Modeling the Internal Princple-Agent problem

Tyler Cowen asks why "Call me naive but . . . " works better than "I am not naive." After all both are intended as signals that the person is not naive. This works because it takes advantage of the fact that a person is not a unitary entity.

Most of the time we can act as if people have a single well defined objective function. Indeed, much of the time we can act as if households have a single well defined objective function.

However, there are times when this model breaks down. Households disagree internally and people have personal internal conflict.

For the sake of this model lets say there are two entities within each person. A purely internal manager and a communications agent.

The communication agent wants to be trusted. The internal manager wants to feel self confident. If the internal manager is indeed naive, then he will become upset at the communication agent's statement "Maybe I am naive" and potentially shut him down.

The communication agent will have to deal with balancing the threat of being shut down with his desire to be trusted. Making the statement "Maybe I am naive" signals to other agents that the threat of being shut down is low and therefore the internal manager is probably not naive.

I believe nternal principle-agent problems are part of why deceit is more difficult and less common than traditional models might suggest.