What is principal-agent problem?

The principal-agent problem occurs when a person or firm, the principal, hires someone else, the agent, to act for them and the agent's interests differ. It matters because the principal cannot fully watch the agent, so the agent's choices may drift from what the principal wants.

How it works

The principal sets goals and pays the agent, but the agent controls the effort and the details. Because the principal cannot see everything, the agent may take the easier path, pursue personal gain or take risks that the principal bears. Two things create the gap: different goals and unequal information.

The usual response is to align interests, for example by tying pay to results, and to add monitoring. Both have costs. Pay linked to a single measure can lead the agent to chase the measure, and monitoring takes time and money. The trade-off is between the cost of the control and the loss from the drift.

Example

Say an owner hires a manager who can pick a safe project that earns 100 or a risky one that earns 300 with probability 0.5 and zero otherwise, an expected 150. The owner prefers the risky one. If the manager gets a fixed salary and loses the job on a failure, the safe choice is better for the manager. A bonus of 10 percent of profit makes the expected bonus 15 for the risky project against 10 for the safe one, so the interests line up. The numbers are an illustration.

How to model it in Machinations

Create a Pool for the owner's profit and a Pool for the manager's pay. Use a Gate to split outcomes between the safe and risky projects according to a probability Lever. Add a Register that computes the bonus from a percentage Lever, and a state connection from the bonus to the manager's choice. Play many times to see the spread of results and change the bonus to find where the manager's choice matches the owner's wish.