Abstract:

With the application of big data, attention has been paid in recent years to the role of unstructured data such as investors' language for asset pricing. We use the most comprehensive online stock bar text data to measure investor sentiment and use it to extend and modify the Chinese three-factor model. The modified four-factor model is subjected to Fama-Macbeth regression tests and GRS regression tests, and the following results are obtained: (1) At least half of the text-based proxies for investor sentiment are significant in the regressions of 25 portfolios. (2) In the Fama-Macbeth regression test, E/P is found to be more significant than B/M. (3) In the GRS regression test, the modified four-factor model is found to pass the GRS test for Size and investor sentiment dimensions. Therefore, we conclude that (1) The textual investor sentiment can be a useful factor in China. (2) E/P is more suitable as a value factor than B/M. (3) The textual investor sentiment factor improves the multifactor model's explanatory strength for stock portfolios' excess returns and is only valid for extreme portfolios.

Keywords:

Asset pricing; CAPM; Chinese three-factor model; investor sentiment; textual investor sentiment.

How to Cite:

Ziwei Li and Yuan Wu. Stock Pricing with Textual Investor Sentiment: Evidence from Chinese Stock Markets. [ref]: vol.21.2023. available at: https://refpress.org/ref-vol21-a196/