Probit Modeling of Chinese High Stock Dividend and Split Announcements

Summary

This report describes a Probit model for predicting Chinese listed companies likely to announce high stock dividends or share splits. It groups predictors into company characteristics such as share price, total share capital, and listing age; growth measures including retained earnings, capital reserves, cash flow per share, and trailing earnings per share; historical distribution and dividend patterns; and constraints introduced by exchange disclosure rules. The report argues that regulatory changes require corresponding adjustments to the prediction model.

For evidence, it cites out-of-sample hit rates near 80% in selected years when predicted probabilities exceeded 90%, and reports a decline in accuracy in 2016 following policy changes. It also cites an 80% prediction accuracy for a 2017 model portfolio. These are reported results rather than independently documented tests; the source provides no sample construction, benchmark, trading returns, or detailed model specification. It says a 2018 stock selection was based on third-quarter reports and the new rules, but the underlying report and company list are not included here.

Key ideas

• The proposed Probit model combines fundamental, growth, historical distribution, and regulatory predictors.
• Exchange disclosure requirements are treated as model inputs that may change prediction behavior.
• The source reports high out-of-sample classification accuracy in some years, with weaker accuracy after a policy change.
• The document does not provide enough methodological detail to independently assess the reported performance.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.