預測成為併購標的之機率 – 變數經產業分佈調整之影響

Abstract

本研究運用羅吉斯迴歸模型，預測台灣上市（櫃）公司成為潛在併購標的之機率。模型基於關鍵併購預測變數所建構，包括標的公司之基本面財務數據指標、技術面股票交易指標、所屬產業併購交易波動指標及產業市占率集中度指標。本研究對基本面財務數據指標與技術面股票交易指標進行產業分佈數值調整，並將調整後的變數模型與原始模型進行比較。樣本涵蓋 2010 年至 2023 年共 21,260 筆觀測值，包括 158 筆併購標的及 21,102 筆非併購標的。 實證結果顯示，台灣併購市場符合無效率管理假說、產業波動假說及產業集中度假說。經過產業分佈調整後之模型具有更佳的解釋能力，且併購標的預測準確率有所提升。此外，本研究採用 Rolling Test 方式，以八年樣本期間建構預測模型，進行 2018 年至 2023 年樣本外測試，分析不同年度模型對台灣併購市場所代表之意涵。

This study employs a logistic regression model to predict the probability of Taiwanese listed companies becoming potential M&A targets. The model is constructed using key predictive variables, including fundamental financial indicators, technical trading indicators, industry disturbance indicators, and industry concentration indicators. This study adjusts the fundamental financial and technical trading indicators based on industry-specific distributional characteristics and compares the adjusted models with the original ones. The dataset comprises 21,260 observations from 2010 to 2023, including 158 M&A target samples and 21,102 non-target samples. The empirical results reveal that the Taiwanese M&A market aligns with the inefficient management hypothesis, the industry disturbance hypothesis, and the industry concentration hypothesis. The model's explanatory power improves after adjusting the predictive variables for industry distribution, leading to enhanced accuracy in predicting M&A targets. Additionally, a rolling test approach is adopted, constructing predictive models using eight-year sample periods and conducting out-of-sample tests from 2018 to 2023 to analyze the implications of these models for the Taiwanese M&A market across different years.