[1]While this statement is true under most circumstances, there are some exceptions. See Friedman (1988).

[2]This position has sometimes been taken by George Stigler, in criticising both welfare economics and (other people's versions of) public choice theory.

[3. The same qualifications apply here as in footnote 1 supra.]

4For some of my reasons for believing this, see Friedman (1980).

[5]For the original statement of the Coase Theorem see Coase (1960).

[6]See Linneman (1982), Peltzman (1973, 1975), Eckert and Hilton (1976).

[7]Downs (1957).

[8]Niskanen (1971).

[9]Stigler has suggested that one reason politicians are so often lawyers is that, since most of the politician's "customers" have legal business to be done, paying the politician's firm above-market rates for its services is a convenient way of concealing payments to the politician for political services. Similar arguments apply to connections between politicians and other widely used firms--banks, for example. Stigler (1975).

[10]This situation may, however, result in majorities organizing to exploit minorities, and produce large organizational costs associated with everyone trying to end up on the winning end of that game. Since I am not presenting a coalition model, I will ignore such problems here.

11The nature of the public good problem is discussed in greater detail in Part IV of this chapter.

[12]If the beneficiaries can organize to pay part of the value of the public part of the good, the result is even more favorable; if, for example, the indirect beneficiaries of innoculation each pay half the value of their benefit, I will get innoculated as long as benefit is at least 4/3 times cost.

[13]See, for example, Posner (1974).

[14] Linneman (1982).

[15] Peltzman (1973).

[16] The argument is given at greater length in Friedman (1986, chapter 14; 1989, Chapter 15).

[17] This argument depends on the medical expenses being uninsurable; otherwise exactly the same utilitarian argument, applied to an individual, will result in his insuring against them. Expenses may be uninsurable because the outcome can be observed before the potential victim can buy insurance. Someone who is born blind cannot insure against blindness.

[18] Friedman (1980).

[19] For a non-technical discussion of this question, see Friedman (1973, 89), especially chapters 4, 27 and 38.

[20]See Stigler (1970) for a discussion of the direction of government redistribution.

[21]For example, a study by the Institute of Economic Affairs found that British National Health Insurance provided higher income clients more for their money than lower income clients; the implicit redistribution was up, not down, the income ladder.

[22] For a more extensive discussion of this point, see Friedman (1982).

[23]The estimate of 7,000-10,000 lives a year was by Arthur Hayes, HHS News, Nov. 25, 1981. Prior to 1981 another beta blocker, propranalol, had been approved (and widely used) for other purposes, but not for preventing heart attacks.

[24] Akerlof (1970).

[25]See, for example, Stigler and Friedland (1962), reprinted in Stigler (1975) as chapter 5.