Estimation of crop production for smaller geographical area: An application of discriminant function analysis

Abstract

Crop production statistics for small areas like Community Development Blocks (generally referred to as blocks) and panchayat level (smaller than blocks) have become essential in view of policy formulation for making development plans at the micro level in India. Presently, the crop production or yields through crop-cutting experiments (CCES) are obtained at the district level and these estimates are aggregated at the state and the country level. Conducting the requisite number of CCEs for a specific area is neither operationally feasible nor economically viable. Following the prediction approach of Singh et al. (2012), we have explored an application of discriminant function analysis of the auxiliary variables related to crop yield for estimation of crop production at the block level in this article. Timeseries data on crop yield, along with auxiliary variables, have been used in the study. The discriminant scores obtained from estimated discriminant functions have been used as an explanatory variable instead of the auxiliary variables such as in the multiple regression model. The proposed procedure based on discriminant function analysis has been empirically illustrated for estimating wheat production at the block level for the Sultanpur district of the state of Uttar Pradesh, India. The results show that the proposed procedure has provided more reliable estimates at the block level as compared to the procedure of Singh et al.