Conclusion: the difficult design of tort law

All liability rules based on negligence struggle with a common dilemma. An increase in care level or a reduction in activity level for one party makes an accident less likely to occur.

Tort rules can only direct efficient incentives with respect to activity level towards the residual bearer, thus failing to enhance the other party’s efficient behaviour. This is Shavell’s theorem on activity level (Shavell, 1980a) ac­cording to which no negligence rule exists which can give both parties efficient incentives with respect to activity level. This follows from the fact that the distinction between care (precaution the judge can observe ex post) and activity level (precaution the judge cannot observe ex post) is due to the introduction of the negligence criterion. The party who can escape liability by simply taking the due level of care will not invest in other unobservable precautions, while the other, the residual bearer, will.

This point can be generalized by observing that a point of discontinuity in the liability curves faced by the parties must be created to entice both parties to choose optimal care and activity levels. With respect to care, this is gener­ally done by identifying a socially optimal care level and by utilizing that level to mark the boundaries between diligence and negligence. Landes and Posner (1987: 70-71) and Gilles (1992) suggest that courts take into account activity levels in their assessment of negligence whenever it is feasible to do so. However, no threshold of ‘optimal activity level’ is generally invoked by legal rules as a liability allocation mechanism.

Optimal activity levels are difficult to specifiy because the value of such activities can only be ascertained from private information of the parties. Unlike optimal levels of care, which largely depend on the objective cost of precaution and the expected gravity of harm, optimal activity levels rely on values that are harder to ascertain by a third-party decision maker since they include the subjective value of the individual that carries out the risk-creating (or risk-bearing) activity. In the absence of such a threshold it is difficult to induce both parties to internalize the full social cost of their activity levels in equilibrium.

This leads us to point out a general characteristic of tort law. Since it is not possible for both parties to bear the accident loss in equilibrium, traditional legal rules concentrate activity-level incentives on one or the other party. Negligence rules under which the victim is the residual bearer (simple negli­gence, contributory and comparative negligence) give efficient incentives with respect to activity level only to the victim, the residual bearer in those cases, while the strict-liability-based negligence rules (under which the injurer is the residual bearer: strict liability and strict liability with negligence de­fences) give efficient incentives with respect to activity level only to the injurer, the residual bearer in those remaining cases.

In theory, a rule of decoupled liability could give both parties efficient incentives with respect to care and activity levels. Decoupling liability (Polinsky and Che, 1991) means making both the injurer and the victim the residual bearers by denying the victim any compensation (as under no liab­ility) and having the injurer pay a fine equal to compensatory damages (as under strict liability), regardless of their level of precaution. However, other functions of tort law (for example, compensatory and so on) would be com­promised by such a decoupling mechanism.