Information Flows And Networks I

Main Menu Previous Topic Next Topic

**Two-Step Search and Option Value**

Note: in order to run the simulation referred to in this slide, go here, to the Java Applet version. You will be directed to download the latest version of the Java plug-in.

In the previous slides of this topic, the decision-maker was faced with a many-step search situation. Consequently, we made sure to keep the other parameters in our model as simple as possible. For instance, the implied discount factor was always at 1.0 - that is, finding a good candidate was no less valuable later in the search process than it was earlier. Also, a priori, all candidates were equally likely to turn out valuable to the company. In other words, the manager was always faced with the same value probability distribution.

In this and the subsequent slides, we drastically reduce the number of possible search steps. This allows us to complicate the simulation in other ways, and explore the influence of several new parameters.

Note that there are only two candidates in the graphic to your left. Candidate 1 has a higher mean value than candidate 2, which is indicated by the circle's larger size. Candidate 2, on the other hand, is a riskier applicant to interview - her value is bound to have a higher variance than that of candidate 1. Accordingly, she is colored red. In this situation, only a couple of strategies are available to the interviewing manager: she could interview either candidate 1 or candidate 2 first, and then consider whether she ought to meet with the other one as well.

The "Manager" agents, then, follow the two strategies available to a perfectly rational interviewer. Manager 2 meets with candidate 2 first. Then, based on the newly-found value of the candidate, she decides to interview candidate 1 if and only if the expected payoff of such a step is positive. Manager 2 follows the opposite procedure, interviewing candidate 1 first.

Click on "Go" and see whose strategy is the better one. At every third time step, each manager's payoff is tallied up, and the graphs below our simulation jump up by a notch. Manager 2 should turn out to have a slight advantage in this situation - if candidate 1 turns out to be better than the well-known value of candidate 2, she chooses to stop, and save on the cost of the next interview. If, on the other hand, candidate 1 turns out to be quite awful, she can still elect to meet with candidate 2. This illustrates an important concept of option value, i.e. the value contained in the early realization of the riskiest choices.