Measuring super-efficiency of financial and non-financial holding companies in Taiwan: An application of DEA models

Data Envelopment Analysis (DEA) has been widely used to assess the efficiency of various manufacturers, but for some specific natural industries, there is a problem of the same efficiency score being 1. Taiwanese banks are all small in scale and their numbers abound. Hence, the efficiency scores of Taiwanese banks turn out many efficient decision making units (DMUs) and create a ranking problem. This paper implements four kinds of DEA models to solve the rating problem and calculates the correct efficiency scores. This paper hypothesizes four kinds of super-efficiency in DEA models as our estimates are based on information obtained from 37 Taiwanese banks for the period from 2004 to 2006. The results show that: (1) Traditional DEA presents more efficient DMUs when estimating Taiwan banking, and in order to solve the problem of ranking, adopting the Modified-Super model is more suitable. (2) The performances of financial holding companies' subsidiary banks are better than the independent banks. (3) The total loans to total asset ratios, ratio of net income, bank share of market loans, number of branches, and group play important roles in efficiency. (4) Banks should segment their own attributes and clients to offer diversified services or specialized businesses, so as to satisfy the needs of different groups of customers.