Abstract
A stochastic frontier cost function is estimated using panel data for the electricity generating industry in India. The impact of distributional and functional form assumptions on technical inefficiency and the sources of inefficiency are investigated by using maximum likelihood, GLS and semi-parametric-GLS approaches and by incorporating firm-specific inefficiency effects in the cost function itself. Average inefficiency in the electricity generating industry in India is found to be high by all three methods. The estimate predicted by the maximum-likelihood approach is, however, lower than that predicted by the other two methods. This could be due to the distributional assumptions made under the maximum likelihood method. Public ownership and low capacity utilization are found to be significant determinants of inefficiency in the electricity generating industry in India.