Principal-Agent Problem

Why it matters

The moment you hand your work to someone whose interests aren’t yours and whose effort you can’t fully see, they’ll quietly optimize for themselves — not out of malice, but because that’s what their incentives pay them to do.

For example: you hire a real-estate agent to sell your house, and you want the highest price they can get. But your agent earns a percentage, and their cut barely moves between a $490K sale and a $520K one — while the extra weeks of open houses and hard negotiating to get that last $30K come entirely out of their evenings. So they nudge you toward the quick, early offer. They haven’t betrayed you. They’re just answering to a different paycheck than the one in your head.

What it reveals. That when one party acts for another, the gap between what they’re paid for and what you actually want is where your real outcome leaks away — and the leak is structural, not a character flaw.
How it changes the read. You stop asking “is this person doing a good job?” and start asking “what is this person actually being rewarded for, and is it what I want?” The behavior follows the incentive, not the instruction.
When to foreground it. Any time someone acts on your behalf and you can’t fully watch them — a broker, a contractor, a fund manager, a CEO answering to shareholders, a salesperson on commission — and the results are fine but you suspect they could have been better.
What you’d miss without it. That the fix is almost never “tell them to try harder” or “watch them more closely” alone — it’s redesigning what they’re paid for so that doing well for you is their best move.
Where it misleads. When interests genuinely line up, there’s no problem to solve — assuming everyone is secretly self-serving manufactures a conflict that isn’t there. And piling on monitoring without fixing incentives just teaches people to game whatever you measure.

Realtime examples

See real, dated analyses where this pattern shaped the read on the news → The principal-agent problem on Main Street Independent

How to invoke it in Ora

You’ve handed something to someone who acts on your behalf — a salesperson, a manager, a broker, a contractor — and you can’t fully see what they’re doing. The results are acceptable, but you suspect their incentives are pulling against yours. You want to know where the misalignment is and how to rewrite the deal so it closes.

Name who’s acting for whom, how that person is paid, and what you can’t observe about their effort, then ask:

“Principal-agent problem: our sales reps are paid on bookings but privately control which deals they chase and how they discount, and we cannot see their effort. Analyze the misaligned incentives and how to redesign the contract.”

Ora names the two parties and what one privately controls, traces exactly where their incentives split, works out the contract that would realign them, and pressure-tests that fix against the gaming it might invite.

One thing to know: the words principal-agent problem are what route you here. A plain version — “our reps keep chasing easy deals, how do I fix it?” — gets a clarifying question back instead, because the bare situation could be a staffing problem, a training problem, or a dozen other things; naming the mechanism is what tells Ora which one you mean. Principal-agent problem, misaligned incentives, agency costs, moral hazard, or incentive design are the words that point it the right way.

Describe each side’s real incentives — how the agent is actually compensated and rewarded, not the job description — because the divergence lives in the pay structure, not the org chart. And say plainly what you cannot observe about their effort or decisions; that hidden action is the other half of the mechanism, and without it Ora can’t tell a misalignment problem from a simple disagreement.

One thing Ora won’t do: assume the person is acting in bad faith. It treats the divergence as a problem of structure — what the incentives reward versus what you want — and looks for the redesign that closes the gap, rather than concluding the agent is disloyal.

How it works

Think again about that real-estate agent selling your house. You’re sure they’re on your side — they only get paid when you do, after all. So picture the afternoon a decent-but-not-great offer comes in, a little under what you’d hoped. You’re inclined to wait for better. Your agent leans in and tells you to take it.

Why? Run the numbers from their chair. Their commission is a few percent of the sale price. Squeezing another $30,000 out of the market — holding more open houses, fielding more lowballs, waiting out the right buyer — might lift their cut by a few hundred dollars after the split. For that few hundred dollars they’d have to burn weeks of evenings and weekends. Meanwhile a bird-in-hand offer closes now, pays them now, and frees them to go list the next house. So the advice to “take it” isn’t a betrayal. It’s the rational move for someone whose reward for the extra effort is a rounding error while the cost of that effort is entirely theirs.

