Humane Studies Review

Volume 7, Number 2 Spring 1992

Public Goods and the Justification for the State

by Dan Garrett

Book Review

A review of David Schmidtz's "The Limits of Government: An Essay on the Public Goods Argument" (Boulder: Westview Press, 1991)
Is state power legitimate? Modern political theorists typically legitimize the state with two major ideas: redistribution and the provision of public goods. In The Limits of Government, David Schmidtz takes on the second of these ideas, the public goods justification for the state. He gives this argument a workover and reaches some surprising conclusions.

The public goods justification for the state, so impressive to economists and political scientists, goes roughly like this. While a market system may allow self-interested individuals to create and allocate many goods optimally, there exists a class of goods -- collective or public goods -- that are not produced adequately in a market system. These collective goods are goods that all individuals want but for whose production it is often not individually rational for people voluntarily to do their part to secure a collectively rational outcome. The state can step in and force us all to contribute toward the production of these goods, and we can all thereby be made better off.

There are other arguments for the legitimacy of the state, but Schmidtz focuses exclusively on this particular argument. He points out that this argument is paternalistic in character, as are most arguments for the state. But the public goods argument is special in that it requires a particularly benign kind of paternalism.

"This paternalism is benign in the sense that the end it helps us attain is not only good for us but is also an end we actually desire. We all want the government to force each of us to contribute and thereby make us better off. Even if no one desires that the government take his or her money, it might nevertheless be true that everyone desires that the government take everyone's money." (p.2)

The benign nature of the normative premise behind the public goods argument explains why political theorists and particularly economists accept the argument so readily. The argument accords perfectly with economists' near-universally accepted yardstick -- the Pareto principle -- that actions or institutions that make everyone better off are efficient and in some strong sense good. It is, at least on its face, as uncontroversial as normative premises get.

Having sketched an explanation for the popularity of the public goods argument, we now ask whether that popularity is justified.

Justifications for the State: Teleological Vs. Emergent
Schmidtz begins consideration of the public goods argument by first categorizing in a general way moral justifications for the state. He wants to understand what kind of justification the public goods argument is in order to see what kind of criticisms are possible. Schmidtz argues that there are two different methods of justification in political philosophy.

"I call the two methods teleological and emergent justification. The teleological approach seeks to justify institutions in terms of what they accomplish. The emergent approach regards justification as a property that emerges from the process by which institutions arise." (p.3)

In debates, we must be aware of the difference between these justifications to avoid needless confusion; one person may be offering a teleological justification of a particular institution and another may be offering an emergent criticism. Their conclusions might " pass each other like ships in the night." (p.7) Supplying both justifications should satisfy almost everyone, but there seems to be no court to handle disputes that arise from a disagreement between the conclusions from each justification. Schmidtz's discussion of this taxonomy is straightforward and, for me, helpful.

Schmidtz goes on to argue that a major class of justifications -- hypothetical consent justifications -- are irrelevant to institutional justifications generally. Emergent justifications must rest on actual historical events, and so a hypothetically consensual emergence is irrelevant. And using a hypothetical consent story as a teleological justification simply obfuscates; the power of the argument rests on the justification of the end-state and not on the hypothetical process that got there. Commentators on Rawls have made this point already; I believe Schmidtz does it more succinctly and more generally.

A Detour: Justifications of Property Rights
In Chapter 2, entitled " Property," Schmidtz points out that the Lockean justification for private property is an emergent one. The difficulty most philosophers have with justifications for property rights concerns original appropriation, or the first removal of the property from the state of nature. Locke said that original appropriation by an individual was legitimate as long as enough and as good were left in common for the rest of society. This so-called " Lockean Proviso," when taken at face value, seems to make little or no appropriation justifiable. And, since the legitimacy of property claims rests for Locke on an emergent justification, illegitimate original appropriation implies illegitimate ownership thereafter.

