Towards a better measure of productivity in India: a case of chemical and chemical products industry

A brief review of earlier studies on the productivity scenario of Indian industry shows that most of the studies analysed are confined to either parametric approach or growth accounting approach of measuring productivity. At the same time, the few studies based on the non-parametric [namely, Malmquist productivity index (MPI)] overlook the returns to scale conditions as well as the bias involved in the estimation of distance functions. Given this backdrop, this study aims to provide a robust measure of productivity, which considers the returns to scale assumptions and correct for the bias involved in the estimation of productivity.

This study empirically tests for the returns to scale that exists in the chemical and chemical products industry in India. The test result suggests that Ray and Desli (1997) approach of MPI is the appropriate one for the present context. Initially, the conventional Ray and Desli (1997) estimation and decomposition of MPI for the period 2001 to 2017 is being used. Subsequently, to correct for the bias in the estimation of efficiency scores used for the estimation of MPI, the bootstrapping algorithm of Simar and Wilson (2007) has been extended into the context of MPI estimation.

The results from the conventional Malmquist productivity estimates testifies to an improvement of total factor productivity (TFP) in seven out of 16 years under consideration. On the contrary, TFP growth is recorded only in the four years throughout the period after the bias correction. A greater discrepancy between the two measures has been found in the case of scale change factor component of MPI.

The technical change (TC) component positively influences TFP, whereas scale change factor (SCF) deteriorates the TFP condition of this industry. It will be appropriate for these firms to identify and operate under an optimal scale of operation, along with reaping the benefits of technological change. From a methodological perspective, researchers should consider the potential bias that arise in estimation of TFP and use a larger sample whenever possible.

This paper brings in a new perspective to the existing literature on industrial productivity. As against earlier studies, this study empirically tests the returns to scale of the sector under consideration and uses the most appropriate approach to measure productivity. The effect of sampling bias on TFP and its components is analysed.