Bank cost efficiency and output specification

There is a longstanding controversy over precisely what it is that banks produce. However, there is little evidence on the sensitivity of bank cost efficiency results when different output definitions are applied. This paper does exactly that. In particular, we compare nonparametric efficiency scores yielded by two output specifications, one mainly identified with the asset approach and the other which also considers deposits as output. Results show that distributions of efficiency scores, estimated nonparametrically by means of kernel smoothing, vary greatly. In addition, firms' positions relative to the mean change according to either output definition, and results do not remain constant over time.