利用資料包絡分析法模型評量台灣半導體產業個股

Abstract

本研究為一篇完整運用多種資料包絡分析法(Data Envelopment Analysis, DEA)模型的論文,以評量台灣 58 家半導體產業個股。使用(劉復華, 2017a)和(劉 復華 & 黃洋城, 2017)所提出之考量固定規模報酬(GBM-bc)和變動規模報酬 (GBM-bv)的 GBM(Gap-basd Measure)模型,(廖子豪, 2015; 劉復華 & 廖子豪, 2015)和(張宏偉, 2016; 劉復華 & 張宏偉, 2016)分別將兩種模型延伸至超高效模 型,分別為 Super-GBM-bc 和 Super-GBM-bv;亦使用了(曾美嘉, 2016; 劉復華 & 曾美嘉, 2016)所提出的 MGBOM(Mixed Gap-based-oriented Measurement)模型以 考量在變動規模報酬下之含有負數數據的混合數據的情形;為完成個股的評比, 本研究提出了其延伸的 Super-MGBOM 模型以進行超高效的計算。以多項財務指 標數據作為評比依據,將指標分為大者為佳(望大)及小者為佳(望小)兩類,作為 DEA 模型上的投入與產出指標,從經營表現和獲利能力兩種層面分析上市櫃公 司之個股的綜合績效。在此兩種層面中,個股依據 DEA 模型所計算出的技術效 率值被分為兩類,因此個股將共被分為四種類型,每種類型的所屬個股擁有不同 的特性和投資風險,以便投資者進行選股。

In this research, we offer the procedure to assess 58 stocks of Taiwanese semiconductor companies by a set of data envelopment analysis models that employed gap-based measure with constant return-to-scale (GBM-bc) and variable return-to-scale (GBM-bv) models introduced by Fuh-hwa Liu at 2017. Fuh-hwa Liu and Zih-Lao Liao, Hong-wei Chang extended the model to measure the super efficiency in both return-to-scale, named as Super-GBM-bc and Super-GBM-bv at 2016. We also employ the model invented by Fuh-hwa Liu and Mei-cha Tseng 2016, MGBOM to deal with the dataset of inputs and outputs are mixed with negative values under the condition of variable return-to-scale. We extended the model to measure the super efficiency of variable return-to-scale, named Super-MGBOM. We selected several indices to assess the stocks. The indices are either desirably larger values or desirably smaller value. In terms of DEA models, they are the outputs and inputs, respectively. We assess the stocks in two dimensions – operating performance and profitability. In each dimension, the stocks are partitioned into two classes in terms of their technical efficiency scores measured by the forementioned DEA models. Therefore, there are four combinations of classes and they have different properties and risks. In the lights of our assessment, investors would able to select the stocks.