Drinkers Earn More Than Non-Drinkers:
Today the Journal of Labor Research published an interesting report, summarized in this 20-page document, noting that people who drink alcohol make more money than people who don't — and people who go out to bars make the most money of all. From the executive summary:
A number of theorists assume that drinking has harmful economic effects, but data show that drinking and earnings are positively correlated. We hypothesize that drinking leads to higher earnings by increasing social capital. If drinkers have larger social networks, their earnings should increase. Examining the General Social Survey, we find that self-reported drinkers earn 10-14 percent more than abstainers, which replicates results from other data sets. We then attempt to differentiate between social and nonsocial drinking by comparing the earnings of those who frequent bars at least once per month and those who do not. We find that males who frequent bars at least once per month earn an additional 7 percent on top of the 10 percent drinkers' premium. These results suggest that social drinking leads to increased social capital.
What a perfect report for a Friday afternoon.

  UPDATE: I fiddled with this a bit to note that the document is more of a "report" than a "study."
Funded by Seagrams and Anheuser-Busch, I bet.
9.15.2006 4:56pm
Also, did they control for significant non-drinking populations that have other issues, like insular religious communities, that might impact earnings?
9.15.2006 5:00pm
liberty (mail) (www):
Or else those that earn a lot unwind by drinking. High stress jobs that pay a lot can push people to drink.

But thats a less fun interpretation.
9.15.2006 5:00pm
Jake (Guest):
Alternatively, people who work hard (and thus earn well) may be driven to drink. And those who are particularly successful can afford to frequent bars more often...
9.15.2006 5:00pm
Ted Frank (www):
The same study shows that Judaism has a much larger effect on income than drinking.
9.15.2006 5:05pm
Similarly to Jake's point, could this study really be finding causation in the other direction: people who earn more have more disposable income and therefore spend more money on things like drinking?
9.15.2006 5:07pm
Clayton E. Cramer (mail) (www):
Hmmm. $3 for a beer in a bar; a six pack will cost you about $7? (I don't drink, so I'm somewhat guessing.) Maybe people with higher incomes can afford to drink in bars, and others can't.
9.15.2006 5:08pm
Stephen C. Carlson (www):
The same study shows that Judaism has a much larger effect on income than drinking.

How about Jewish drinkers, then?
9.15.2006 5:08pm
Gabriel Rossman (www):
The study doesn't control for whether you live in a city (where earnings are higher and bars are more accessible). Likewise, as Jake said, people with lower earnings may not be able to afford to drink as much (or to drink at bars, which is much more expensive than drinking at home).
The real test would be to use longitudinal data. Fortunately, many panel studies include questions on both alcohol and earnings and some of them even have social capital or network questions.
9.15.2006 5:09pm
Beerslurpy (mail) (www):
People who earn more have more stress, and thus a greater incentive to particpate in substance abuse.
9.15.2006 5:14pm
happylee (mail):
As much as I love the paper, the topic and the authors, it seems to me that the paper embodies all that's wrong with modern economics (I should say "economics"). Uh, let's see, grab some data, plug it into a spreadsheet, run it through some regressions, and, viola!, the veil is dropped and new knowledge reveals itself to us like some Playboy centerfold.

One day, back when I was a spreadsheet monkey, my boss came into my office after a day of getting ripped to shreds in cross-examination at trial. Ooops. Lo and behold, dressing the pig up in lots of pretty equations don't make it a gurl! Imagine that. Ditto for cause and effect, especially in human relations.

That said, I've passed this paper around with an invitation to come out and play.
9.15.2006 5:16pm
liberty (mail) (www):
"The same study shows that Judaism has a much larger effect on income than drinking."

Ah, but its easier to pick up a drinking habit than to convert to Judaism.
9.15.2006 5:19pm
jvarisco (www):
I would be curious, assuming the study holds up, if drinking is also correlated to lower life expectancy, and if so whether drinking increases total lifetime earnings or not.
9.15.2006 5:25pm
frankcross (mail):
This is not a study, it is a review of the literature. The challenges in the comments are legitimate ones, but they may have been considered in the underlying studies. You would have to actually read them to know if the criticisms are valid.
9.15.2006 5:34pm
Despite all the naysaying, I'd note that anecdotally, I've heard some people in business and law suggest the same sort of thing (that it can be important for the employees to be able to go out and have a few drinks with coworkers and clients). And as a general matter, for good or ill, alcohol consumption is typically an important element in many social events and interactions.
9.15.2006 5:34pm
One's personality must be playing a role. People who are outgoing are more likely to meet with friends in a bar. Also, outgoing persons are more likely to take risks when it comes to their professional lives, therefore increase their overall chances to succeed professionally.

