DEA and benchmarks - an application to Nordic banks

In this paper, Data Envelopment Analysis (DEA) is developed to analyze the efficiency of a single bank. The inputs are given in terms of cost of personnel, cost of material and expected cost of credit losses. Outputs concern lending, deposits and gross revenues (interest margins and non-interest income). The data covers 48 large Nordic banks during the two years 1992 and 1993. Fourteen banks are from Denmark, thirteen from Finland, twelve from Norway and nine from Sweden. For each of these banks, the DEA method is used to form a "reference bank", which is a convex combination of the best competing banks (those at the efficiency frontier). The three inputs and the three outputs of the reference bank will be used as benchmarks. This procedure implies that one can only say that one single bank is less efficient than its reference bank, not less efficient than another bank. The results show that 4-7 Nordic banks were situated at the efficiency frontier for those two years. These banks should then be used to form reference banks for other banks, and to set benchmarks for them. Such benchmarks would have been slightly different, dependent on the "window" to be used, 1992, 1993 or 1992 + 1993.