Three major justifications for the welfare state, distinct but related, are (1) social welfare functions may be increased by redistribution (see my previous post on “the conservative social welfare function,”) (2) the welfare state serves as a form of social insurance against the ill chances of life, and most impertinently ambitiously (3) welfare and aid to the poor generally may be a public good via its effect on the utility functions of people who are at least mildly altruistic. Jonathan Gruber’s Public Finance and Public Policy, 3rd ed. offers the following defense of argument (3), stopping at argument (1) along the way:

Why is the government involved in the business of redistributing income?… If society cares equally about the utility of all its members, then social welfare may be maximized by redistributing from high-income individuals (for whom the marginal utility cost of losing a dollar is low) to low-income individuals (for whom the marginal utility gain of getting a dollar is high). Arguments for redistribution are even stronger if society cares in particular about low-income persons, a philosophy embodied in the Rawlsian social welfare function…

The private sector, however, is unlikely to provide such income redistribution, since redistribution faces the same free-rider problems encountered in private provision of other public goods. The consumption of the poor is a public good: I would like the poor to consume more, but I would prefer if others provide them the means of doing so, since I would then get the benefits of seeing the poor consume more but not bear the costs of their increased consumption. If everyone feels this way, then there will be too little private redistribution because everyone will be relying on others to contribute… There may be a role for a government in solving this free-rider problem by taxing its citizens to provide public redistribution. (Gruber, pp. 490-491)

Let me unpack this.

Recall that a public good is defined by two characteristics: (a) non-rivalry, and (b) non-excludability. Non-rivalry means that one person’s use of a good does not preclude another person’s use of it. Non-excludability means that it is not feasible— as distinct from merely not legal as a matter of policy– to exclude anyone from using the good. In this sense (to illustrate the concept) public schools are not a public good, though the general public refuses to hear this message and public finance economists often try to weasel out of it because of its unpopularity. Nonetheless, the fact is logically inescapable, for it is quite feasible– though perhaps illegal, but that’s beside the point– to exclude a child from a public school classroom. Also, classroom seats may be scarce/rivalrous at the margin, and a teacher’s grading time is certainly a rivalrous service: I can grade student A’s exam or student B’s exam, not both.

A welfare payment is certainly not a public good. It is rivalrous: if you receive cash from the government, I can’t receive that same cash. It is excludable: it is clearly feasible not to send the welfare payment. How, then, can Gruber claim that redistribution might be a “public good?” In a rather subjective sense. If a poor person eats a meal, no one else can eat that meal, or get any immediate benefit from it. But if a certain kind of altruism is built into others’ utility functions, these altruists all get some satisfaction from the poor person eating the meal. Given that the poor person eats, these altruists can’t be prevented from thus enjoying, second-hand, his meal. Therefore, this benefit is non-excludable. Nor does one altruist’s enjoyment of the poor person’s meal prevent another person from enjoying it. Therefore, this benefit is non-rival. Being non-excludable and non-rival, the external benefits of helping the poor

I have many objections to this interesting argument. First, it seems improper, somehow, for public policy to take into account such subjective factors. If one does allow it, the practice soon leads to unwanted conclusions. Suppose that, instead of altruism towards the poor, the general population felt hostility towards some group, but didn’t bother to harm that group much because of a free-rider problem. (“I wish somebody would go beat up those nasty wogs, but I can’t be bothered to do it myself.”) If we accept Gruber’s “public good” argument for the welfare state, we should also have to argue, it seems to me, that the brutal mistreatment of unpopular minorities is a public good. Second, the attitudes imputed to the public are not observable. If they were observable, the welfare state could be financed by a Lindahl tax enjoying universal consent. Of course, this argument applies to some extent to all public goods– how much people like a public good can’t be measured effectively in the absence of revealed preference and the price mechanism– but at least in the case of other public goods, the physical use of the good– walks in the park, driving on the roads, listening to public radio, whatever– is observable. Third, the argument is, in its strange way, simultaneously flatters and insults taxpaying citizens, with a certain insolence in both respects. It tells the citizen: (a) we know that, whatever you may say to avoid paying taxes, you really do care about the poor, but (b) we know that, left to your own devices, you won’t give as much as you wish that people like you would give. If some citizen sincerely says, “No, I really don’t care about the poor at all,” the public goods case for the welfare state fails. On the other hand, if most people, when it comes to charitable matters, follow Kant’s advice and act by maxims they desire to be universally practiced, the public goods case for the welfare state fails again.

But my biggest objection to the “public goods” argument for the welfare state is that it assumes what might be called a citizenist, or perhaps a territorialist, social welfare function. That is, it imputes to citizens a certain degree of altruism towards their fellow citizens, but not an equal degree of altruism towards the rest of mankind. If citizens would like the poor in general to consume more, regardless of nationality, then their first priority would probably be open borders, though possibly, if they have a very different understanding of how economy and society work than I do, they might support more foreign aid instead. At any rate, helping poor people resident in the US would not be a very high priority. Even if citizens are assumed to have a citizenist social welfare function, that should really point them towards the citizenist case for open borders and keyhole solutions (like DRITI) that hold natives harmless. To offer the public goods case for the welfare state, and at the same time to support migration restrictions, seems to make sense only from a decidedly territorialist perspective, i.e., if citizens feel altruism towards those present in a country, or at least they’re squeamish about observing dire poverty, but place little or no value on the welfare of those not on the country’s territory. They don’t want to be made to feel pity towards the less fortunate: hence they support the welfare state, and at the same time, the border as blindfold.

It’s irritating to have such attitudes imputed to me as a citizen-taxpayer. Even more irritating is the suggestion that such attitudes are implicitly granted the moral high ground. Gruber may well be right that attitudes such as he describes are an important reason why the welfare state exists. But since some of us don’t share the territorialist social welfare function, the welfare state cannot properly be regarded as a public good. And from a universalist utilitarian or Rawlsian perspective, the territorialist attitudes on the part of citizens that undergird support for the welfare state may be among the chief barriers to rational pursuit of the welfare of mankind.