Rapid Consumption Survey Methodology: An Empirical Study in Bangladesh

Abstract

Every unidimensional poverty measure is mostly based on consumption expenditure. In the household consumption expenditure survey, the questionnaire needs to record detailed information about food and non-food items. The lengthy administering time hampers the survey very often. This study applies a rapid consumption methodology on two-round panel data from the Bangladesh integrated household survey conducted by the International Food Policy Research Institute. The surveyed items for each household are partitioned into the core and optional modules. Any randomly taken optional module is assigned to a household, and missing modules are then estimated by the imputation method. In estimating the poverty indicators, the multiple imputation with chained equation model shows better estimates than the ordinary least squares (OLS) and median imputation techniques. The estimated poverty headcount ratio (FGT0) is 26.72%, whereas the full consumption data show FGT0 as 27.2%. These results suggest that the rapid consumption survey methodology in a risk-prone area can be a potential approach to estimate the consumption and poverty at a lower cost and lower administrative time.