Loser-pays cost rules

Why does making the loser pay the winner's legal costs deter sound claims along with the frivolous ones?

The argument for making the loser pay is almost irresistibly clean. If your claim is good, you win and are made whole — costs and all. If it is rubbish, you lose and pay for the trouble you caused. A filter that charges the wrong and refunds the right: what could it screen out except nonsense?

And yet where it is used, the complaint is not that frivolous suits survive. It is that people with strong claims will not file. Something in the clean argument is wrong, and it is worth finding where.

Put numbers on a claim so we can see the shape rather than argue about it. Say the amount at stake is 100, your own lawyers will cost 40, and the other side's will also cost 40. You think your chance of winning is p.

Where each side bears its own costs, filing is worth 100p - 40. That turns positive at p = 0.4: you file if you think you have a better than two-in-five chance.

Under loser-pays, you keep the 100 and recover your 40 when you win, and pay both bills when you lose. The value is 100p - 80(1 - p), which is 180p - 80, positive once p exceeds 80/180, about 0.444. So the merit threshold moved — from 40% to roughly 44%. Real, but small. If that were the whole story, loser-pays would be a mild filter and the complaint would be unfounded.

Now look at what else moved. Under each-pays, your worst outcome is losing 40. Under loser-pays it is losing 80. The average barely shifted, but the downside doubled. And here is the step that matters: a claimant is not an expected-value calculator with unlimited reserves. For an individual, 80 may be the difference between an inconvenience and losing the house, while the defendant — an insurer, an employer, a hospital — meets both figures out of a budget line and is close to risk-neutral by construction.

So ask what the rule actually sorts on. Not merit alone. It sorts on merit multiplied by the capacity to absorb a two-sided loss. A claimant with a 70% case and no reserves may rationally decline; a defendant with a 30% defence and a balance sheet may rationally fight. The filter is real, but one of its two dimensions has nothing to do with whether the claim is sound — and the deterrent bites hardest on exactly the cases the general rule was framed to help: the meritorious but uncertain.

A further effect runs the same way. If the other side may pay my bill, my incentive to economise on it weakens, so both sides spend more per case and the sums being shifted grow. Fee shifting does not only redistribute costs. It inflates them.

Now, is the shifting itself the deterrent, or the uncertainty about the amount? Jurisdictions answer differently. In England and Wales costs are assessed after the fact by a judge, subject to proportionality, so a claimant cannot know the exposure when deciding to file. Germany also has loser-pays, but recoverable amounts are fixed by statutory scales keyed to the value in dispute, so both sides can look up the worst case in advance — and legal expenses insurance is widely held, converting the tail into a premium. Same nominal rule, very different felt risk.

That points to the fix. Where loser-pays coexists with easy access, something is absorbing the variance: scheduled fees, insurance, or a carve-out. England and Wales adopted one such carve-out for personal injury, costs shifting against a losing claimant only in limited circumstances, alongside conditional fee agreements (I recall these as Jackson-era reforms effective in 2013, and state the dating as recalled).

Three refinements, and one correction of the framing.

The correction first: "loser-pays deters sound claims" does not mean it fails as a filter. It filters — but on two variables at once, and a court can observe neither. "This claim is weak" and "this claimant cannot carry the tail" both arrive as an absence.

First refinement: the deterrence is two-sided, and this is the strongest thing to be said for the rule. A defendant with a hopeless defence now faces paying the claimant's costs for dragging it out, which pushes toward earlier settlement of clear cases. That is a genuine benefit and it is why the arrangement has defenders, not merely inertia.

Second: the effect on settlement is theoretically ambiguous, not obviously good. Raising the stakes on both sides makes agreement more attractive when the parties agree about p, but when each is privately confident, fee shifting magnifies the disagreement rather than resolving it, and analytical work going back to Shavell in the early 1980s finds it can raise the trial rate. I state that attribution as recalled and quote no litigation or settlement rates: published figures vary too much with definitions and courts included.

Third, the rule's survival is political. The American rule's cost is that a defendant who wins outright still pays for the defence — a standing grievance of firms facing many small claims. The English rule's cost falls on individual claimants of modest means, who are numerous, unorganised, and not in the room when costs rules are reviewed. Which side of that trade a system sits on is mostly not settled by evidence about efficiency; it is settled by whose losses are visible.

That yields the test. If capacity to bear the downside is doing the work, claim rates within a loser-pays system should rise sharply where the tail is capped or insured — scheduled fees, one-way shifting, widely held cover — while the shifting rule stays formally identical. The refuting observation: a loser-pays system whose claim rate is unchanged by capping or insuring the exposure, which would put the deterrent in the shifting itself rather than the variance.

Every term the collection defines is gathered in the glossary.