Breaking the Negotiator's Dilemma

In 1992, David Lax and Jim Sebenius (who I was lucky enough to have as a professor) noted that the classic Prisoner's Dilemma has a natural application for negotiation work.

For anyone not familiar with Prisoner's Dilemma, it's a concept from game theory about how people can be trapped in situations where rational behavior leads to bad outcomes.  The simplest form of the dilemma involves two suspects in a major crime who have been arrested and are being questioned.  The police have enough evidence to convict them for a minor crime but offer each one a degree of leniency if he will confess to the larger crime and implicate their partner.

If neither suspect confesses, each gets 1 year in jail for the minor crime.  If both confess, they each get eight years.  If one confesses and the other doesn't, the one who confessed gets off clean while his partner goes to jail for twelve years.

The worst combined outcome for the suspects is if both confess -- they serve a combined sixteen years.  The best combined outcome is if neither confesses -- they serve a combined two years.  The dilemma's fame derives from the fact that the rational decision for each prisoner is to confess, leading to the worst outcome.  The problem is that whatever one prisoner does, the other is better off confessing.  If the first prisoner doesn't confess then the second must choose between a year in jail and walking out a free man.  If the first prisoner does confess than the second must choose between eight years and twelve.  In each case he gets a lower sentence by confessing, meaning that confessing is a dominant strategy.  Since the same calculus applies to both prisoners, the rational outcome is also the worst.

A lot has been written about various Prisoner's Dilemmas, both theoretical and in real-world applications.  In general we speak of decisions to cooperate or defect, where a choice to cooperate improves the combined result and a choice to defect maximizes the individual result at the expense of the other player(s). In negotiation the tension arises over whether one should seek to create value or to claim it.

Creating value requires sharing information.  In particular, it means being open and clear about which elements of a potential deal are relatively more valuable and by how much.  For example, let's assume that a supplier and a customer are negotiating a parts deal.  In addition to price, they are discussing the length of the contract and volume minimums.  At the moment the deal that's on the table is a two-year contract with a minimum annual commitment of two hundred units.

The supplier wants a longer commitment from the buyer with large volume minimums; the buyer wants a shorter commitment and low minimums.  But it's likely that they don't value them equally.  It may be, for example, that each party would be happier with a one-year contract with a three hundred unit minimum.  The more openly both parties share their interests the more likely they are to find this sort of value-creating trade.

Capturing value, however, often requires holding back information.  Let's say you're the customer and the minimum commitment isn't a problem -- you know you're going to buy more than double the current minimum.  If the supplier indicates that she'd like a higher minimum your value capture instinct will be to suggest that this is possible but a very painful concession that will have to be made up for with a price decrease or some other substantial concession on the supplier's part.

If one party is completely open and the other isn't, they will probably find many of the value-creating opportunities, but the less open party will capture most of that created value.

In the classic PD formula, "create value" is cooperate, while "capture value" is defect.  Each party is individually better off emphasizing value capture, regardless of what the other party chooses.  Unfortunately, this leads to the worst aggregate outcome -- both parties emphasize capture and leave money on the table -- in some cases, more money is left on the table than was captured.

So how do we, as negotiators avoid this trap?  It's not enough to read Getting to Yes and committing to focusing on value creation; we need to maximize the chance that our counterparts will as well.

The simplest and probably most reliable way to solve a prisoner's dilemma is with iteration.  If two parties are playing a single round of a PD "game" then, as we noted above, "defect" is the dominant strategy.  (By dominant we mean that it defeats all other strategies in all circumstances.)  But what if we are going to play hundreds or thousands of rounds and our behavior can be tracked from round to round?

A little over thirty years ago, political scientist Robert Axelrod held two "tournaments" to test various strategies for a iterative (i.e. repeating) prisoner's dilemma.  Entrants defined the rules by which their agents (essentially computer programs) would operate, and each agent would play multiple rounds against each other agent.  The rules could not take advantage of knowing that a given round was "final", so the effect of repeated rounds was always present.

The winning strategy was a very simple one: tit for tat.  A tit for tat approach involves cooperating on the initial round and then choosing whatever the other party chose for all subsequent rounds.  If you cooperate with tit for tat, it keeps cooperating back.  If you defect, it defects but if you return to cooperation it does as well.  It retaliates but is quick to forgive.

A tit for tat strategy will still lose, very slightly, to a defect strategy because the defect strategy scores big in round one and then every other round is a tie.  But in a world with lots of different strategies, including some that seek mutual cooperation, it is extremely effective.  Two tit for tat agents meeting will cooperate in every round, maximizing their total benefit.

Tit for tat (and similar approaches) are even more effective with people because we can signal our intentions with words as well as actions.  If you know your counterpart will cooperate as long as you do but will retaliate if you defect, your dominant strategy is to cooperate (assuming you have enough future negotiations to come).

So far, so good -- but how do we make our negotiations iterative?  Fortunately there are some practical steps we can take.

1. Form long-term relationships, especially in areas where value creating opportunities are likely to be substantial.

If you're choosing vendors for your business, think about the scope for value creation that will exist.  For office supplies that potential may be low, so whoever provides what you need at the best price is likely the best choice.  But the more complex the interaction and the more variability (number of terms, potential to meet needs with different products/services) involved in your purchases, the more value creation is likely to dominate value capture as a source of good deals.  In those cases it's worth investing in a partnership where your vendor sees each deal as a step towards a dozen more.  (Unfortunately, many companies use long term relationships solely for gaining leverage for capturing value -- the most successful ones use it to create a pattern of value creation.)

2. Build your negotiation brand/reputation.

If you're known in your industry as someone who will share information openly and look for win-win solutions but who will retaliate (proportionately) against someone who tries to use that against you, many negotiating partners will choose to cooperate with you. In some cases it will be because that's enough to make it the correct strategy; in others it will be because that's the way they prefer to do business (they emphasize capture only because they assume the other side is doing the same).

3. Make this negotiation iterative.

Even if you're only negotiating a single deal and it's with someone who doesn't know your reputation (or you haven't built one yet!) you can create iteration within a single deal provided there are enough issues for it to be worth the effort.  You don't have to share all of your information at once or none at all.  Instead, lead off by saying something like, "With so many factors to consider in this deal it seems clear to me that we can find some great win-win opportunities if we're able to discuss our interests openly.  At the same time, I recognize that this can be difficult so I'd like to start by telling you about some of our key issues.  Then you can do the same, and we can build on that."  Each party can then share

4. Be willing to walk away from deals that are slightly better than your BATNA.

Guhan Subramanian noted that a key assumption of the Negotiator's Dilemma is that people will accept any deal that is better than their no-deal alternative.  If, however, people will walk away from deals unless they create substantial value it becomes more attractive for each party to emphasize value creation.  "Cooperate" can easily become an optimal strategy.  Subramanian points out that in the private equity world, this is already the case -- the big private equity firms have high value thresholds for the deals they do, which leads to more value creation behavior in negotiations.

This plays into the value of your reputation as a negotiator.  If you're willing to walk away from small gains (especially in cases where you have reason to suspect that the pie is at least potentially large) and your counterparts know it, they take a much bigger risk emphasizing value capture and counting on you to provide the information necessary for creation.