Schmidtz comes to a different conclusion, by drawing insights from the " tragedy of the commons." If no one appropriates the property, society will overuse and ultimately destroy the commonly owned property. Since the Lockean Proviso was intended to assure that people could always have the use of some property, the fact that common ownership leads to no one's ability to use the property means that common ownership itself violates the Lockean Proviso. " Leaving goods in the commons fails to satisfy the Proviso." (p.21) Schmidtz makes another point: " It is not that appropriation of land inevitably decreases the amount of land available to others. It only inevitably decreases what is available for original appropriation by others, which is not the same thing at all." (p.27) Thus, rather than severely limiting the extent of justified original appropriation, the Lockean Proviso, as Schmidtz interprets it, justifies widespread original appropriation of resources.

Public Goods Problems as Prisoner's Dilemma
In Chapter 4, " The Prisoner's Dilemma," Schmidtz returns to the public goods argument. He sets up the public goods problem as a prisoner's dilemma. There are two reasons force may be necessary to produce an optimal quantity of a collective good. First, if asked to contribute voluntarily, an individual may realize that the others in the community are contributing enough to produce the good and decide to enjoy the good for free. Using the normal economics nomenclature, Schmidtz calls this the " free rider" problem. Second, an individual might not contribute because she thinks the project will be underfunded anyway. She lacks the assurance that other people will contribute, and so she fails to contribute to avoid wasting money. This Schmidtz calls the assurance problem. These two problems are conceptually separable. It is with Schmidtz's deft handling of the assurance problem that he begins to develop his strong case.

Schmidtz considers a hypothetical attempt to fund an AIDS research project. He suggests an assurance contract, a money back guarantee of sorts that states something like,

The contract is enforceable against a contractor if and only if the rest of the group agrees to contribute enough to ensure that the project's total funding is sufficient to produce a return R that exceeds the contractor's cost c. (p.66)
This contract gets around the assurance problem; every would-be contributor can be assured that their pledge will only be taken if enough pledges come in to guarantee the project's viability. There would still be the free rider problem, however. (see note1)

Here Schmidtz inserts three possible reasons why any particular alleged free-rider problem may not be a problem. 1) People may enjoy participating in the venture. 2) If exclusion of some benefits is possible, it could be that it is worthwhile to contribute to gain the benefits that would have been lacking. For example, if getting the AIDS vaccine were tied to contributing to the research project, then it might be a best response to contribute. (see note 2) 3) Finally, there is the option of a unanimous commitment clause. The project gets done if and only if everyone contributes. Since everyone is assumed to value the project more than they are asked to contribute, the dominant strategy for everyone is to contribute.

It is at the beginning of Chapter 5, " From Prisoner's Dilemma to Public Goods," that Schmidtz starts turning the knife.

According to this argument, the state is coercing people for their own good -- and they know it. Moreover, they want the state to do this, because they cannot get what they want with voluntary provision mechanisms. Thus, one of the most attractive features of the public goods argument is the minimal nature of the normative assumptions it must make in order to ground a justification of the state. The public goods argument seems to presume only the legitimacy of helping people to do what they want to do but cannot do without the state's help. Yet this picture of the normative side of the coin, so readily accepted by political theorists, turns on an assumption that coercive public goods provision really does make us all better off. (p.82)
However, Schmidtz's result from the previous chapter showed that, under the assumptions of the public goods argument, private provision is possible. Therefore, this version of the public goods argument which rests on such a benign paternalism fails.

If we acknowledge that people differ in their like or dislike for almost everything, then we have to realize that some of the people who don't contribute in a voluntary scheme are honest holdouts. Governmental provision, while it may avoid the problem of free riders, forces honest holdouts to pay for something they do not want. Thus, the normative side of the public goods argument must justify forcing some people to pay for other people's projects. This normative task is far more ambitious than a simple reading of the public goods argument would suggest is necessary. " Merely pointing out that coercing these people is the most efficient way for others to get what they want does not even begin to discharge the burden of proof." (p.84) Schmidtz concludes this section,

