The Principal-Agent Problem

When you delegate to someone else, your incentives and theirs diverge. The agent optimizes for agent rewards, not principal goals. This single mechanism explains most organizational dysfunction.

You hire someone to do something. They don't do exactly what you wanted. Why?

The naive answer: they're lazy, incompetent, or dishonest. Sometimes true. But even perfectly competent, hardworking, honest agents diverge from principal goals.

The structural answer: principal-agent problem.

The Basic Structure

Principal: The one who wants something done. Shareholder, voter, patient, client.

Agent: The one who does it. Manager, politician, doctor, lawyer.

The principal can't do the task themselves (lack of time, expertise, access). They delegate to an agent who can.

The problem: the agent's incentives don't match the principal's goals. The agent optimizes for what rewards the agent.

Why Divergence Is Inevitable

Information Asymmetry

The agent knows more than the principal. That's why you hired them. But this means the principal can't fully monitor the agent's work. The agent has private information and private actions.

Effort Is Costly

Effort costs the agent but benefits the principal. If the agent could get paid without effort, they would (on the margin). Principals want agents to work hard; agents want to conserve effort.

Risk Preferences Differ

Principals often want risk-taking; agents prefer safety. A failed project hurts the agent's career more than the principal's portfolio. Agents are conservative when principals want boldness.

Time Horizons Differ

Agents often have shorter time horizons. A CEO might maximize quarterly earnings and exit before long-term costs hit. The principal bears consequences the agent escapes.

Examples Everywhere

Corporate Governance

Shareholders (principal) hire managers (agent). Managers maximize manager welfare: high salaries, job security, prestige projects. Shareholder returns come second.

Politics

Voters (principal) elect politicians (agent). Politicians maximize politician welfare: re-election, power, post-office opportunities. Voter interests come second.

Healthcare

Patients (principal) hire doctors (agent). Doctors maximize doctor welfare: income, time, malpractice avoidance. Patient health comes... not first.

Academia

Society (principal) funds researchers (agent). Researchers maximize researcher welfare: publications, citations, tenure. Truth-seeking comes second.

Real Estate

Buyer (principal) hires agent. Agent maximizes agent welfare: fast closing, minimal effort. Best price for buyer comes second.

Attempted Solutions

Monitoring

Watch the agent. Track their work. Require reports. Problem: monitoring is costly, often impossible for specialized work, and agents game metrics.

Incentive Alignment

Make agent rewards depend on principal outcomes. Stock options, performance bonuses, outcome-based pay. Problem: hard to measure outcomes, creates gaming, shifts risk to agents who then demand higher pay.

Selection

Choose agents whose preferences naturally align. Hire mission-driven people. Problem: preferences are hard to observe ex ante, and people adapt to incentives.

Competition

Create alternatives. Threat of replacement disciplines agents. Problem: switching costs, information loss, and agents may collude or have market power.

Reputation

Make future opportunities depend on current performance. Problem: short-term agents don't care, and reputation can be manipulated.

Why Perfect Solutions Don't Exist

Every solution has costs. Monitoring costs money and distorts behavior. Incentive alignment creates gaming. Selection is imperfect. Competition has frictions.

More fundamentally: if the principal could perfectly specify and monitor what they wanted, they wouldn't need an agent. The information asymmetry that creates the need for delegation is the same asymmetry that prevents controlling it.

The principal-agent problem isn't a bug to be fixed. It's a structural feature of delegation. You can mitigate it but never eliminate it.

Implications

• Expect divergence. When you delegate, assume your agent's actions will serve their interests more than yours. Plan accordingly.
• Minimize principal-agent distance. Each layer of delegation adds divergence. Shorter chains mean less drift.
• Combine mechanisms. No single solution works. Use monitoring AND incentives AND selection AND competition.
• Accept imperfection. The cost of perfect alignment exceeds the cost of some divergence. Tolerate manageable slippage.

— Decoded by DECODER