Technical Working Paper 01-102

Simulated Maximum Likelihood Estimation of A Consumer Demand System Under Binding Non-Negativity Constraints: An Application to Chinese Household Food Expenditure Data, M. Hasan, R. Mittelhammer, and T. Wahl January 2002

Abstract

This paper examines simulated maximum likelihood (SML) estimation of a ten-commodity consumer demand system under binding non-negativity constraints. First an almost ideal demand system (AIDS) under binding non-negativity constraints is defined. Then two new simulators (ORDGHK and MIX) are introduced that can be used to overcome the formidable computational problems involved in calculating the high dimensional integrals of the multivariate normal distribution required when estimating multivariate Tobit-type models based on maximum likelihood methods. Finally, the SML estimation method is employed to estimate the ten-commodity food demand system based on Chinese Household Expenditure Survey Data.