This is the principal-agent problem, the structure economists have studied since the 1970s. There’s a principal — you, who wants the outcome — and an agent — the person acting for you, who has their own interests and who knows things about their own effort that you can’t see. You can watch the result (the house sold) but not the work behind it (how hard they really pushed). And because their paycheck rewards something subtly different from what you want, they’ll quietly optimize for the paycheck.

Here’s the part that turns a hunch into a fact. Economists checked it. When real-estate agents sell their own homes, they leave them on the market longer and hold out for more money than when they sell yours — because now the full upside of waiting lands in their own pocket. Same person, same market, same expertise. The only thing that changed was whose interests the sale served, and the behavior changed right along with it. That’s the whole lens in one finding: the agent wasn’t disloyal when selling your house. Their incentives simply weren’t yours.

Once you see that, you see it everywhere a hidden hand acts for someone else. A fund manager paid on assets-under-management wants your money parked with them, which is not quite the same as wanting it to grow. A contractor paid by the hour is not paid to finish quickly. A CEO whose bonus rides on this quarter’s share price is not, in that moment, being paid for the company’s next decade. None of them has to be a villain. The wiring does the work.

And that points at the only fixes that actually bite. You can’t just instruct the gap away, and watching someone’s every move is expensive and breeds resentment. What works is changing what the agent is paid for so that serving you becomes their best move: a bonus that kicks in above your target price, a slice of equity so they win only when you win, a commission tied to the outcome you care about instead of the activity that’s easy to bill. Realign the reward, and you don’t have to police the effort — the agent polices it for you, because now your win is theirs.

Framework & implementation

This section uses Ora’s own terms for the parts of an analysis, so that if you open the actual mode and lens files they line up. Each is glossed in plain language on first use.

Pipeline execution

The principal-agent problem is one of the mental models in Mechanism and Incentive Analysis’s ANALYTICAL PERSPECTIVES block, listed under “always loaded” — so it is active on every run of that mode, whether or not the prompt names it. (Mechanism and Incentive Analysis is the information-and-incentive sibling of Strategic Interaction: where that mode reads a game of observable moves, this one handles the case where what each party privately knows or privately does is the crux.) It runs at Gear 4, Ora’s most thorough setting: a Depth analyst and a Breadth analyst read the situation independently, each critiques the other’s reading, both revise under that critique, and a consolidator merges what survives. This lens — the broadest frame in the mode’s incentive family — threads through those stages like this.

Detection. The lens engages on the cases in its Detection Signals — a delegation relationship where the principal cannot fully observe the agent; an intermediary (manager, contractor, advisor, broker) acting on the principal’s behalf; results that are acceptable but where the principal suspects the process was self-serving; incentive structures that reward activity or output rather than the outcome the principal actually cares about; an agent whose expertise the principal cannot independently evaluate. The precondition is the mode’s first Critical Question (CQ1, asymmetry named): one party acts for another, holds material private information or takes hidden actions the principal cannot verify, and may have interests that diverge.

The Depth and Breadth analysts. Two models read the situation in parallel. The Depth analyst commits to one reading and defends it — it runs the lens’s Application Steps: map the relationship (who is principal, who is agent, what each one wants), locate where their incentives diverge by reading the actual compensation and career rewards rather than the job description, and pin down precisely what the principal cannot observe about the agent’s effort or decisions. The Breadth analyst works the same situation at the same time, scanning the mode’s other information-and-incentive failure modes — pure moral hazard, adverse selection, the winner’s curse, signaling, screening — and ruling each in or out rather than assuming this one. Neither analyst sees the other’s work.

