Testing the "Theory of the Pie" approach to Negotiation

This experiment will test whether people can be taught a better way to negotiate. In the experiment, two parties will be tasked with allocating an amount of cash between themselves. The two parties have an opportunity to invest together and thereby earn a greater amount of interest than if they each invest separately. The question is: how will they split the proceeds? There is a typical default solution where parties agree to a proportional split of value created (based on the relative amounts invested). Instead, we believe an equal split of the extra value created better represents a fair solution and one that aligns with the equal contributions of the two parties--even though they may be different in size. The basic idea is explained in this video (https://tinyurl.com/BNYALEIRB) At the big picture level, we want to test the extent to which a new framework for negotiation leads to a different outcome. At the more granular level, we want to see how much coaching is required to achieve this result. This leads us to consider four potential interventions: (1) No coaching . (2) General introduction to the pie framework (similar to the video) but no specific information to the case at hand. (3) General introduction to the pie framework (similar to the video) along with a specific application to the case at hand. (4) General introduction to the pie framework (similar to the video) along with a specific application to the case at hand and an example where proportional division hurts the larger player.

2021-07-08

2024-07-07

We are looking to see whether taking the pie perspective leads to a more fair outcome. Are the parties in the traditionally weaker position able to get more or even half of the pie when they employ the pie perspective. Are they able to move people away from proportional division?

A second and related question is whether employing the pie perspective will lead to more or less agreements. It is possible that having two conflicting views of fairness (proportional division and split the pie) will lead to fewer agreements. It is also possible that employing the pie perspective will allow both parties to converge on a view of fairness and thereby facilitate more agreements. We do not have a view as to the sign of this effect. We also note that there are relatively few cases of no agreement and so there may not be sufficienbt power to test this hypothesis.

As mentioned above, we are also interested in a second experiment in which proportional division hurts the larger party. In this case, we hypothesize that split the pie or no agreement are the more likely outcomes.

One subject is assigned the role of Alice and the other is assigned the role of Bob. Here is an example of the instructions provided to Alice. The Bob role is parallel.

You are planning to buy a one-year certificate of deposit (CD) with the $5,000 you have to invest. Because that is all you have to invest, the bank is offering you a 1 percent interest rate. That means you can earn $50 of interest.

You and Bob were discussing investment options. In the process, you learned that he has $20,000 that he plans to invest in a one-year CD. With the $20,000 he has to invest, Bob can get 2 percent interest. That means he can earn $400 of interest.

The good news is that you can both do better if you pool your funds and invest together. The bank will offer a better rate—3 percent—if you purchase a $25,000 CD. Neither of you have additional funds to invest other than the $5,000 and the $20,000, and there are no other potential investment partners.

With a 3% rate, the two of you can earn a $750 in interest.

Amount Invested Interest Rate Interest
Alice $5,000 1% $50
Bob $20,000 2% $400
Alice & Bob $25,000 3% $750

However, to invest together you have to come up with a way to divide the $750 interest earned. Any agreement you reach is binding. If you can’t agree, each of you will buy a CD on your own.

The two of you will be engaged in a negotiation to determine the split. You have 5 minutes to read these instructions and another 15 minutes to reach an agreement. If no agreement is reached at the end of 20 minutes, your result is no agreement. All of the information above has been provided to both parties.

Parallel information is provided to Bob. Under version Q, no additional information is provided. Under version R, Alice is provided with the private information below and no additional information is provided to Bob. Under version S, the private information below is provided to Alice and Bob is given information about the value of anchoring. Under version T, Bob is given this private information below, but Alice is not. The point of this last intervention is to see if when Bob appreciates that the proportional solution is not fair, does he or she share this information with Alice.

We also plan to run a second negotiation experiment in which the numbers are as below. The difference here is that proportional division now hurts rather than helps Bob. How will the parties reach an agreement in this situation?


Amount Invested Interest Rate Interest
Alice $5,000 1% $50
Bob $20,000 2% $400
Alice & Bob $25,000 2% $500

Private instructions to Alice in versions R and S

In this negotiation, you should act as if you are doing this for real money. You would walk away with whatever you agree to. But if you end up with no deal, you only get the 1% interest on your $5,000, which is $50. For pair person it will be real money. As an extra incentive, we will choose one pair at random from this session and provide them with an Amazon gift card worth whatever dollar amount of interest they ended up with.

Some Bobs believe that the two sides should split the $750 in proportion to the size of the investments. Thus Bob may propose that he gets $600 and you get $150. Bob may justify this by the view that this is fair because you are both earning the same 3% interest rate. Or, this is fair because you are sharing the interest in the same proportion as your investments.

A more principled view of fairness calculates the gain created by the negotiation and splits it evenly. If the two of you don’t reach an agreement, you will earn a combined $50 + $400 = $450 in interest. If you do reach an agreement, you will earn $750 in interest. Thus, reaching an agreement creates $300 of value over what can be gained separately. Bob can’t get any of that extra $300 without your agreement, just as you can’t get any of that extra $300 without his agreement. Since you are equally essential for creating the $300 in value, you should split this total right down the middle, $150/$150. Thus, you should get $200, which is $150 (half of the extra $300) + $50 (what you would get by yourself without a deal), while Bob will get $550 = $150 (his 50% of the pie) + $400 (what he would get by himself without a deal).

To succeed in this second approach, you might want to frame the negotiation using the $300 as the negotiation pie. The negotiation is truly about how to split $300, not $750. In making these arguments, you may share these insights or keep them to yourself. You may negotiate however you prefer. You are not being told to behave in any specific way and you should not say that you are being told to behave in any specific way. If you employ this “splitting the pie” approach (200, 550), you will have to convince Bob by the force of your own logic.

Subjects brought to the lab will be assigned roles in the negotiation at random by a coin flip.

There is only one level of randomization.

No

4

250

100 pairs from classroom experiments, 150 pairs from behavorial lab.