Censored Demand System Estimation with Endogenous Expenditures in clustered samples: an application to food demand in urban Mozambique

We address the issue of endogenous expenditures in the context of a censored demand system by an augmented regression approach estimated with a two-step estimator. An application to food demand by urban households in Mozambique shows that accounting for endogeneity is potentially important in obtaining reliable point estimates of price and, in particular, expenditure elasticities. Furthermore, a bootstrap approach to obtain confidence intervals when data are clustered - as is the case with most household surveys - is devised. Based on a Monte Carlo exercise we speculate that previous studies in failing to account for the clustered nature of the data overstate the precision with which elasticities are estimated.