1Nicholas Kaldor: Welfare Propositions of Economics and Interpersonal Comparisons of Utility, 49 Economic Journal 549-552 (1939). See also John R. Hicks, Foundations of Welfare Economics, 49 Economic Journal 696-712 (1939) and Tibor de Scitovszky, A Note on Welfare Propositions in Economics, 9 The Review of Economic Studies pp. 77-88 (1941).

2Alfred Marshall, Principles of Economics, Bk III Chapter VI pp. 124-137 (8th ed. 1946).

3Kaldor, op. cit.

4David D. Friedman, Price Theory: An Intermediate Text, Chapter 14.

5Gary S. Becker, A Theory of Social Interactions, 82 JPE pp. 1063-1091 (1974).

6In some cases, most notably altruism towards children, the idea that the beneficiary knows what makes him happy becomes implausible; one must then redefine the beneficiary's utility function to correspond not to what he does want but to what he would want if properly informed. Such complications do not affect the argument of this paper.

7 For an interesting discussion of some of the problems with explaining observed behavior in terms of the Becker analysis of altruism, see Howard Margolis, Selfishness, Altruism & Rationality: A Theory of Social Choice, Chapter 2.

8I am assuming here that we do not have a corner solution at T*=0. In other words, I am considering a case where the altruist chooses to make at least some transfer. This need not always be the case. One could have an altruist who, although he valued the utility of the beneficiary and would be willing to pay something to increase the beneficiary's consumption by a dollar, would not be willing to pay as much as a dollar to do so, and would therefore choose to make a transfer of zero.

9William Landes and Richard Posner, Salvors, Finders, Good Samaritans, and Other Rescuers: An Economic Study of Law and Altruism, VII JLS 83-128 (1978).

10Landes and Posner (1978) note the existence of this problem in footnote 27. The authors use the term efficient to describe both the level of rescue inputs that would have been efficient if altruism did not exist and the higher level that is (Marshall?) efficient given the existence of altruism, but do not explore the discrepancy.

11Note that this result does not depend on the particular rule for interpersonal utility comparisons (assume a dollar has the same utility for everyone) implicit in Marshall's definition of improvement.

12Each dollar of tax payment is also being multiply counted, for the same reason; taxpayers are presumably altruistic towards each other as well as towards potential welfare recipients. If consumption by a beneficiary has declining marginal utility in the utility function of the altruist (either because of declining marginal utility of consumption for the beneficiary, or declining marginal utility of beneficiary's utility for the altruist) then the reduction in the utility of the altruist due to payment of one dollar by (high income) tax payers will be more than balanced by the increase due to receipt of one dollar by low income recipients. Whether voters will support such transfers will then depend both on the degree of their altruism and on their opinion about the excess burden imposed by the transfer--how much more than a dollar it costs the taxpayers to provide a one dollar benefit to the recipient.

13Private transfers of this sort are in some respects superior to governental transfers. Donors are likely to be more altruistic towards fellow members of their community--possibly friends and neighbors--than towards anonymous beneficiaries. In addition, it may be much easier to limit the costs of the moral hazard resulting from transfers--recipients who choose to accept welfare instead of looking for jobs, or unmarried mothers who continue having children in the knowledge that someone else will pay for them--within a small and homogeneous community.

14The same externality argument which implies that there will be too little transfer when the level is chosen by the altruistic donor may also imply too much expenditure on causing transfers when the level of such expenditure is chosen by the recipient. Imagine that I am a non-altruistic potential beneficiary lobbying in favor of a welfare program to be paid for by non-altruist taxpayers. I will spend money lobbying up to the point where a dollar spent results in an additional dollar of transfer. In doing so, I ignore the cost to the taxpayers from whom the transfer comes. In equilibrium, the marginal dollar I receive costs two dollars--a dollar paid by me for lobbying plus a dollar transferred from the taxpayer. This phenomenon, generally referred to as rent seeking in the recent literature, provides one of the strongest arguments against permitting governments to redistribute income.

15Part of the reason that this seems paradoxical when applied to real examples is that we are not willing to regard the children's behavior as merely reflecting given and unchangeable tastes. Selfish or altruistic behavior may reflect, not selfish or altruistic tastes, but a judgement by the child about what behavior will be rewarded by the parent.

16This particular application of the analysis was suggested to me by the Editor of this journal.

17Henry Simons, Personal Income Taxation (1938). The definition of income is on page 50, the discussion of gifts is on pp 57-58.

18[ ]For simplicity, I am ignoring any effect of the tax on the producers; the supply curve for apples is assumed to be perfectly elastic.

19I ignore here a possibility raised earlier--that the transfer itself may involve some excess burden due to the beneficiary modifying his behavior in an attempt to increase the amount of the transfer. This problem will normally not arise as long as the altruist can directly observe the beneficiary's utility, as assumed here.

20In some cases, current legal doctrine treats such transfers as gifts, and consequently non-taxable. Examples are Commissioner v. Duberstein, 363 U.S. 278 (1960) and Stanton v. United States 186 F. Supp. 393 (E.D.N.Y. 1960). On the other hand, Olk v. United States, 536 F .2d 876 (9th Cir.), cert. denied, 429 U.S. 920 (1976) appears, from the perspective of this paper, to err in the opposite direction. The case involved the status of "tokes"--gifts given by gamblers to casino dealers. The court accepted a finding of the district court that "The tokes are given to dealers as a result of impulsive generosity or superstition on the part of players, and not as a form of compensation for services. ..." but still ruled that they qualified as taxable income.

21 In many cases, "gifts" may be given to influence behavior even though, at the point at which they are given, they no longer have any influence on the behavior of the recipient. An obvious example is an employer who is better able to recruit employees because of his reputation for giving generous gifts at retirement. The analysis of this paper implies that such gifts should be treated as taxable income.

22This would apply in the case of envy, where one person's utility is a decreasing function of another person's utility, as well as in the case of altruism.

23This is somewhat of an oversimplification, since the particular rule used to weight the changes in the utility of the people affected will depend on the details of the altruist's utility function. Even if all utilities enter into it in the same way, the utility that the altruist receives from a one utile increase in someone's utility will in general depend on how great that utility already is; the altruist may well have declining marginal utility for other people's utility. But the correct rule will be similar to that proposed by Marshall although with different weights, and Marshall's rule can be interpreted (as he himself, from a utilitarian perspective, interpreted it) as a way of approximating the operation of the more complicated rule. Similar qualifications would apply to the "veil of ignorance" case.