Expenditures per drink, amounts consumed, context of drinking and alcohol-relater\d problems in the 2005 National Alcohol Survey: The policy implications

Earlier we found mean expenditures per drink of the top 10% of drinkers is lower ($0.79/drink) than the bottom 50% by volume ($4.75/drink). The 2005 NAS (n = 6,919), a telephone survey of US adults, asked 30-day expenditures for wine, beer and spirits and associated consumption. We investigated factors affecting alcohol expenditures and conversely, the role of expenditures and other factors in predicting alcohol-related problems. Using linear regression among current drinkers (n=3,951) we modeled expenditure per drink with demographics including gender, age, ethnicity, employment status, income, and amount consumed in on- vs. off-premise venues, plus number of social and health consequences. Alternative specifications considered the effect on number of alcohol problems of volume and mean expenditure per drink, accounting for the same influences. Considering results with alcohol measures that adjusted for a person's mean drink size and strength, those reporting more problems were as expected more likely to be male, younger, have lower incomes and ages, be Hispanic and not only drink higher volumes, the strongest influence, but also spent more per gram ethanol (or per self-reported drink in separate models; both ps < .01), controlling for on-premise drinking (p = .07). While descriptively, heavier drinkers tend to pay much less per drink than lighter drinkers, once many personal characteristics are controlled, those reporting higher expenditures (willing to spend more per drink) express more alcohol related problems. Few general population studies have included alcohol expenditures. We consider data limitations and the implications for price and tax policy.