I'd like to know if, in comparison, generally isolated and shy people are more conservative when it comes to drinking habits. It seems obvious to me that a correlation between personality and earnings could be established, what about bar frequentation?

Lets not forget that one can frequent bars (hence build/maintain their 'pub' social network) and abstain drinking. Could there be confusion between bar frequentation and alcool drinking?
9.15.2006 5:35pm
A. Zarkov (mail):
Look at the statistical analysis presented, and the large number of independent variables. The late Leo Breiman (professor of statistics at UC Berkeley) wrote on the problems of trying interpret a coefficient in a high dimension regression analysis. He discussed an example on sex discrimination in employment. You run a regression on salaries with a bunch of independent variables thrown in like age, etc. If the coefficient for "sex" (nowadays people say "gender," by I don't) tests as being non-zero, you file a lawsuit. His little note was called "The Dirty Little Secrets of Statistics," or something to that effect. Unfortunately that note disappeared from his web site, and I'm not sure it was ever published. I hope I can find it in my notes because it was a gem. It shows how misleading regression analysis can get when you have more than two or three variables.
9.15.2006 5:37pm
Dick King:
I have heard from more than one source [but not from any particularly reliable one on this issue] that the reason Russians drink so much, some entirely to excess, is that in their culture somoene who won't let himself get drunk in the company of his peers is presumed to have something to hide.

This story would be an interesting explanation of the culture of three-martini lunches among businessmen closing deals. I don't know whether they really happen in significant numbers, and populists love to rail at them, but maybe they serve a business purpose and therefore should be fully deductible.

9.15.2006 6:12pm
Peter Wimsey:
It would be interesting to see how drinking/nondrinking corresponds with educational level. I suspect that nondrinkers tend to be less educated than drinkers, based unscientifically on people that I know.
9.15.2006 6:24pm
Jon Rowe (mail) (www):

Maybe people with higher incomes can afford to drink in bars, and others can't.

I doubt it. In my observations, people with higher income and greater wealth (unless they are obscenely rich and even some of those), tend to be more provident with their $$. And poorer people tend to be more improvident. People who can least afford to smoke cigarettes and play the lottery are every bit as likely if not more so to do so. Being poor or of modest income doesn't stop people from blowing their $$ in bars.
9.15.2006 6:36pm
Just returned from the bar, where I brought this article up and took a survey of the 12 people sitting at the bar at 6:00 on a Friday... Of the 12, 5 were "once a week drinkers," 2 were tourists, and the other 5 were regulars... I didn't ask for actual incomes (that wouldn't be very cool) but all 5 regulars had decent paying jobs (1 teacher, 2 well-to-do farmers, 1 state representative, 1 owner of a small time construction company) Of the "once a weekers" 3 were builders, 1 owned a garage, and 1 custom instrument builder.

Totally anecdotal, impossible to extropolate from, but there we are... I'm going to grill a chicken...
9.15.2006 7:52pm
I suspect that the British word for those who frequent bars once per month is tee-totaller.
9.15.2006 7:59pm
frankcross (mail):
Apologies for my first post, for some reason my Acrobat screwed up and cut it off before the study.

Re Zarkov's point. Don't jump from one extreme of worshipping stats to the other. Multiple regression is very reliable when properly done. But I think Zarkov's criticism may apply here. There is some strange independent variable selection (number of siblings?) not grounded in good theory that I can see.
9.15.2006 10:48pm
B. R. George (mail):
Is there a reason to exclude the most obvious explanation? I'm thinking of something along the lines of `earnings increase drinking because they increase disposable income'.

After all, having more cars is positively correlated with earning more, and nobody would conclude from that report that cars cause success.
9.15.2006 11:32pm
Mary Katherine Day-Petrano (mail):
I think this report is caca-laca. For one thing, do we even know if the drinkers sampled might be young "earners" whereas the abstainers might be very old and ready to retire? No, we don't know, because conveniently, in order to produce the desired conclusion that 'drinking is a good thing,' the Ph.D's left those significant facts out. We really do not know much about the relevance or representativeness of the sample.

Also, this is a great word the reporters have chosen, "earnings." So what does THAT mean? Is that gross revenues? Gross income? Wages? Self-employment income? After taxes? and, I can't help myself, I want to know if the "earnings" of the drinkers are calculated before or after all the $$$ spent on the booze is subtracted? I can certainly envision how "earnings" might have been manipulated in this report thusly: abstainers and drinkers gross "earnings" are the figure used, but the abstainers do not spend any $$ to buy all that expensive booze, whereas (in order to build "social capital") the boozers, I mean the drinkers, have to spend lots of $$ on all that booze they drink, and even more $$ to treat all their rather large circle of alcoholic "social capital" friends to several rounds.

