The Principal-Agent Problem: Your Agent Earns ₹2,000 From Your ₹5,00,000
You are selling your house. Your estate agent earns 2% of the sale price.
A buyer offers ₹1 crore. Your agent thinks that holding out for two more weeks would probably get ₹1.05 crore.
Run the numbers from each side. Two more weeks of work earns the agent an extra ₹1,00,000 × 2% = ₹2,000. Barely worth the effort — better to close now and start on the next property.
For you, those two weeks are worth ₹5,00,000, less the 2% commission. Enormously worth waiting for.
Your agent is not corrupt. They are responding sensibly to their own incentives, which are 98% different from yours. This gap — between what you want from someone and what they are actually rewarded for — is the principal-agent problem, and once you can see it you will find it in nearly every professional relationship you have.
The three conditions
The problem appears whenever three things are true at once, and none of them involves anybody behaving badly.
Someone acts on your behalf. You are the principal; they are the agent. You hire them because you cannot or should not do the thing yourself.
They know more about the work than you do. This is usually the whole reason you hired them. A mechanic knows whether the part really needed replacing; you don't. This is called information asymmetry, and it is not a flaw in the arrangement — it is the point of it.
Their interests differ from yours, even slightly. Not oppositely. Slightly. The estate agent wants a good sale, just a bit less than you want the best one.
Notice that the second and third conditions interact badly. You cannot easily check whether the agent served your interests, because they know more than you — which is why you hired them. The expertise that makes them useful is the same expertise that makes them hard to evaluate.
Where you meet it
Fund managers. Typically paid a percentage of assets managed, sometimes with a share of the gains. A fee on assets rewards gathering more money more than performing well with it. A share of gains with no share of losses rewards taking risk with your capital — the upside is shared, the downside is mostly yours.
Company executives. Paid partly in shares vesting over a few years, they have a real reason to prefer decisions that lift the price within that window over decisions that build the company over fifteen years. Cutting research and buying back shares can be exactly rational for the agent and quietly bad for the principal.
Contractors paid by time. Finishing early ends the income. Nobody needs to pad anything deliberately for the incentive to point the wrong way.
Doctors under fee-for-service. Paid per procedure, the reward attaches to doing more, not to the patient being healthiest. Most doctors resist this; the pull exists regardless.
Elected representatives. Voters are the principal, officials the agent, and the election cycle is short while many consequences are long. Decisions whose costs arrive after the next election are structurally underweighted.
Your own employees — and you, to your employer. Everyone in an organisation is an agent for someone. This is not an accusation. It is the ordinary condition of working with other people.
What actually reduces it
There is no clean solution — you cannot both delegate to an expert and fully verify their expert judgement. But some approaches genuinely help, and it matters which one fits your situation.
Tie the reward to the outcome you actually want — but only if you can measure it. This is the obvious fix and it has a sharp limit. Where the real goal is measurable, aligning pay to it works well. Where it isn't, you end up paying for a proxy, and the agent optimises the proxy. That is Goodhart's Law arriving on schedule, and it can leave you worse off than a plain fixed fee, because now the misalignment is funded.
Make the relationship repeated and visible. An agent who needs your future business, or who depends on referrals, has a real stake in your outcome. This is why reputation systems, professional licensing, and long relationships reduce the problem — they convert a one-off interaction into a repeated one, exactly as in the prisoner's dilemma.
Reduce the information gap where it is cheap to do so. You will never know as much as your mechanic. You can get a second quote. Independent inspections, audits, and second opinions all work by importing expertise you don't have rather than by acquiring it.
Match the agent's time horizon to yours. Much executive misalignment is not about greed but about clocks. Longer vesting periods and deferred compensation move the agent's horizon closer to the principal's, which addresses the cause rather than the symptom.
Prefer fixed fees when outcomes are hard to measure. Counterintuitive, but sound: if you cannot measure the real result, a fixed fee at least leaves the agent neutral, whereas a badly chosen incentive actively pays them to do the wrong thing.
The practical habit is simply to ask, of anyone advising you: what happens to them under each option they're presenting? Usually the answer is nothing sinister. Occasionally it reveals that the advice you're receiving is, entirely honestly, advice about their situation rather than yours.
