Alcohol consumption by men in China: A non-Gaussian censored system approach

Consumption of beer, wine and spirits by men in China is investigated, using data from the 2006 China Health and Nutrition Survey. Censoring of consumption levels is accommodated by estimating a censored equation system, using quasi maximum-likelihood and copula methods which allow the specification of non-Gaussian error distributions. Findings suggest that a misspecified error distribution can obscure the effects of explanatory variables on alcohol consumption. The procedure produces very different empirical estimates from a more conventional (Gaussian) estimator. Income does not affect alcohol consumption, with socio-demographic factors such as education, employment, and marital status playing more definitive roles. Regional differences are also found.