The Value of Information in Agency Models

  • CK's website
  • The Value of Information in Agency Models
  • Abstract:
    In an agency model where the principal cannot observe the agent’s productive action, it is assumed that there are mutually observed signals conveying partial information about the hidden action and some of the signals can be used as a proxy in writing a contract to incentivize the agent to make a desired action. A natural question that arises is among available signals, of which one the principal should make use for incentive provision. This manuscript establishes a simple criterion of ranking signals. I show that signal X is more precise than another signal Y in the sense of Lehmann (1988) if and only if X is more efficient in the sense that the principal can incentivize the agent at a less cost. This equivalence result provides a link to the previous criterions such as Holmstrom (1979) and Kim (1995). Furthermore, I show that this new criterion is applicable to the relational contract environment.