Many of the enterprises governments engage in cause people's deaths in one way or another. Therefore, those governments that like to give some of their actions the appearance of economic rationality have to fix on a monetary value for a human life. A blossoming literature explains to them the correct way to make this valuation. But, though it blossoms still, I think the roots of the literature are insecure.

Before the discussion can get under way, the whole notion of the ‘value’ of benefits and costs – values that can be added and subtracted to establish the rightness or wrongness of a government's project – needs to be supplied with both an exact definition and a justification. In this chapter, I shall consider only the familiar interpretation of cost–benefit analysis as a ‘compensation test’. Under this interpretation the values of costs and benefits are defined by means of monetary compensation, in a manner I shall specify in a moment. This definition is justified by arguing that, if the benefits of a project exceed its costs when valued this way, the project is a good thing. I must explain.

I think that by now most people recognize the invalidity of the compensation test except when compensation is actually paid, so let us assume it is. Then the test boils down to something rather harmless, the ‘Pareto criterion’, and it works as follows. Each person who would lose from the project is asked what is the minimum payment of money he would consider full compensation for his loss. This amount is defined as the value of his loss. If the project is carried out it will be paid him, so he is deemed not to have suffered in the end.