Pay a salesperson pure commission and you load all the risk of a bad quarter onto them; pay pure salary and they have no reason to hustle. Every contract sits somewhere between, and the same tension, incentives against risk, runs through employment, insurance, and litigation.
Contracts and the law
Contracts and legal disputes are games of hidden action and private information. Optimal incentive pay balances motivation against the cost of imposing risk (Holmström), and most lawsuits settle to avoid trial costs, except when the two sides disagree too much about who will win.
Moral hazard is a problem of hidden...
Two parties usually settle a lawsuit to avoid the cost of...
Predict, then play
As principal, split pay between a base the agent gets either way and a bonus for high output. Start with a risk-neutral agent, then raise risk aversion. Watch the motivating bonus change and check whether the contract still gives enough certainty-equivalent value for the agent to accept.
Now switch to litigation. Set each side’s belief about the plaintiff winning and see whether their acceptable settlement ranges overlap.
Why did that happen?
Take a guess, then reveal the reason
The principal cannot see effort, only noisy output, so pay must be tied to output to motivate, and with a risk-neutral agent the cheapest way to do that is all bonus and no base. Now make the agent risk-averse, as most employees are. Output still depends on luck, and an agent who dislikes bearing that luck demands a risk premium to accept a high-powered contract. The optimal contract trades these off: steep enough to induce effort, flat enough to avoid overpaying for noise. Make the agent bear all risk and you over-punish them for bad luck and pay dearly for it; make pay flat and effort collapses. Litigation is the mirror image with private information: both sides usually prefer to settle and split the saved trial costs, but if each is too optimistic about winning, or holds private information the other lacks, their acceptable ranges stop overlapping and the case goes to trial.
The real strategy
Holmström (1979) showed the optimal incentive contract balances the benefit of motivating effort against the cost of imposing risk on a risk-averse agent, and that pay should depend on any signal informative about effort (the informativeness principle), filtering out noise the agent cannot control. In litigation (Shavell, 2004), a plaintiff who expects to win with some probability, net of trial costs, has a minimum acceptable settlement; the defendant has a maximum. When these overlap there is a settlement surplus and the parties settle; when optimism or asymmetric information pushes them apart, the ranges miss and the case is tried.
Suppose a plaintiff wins with probability (an expected ), and each side faces in trial costs. The plaintiff’s trial value is : they will settle for anything above that. The defendant’s expected loss at trial is : they will pay up to that to avoid it. Any settlement in beats trial for both, a surplus from the saved costs, so rational parties settle. Now let the plaintiff believe they will win with probability and the defendant believe only : the plaintiff now demands more than the defendant will pay, the ranges no longer overlap, and the dispute goes to trial. Divergent estimates or asymmetric information, not irrationality, drive cases to court.
Work it out
Risk-sharing, and when suits settle
On the contract side, imagine output is effort plus noise, and the agent is risk-averse. A pure commission (all risk on the agent) maximizes the incentive to work but forces the principal to pay a large risk premium for outcomes driven by luck. A flat salary removes the premium but kills effort. The optimum is an intermediate slope: enough pay-for-performance to motivate, filtered so it responds to informative signals of effort and not to pure noise. Over-punishing a risk-averse agent for bad luck is both unfair and expensive.
On the legal side, take the numbers above. The plaintiff’s trial value is and the defendant’s is , so any settlement between them saves both the in combined trial costs, and they settle. But if the plaintiff thinks they will win of the time and the defendant thinks only , the plaintiff wants more than while the defendant will pay at most about ; the ranges do not overlap and the case is litigated. So most disputes settle (the ranges usually overlap), and trials cluster where beliefs diverge sharply or one side holds private information, exactly the pattern the model predicts.
Key idea
Optimal incentive pay balances motivating effort against the cost of loading risk onto a risk-averse agent, so full pay-for-output is rarely best. Most lawsuits settle to split the saved trial costs; they go to trial mainly when optimism or asymmetric information makes the acceptable ranges miss.
Check yourself
Sources
- Holmström, B. (1979). Moral hazard and observability. Bell Journal of Economics, 10(1), 74-91.
- Shavell, S. (2004). Foundations of Economic Analysis of Law. Harvard University Press.