Skip to main content
The Incentives Lab
← All 10 lessons05 / 10
Lesson 5 of 10

The Principal–Agent Problem.

Whenever someone acts on your behalf, their interests are not automatically your interests.

You (the principal) hire someone (the agent) to do something. Your goals and theirs overlap, but never perfectly. Real-estate agents make 3% of your house price — they'd rather close in a week than fight for the last $20k.

Good incentive design narrows the gap between principal and agent. Bad design widens it, then blames the agent for being self-interested.

Go deeper

The applied version.

The principal–agent problem is the original sin of organizational design, and almost nothing in modern management gets it right. The standard playbook — fixed salary plus a vague bonus plus an annual review — is structured to maximize the gap between what the principal wants and what the agent will rationally do.

The diagnostic is brutal and simple. For every role in your organization, ask: "On a random Tuesday afternoon, with no one watching, what is the rational thing for this person to do?" If the answer matches your strategy, you've designed well. If not, you have a principal–agent problem masquerading as a culture problem.

The deeper architecture has three moves. First, outcome alignment — pay for the result the principal actually wants, not the activity that's easy to measure. Second, information parity — close the gap between what the agent knows and what you know, because every information asymmetry becomes an incentive asymmetry. Third, skin in the game — make the agent carry some downside, not just upside, on the outcome.

Every great compensation system in history (Nucor's profit-sharing, Berkshire's CEO comp, the partner model at top law firms) is a carefully tuned principal–agent solution. Every infamous failure (mortgage brokers in 2008, real estate "flat 6%", PE rollups paid on IRR not absolute return) is a principal–agent gap left open.

Real example

Mutual fund managers paid on assets-under-management have an incentive to grow the fund, not to maximize your return. The two often pull in opposite directions.

Inside the Academy
Week 5 — Compensation as Code

You'll redesign one compensation structure in your organization — and present it for cohort review the way you'd present it to your CFO.

  • Outcome / Information / Skin teardown method.
  • Nine compensation patterns we've shipped at Fortune 500 and venture-backed firms.
  • Optional 1:1 with a Lab practitioner to pressure-test your design.
Quick check

Which arrangement reduces the principal–agent problem most?

Practice — make it yours

Pick a relationship in your life where someone acts on your behalf (advisor, contractor, manager, agency). Where do their interests diverge from yours?

Hint: Look at how they get paid, evaluated, or promoted — that's what they actually optimize for.

Full glossary entry: principal agent problem