The impact of problem drinking on employment
利用美国南部六州4898名女性和3224名男性的随机样本，通过双变量probit模型控制不可观测因素的相关性，发现问题饮酒与就业之间无显著负相关，提示问题饮酒的社会成本可能被高估。
Social cost studies report that alcohol use and misuse impose a great economic burden on society, and over half of the total economic costs are estimated to be due to the loss of work productivity. Controversy remains, however, as to the magnitude and direction of the effects of alcohol consumption on productivity. Furthermore, most of the studies have looked at the relationship between problem drinking and wages. This paper investigates the impact of problem drinking on employment by analysing a random sample of men and women of prime working age from six Southern states in the US (Alabama, Arkansas, Georgia, Louisiana, Mississippi and Tennessee). The data set contains 4898 females and 3224 males, with information on both employment and problem drinking. To eliminate the bias that may result from single-equation estimation, we used a bivariate probit model to control for possible correlation in the unobservable factors that affect both problem drinking and employment. We find no significant negative association between problem drinking and employment for both men and women, controlling for other covariates. The findings are consistent with other research and highlight several methodological issues. Furthermore, the study suggests that estimates of the costs of problem drinking may be overstated owing to misleading labour supply relationships.