摘要：We apply a Bayesian Markov Chain Monte Carlo (MCMC) technique, along with data augmentation to accommodate censoring in the dependent variables, to the estimation of a large expenditure system of food expenditures. Our finding of significant error covariance estimates justifies estimation of the system in improving statistical efficiency. Income, household composition, regions and other socio-demographic variables are found to play significant roles in determining household food expenditures.
