urned, and to which a satisfactory answer has yet to be reached. 1 A similar conclusion must hold with regard to patients who, though knowing that they are sick, fail to demand treatment. Given their pre ferences, information, fears, etc, there is no a priori reason for supposing that they are behaving irrationally, nor that they would behave differently nder an alternative system. On the other hand, if there exists another set of individuals who would prefer them to receive more care than they choose, then there may again be marginally relevant externalities which should be taken into account in describing the nature of the optimum The point here, however, is not that the individuals in question are behaving irrationally, violating any of the postulates of utility theory by, for example, acting inconsistently with their own preferences, but that they re behaving inconsistently with the preferences of an entirely different set of individuals. 2 3. Similar conclusions hold with respect to the emotionally disturbed, children, and emergency cases. If there is evidence that these individuals are actually behaving irrationally or are not in a position to choose, then it must follow that welfare economics, based as it is on an assumption of rationality, has nothing at all to say about their subjective utility maxi mization. It cannot therefore be used to assess alternative forms of pro vision. On the other hand, welfare economics can be used to describe the characteristics of an optimum if there exists an external demand for the care of such people. Since the problem of external demands emerges as a general problem in health economics its discussion is, however, postponed until a later point in this paper. It is, however, dangerous to overstate the degree of irrationality in the behaviour of patients, and the discussion here is not intended to lend support to any presumption that individuals, even those who are emotionally disturbed, are in general irrational in matters of personal health Uncertaint Four points in particular have been raised, which may affect the ability of an open market to satisfy the necessary optimal conditions, all arising 1 For some attempts in connection with the medical care industry see Newhouse [36 and Weisbrod [46]. For a different context (universities)see Culyer and the costs of the appropriate treatment nottoo high, the Hypo (almost)entirely logically, are Arrow [1]. Lees and Rice [28], Arrow [2], Pauly [38], and Crew [11]. Reproduced with permission of the copyright owner. Further reproduction prohibited without permissionReproduced with permission of the copyright owner. Further reproduction prohibited without permission