The Determinants of Binge Drinking: Fresh Eyes to an Old Problem
Presented by:
Reka Sundaram-Stukel
Department of Agricultural & Applied Economics
University of Wisconsin - Madison
Wednesday, September 26, 2018
12:00 pm-1:30 pm
Taylor-Hibbard Seminar Room (Rm103)
To investigate how alcohol and its correlates affect socio-economic and health outcomes by estimating the demand for problem drinking, we use the 2016 waves of three annually conducted national population surveys in this research: the Behavior Risk Factor Surveillance (BRFSS), the National Survey of Drug Use and Health (NSDUH), and the Consumer Expenditure survey (CES). Our final data integrates expenditure shares on alcohol from the CES with state level alcohol prices from the BRFSS data. Both the NSDUH and BRFSS data allow us to identify the at-risk binge drinkers used through the analysis. The BRFSS sample considers 457,202 respondents 18 years and older, out of which 17% could be classified as at risk binge drinkers having a mean age of 39 years and mean income of around $50,000. Perhaps the most striking result is that the correlates of binge drinking don’t dramatically differ for the poorest income quartile as compared to the richest income quartile in the BRFSS sample. The latent demand (prevalence) estimations revealed two patterns. One, that white males are 5% more likely to have a latent demand for binge drinking than other races. Second, being employed increases the latent demand for binge drinking by 4% while the unemployed or unable to work reveal a lower likelihood to binge drink. The zero inflated Poisson estimation of frequency and intensity of drinking showed that significant positive income (TODO numbers here) and negative education effects (10% less likely to engage in frequent/intense drinking episodes) on the consumption patterns for binge drinking. Other co-morbid risk loving behaviors such as not fastening seat belts (10% more likely), smoking (20% more likely) are strongly correlated with binge drinking behavior. The NSDUH data precisely identifies binge drinkers through detailed clinical prescreening questions, and out of a sample size of 44,000 respondents (18 and older), 10,000 have a latent demand binge drinking. Because the NSDUH questions are more detailed, we find a much stronger income effect (10% increase in frequency and intensity of drinking) but a relative insensitivity to education. Our structural models show that education has a two pronged effect: increased frequency of consumption (11%) and decreased intensity of consumption (2.3%).