In the last two decades, numerous studies have used parametric and non-parametric techniques to estimate commercial bank productivity and efficiency in the Indian context. However, none of them has considered undesirable outputs in their analytical framework. The present paper makes an attempt to fill this gap and evaluates the performance of 49 Indian commercial banks for the period 2006-07 to 2010-11 using Seiford and Joe Zhu (2002) approach, which is essentially a variant of the popular Banker-Charnes-Cooper (BCC) model.

Consequent to the promotion of mass banking during the two decades after the initiation of bank nationalization, the quality of the assets of Indian banks declined sharply as the system of directed allocation of credit to priority sectors worsened the quality of banks’ loan portfolio to a considerable extent.

Against this backdrop, the present paper attempts a risk-adjusted performance benchmarking of the Indian public and private sector banks using a non-parametric efficiency analysis framework.

Literature Review

In their 1978 seminal paper, Charnes et al. (1978) provided a generalization of Farrell’s single input-single output technical efficiency measure to the multiple output-multiple input case and their contribution resulted in the genesis of Data Envelopment Analysis (DEA). The methodology originally developed by Charnes et al. (1978) was later extended by Banker et al. (1984).

One of the earliest attempts to incorporate undesirable outputs was by Pittman (1981). Fare et al. (1989) modified the Farrell measure of efficiency to accommodate both desirable and undesirable factors in the context of a productive system where the two sets of factors are treated differently. Golany and Roll (1989) suggested the multiplicative inverse method which was later applied by Lovell and Pastor (1995) and Athanassopoulos and Thanassoulis (1995).