The Principal-Agent Problem

Delegation is necessary, but different payoffs and incomplete observation can make the agent’s best move different from the principal’s.

The problem begins when someone else acts

A shareholder supplies capital but does not approve every purchase, hire, pricing decision, or maintenance job. A board delegates operations to an executive team. A client hires a contractor. In each case, the principal wants an outcome and gives an agent authority to choose actions that the principal cannot fully observe.

The principal-agent problem is not a synonym for fraud. It appears when the agent knows more about the action, bears only part of its consequences, or is rewarded by a measure that differs from the principal’s objective. The classic Jensen and Meckling analysis formalized agency costs as the costs of monitoring, bonding, and residual divergence created by this separation of ownership and control.

Three different information problems

Hidden action occurs after the contract is signed. A manager can reduce preventive maintenance, a loan servicer can choose a payment allocation, or a salesperson can spend time on easy accounts rather than difficult ones. The principal observes an outcome later but not the effort, sequence, or alternative that produced it.

Hidden information exists before the decision. An executive may know that a target acquisition has weak systems, a supplier may know its process is near a quality limit, or an adviser may know that a recommended product earns a higher commission. The principal can ask for representations and tests, but cannot assume that a report reveals every relevant fact.

Risk transfer changes the agent’s preferred action. A diversified shareholder may accept a project with a wide range of outcomes, while a manager whose career is concentrated in one company may prefer a safer project with lower expected value. A bonus tied to quarterly revenue can make growth look attractive even when the required discounts, returns, or support burden reduce long-run cash generation.

Wells Fargo shows how a measure can become an instruction

The CFPB’s 2016 Wells Fargo enforcement record describes employees opening unauthorized deposit and credit-card accounts to meet sales targets and receive compensation. The Bureau’s account is a documented enforcement finding, not proof that every employee had the same motive. It does show a complete agency mechanism: senior goals were translated into local measurements, employees had the ability to create reported sales, customers bore fees and confusion, and the control system failed to detect or stop the practice soon enough.

The case also illustrates why a principal cannot evaluate an incentive using the target alone. “Accounts opened” was an observable number, but it was a poor proxy for authorized customer value. Better controls would have included consent, complaints, reversals, account use, and consequences for managers whose results depended on invalid sales. Monitoring changes the cost of the action, but it also creates expense and can move the gaming to a less visible metric.

Governance is a portfolio of imperfect controls

• Contract design. Pay can depend on revenue, cash generation, risk-adjusted returns, customer retention, safety, or quality. Each measure captures something and leaves something out.
• Ownership and exposure. Equity ownership, clawbacks, deferred pay, guarantees, and liability can make the agent bear more of the outcome, but concentrated exposure can also produce excessive risk aversion.
• Monitoring. Boards, audits, customer verification, segregation of duties, and independent testing can reveal hidden action. They cannot observe every choice and may create false assurance if they check only the recorded sample.
• Competition and exit. A principal can replace an agent, move assets, or terminate a contract. These remedies work only when performance is measurable and alternatives are available.
• Authority boundaries. A control is weak when the person who sees a problem cannot stop the process, or when the person who can correct it is paid on a different outcome.

Why alignment can fail even with good intentions

Agents may pursue the principal’s objective and still make different choices because they face different information, time horizons, and downside. A manager may delay a necessary investment because the benefit arrives after a planned retirement. A board may approve an acquisition because the strategic case is plausible and the information available at the time is incomplete. Hindsight can identify a bad outcome without proving that the decision was an agency failure.

Incentives also create selection and measurement effects. A target can attract effort toward what is counted and away from what is not. A stock-based award can align managers with shareholders while encouraging leverage or short-term price support. A customer-satisfaction score can improve service or teach employees to avoid difficult customers. The instrument must be read with the operational process it changes.

What the concept does not prove

• A poor result is not automatically evidence of self-dealing. Demand, technology, regulation, and bad luck can produce the same result.
• More variable pay is not automatically better alignment. It can increase risk-taking, short-termism, or manipulation.
• Manager ownership is not a complete solution. Owners can still receive private benefits, misjudge risk, or prefer empire and control.
• An independent board or audit is not proof that all relevant conditions were observed. It establishes what was reviewed and under which procedure.
• Different principals may want different outcomes. “Shareholder value” can conflict with creditor protection, customer safety, employee continuity, or a long-term investment horizon.

What investors can test

• Map the decisions management controls and the consequences shareholders, customers, creditors, or employees bear.
• Compare incentive measures with the physical and financial work that creates durable value: cash collection, maintenance, quality, safety, retention, and return on invested capital.
• Read targets, clawbacks, related-party transactions, acquisitions, and capital allocation together. The contract often explains which choices are rewarded before the result appears.
• Look for independent observations: customer complaints, reversals, audit findings, employee turnover, warranty claims, covenant breaches, and cash conversion.
• Check whether the person who can correct a problem has the information, authority, budget, and time to act before the cost becomes irreversible.

The principal-agent problem is a reminder that delegation changes the information and payoff structure of a business. Good governance does not eliminate self-interest. It makes important choices observable, gives the agent a reason to protect the principal’s objective, and preserves a credible correction when the measure starts to diverge from the work that matters.