So, in fact, this report may not even reach a valid conclusion (even though the desk top publishing was first rate), because it may be that after the drinkers booze-related spending is subtracted fom the "earnings" (if "earnings really mean gross "earnings"), then we would learn that the drinkers are really taking on debt (loans) to support their overspending habits. But the reporters don't worry about that little glitch, do they?

See, I think EV is trying to teach us that we are not supposed to take statistical reports at face value, since they can look scientific, but really be a whole bunch of hogwash to influence more people to support the idea that drinking is a *good* thing. (But I am certainly not poaching on EV's mathematical greatness, not me, because everyone knows I can't do algebra).

And this brings back memories of every day when I walked across the street between my law school and business classes; the problem on the business side of campus was always leaving out the cost of those *other* contingent expenses drinking can cause -- such as having to pay the personal injury verdict for that $1 M drunken driving accident the "social capital" career-climbiing drinker caused when he got a little too drunk, fumbles with his keys, and tried to drive home.

On another note, I wonder how the report on drinking "earnings" would compare to building "social capital" leading to higher "earnings" for the reefer set. In the Medical Marijuana States, of course.
9.15.2006 11:58pm
Mary Katherine Day-Petrano (mail):
Orin, I did not mean to give EV all the credit for your post, but I just never saw one like this on your blog. It is rather EV-esque, and somehow I think EV encouraged you to take the heat for this post. So credit is due you both, but I want to know if this is some sort of critical thinking quiz?
9.16.2006 12:04am
The River Temoc (mail):
the reason Russians drink so much, some entirely to excess, is that in their culture somoene who won't let himself get drunk in the company of his peers is presumed to have something to hide. This story would be an interesting explanation of the culture of three-martini lunches among businessmen closing deals.

As someone who has worked extensively with Russians and non-Russians, I can tell you that there's no tradition of three-martini lunches among businessmen closing deals in the West. In Russia, the tradition is drinking beer or vodka, not martinis. Although there was a damn good martini bar in Denver I went to on a deal once.
9.16.2006 1:07am
They've found a correlation. To test which is cause and which is effect, perhaps they should see whether those who increase/decrease drinking increase/decrease total income, or whether those who increase/decrease total income increase/decrease drinking.

I suspect the causation is the other way. As with most forms of recreation (including legal and illegal drug use), increasing income usually makes for increased disposable income, and thus additional economically practical recreational opportunities. Speaking only for myself, whenever I've been in a financial pinch, the booze is the second thing to go. (Eating out is the first — no addictive threshold to overcome, and I'm a durn good cook.)
9.16.2006 2:37pm
Mary Katherine Day-Petrano (mail):
"total income" -- abb3w, see you fell for the scam. You did not read what the report says, they are not measuring "total income," but an undefined ambiguous "earnings." Before one can offer any useful relevant critique, one must carefully read what one is critiquing. It would be impossible to test cause and effect without first addressing the points I raised in my eearlier post above.
9.16.2006 3:12pm
EricRasmusen (mail) (www):
One does have to look at this kind of study. The basic question to ask yourself is "What kind of people are total abstainers?" To do the study correctly, you should give some descriptive statistics on that, and some backgrounds that can't be gotten from the data-- that Moslems and Mormons don't drink for example.

Some things I see missing from the regression are urban/rural, as noted above, health (does the person abstain because he has severe health problems, which also means he lives on disability payments), and past drinking (did the person ruin his life with alcohol, join AA, and now is putting himself back together).

That said, even if the result survived, only its size would be surprising,a nd we wouldn't infer causality. I wouldn't be surprised if people who prefer pork to beef also earn more--- such things are correlated with a lot of omitted variables.
9.16.2006 7:12pm
Mary Katherine Day-Petrano (mail):
Well, autistics like to drink champagne, with strawberries ...
9.17.2006 1:30am
triticale (mail) (www):
The premise is that I can acquire social capital by consuming alcohol in the presence of other drinkers. The consumption of alcohol is not essential to the results. I can acquire more social capital by consuming a boozeless tonic and lime at the tavern in the company of others as I did to positive effect at a recent blogger bash than by consuming a wee dram of single malt at home as I'll do of an evening if I have no driving ahead of me.
9.17.2006 10:50am