Both must serve the mode’s CQ2 (selection vs. hazard), and here this lens plays a distinctive role: the principal-agent problem is the broader frame that spans both faces of the asymmetry. Its pre-contract face — does the principal even want this agent, given a quality the principal can’t see before signing — is adverse selection. Its post-contract face — does the agent behave differently once the principal bears the downside — is moral hazard. The analysts hold the full frame while still naming which face is doing the damage in the case at hand, because the two faces take different fixes.

Cross-adversarial evaluation. Each analyst’s reading is handed to the other to critique against the mode’s criteria. The lens’s signature failures are caught here, keyed to its own Critical Questions: assuming the agent is self-serving when their goals actually align with the principal’s (is the misalignment genuine, or are the incentives more aligned than they appear? — the mode’s assume-away-asymmetry failure in reverse), and reading bad faith into behavior that is simply the rational response to a bad incentive (is the apparent self-serving behavior caused by misaligned incentives or by bad faith? — the distinction the mode will not let blur). The evaluator also presses the lens’s does the proposed alignment mechanism create new misalignments?, refusing a redesign that quietly trades one agency cost for another, and its proportionality check (is the cost of mitigation proportional to the cost of the misalignment?).

Revision and claim-check. The reviser addresses the fixes. Where the reading rests on a factual claim — how the agent is really compensated, what the principal can and cannot observe, who actually controls a decision — that claim is marked a flagged claim and sent to a web-search tool; it has to resolve against outside sources before the revised draft moves forward.

Consolidation and output. The consolidator merges the two revised readings into one information-and-incentive corpus, and the formatter places it into the mode’s set sections. The principal, the agent, and what the principal cannot observe land in Parties and the asymmetry. The lens itself — named and shown operating on the specific relationship — lands in Named mechanisms in play (the mode’s CQ4 forbids name-dropping: it must be demonstrated, not merely cited). Because the frame spans both faces, Selection vs. hazard states its relation to its two siblings — adverse selection as the pre-contract face, moral hazard as the post-contract one — and names which face is active here. The divergence itself — effort substitution, metric-gaming, rent extraction from the information gap — lands in The distortion. When the prompt asks for a fix, the contract redesign (outcome-based pay, equity, monitoring, reporting) lands in Mechanism (for a design), which must carry both the participation constraint (will the agent still take the job once the terms are rewritten?) and the incentive-compatibility constraint (is acting in the principal’s interest now the agent’s own best move?) — a design that satisfies only one is reshaped, and metric-gaming is named as the residual gaming surface any outcome-based scheme leaves open. The conclusion lands in Read, with the mode’s CQ5 posture marker stating plainly whether it is explaining a misalignment or proposing a fix.

What the analysis will not assert. It reports the incentive divergence and what would close it. It does not assume the agent is acting in bad faith — the whole point is that rational self-interest under a bad contract produces the behavior, and the lens’s own trust-erosion caution warns that heavy-handed monitoring can manufacture the very disloyalty it suspects. And it holds the principal-agent frame open across both faces rather than collapsing a pre-contract selection problem and a post-contract hazard problem into one undifferentiated “incentive problem,” because they demand different remedies.

Origin and evidence

The structure was formalized in the early 1970s. Stephen Ross’s 1973 American Economic Review paper, “The economic theory of agency: The principal’s problem,” set out the problem in its now-standard terms — a principal who must design a fee schedule for an agent whose actions the principal cannot fully observe. Michael Jensen and William Meckling’s 1976 Journal of Financial Economics paper, “Theory of the firm,” gave the idea its most influential application: they recast the modern corporation as a nexus of principal-agent contracts, coined agency costs for the residual loss the misalignment imposes (the cost of monitoring, the cost of bonding, plus the value that still slips away), and showed how ownership structure itself is a response to the problem — a paper that became one of the most-cited in all of economics. Bengt Holmström’s 1979 “Moral hazard and observability” worked out the optimal contract when effort is hidden, establishing that the principal should pay on any signal that carries information about the agent’s effort — the formal core of why outcome-based pay works and where it breaks. Kathleen Eisenhardt’s 1989 Academy of Management Review survey, “Agency theory: An assessment and review,” carried the framework into management and organizational science and remains the standard synthesis. The lens’s two faces are themselves founding results in the economics of information: adverse selection (Akerlof) and moral hazard, the pre- and post-contract problems the principal-agent frame unifies.

