This is circulating again... it is important to point out that the topline number is nonsense. Based on survey data, if you drink more than TEN drinks per week you are definitely in the top 10%. The problem is we can't find about half of alcohol consumption in surveys. 1/6

We (w @nirupama_rao and Yinan Wang) compared the distribution of reported drinking from NESARC (a self-reported survey --used by the viral article) and NielsenIQ (household purchases). They are similar to each other but nowhere near 70 drinks per week. 2/6

The problem with surveys of alcohol consumption (like NESARC) is that when we take people at their word -- we're missing about 50% of alcohol sales. Some combination of: poured down the drain, not understanding serving sizes, and not truthful reporting. 3/6

Household level retail purchases from NielsenIQ line up nicely with NESARC but are still missing about 80% of alcohol according to tax data on production around 40-60% of alcohol is consumed away from home (bars and restaurants) but that still gives a HUGE gap. 4/6

One approach is to just scale everyone by 2-3x to match the production totals (that's how they get 70 drinks per week!). Of course, almost nobody actually reported drinking that much! In terms of what people admit to -- 10 drinks per week puts you in the top 5% of drinkers! 5/6

Getting this right is important because the eye-popping numbers (10% of people drinking 10+ drinks per day) make people who are drinking 10+ drinks per week think they don't have a problem. (They do!) 6/6

Anyway this is all from the appendix of my paper with @nirupama_rao and Yinan Wang. chrisconlon.github.io/site/sin_tax_appendix.pdf 7/6

Our paper (forthcoming at @restatjournal) looks at how skewed the distribution of sin taxes was across households (including alcohol and cigarettes). We found 10% of households paid about 80% of taxes and 2.5% of HH's paid large amounts of both taxes. 8/6 chrisconlon.github.io/site/sin_tax.pdf