000 fundamental signals by considering various combinations of the accounting variables in Chinese stock market. I take two standard approaches, time-series intercept tests and cross-sectional coefficient tests, to identify anomalies from this signal library. I find that 142 signals can pass both of the two tests even after accounting for sample variation with bootstrap. I also take several aggregation techniques to extract information from these 142 signals and find that PCA (Principal Component Analysis) performs best. Furthermore, I construct a new factor based on the 142 signals and augment the Fama-French three-factor model to form a four-factor model (A4). The four-factor model performs better than Fama-French three-factor model, Carhart four-factor model, Q4 factor model, Fama-French five-factor model and at least as well as Fama-French six-factor model in terms of accomodating the hedge portfolio returns of over 8000 fundamental signals.

Keywords: Data-Mining, Anomalies, Cross-Section of Returns, Chinese Stock Market

JEL Classification: G10