By Nick Colas of Convergex
Let’s Make a Deal, Monty!
Today we’ll reprise a favorite topic from prior years of this note: the Ultimatum Game. This most-replicated of all behavioral finance studies highlights the significant role human notions of fairness play in what classical finance says should be a straightforward set of decisions. The game itself is simple: two people have to split a pot of money with one person offering a percentage and the other accepting or declining. The former results in both walking away with their agreed take; the latter means no one gets anything. Offers should be minimal – any money is good, right? But in practice offers have to be +35% to get a deal done. Recent research highlights just how capricious real people are in their decision-making. For example, attractive people can cut more advantageous deals, prolonged eye contact is a no-no, and negotiating through social media avatars seems to ease the process of getting to more agreements.
“The secret of life is honesty and fair dealing. If you can fake that, you’ve got it made.” That witticism comes from Groucho Marx, and it’s one my favorite quotes on the slippery subject of fairness. We learn early on that “Life’s not fair”, even if we spend the rest of our existence hoping that somehow we can make it so. At best we fake our way through it, grabbing bits of ersatz equality where we can.
Classical finance solves the fairness quandary with one of its famously simple models: anything with positive utility is desirable. Someone handing us a dollar with no extra effort on our part is a happy occurrence. End of story.
Behavioral finance takes issue with that construct, and it has a powerful case study in its corner: the Ultimatum Game. The landmark paper on the topic (Guth, Schmittberger and Schwarze) dates to 1982 and it has provided researchers with fertile ground for everything from undergrad psych papers to Ph.D. theses ever since. The basic format is as follows:
- A researcher has two subjects who do not know each other flip a coin. The winner is handed a nominal sum, say $100.
- The winner of the toss, cash in hand, is told they must offer a portion to the loser. It is up to them how much to share.
- The catch to the game, explained to both participants, is that if the recipient of the offer turns it down then no one gets any money. If, on the other hand, the recipient accepts the offer then both get the cash. Either way, the game is over and the subjects are free to leave.
- Classical finance has a clear strategy for the person making the offer: $1. After all, the person evaluating that amount is still getting more than they would otherwise have.
- In what are now hundreds of research papers on the topic, 1 percent never yields a positive outcome. The recipient of the offer turns it down. In practice, 20% is the lowest generally acceptable offer, and in some cultures the offer has to go over 50%. The median accepted offer is generally 35-45%.
- The reason that offers have to be close to half, says the behavioral finance researchers, is because humans value fairness much more than classical finance understands. Hand someone a dollar out of the blue, and they are happy. Hand them a dollar when someone else is getting $99 because of the chance of a coin flip, and they aren’t.
That’s the basic insight of the Ultimatum Game, but the research continues and for the remainder of this note we’ll summarize several recent updates to the work.
One question I get when I discuss the game is “What happens when the stakes are really high? If the amount was $1 million, would someone really turn down $100,000?” That’s a tough experiment to do, mostly because of the inherent cost of running a typical 25-100 person sample size. Still, in 2011 some researchers (Andersen et al, link at the end of this note) went to rural India and tried the game on +450 villagers. The stakes went as high as 160 hours of work –something like $5,000 to an American. The results are pretty much what you’d expect. Those offering the split do offer lower percentages then when the amounts are more trivial, but rarely very low amounts. Those evaluating the offers take the lower amounts, with the cutoff for +90% certainty of acceptance around 10 days wages.
With Americans living more of their lives online, one interesting question is what this does to social cohesion. One 2014 paper (Greiner et al) looked at what happens when you play the Ultimatum Game on Second Life, the online virtual world. The surprising answer, from the abstract: “In Second Life we detect a shift to more cooperation when there is no communication.” In other words, more deals get done (offers are deemed “Fair” and accepted) when the players do not chat or otherwise communicate. “The higher cooperativeness in the virtual world lowers the need for additional communication between avatars in order to achieve efficient outcomes.”
Since the Ultimatum Game essentially measures an emotional arbitrage – how far people will go to harm themselves for the sake of fairness – one logical avenue of research relates to the person making the offers. In a 2015 paper (Ma, et al), researchers in China paired male evaluators with female “proposers” (the term used in the literature to denote the person making the proposed split). In findings that you probably won’t find all that surprising, male evaluators accepted lower offers from women they deemed “Attractive”. We await the reverse study to see if this works the other way.
It isn’t just a “Beauty Premium” (not my term - that’s what it’s actually called in the literature) that draws the attention of evaluators in the Ultimatum Game. Make too much eye contact as you offer a split, and you’ll be perceived as overly aggressive (Tang, Schmeichel 2015). That will make the evaluator to a “More dominant choice” – essentially upping the required split to make a deal happen.
On the flip side, some research focuses on how to manipulate the proposer into offering more (Reed, et al 2014). By showing them clips of angry faces and written demand to make an outsized offer (70%), researchers were able to shift proposed splits higher. Their interpretation is that humans respond to visible anger in others, even if the threat is not actually present.
In summary, the Ultimatum Game is a unique window into how humans actually make economic decisions and it is a distinctly opaque aperture. The best we can probably do is to accept that context plays a far larger role in economics than classical models would suggest. And don’t stare at someone or make angry faces if you want them to work with you.
Sources:
Original study: http://teaching.ust.hk/~bee/papers/guth.pdf [4]
Literature review: http://www.u.arizona.edu/~martind1/Papers-Documents/Seventeen_JEBO_2014.pdf [5]
High Stakes Game: http://rady.ucsd.edu/faculty/directory/gneezy/pub/docs/ultimatum_aer_published.pdf [6]
Online play: http://www.sciencedirect.com/science/article/pii/S0167268114000171 [7]
Facial attractiveness: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4354284/ [8]
Eye Contact: http://link.springer.com/article/10.1007/s10919-015-0206-8#/page-2 [9]
Angry Faces: http://pss.sagepub.com/content/25/8/1511.short [10]
