Income convergence in India: evidence from nonparametric panel data

Abstract

Purpose The purpose of this paper is to investigate convergence hypothesis in a balanced panel of 22 Indian states for the time period of 1980-81 to 2010-11 by applying nonparametric model setting in a panel framework. Design/methodology/approach The present study uses nonparametric and semi-parametric panel data methods to test the absolute and conditional convergence, respectively, and examines the income convergence using nonparametric panel data methods with state specific effects taken into consideration. These models are being estimated by the iterative process for a balanced panel of state wise per capita income and other conditioning variables for the time period of 1980-81 to 2010-11. For removing the fixed effects, the authors follow within transformation procedure according to the feasibility of the problem. Since convergence is estimated by regressing dependent variable on initial level of independent variable (as growth rate of income and per capita income in this case). So using usual transformation for removing the fixed effects is not feasible because by doing so the authors may end up with singular matrices on both sides of the regression model. Findings The results reject the null of parametric specification for both absolute as well as conditional convergence model. As to the outcome of the empirical analysis, the findings reveal that the Indian states are diverging in absolute sense and converging on conditional basis. Convergence happens to be consistent and conditional upon public expenditure, power generation share of primary and tertiary sector to Gross State Domestic Product. Originality/value The originality of the study is in its application of advanced methodology to highlight the model misspecifications while testing the convergence hypothesis in earlier literature.