Estimation of firm-level productivity changes in the Indian power sector: Disentangling unobserved heterogeneity by a transformed fixed-effect stochastic frontier model

TL;DR

This paper analyzes firm-level productivity changes in the Indian power sector from 2000 to 2009, accounting for unobserved heterogeneity, and finds a decline in productivity despite technological improvements.

Contribution

It introduces a transformed fixed-effect stochastic frontier model to control for unobserved heterogeneity in firm-level productivity estimation.

Findings

Overall productivity declined by 1.6% annually.

Technological change contributed an 8% annual increase.

Inefficiency increased by 3.1% annually, except for smaller gas generators.

Abstract

We measure firm-level productivity changes in the Indian electricity sector during a period that witnessed several pro-market regulatory changes. Using information collected from multiple sources we construct a unique panel of generating firms and transmission and distribution utilities spanning the years 2000 to 2009. We employ a recently developed improvement in the Stochastic Frontier panel method that allows controlling for time-invariant unobserved heterogeneity. Using the method we jointly estimate inefficiency and exogenous determinants of inefficiency. We estimate a flexible translog production model and compute decomposition of productivity into components of changes in technology, efficiency, scale and price effect. During this period, especially post Electricity Act 2003, we observed a general decline in firm-level productivity at the mean rate of -1.6% per year. A positive and large technical change is observed in the sector at the rate of 8% per year, attributable possibly to newer capacity addition. Except for smaller gas based generators, inefficiency is observed to be increasing at the mean rate of 3.1% per year in the sector. Consistent with extant findings we also document no significant impact of un-bundling on firm-level efficiency.