New data on how Americans drank themselves to death during the pandemic
It's not great
It’s been a rough couple of years, what with the pandemic and the rise of American fascism and all. Many Americans have been weathering the ongoing storms with the aid of a time-tested coping mechanism: booze. This was apparent from the get-go, when researchers noticed a sustained spike in alcohol sales following the earliest lockdowns in March of 2020.
At the time there was concern that this sales spike would lead to increased health problems and mortality down the road. Alcohol is a carcinogen, it’s lethal even at relatively low doses, and being drunk dramatically raises the odds that a person will wreck their car or murder someone else. Now that we’re several years out from the dawn of the pandemic we’re starting to get some good data. In a working paper posted by the National Bureau of Economic Research, University of Chicago economist Casey Mulligan finds that between March 2020 and June 2021 there were about 14,000 alcohol-caused deaths above what you’d expect if the pre-2020 trend had continued. That mortality spike is plainly visible when you plot the data, as in the chart below.
Prior to 2020, CDC mortality data show there were typically a little north of 3,000 alcohol-caused deaths per month in the United States, including deaths from alcohol poisoning, acute pancreatitis, liver disease, and other booze-driven ailments. But by mid-2020 the CDC was logging about 1,000 deaths per month above the previous level, a mortality increase of roughly 50 percent. By June 2021 the number of excess deaths shrunk somewhat, but still remained well above the historic trend. More recent mortality data tends to be incomplete, so the full story for the remainder of 2021 isn’t known yet.
Mulligan notes that this is an undercount of the true toll of excess alcohol-related death during the pandemic: drunk driving deaths aren’t included in these figures, and we know from other sources that fatal drunk driving accidents increased in recent years.
Mulligan, it should be noted, was a member of Trump’s Council of Economic Advisors until 2019. His main project in the paper, which also looks at pandemic-era drug mortality, is to argue that these excess drug and alcohol deaths were driven in large part by the generous unemployment benefits and stimulus checks the federal government provided in 2020 and 2021. There’s something to this I think — plenty of research shows that when drugs and alcohol become cheaper or easier to purchase people consume more of them, with a certain percentage of those users overdosing and dying as a result.
But I suspect Mulligan’s case is overstated. To make it he builds some extremely complicated models that rest on lots of assumptions about things like how price changes affect demand for intoxicants, how people spend unexpected stimulus money, how much alcohol consumption happens at home, how much lag time you’d expect between drug use and death, the price ratio between prescription opioids and illicitly-manufactured ones, how opioid use affects worker productivity, the overall employment rate of illicit drug users, the average years of life lost by the typical drug overdose victim, the price ratio of alcohol purchased in bars to alcohol consumed at home, how the employment rate of bartenders mirrors price changes in alcohol, how the reduction of vodka prices in Russia in the early 1990s affected fatal alcohol poisoning rates, how Estonia’s entry to the European Union affected alcohol-related mortality in Finland, how well high-functioning alcoholics are able to function at work, and the value of time itself.
To put it lightly there are a lot of assumptions happening here, and when you’re stacking them one on top of the other like this to build an equation I wonder how much that equation can really tell you about the state of the world. There’s a danger that you end up just rubbing a bunch of datasets together and fiddling with the parameters until you get a result that aligns with your expectations.
Economists do this sort of thing all the time and Mulligan may in fact be right, but as this paper hasn’t been through peer review I’m not inclined to take his word on things, especially since he’s a fairly ideological guy who gladly put his expertise to work for the most fundamentally dishonest president in the history of the union.
But the baseline numbers he reports — coming straight from the CDC itself — are solid. And it’s really useful to see how those numbers differ from the pre-pandemic trends.