The problem motivating this research is one of how to properly structure bargaining under incomplete information. Such a problem arises naturally in the context of the so-called "transfer price problem," i.e., the problem of designing mechanisms for the pricing of goods and services that are transferred from one division of a firm to another. A typical feature of this situation is that the relevant data, e.g., the costs of the supplier division and the benefits of the customer division, are not well known to the participants. For example, it would be normal for the supplier's costs to be better known to the supplier than the customer, and for the customer's benefits to be better known to the customer than the supplier. This research project continues and extends work undertaken previously by the two investigators on sealed-bid mechanisms for bargaining. The sealed-bid procedure is one widely used method for structuring bargaining under incomplete information because it has a simple linear equilibrium with desirable welfare properties. Unfortunately, there are an infinite number of other equilibria with welfare properties that range from second-best optimal to zero. As a practical matter, this mechanism is not very appealing, since there is no guarantee that the "good" equilibrium will be picked out of the infinity of other equilibria that exist. The investigators conducted the first controlled laboratory experiments on the sealed-bid mechanism. They demonstrated that buyers and sellers bargaining under a sealed-bid mechanism on average chose the simple, linear bidding strategies that resulted in "good" equilibria. This project pursues a number of important unresolved questions raised by these experiments. For instance, what exactly account for the emergence of step-function-like bidding behavior as the horizon of the experiment increases? How can we account for the asymmetrical behavior of the buyers and sellers in these experiments? Why did unstructured face-to-face bargaining perform better than all other forms of bargaining? Does "cheap talk," i.e., talk that does not have any binding impact on the eventual negotiations, influence the outcome of the process? The project makes significant methodological and substantive contributions to our understanding of bargaining behavior. Methodologically, it demonstrates whether commonly used experimental practices such as identifying buyers and sellers or repeating the experiment a small number of times bias the results of bargaining experiments. Substantively, it derives rigorously and tests experimentally for the presence of regularities or "laws" in bidding behavior under incomplete information.
