Demand Analysis with a Thin Price Sample

Abstract: For about 125 items of food, the Consumer Expenditure Survey (CES) schedule of the Indian National Sample Survey asks the interviewer to obtain both quantity and value of household consumption during the reference period from the respondent. This would appear to put a great burden on the respondent. But it is likely that the price usually paid is almost the same within each first stage unit (fsu). The present work proposes a new sampling scheme to estimate demand elasticities of essential food items. While the conventional sampling method used in practice (e.g. in NSS consumer expenditure survey) involves seeking price information from many households sampled from a fsu, the proposed procedure involves only one household chosen randomly from every fsu for price data collection and thus requires much less interview burden. Using unit records for vegetable items in the NSS's 2011-12 CES, our results show that in spite of requiring much less data, the new scheme captures the household food consumption behavior as precisely as before.