Abstract
ObjectiveOur primary goal is to move towards establishing a causal linkbetween binge drinking, mental health, employment and income.IntroductionOne of the key questions in health economics is what is thedirection of causality: does poverty cause poor health outcomes; doeslow education cause poor health outcomes; does poor health resultin lack of productivity; does poor health cause poor educational andincome outcomes; and how is this all related to mental health if at all.We are used to breaking down data into fragments as researchers:an investigator who is predominantly focused on health outcomeswill approach the problem with disease as the dependent variable andincome as the conditioning variable. However, if we are interested inincome inequality we will reverse the direction and income will bethe dependent variable with health status as the conditioning variable.The representation above allows us to visualize data as a functionof multiple fragments. For example if we want to understand howdepression is related to income, one can look at the figure to observethat with lower income there is a higher likelihood of being depressed.With this simple illustration we can see that establishing causal linkscan be very tricky, if not incredibly challenging.MethodsTwo methods are: applied descriptive analysis and estimation.We approach this without causality in mind, but with an intentionto explore how behavior responds to income, education, labor andhealth. Our descriptive approach looks at trends in binge drinking andmental health as it affects key economic outcomes such as education,employment, and income. For each outcome we then run a simpleprobit model controlling for a variety of characteristics. The keyco-variates in these models are income, employment and health.It is very useful to look at these simple probits because often itis hard to separate the effects of income on health, employment onincome, health on employment, education on employment, health andincome, and finally income, employment, health and education onmental health and substance abuse.ResultsOur estimated results are rather interesting. Examining themarginal probits, e.g. figures 1.3, and 1.5, we show that there isn’ta significant income effect, nor do we find significant education oremployment effects associated with binge drinking. In fact we findthat in Wisconsin binge drinking is a health burden for those whoare eligible to drink irrespective of education and that the effect issignificant; we also find that higher levels of education increase theprobability of being unemployed but not significantly. The secondset of probit estimates, e.g. figure 1.7, show that poor health is indeedassociated with outcomes lower employment as compared to othergroups, and higher probability of depression. The last set of probits,e.g. figure 1.1, show that retired, self employed and employed areless likely to be depressed but not significantly so, and those who areunable to work have a higher estimated probablilty to be depressed.Income doesn’t appear to have a significant estimated effect ondepression.ConclusionsOur analysis provide insights into the question of socio-economicstatus (SES), binge drinking, and depression in three important ways.First, we explore the relationship between SES and binge drinkingand we find that binge drinking is SES invariant. Second we findthat depression is not associated with income it does have a strongrelationship with employment status. We are in the process ofunpacking the effects of SES, binge drinking and depression to movebeyond associational inferences to causal inferences.