Handling Missing Values in DEA Models – A Case of Taiwan’s Machinery Tool Industry

• Main Authors: Chia-Jung Kao, 高嘉嶸
• Other Authors: Hsu-Hao Yang
• Format: Others
• Language: zh-TW
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/58370563377738726303

碩士 === 國立勤益科技大學 === 工業工程與管理系 === 99 === Machinery is often referred to as the “mother of industry.” According to the statistics of 2010, Taiwan is the world’s fourth largest country in terms of exporting values of machine tools, and the fifth largest one in output values. Traditional data envelopment analysis (DEA) evaluates the relative efficiencies of a set of decision making unit (DMUs) that us input(s) to produce output(s). If the values of inputs or outputs are missing, traditional DEA models cannot handle this kind of situation. Due to the presence of missing values, this study applies fuzzy DEA (FDEA) and interval DEA (IDEA) to measure the efficiency of Taiwan’s machine tool industry. The results of FDEA and IDEA are followed by index ordering and interval classification to rank the efficiencies. We collect the data of 17 firms over the period 1998-2008. The input variables are fixed assets, operating expenses, and the number of employees; the output variables are net sales and earnings per share. Five missing values are present in 2000, where they belong to LEADWELL、Victor Taichung、Ta Liang、KENT industrial。According to the empirical results, the rankings of the big-M method and FDEA are highly consistent, and four companies are classified as E++ by IDEA. Regarding to those companies with missing values, the rankings are always Ta Liang＞Victor Taichung＞KENT industrial＞LEADWELL, no matter what type of method is used. An interesting result is that Ta Liang is the one with missing value, its efficiency value is crisp however. This fact appears to deserve future investigation.