以兩階段資料包絡分析法探討台灣壽險業之效率前緣

現今台灣壽險業市場非常競爭，對壽險業者願景而言，都希望能以最小投入追求企業產出效率利益最大化，然而如何達成此一願景，各學者之見解有很大的不同。本研究藉由兩階段DEA分析台灣壽險業2014年至2016年22家壽險公司之相對經營效率，第一階段以員工人數、負債及股東權益、固定資產及其他資產為投入變數；人壽保險保費收入、醫療保險保費收入及意外保險保費收入為中間變數；第二階段以第一階段的中間變數為投入變數；銀行存款、營業收入及投資報酬率為產出變數，探討22家壽險公司相對經營效率、學習標竿及仍待改善空間。本研究以2014年至2016年共22家壽險公司，66筆決策單位(DMU)進行分析，資料來源為保險業資訊公開觀測站，在第一階段結果顯示19個DMU已達成強勢效率單位，並成為標竿學習對象，另有15個DMU為中緣效率單位、13個DMU為邊緣效率單位、19個為邊緣非效率單位，這些DMU都必須與強勢效率單位(標竿)進行學習，學習的方式可以在既有的產出減少投入，或在既有的投入增加產出。在第二階段結果顯示16個DMU已達成強勢效率單位，另有6個DMU為中緣效率單位、16個DMU為邊緣效率單位、28個為邊緣非效率單位，這些DMU也必須與強勢效率單位(標竿)進行學習，學習的方式與第一階段一樣。第一階段結果顯示以國泰人壽、南山人壽、新光人壽、富邦人壽、三商美邦、康健人壽、友邦人壽及巴黎人壽在2014至2016年都呈現效率前緣。代表此階段其他落後的壽險公司都可以這些公司為標竿學習對象。第二階段結果顯示以中華郵政及安達人壽連續三年呈現效率前緣，表示其他壽險公司也可依此為標竿學習對象。本研究之結果可提供壽險公司經理人做為自我改善或標竿學習之依據，也可做為公司擬定部門策略規劃之參考。

關鍵字

壽險業；兩階段資料包絡法；效率前緣；差額變量；標竿學習

並列摘要

Nowadays life insurance market in Taiwan is very competitive. Particularity for life insurance firms, to minimize the costs and to maximize the profits as the first priority. However, different scholars tend to handle this issue from different perspectives. This study adopted a Data Envelopment Analysis (DEA) to examine relative management efficiency, benchmark learning and slacks for 22 insurance firms in Taiwan in the period of 2014-2016. This study adopted a two-stage DEA approach. In the first stage, numbers of employees, debts and shareholders’ equity, fixed assets and other assets were selected as the input variables, while premium life insurance incomes, medical insurance incomes, and accident insurance incomes were used as the intermediate variables. In the second stage, those intermediate variables were used at the input variables, while bank deposits, sales revenues and return on investment(ROI) were selected at the output variables. This research analyzes 22 life insurance companies with 66 Decision Making Units (DMU) in the period of 2014-2016. The data of this study was obtained from Insurance Information Disclosure Observatory. The outcome of the first stage shows that there are 19 Robustly Efficient Units (REUs) which are benchmark learning partners, 15 Median Efficient Units,13 Marginal Efficient Units (MEU), and 19 Marginal Inefficient Units (MIUs). The results of the second stage’s results indicate that there are 16 REUs, 6 Median Efficient Units, 16 MEUs, and 28 MIUs. These DMUs which are not REUs could either decrease input without changing outputs or increase output without changing the inputs to maximize profits by referring REUs. The results of the first stage show that 8 companies are near the efficient frontier, including Cathay Life Insurance, Nan Shan Life Insurance, Shin Kong Life Insurance, Fubon Life Assurance, Mercuries Life Insurance, Cigna Taiwan Life Assurance, American International Assurance, and Cardif Assurance Vie from 2014 to 2016. The analyses of the second stage indicate that Life Insurance Division of Chunghwa Post and Chubb Life Insurance are in efficient frontier in three consecutive years. In other words, these companies can be benchmark learning partners for other in-efficient companies. The results of this research can provide reference for firm’s continuous improvements, benchmarking learning, and strategic planning.

並列關鍵字

Life insurance industry ； Two-stage Data Envelopment Analysis ； Efficiency Frontier ； Slacks ； Benchmarking Learning

參考文獻

Golany, B., & Roll, Y. (1989). An application procedure for DEA, Omega, 17(3):237-250. DOI: 10.1016/0305-0483(89)90029-7

Kao, R., & Hwang, S.N. (2008). Efficiency decomposition in two-stage data envelopment analysis: An application to non-life insurance companies in Taiwan, European Journal of Operational Research, 185(1), 418-429. DOI:10.1016/j.ejor.2006.11.041

方顯光、余惠芳、王煒竣（2015）。後金融海嘯時期之台灣壽險公司經營效率之分析－兩階段資料包絡分析法之應用。華人前瞻研究，11卷2期，93–112。