Zirui Hu

Based on mergers and acquisitions data from Chinese firms listed on the Shanghai and Shenzhen stock exchanges spanning 2011 to 2021, we construct a dataset with 14 feature variables, covering deal characteristics, firm financial and non-financial traits, as well as external environmental factors. Employing machine learning techniques and Shapley Additive explanations value interpretation, we access the contribution of these variables for explaining the success or failure of M&A transactions. Our results highlight the superior performance of machine learning models, particularly the Gradient Boosting model and Random Forest model, compared to the traditional logistic model in predicting the success of M&A transactions. This holds true in both out-of-sample and out-of-time testing scenarios. Deal characteristics, especially the use of stock payments, emerge as pivotal determinants affecting M&A outcomes in the Chinese capital market, with a higher vulnerability to failure. The complexities arising from a politically uncertain environment magnify the intricacies of transactions, thereby increasing the likelihood of M&A failures. Interestingly, our analysis reveals the insignificance of market response in predicting the ultimate success of announced transactions, aligning with the Efficient Market Hypothesis. Furthermore, we identify an interactive effect involving the payment method of M&A, transaction size, and policy uncertainty. This study provides valuable insights into critical determinants of M&A transactions, offering strategic guidance to enhance success likelihood and facilitate sustainable corporate development in the future.