A quick glance at the headlines about beverage alcohol in recent weeks clearly points to Americans drinking more over the past month. The New York Post, long a paper of subtlety, used the headline “Americans are drinking a crazy amount of alcohol during the coronavirus lockdown.” Another article from The Conversation, which claims “academic rigor” in its tagline, used the headline “America is drinking its way through the coronavirus crisis – that means more health woes ahead.” In addition to these alarmist headlines, all of these stories have one thing in common, they point to scan data as their primary evidence.
Anyone who is part of the beverage alcohol industry (and anyone who has read these posts for a while) knows that scan data is only a partial picture of the market. So when we step back and try to take a more comprehensive view of the past month, are Americans really drinking more? The answer: possibly a tiny bit, but these answers are much less clear than many articles have made them out to be.
Let’s start with what scan data is and what it isn’t. First, scan data typically only covers off-premise, so when you see a 10%, 20%, or 50% sales jump, that’s not total sales going up, but sales going up in one channel. That means it’s a partial picture, and any growth in on-premise needs to be weighed against it.
Secondly, scan is a measure of purchase, not consumption. Now in the long-run, these are very similar, but there’s a reason that most analysts didn’t publish many one-week reports on scan data prior to this pandemic. In the overall beverage alcohol market, it’s likely this disconnect will affect spirits and wine the most and beer less (since you can stock up on spirits and wine and then store them indefinitely), but many of the most sensationalist headlines relied on purchase data when stocking up was at its peak. More reliable measures of demand shift for off-premise are likely in the weeks that followed and those will be more closely related to consumption via off-premise channels.
Looking at the three weeks ending April 12, Bump Williams Consulting finds beer sales up an average of 16.6%, wine up 26.0%, and spirits up 31.8%. Note that for the eight weeks ending February 23, beer sales were up an average of 3.4%, wine down -2.6%, and spirits up 4.1%, so these are 13.1%, 28.6%, and 27.7% above that existing trend, respectively.
Nielsen data for roughly the same period (three weeks ending April 11), show very similar results, though I only have dollar sales to compare. Beer is up 18.7% in those three weeks, wine up 32.0%, and spirits up 31.1%. The 52-week growth for each through February 29 was 3.7%, 1.0%, and 5.0%, so those are 15.1%, 31.0%, and 26.1% above trend, respectively.
On-premise data is typically less comprehensive than scan data for its channel, and so reliable data is hard to get for the total U.S., but Nielsen CGA has some data in the markets they cover. They show dollar velocity down 72% during the week of April 11 versus what would be expected. That’s stronger than our survey of members about their distributed draught beer sales, but that makes sense, since packaged alcohol might be easier to deliver than draught via crowlers/growlers and you might see differential trends between craft beer and beer, wine, and spirits more generally.
Want more on-premise insights like this? BA members have access to the Nielsen CGA On-Premise Knowledge Center. Find the link to access it here.
Side note: We asked about “distributed package” in our survey and respondents answered +8%. We can use the difference between that number and the +13% in scan data to estimate onsite package trends. Using a rough breakdown of packaged craft being 92% off-premise and 8% on, the math works out to packaged craft down 53% (47% of its former volume). If we use this to calculate a total on-premise number for craft, it’s closer to down 87%, but craft has a much higher draught to packaged mix than overall beer.
Going back to our original thread, if we use 75% as a volume number for on-premise (it’s hard to precisely adjust dollar sales to volume, so this is an assumption), and assume 20% of beverage alcohol volume is on-premise and 80% off-premise (sources: Beer Institute and DISCUS/Liquor Handbook), apply the volume growth trends from the Bump Williams Consulting analysis, and then weight by 2019 volume share…. you get a 2.2% growth in volume sales compared to trend over those three weeks. So all it takes is an assumption that the additional 2.2% sales bump are stock up purchases that aren’t immediate consumption, and you can argue that total beverage alcohol consumption hasn’t gone up at all. If you assume the stock up percentage is more than that, it may have even gone down.
By the way, if we do this for consumer spending, beverage alcohol spending has certainly gone down. On-premise is roughly half of total dollars, so with static volumes or even slightly positive volumes but shifting heavily to off-premise, consumers are spending less on beverage alcohol, a loss mostly realized by on-premise retailers.
There is support for that conclusion in consumer surveys. A Morning Consult survey conducted April 3-5 found that 19% are drinking less “due to social distancing and self quarantining practices,” 16% are drinking more, and 55% are drinking about the same amount. So that’s a net -3%. Similarly, a YouGov survey conducted March 23-24 reported that 25% of drinkers are drinking less than usual, 52% no more or less than usual, and 20% are drinking more than usual in the past two weeks. So that’s net -5%.
Neither of these surveys should be seen as definitive. Moving beyond common survey issues like sample error, or even the specific issues with beverage alcohol surveys, which tend to under report consumption, these surveys don’t indicate the previous drinking levels of the more and less groups, and they don’t provide insights into how much more or less. But they also don’t clearly point in the direction of more consumption.
To sum, it’s very clear that beverage alcohol buying patterns are shifting. What’s much less clear is if total consumption levels have changed at all. Yes, some Americans are drinking more, but some are drinking less, and the net effects of those changes aren’t clear. You could say the same thing during most periods of time. In addition, there are clearly shifts in consumption level across categories and package types that will have a differential effects on certain parts of the industry. I’ve covered this extensively already in relation to craft brewers, whose sales are clearly down.