We can use our analytical framework of bilateral accidents to analyse different policies which influence the levels of care. Consider, for example, a no fault liability scheme. In the simplest example of a no fault liability scheme, the government taxes injurers, and then places the proceeds in a common pool. When an accident occurs, victims are not required to sue injurers, but instead they are compensated for their injuries by drawing from the pool. Injurers therefore do not pay any damages directly, but simply pay taxes in a lump-sum fashion, whether they cause an accident or not.

The cost function of injurers is as follows. Suppose that the government knows what the level of damages h would be in the event that an accident occurs. Injurers are taxed this amount, whether or not an accident occurs. Therefore, the cost function of the injurer is:

Now the cost function of victims is:

Optimal choices of care. The optimal choice of care for the victim is:

and the optimal choice of care by the injurer is:

The probability of an accident occurring is therefore:

and the aggregate expected social costs under this scheme are:

From an efficiency point view, this no fault liability scheme is the worst possible scheme. While victims are always compensated for their damages, the incentive structure that this scheme creates means that the probability of accidents increases, and expected social costs also rise. This outcome can be shown in the diagram below.

Suppose instead that the government does not observe the true amount of harm, but instead underestimates them and taxes injurers an amount T < h. The injurer's cost function is still the same, but the victim's cost function is now:

where h - T > 0 by construction. The injurer still takes no care, but the victim's cost function, given injurer behaviour, is:

The victim chooses his level of care to satisfy the marginal condition:

In general, the level of care that
satisfies this condition will not be zero, and so the probability of an accident and expected social costs will be lower when T < h. In the extreme case where T = 0, we are back to the legal rule of no liability. Thus, in general, a rule of no liability is preferred to this no fault liability scheme. The reason is that incentives for both injurers and victims are completely undermined in a no fault liability scheme, whereas with a no-liability rule (and also with a strict liability rule), at least one of the parties takes some positive amount of care.