The principal-agent problem
agency cost · agency problem · principal agent
What happens when one party acts on behalf of another whose interests differ from theirs and who cannot see everything they do. The gap between those interests has a cost, and that cost does not go away — it gets paid in monitoring, in incentives, or in outcomes.
In practice
Nobody in the company loses sleep over a bad month except the owner. That is not a character difference. The owner holds the residual, and everyone else holds a salary.
The common mistake
Treating it as a problem about honesty. It arises between entirely honest people whenever interests differ and observation is incomplete, which is every company with more than one person in it.
Every business with an employee has this problem. Most owners experience it for years before they have a name for it.
The structure: one party — the principal — needs something done and hires another party, the agent, to do it. The agent knows things the principal does not. The agent's interests are not identical to the principal's. And the principal cannot watch everything.
Give those three conditions and behavior diverges. It does not require anyone to be dishonest.
Why honest people diverge anyway
An owner holds the residual. Everything left after costs is theirs, and so is everything lost. That makes the owner care about outcomes at a level nobody on payroll can match, because nobody on payroll has the same exposure.
A salaried employee has a different and equally rational position. Their downside on a bad quarter is bounded, their upside on a great one is usually bounded too, and their effort is largely unobserved. Doing the job well matters to them — most people want to do good work — but the specific thing the owner is optimizing for, residual value over years, is not what their week is arranged around.
So the divergence is structural. Replace the whole team with more conscientious people and it is still there.
The three ways it gets paid for
Agency costs are not avoidable. They are only ever moved.
Monitoring. The principal spends to see more — reporting, reviews, check-ins, dashboards, approval steps. This works and it is expensive, and it scales badly, because the owner's attention is the constraint being spent.
Incentive alignment. Give the agent exposure to the outcome: commission, profit share, equity. This is the strongest of the three and it is partial. You can align an agent to a number, and then you get what that number measures rather than what you wanted.
Bonding. The agent spends to prove they can be trusted — the contractor who sends detailed updates nobody asked for, the employee who documents their work. It is a real cost and it is borne by the agent.
Add the three together and you have the agency cost of that relationship. A well-run company has low agency costs, and that is a different claim from having good people.
Where it shows up in a business
This is the mechanism under a set of problems that usually get discussed separately.
Owner dependency is agency cost that was never paid. The owner keeps the work because supervising someone else doing it costs more attention than doing it, which is true, and compounds until the business cannot run without them.
Delegation fails for the same reason it is necessary. Handing over a task transfers the doing and not the caring, so the gap has to be closed by one of the three mechanisms, and "I told them what I wanted" is none of them.
Proxy work and performance theater are what monitoring produces when the thing monitored is easier to observe than the thing wanted. The agent is not cheating. They are responding to what is measured.
And the sale of a company is the moment the whole account comes due. A buyer is about to become a principal with no operating knowledge, which is exactly why a business that runs out of one person's head is worth so much less than the same revenue with a system underneath it.
What to do about it
Three moves, in order of how much they return.
Give people the residual where you can. Exposure to the outcome is the only mechanism that works while you are not looking. Commission, profit share, and equity for the people whose decisions actually move the number.
Measure outcomes rather than activity. Activity is easier to see, which is why it gets measured, and measuring it is how you buy activity.
Reduce what has to be observed. A process with fewer judgment calls has fewer places for interests to diverge. This is most of what systematizing a business actually does, and it is why it raises the sale price.
The problem is never solved. It is priced, and a company that has priced it correctly looks from the outside like one with unusually good people.
Concept web
Open the full webQuestions
What is the principal-agent problem?
It arises when one party acts on behalf of another whose interests differ and who cannot observe everything they do. The gap produces a cost that is paid in monitoring, in incentives, or in worse outcomes, and honest people do not remove it.
What are agency costs?
The total of what the principal spends on monitoring, what is given up in aligning incentives, and what the agent spends proving trustworthiness. They can be moved between those three, and reduced, but not eliminated.
How does the principal-agent problem affect selling a business?
A buyer becomes a principal with no operating knowledge, so a business that depends on the owner's judgment carries very high agency cost for them. That is a large part of why the same revenue sells for less without a system underneath it.