The Duc de Richelieu remarked that the fate of empires was often changed by an extra bottle of Johannisberg. American businesses evidently agree. They spend roughly $20 billion each year wining and liquoring clients and business partners. That’s about 12 percent of the retail alcohol market. It’s not largesse. The liquor is expected to pay its own way, by causing customers and suppliers to make deals they wouldn’t, and quote better prices than they would, when sober. The anchors known as trial balloons are often broached over drinks, and occasionally the outline of a deal is sketched on a cocktail napkin. The IRS allows businesses and individuals to write off alcoholic “entertainment” as long as it is “ordinary and necessary.” Nobody seems to doubt that it’s both.
When the economy goes south, alcohol-lubricated deal making is one of the last things to be cut. As the New York real estate market tanked in 2008, Prudential Douglas Elliman was offering high-net-worth customers condo tours awash in free Talisker and Lagavulin whiskey— which sell for $60 and $77 a bottle. “A little bourbon” could be good for sales, suggested one broker, who sounded confident that the liquor budget would be recouped and then some. Real estate journalist Christine Haughney wrote, “Just as a few drinks may coax timid traders onto a dance floor, it could help them muster the courage to buy multimillion-dollar apartments.”
Are the prices people agree to under the influence different from those they’d accept fully sober? A British team at the University of Leeds and Oxford did an experiment in which social drinkers consumed alcohol or a placebo, then filled out a battery of psychological tests, including a series of choices between gambles. The experimental cocktail was a tangy blend of tonic water, Tabasco sauce, and alcohol—or, for the control group, a virgin counterpart. The amount of alcohol, scaled to body weight, was the equivalent of three strong drinks. (Yes, it’s hard to believe that the subjects couldn’t determine whether they were in the alcohol group or the control group. The perennial problem with alcohol experiments is that it’s impossible to devise a credible placebo.)
In folk wisdom, alcohol promotes risk taking. You have to wrestle the keys from a drunk friend who’s sure he can drive; “free” liquor in casinos encourages customers to bet more recklessly. Yet in many respects, there wasn’t much difference between the intoxicated and placebo groups in the British experiment. Alcohol did not wash away prospect theory. Both drinkers and nondrinkers were loss-averse with gains, risk-seeking with losses. Both groups reliably favored “P bets,” those with the better chance of winning.
The one significant difference that the British team found was a very specific one. It occurred when participants were presented with “difficult” Lichtenstein-Slovic-type choices involving large losses.
If you want to try the experiment yourself, you’ll need to make three drinks. For each, carefully measure 3.6 ounces of 80-proof vodka into a highball glass and top off with tonic to make 10 ounces of beverage. (The 3.6 ounces applies to a 150-pound person. Scale the alcohol amounts accordingly.) You have fifteen minutes to consume all three drinks. Then wait ten minutes and answer these two questions (in both, I’ve converted the British psychologists’ “points” into dollars):
Question 1. Which would you rather have?
(a) a 50 percent chance of winning $10; otherwise you lose $10
—or—
(b) a 66 percent chance of winning $20; otherwise you lose $80.
Question 2. The choice is
(a) a 50 percent chance of winning $10; otherwise you lose $10
—or—
(b) a 66 percent chance of winning $80; otherwise you lose $80. Write your answers here: 1. ___ and here: 2. ___. Now for the most important part of the instructions: Do not drive, ride a bike, operate machinery, or do anything stupid for at least two hours.
By design, both questions present difficult choices. No answer is indisputably better. Option (a), identical for both questions, is a fair coin flip. Because losses are regretted more than gains are valued, that’s subjectively a losing bet for almost everyone.
The two (b) gambles are P bets, and we know that everyone likes P bets. Choose (b), and you will probably walk away with some quick cash. (As Guy Grand asks in The Magic Christian, “How ’bout it, pal—got a taste for the easy green?”)
Hold on, there’s a catch. Both of the (b) bets carry a worrisome penalty of $80. That makes these bets less attractive. With the first and possibly the second question, participants are forced to choose the lesser of two evils.
There is only one difference between Questions 1 and 2. It is the amount of the win in option (b). It’s $20 in the first question, and a more generous $80 in the second. Logically, you would expect more people to choose (b) in Question 2 than in Question 1.
That is indeed what happened. Most subjects (sober or intoxicated) picked (a) in Question 1 and (b) in Question 2. However, the shift was larger with the sober group. They were more responsive to the change in win amount than the alcohol group was.
The British team’s overall conclusion was this nuanced one. When weighing the prospect of large losses, drinkers had a diminished capacity to factor in the amount of gains. When the gain in (b) was raised from $20 to $80, many drinkers didn’t change their choice. It was as if they didn’t notice.
Alcohol narrows an already limited scope of attention, a phenomenon that’s been dubbed alcohol myopia. This places yet tighter bounds on rationality. Both questions 1 and 2 make competing demands on the chooser’s attention. Since subjects had been asked to choose (rather than name a price), they would have focused attention on the probabilities of winning (50 versus 66 percent). Second, they had to worry about the downside risk. The biggest loss possible was $80, and this is so much more than the $10 loss in (a) that the $80 loss commanded the attention. Much like a magician’s “big move,” the $80 loss created misdirection. The subjects were preoccupied with weighing how bad a 1-in-3 chance of an $80 penalty is, and whether that should overturn an innate preference for a P bet.
Buzzed or sober, this left few cognitive resources for giving the (b) option gains the consideration they deserved. There were just too many numbers to juggle. The intoxicated subjects were especially overwhelmed. They ended up not paying much attention to gains. This sometimes led to decisions that appear risky, and other times to choices that were anomalously conservative.
There are many counterparts in business. Quoting a price to a potential client is a gamble. It can never be known exactly how much work will be involved, how demanding the client will be, what can go wrong with the job, and what the relevant chances are. A three-martini lunch converts complex problems into deceptively simple ones. Any prices quoted are apt to be “wrong” in the sense of not reflecting all the relevant information. A vendor may lose profitable business by pricing himself too high and saddle himself with ruinous contracts by pricing too low.
One bit of singles bar wisdom might help you to remember this rule. Going home with a stranger is a gamble (posing small chances of date rape, an STD, and/or a bad marriage followed by a messy divorce), but normally a favorable one. Drinkers forfeit the ability to discriminate between the worthwhile risks and the bum bets. After a few drinks, they all look good.