QUANTILE REGRESSION APPROACH FOR THE ANALYSIS OF PRODUCTION FUNCTION OF FOODGRAIN IN INDIA

Ordinary least square method (OLS) is an efficient method for estimating the parameters under the presence of assumptions of error term. Moreover, the problems like multicollinearity, auto-correlation, outliers, influential observations, heteroscedasticity etc., do occur and then OLS estimates are less efficient. Under these conditions, quantile regression technique is advisable, as it does not require any distributional assumption about error term. In this paper, an attempt has been made to propose a quantile regression model using Cobb-Douglas production function for analyzing decadal data of foodgrain production as endogenous variable with respect to net irrigated area, net sown area, net area under cultivation, consumption of fertilizers, consumption of pesticides and consumption of electricity in agriculture as exogenous variables. The behaviour of the data indicated that some observations were outliers and influential observations. Elasticity of production and marginal value productivity has also been obtained and the best model has been selected on the basis of sign, size and significance of the parameters. Moreover, on the basis of elasticity of production, marginal value productivity and quantile 0.90th regression, net sown area and area under cultivation played a vital role in increasing foodgrain production.