本研究試圖利用潛在類別分析探討電子業公司財務績效的公司價值分群,以往潛在類別分析大多使用在社會科學的研究,因為在這領域有很多潛在變項是無法直接測量。然而在金融領域也有許多的潛在變項,例如投資人無法直接觀察公司價值的好壞,僅能利用客觀的財務指標去評估公司。本研究以台灣證券交易所核准上市電子業公司293家為對象,利用潛在類別分析與集群分析針對10個財務指標進行分群,並以下一季股價報酬率做為分群的效標變數。研究發現潛在類別分析確實能成功運用在公司價值的分群,可以將樣本公司區分為健全、一般與欠佳三種類型,且在預測未來股票報酬率的表現上優於集群分析的分群效力。
This study uses latent class analysis to analyze the performance of listed electronic companies based on 10 financial ratios. We compare the difference of clustering of companies between latent class analysis and cluster analysis. The results show the latent class analysis actually can be applied to clustering of companies based on the binary data of financial ratios and classify the sample companies to three different patterns of companies: excellent, normal and questionable. The clustering validity of latent class analysis is superior to cluster analysis. The rate of return of stock prices of the excellent firms is larger than excellent companies of cluster analysis and the rate of the questionable firms of latent class analysis is lower than questionable ones of cluster analysis.