Unemployment, Alcohol, and Tobacco Use: Separating State Dependence from Unobserved Heterogeneity

Previous literature presents mixed evidence on the effect of alcohol consumption on labor market outcomes. On one hand, heavy alcohol consumption has been shown to have detrimental effects on labor market outcomes. On the other hand, moderate consumption is positively associated with wages and employment. Despite substantial reduced form evidence, previous literature has not been able to separately identify the causal pathways linking moderate versus heavy alcohol use to labor market performance due to the lack of natural experiments that only target moderate versus heavy drinking, as well as limitations of available structural methods that model state dependence and unobserved heterogeneity. This study develops a multiple-equation dynamic discrete choice ordered logit model, which allows separate identification of the contribution of state dependence (within and between outcomes) and unobserved heterogeneity. I apply this newly-developed model to differentiate the effects of moderate and heavy drinking, after accounting for other correlated unobserved heterogeneity. This study finds that moderate alcohol use increases employment, which is consistent with moderate alcohol consumption being a venue for social capital accumulation. Policies that target alcohol consumption separately by dosage level may be beneficial to employment in ways that have not previously been expected.