The Theory of Benchmarking and the Measurement of Industrial Organization

Abstract

If more productive …rms grow relatively fast, an industry performs better, even when no …rm exhibits technical or e¢ ciency change. In other words, the two well-known sources of productivity growth-technology and e¢ ciency-can be augmented by a third one, namely the industrial organization e¤ect. In this paper the e¢ ciency of an industrial organization and its contribution to performance are measured by benchmarking all …rms on the industry. More precisely, e¢ ciency is measured by the proximity between a …rm and the best practices. Aggregation of …rm e¢ ciencies is imperfect. The bias is used to measure the e¢ ciency of the industrial organization. In benchmarking, change transmitted by a …rm represents productivity growth and change transmitted by the best practices represents technical change. Although I use a nonparametric framework, which requires only input and output information, duality analysis reveals the Solow residual. In discrete time Malmquist indices capture the measurement of the industrial organization e¤ect, e¢ ciency changes, and technical change. The industrial organization of Japanese banking is analyzed. The dynamic industrial organization e¤ect of entry and exit can be accommodated.