国有企业并购风险预警及其影响因素研究——基于数据挖掘和XGBoost算法的分析

Early Risk Warning for State-owned Enterprises’Merger and Acquisition and Its Influencing Factors——Based on Data Mining and XGBoost Algorithm Analysis

随着国企混合所有制改革的不断深入,出现了大量国企并购行为,然而很多国企由于盲目并购而导致并购绩效不佳。为了使国企更好地开展并购活动,对并购风险预警是十分重要和必要的,直接关系到并购双方的利益,甚至影响国企改革的成效。通过设计影响国有上市公司并购风险指标评价体系,利用Python爬取网页和文本数据,应用机器学习XGBoost算法构建预警模型实现风险的计量、监测、预警和管理,并将结果与其他经典模型作对比实验以评价预警效果,最后运用多元线性回归模型研究并购风险显著性因素。实证结果表明,基于XGBoost算法的预测结果精确度为80%,在所有模型中表现最优,具有更强的可靠性和适用性;投入资本回报率、营业利润率、支付对价净利润比对于并购风险的预测更加重要和有效;总资产周转率、投入资本回报率、股权制衡度、审计质量更加有利于抑制并购风险。

With the steady progress in the reform of the state-owned mixed ownership system,a large number of state-owned enterprises(Hereinafter referred to as SOEs)have carried out mergers and acquisitions(Hereinafter referred to as M&A).However,many SOEs failed to complete the M&A due to blind M&A.In order to enable SOEs to better carry out M&A activities,it is essential to be wary of M&A risks,which is directly related to the interests of both parties,and even affects the effectiveness of SOEs reform.This paper designs the evaluation system of M&A risk index of state-owned listed companies,takes Python to mine webpages and text data,uses machine learning XGBoost algorithm to construct early warning model to carryout risk measurement,monitoring,early warning and management,and compares the results with other classic models to evaluate the early warning effect.Finally,multiple linear regression models are used to study the significant factors of M&A risk.The empirical results show that the accuracy of the prediction result based on XGBoost algorithm is 80%,which is the best among all models,showing stronger robustness and applicability;Return on invested capital,operating profit margin and payment-to-price net profit ratio are more important and effective for the forecast of M&A risks;Total asset turnover,return on invested capital,equity balance and audit quality are more conducive to suppressing M&A risks.