Game Theory, the Prisoner's Dilemma and... Winery Membership Organizations, Part 1
May 22, 2014
Game theory describes a branch of science at the intersection of economics, psychology and mathematics which explores models of interaction between rational actors, seeking to explain why and when these actors (be they individuals, companies or even nations) will choose to cooperate or to betray each other. Many of these games are iterative, a fancy way of saying that they happen again and again, like many actions in life, where the actors can learn from their previous actions and the previous actions of their competitors.
One of the classic examples of game theory is the prisoner's dilemma. Imagine the situation where two co-conspirators are arrested on light evidence, and each independently offered the opportunity to inform on the other in return for escaping jail time. If neither chooses to inform, the prosecution doesn't have much of a case and so both get light sentences (say, 1 year). If both choose to inform, both get moderate sentences (say, 3 years). If one chooses to inform and the other doesn't, the one who informs gets no jail time, but the one who doesn't, and sees his co-conspirator testify against him, gets 5 years. You can set up the four possible actions in a grid:
|A Testifies||A Stays Silent|
|B Testifies||A gets 3 years
B gets 3 years
|A gets 5 years
B gets 0 years
|B Stays Silent||A gets 0 years
B gets 5 years
|A gets 1 year
B gets 1 year
At first glance, the actions that the actors should take in this game seem pretty clear: whatever one prisoner does, the other comes out better if he testifies, serving no time (vs. 1 year) if the other person stays silent, and 3 years (vs. 5 years) if the other prisoner testifies. And yet the best outcome for the duo happens if both behave irrationally (or perhaps trustingly) while the worst outcome occurs when both behave rationally (or self-interestedly). Real-world applications of this abound, from arms reduction treaties to curbs on greenhouse gases to the production of individual countries in the OPEC oil cartel.
Graphic courtesy Wikimedia Commons, which has a great interactive Prisoner's Dilemma
example -- from which the screenshot above was taken -- here.
How is this applicable to wine associations? I'm happy you asked. I've been spending a lot of my time thinking of this recently thanks to my positions on the board of directors of two organizations: the Paso Robles Wine Country Alliance (PRWCA) and the Rhone Rangers. In both cases, I believe that the organizations provide a valuable service to their members, but there are significant free-rider problems that discourage membership. Think about it this way. If the Rhone Rangers is successful in its marketing and makes Syrah easier to sell, all Syrah producers benefit, not just the ones who are members. Similarly, if the PRWCA is successful in its promotion and brings more people to area tasting rooms, or raises the profile of Paso Robles so its wines sell better off retail shelves and wine lists, any Paso Robles winery will benefit, whether or not they have paid their membership dues.
And the dues, for the PRWCA at least, are not cheap. We've had two important local wineries drop out of the alliance this year, each saying that they were going to reallocate their marketing dollars to efforts that more directly benefited their bottom lines. These decisions shot a significant (though not crippling) hole through the PRWCA marketing budget. Were the wineries behaving rationally? Actually, yes, they almost certainly were, though if their behavior was generalized everyone, including them, would be worse off. It's a prisoner's dilemma-type example!
The main reason that wineries band together is to gain efficiency with the money they spend. It's generally accepted that an advertising campaign gains efficiency with repetition and with consistency. So, a single marketing campaign, well targeted and well run, is typically more effective at driving behavior than ten different advertising campaigns each one-tenth the size of the original campaign. And the PRWCA gains additional efficiency because of its expertise -- unlikely to be found in-house at any individual winery -- both because of the people running it (thank you, Jennifer Porter) and because of the outside consultants it is able to afford.
Let's look at an example that will require a little math. I'll round the numbers to help them make sense, but it doesn't really matter what the numbers are: the conclusion still holds, as long as we agree that wineries are unlikely to be as efficient spending individually as the group would be spending their money in a coordinated campaign. For ease of calculation, I will assume that there is a 50% loss in overall spending efficiency when a winery splits their money from the group's to spend it individually. And we'll round numbers to 200 wineries, with a contribution of $5000 each (leaving a total budget of $1,000,000). I'll look only at the power of advertising to drive people to local tasting rooms, and assume that each visitor makes 5 tasting room visits when they're in town, and assume that the PRWCA gets one person to make the decision to come to town for each $5 they spend.
OK, back to specifics. Let's look at the impact of the PRWCA's spending of $5000 -- one winery's portion of the total budget -- as a part of their master marketing campaign. This $5000 brings 1000 customers into town. These customers make 5 visits each, or 5,000 visits total, split among the 200 wineries. Each winery receives 25 of these tasting room visits.
Now, let's look at the scenario where Winery X takes the $5000 that they were going to give to the PRWCA and reallocates it to running radio ads in Fresno, Bakersfield and Orange County. This advertising is only 50% as efficient as the PRWCA's marketing, so the winery might expect to pay $10 per customer (double the PRWCA's cost). But the people that this advertising drives to Paso Robles will all start at Winery X's tasting room. So they get 500 visits for their $5000. Other tasting rooms in the region still benefit, as these visitors make on average 4 more visits to other tasting rooms when they're in town. But those 500 customers, who make their additional 2,000 total visits to other Paso Robles tasting rooms, account for only about 10 new visits to each of the other 199 wineries -- less than they would have each received had the same money been spent by the PRWCA. So Winery X ends up 475 customers ahead, whereas every other winery in the area ends up 15 customers behind. Winery X is behaving rationally, and can even point to the fact that its advertising is helping their neighbors gets customers.
Like in the prisoner's dilemma example, the problem comes in the aggregate. What seems like a small loss per other winery looks a lot larger when you multiply that loss by all the wineries affected: the region loses 2,985 tasting room visits from the 199 other wineries, and only gains 475 for Winery X.
If every winery were to make the same decision to spend their $5000 individually, with the same results, they too would get 500 customers to start in their tasting room, and would each contribute 2000 other tasting room visits to the region. Across the 200 wineries, that pool grows to 400,000 visits. Divided equally, each winery gets 2000 of these, plus the 500 from the advertising they paid for themselves. That's 2,500 customers total. If the same $1,000,000 had been spent by the PRWCA, it produces 200,000 customers who make 1,000,000 visits total: or 5,000 visits per winery. By spending their money rationally (an economist might equally say selfishly) they have cut their total number of customers in half.
Of course, not every winery makes this decision. And that's the most frustrating thing for those of us who do contribute to the group's spending. Winery X receives most of the benefits of the marketing that an organization like the PRWCA is doing with the member wineries' money... whether or not they are members. Sure, there are a few ways that they lose out: they're not on the organization's printed map; they're not included in the group's media outreach; they're not a part of the trade outreach that the organization does; and they lose a modicum of goodwill from their neighbors. I actually think that these benefits on their own probably pay for the costs of membership. But if their principal driver of revenue is their tasting room traffic, they still probably come out ahead, at least in the short term.
How does one quantify the benefits that do accrue directly to the members from their membership? And how does a regional organization best respond to this? Game theory has answers for this, too. I explore how to quantify the value of membership in part 2 and will delve into game theory's suggestions for how an organization should respond to those who drop out in part 3.