A firm's research division consists of a manager and a scientist. The scientist is respon- sible for proposing new research projects. The quality of the proposed project depends on the scientist's effort. If the scientist exerts high effort, then the proposed project is worth 1 if it is developed. If the scientist exerts low effort, then the proposal is worth 1/2 if it is developed. High effort costs the scientist 1/3.

The manager is responsible for selecting which proposed projects to develop further and directing their implementation, not for proposing new projects. However, the manager has extensive scientific training, and is therefore capable of proposing promis- ing projects. In fact, the manager is even more skilled than the scientist?a project proposed by the manager is worth 3/2 if it is developed. The manager's other job responsibilities limit his or her ability to propose new projects. With probability p, the manager can complete his or her other tasks easily, using the remaining time to start a new project proposal. With probability 1 ? p, the manager's other responsibilities require his or her full attention. In this case, the manager is unable to make a new proposal. The game between the scientist and manager is represented by the following tree.

First, the scientist decides whether to exert high effort (branch H) or low effort (branch L) and makes a proposal. Second, a coin flip determines whether the manager proposes a new project idea (probability p). If the manager does not propose a new project (probability 1 ? p), then the game ends. If the manager does propose a new project, then he or she decides whether to select the scientist's proposed project for further development (branch S), or whether to select his or her own (branch M).

The scientist incurs cost 1/3 whenever he or she exerts effort, and gets a benefit equal to the value of his or her proposed project, if it is selected to be developed. If the manager does not select the scientist's proposed project, then the scientist's gets no additional benefit. The manager's payoff is equal to the value of the project that he or she decides to develop, whether it is the scientist's or his/her own.

Scientist H L o coin flip O cam Flip I-p V-p M M S O O 3/2