A quasi-maximum-likelihood estimator is proposed and applied to a censored Translog demand system for foods, using a sample of food stamp recipients in the United States. The procedure produces remarkably close parameter and elasticity estimates to those of the simulated-maximum-likelihood...