Applications and common uses

The principal-agent problem is the master frame for almost any delegated relationship — used both to explain why a delegation underperforms and to design the contract that fixes it.

Corporate governance. The original Jensen-Meckling case: shareholders (principals) own the firm but managers (agents) run it, and the two don’t automatically want the same things. Equity grants, stock options, performance pay, independent boards, and takeover threats are all devices for binding the manager’s payoff to the owner’s — each a way to make running the firm well the manager’s own best move.
Finance and investing. A fund manager paid on assets-under-management is rewarded for gathering money, not necessarily for growing it; an investment adviser earning commissions has a reason to favor the products that pay them. The fixes are structural — fiduciary duty, fee transparency, performance-linked compensation — not appeals to put the client first.
Sales and distribution. Commission structures are pure principal-agent design: pay on bookings and reps chase volume and discount hard; pay on margin and they defend price. The art is choosing the metric closest to what the firm actually wants while staying alert to how the metric will be gamed.
Procurement and contracting. A contractor paid by the hour is not paid to finish; one paid a fixed price is not paid to maintain quality. Cost-plus, fixed-price, and milestone contracts each align some incentives and distort others, and the choice is a deliberate trade.
Public institutions and regulation. Voters and elected officials, citizens and bureaucracies, regulators and the firms they oversee — each is a delegation under hidden information and hidden action. The frame diagnoses regulatory capture, the revolving door, and bureaucratic slack as agency problems, and points the fix at the incentive structure rather than at exhortations to serve the public.

In every case the diagnosis is the same: the agent is answering to the incentive, not the instruction, so the lever is the contract — realign what the agent is paid for, and you don’t have to police the effort.

Failure modes and when not to use it

The lens’s characteristic ways of going wrong are catalogued in its Common Failure Modes:

Metric gaming. Outcome-based incentives produce optimization for the metric rather than the outcome it was meant to stand for. The tell is a metric that improves while the underlying result does not — the call-center timer that drops average handle time by cutting customers off. The fix is to choose metrics closer to the ultimate outcome and to use several at once so no single one can be gamed in isolation.
Monitoring overhead. Watching the agent more closely until the cost of governance exceeds the agency cost it was meant to save. The tell is total cost rising without a commensurate gain in behavior. The fix is to shift weight from monitoring toward incentive alignment — pay the agent for the outcome instead of paying someone to watch them produce it.
Trust erosion. Monitoring so heavy that it signals distrust and provokes the very withdrawal of effort it was meant to prevent. The tell is the agent pulling back the discretionary, hard-to-measure effort that was never the problem. The fix is to pair what monitoring is genuinely needed with real autonomy on the dimensions that don’t require it.

When not to reach for it. When the agent’s interests genuinely align with the principal’s, the lens manufactures a conflict that isn’t there and corrodes a working relationship — the mode’s assume-away-asymmetry failure run in reverse. When the principal can in fact observe the agent’s effort directly, there is no hidden action to design around and a plain performance standard does the job. And when the apparent shortfall is a matter of pre-contract quality the principal couldn’t see before signing rather than behavior after it, the active face is adverse selection and the fix is screening the agent before the deal — not rewriting the incentives of one already hired.

Mechanism and Incentive Analysis — the analysis that hosts this lens; reads situations where hidden information and incentive structure drive the outcome, and designs the rules that fix them.
Adverse Selection — the pre-contract face of the problem: the agent’s hidden quality, known before signing, determines whether the principal wants the contract at all.
Moral Hazard — the post-contract face: the agent’s behavior turns riskier after the deal, once the principal bears the downside.
Information Asymmetry — the underlying structural property the whole frame rests on: one party knows or does what the other cannot see.