(1) If we know there are no honest holdouts, we do not need coercion. We can simply require unanimity as a precondition of the contract's validity. (2) If there are honest holdouts but we can tell them apart from others, we do not need coercion. We can require unanimity among those who are not honest holdouts. (3) If we believe there are honest holdouts but cannot identify them, then we may need coercion, but in that case justifying coercion requires us to justify using some as means to the ends of others. (pp.86-7)
Comments and Objections
I will offer three comments on or objections to Schmidtz's arguments in Chapters 4 and 5. First, his conclusion that contributing is a dominant strategy given the assurance contract and the unanimity clause rests on a very specific method of seeking voluntary funding. Namely, the entrepreneur(s) getting people's signatures on the contribution contract must be committed to making either A) only a single try to get funding, or B) multiple tries at unanimous funding. In the case of B, however, a problem arises if the entrepreneur allows for the possibility of excluding early holdouts from the effort in subsequent tries. That is, suppose the entrepreneur, in passes subsequent to the first, revises the contract so that the honest holdouts (m) are deducted from the total number of persons (n) to give the number needed for unanimity. The problem is that this n-m contract provides the opportunity for a potential free rider to pretend she is an honest holdout in order to free ride in a subsequent attempt to fund the project. Even if she really values the project more than her contribution, she might refuse to contribute in the hope that a second contract will be drawn up that is not conditioned on her participation. This would not be a problem if she can be sure that the public goods supplier can make only one pass at the community. If the public goods supplier can make a second pass with a contract that does not include a contribution from her, it might be in her interests to say no the first time.

Schmidtz does discuss this issue in the context of the supplier's going ahead with the project even if there is one holdout. Since the contract states that the supplier must have 100% participation to do the project, he is contractually forbidden from doing the project anyway. I'm concerned with the lack of commitment to take only one pass at the community. If multiple passes are allowed, Schmidtz's payoff matrix does not fully describe the strategies available in a repeated game. (Note that iteration usually makes the likely outcome better than it would be where there was no iteration. See Robert Axelrod's The Evolution of Cooperation [New York: Basic Books, 1984] for an extended argument along those lines. Nevertheless, in Schmidtz's scenario, the possibility of repetition makes the optimal outcome less likely.)

This objection -- that the possibility of making a second non-unanimous try to get the project funded spoils Schmidtz's equilibrium outcome -- can be overcome certainly in some cases and perhaps generally. Commitment mechanisms might be imagined that would limit the potential supplier to making a single " take it or leave it" offer. Furthermore, the time cost of delay to our potential free rider and/or social pressures directed against her might make contributing a best strategy for her. There are no doubt other ways to solve this problem, but I think it is still an objection worth considering.

My second objection relates to the discussion between reviewer Robert Sugden and author Anthony de Jasay (Social Contract, Free Ride, [New York: Oxford University Press, 1989]) in the Winter 1991-92 Humane Studies Review (Vol. 7, No. 1). Jasay claims that public goods are " indivisible," while Sugden takes many of them to be divisible. (I prefer the terms " lumpy" and " continuous," respectively.) I take Jasay to mean that, in general, supplying a particular public good means supplying a fixed quantity of something, like a " four-lane bridge" or a " 100-foot lighthouse." Less than that is insufficient, and more is unnecessary. Sugden is concerned with a public good for which each person's utility is a continuous and increasing function of the quantity and/or quality of the good supplied. (Perhaps " public parks" qualify.) I'll allow that there are both types of public goods, continuous and lumpy. Do continuous goods make any special problems for Schmidtz?

I think that the production of a specific quantity of either continuous or lumpy goods poses no problem for Schmidtz's mechanism. The production of some given quantity of a public good is, after all, what Schmidtz's mechanism seeks to do. This seems to pose no problem for lumpy goods, which are always available only in discrete " lumps" . The reason that production of a continuous good is more problematic, however, is that with its production someone must know and/or decide how much of the good is to be produced. For example, a discrete " Welcome to Spoonerville" sign on the main road into Spoonerville (a lumpy good) has a fairly clear optimal quantity connected with it, either 1 or 0. How much open access park space Spoonerville should have is more subtle. Schmidtz does not address the optimal quantity issue (see note3); we must ask whether this harms his argument.

The public goods argument is that the state must step in to produce a unanimously desired public good that would otherwise not be produced or would be underproduced through voluntary action. There is no mention of how the state discovers or knows the (presumably) optimal quantity to produce. (To decide how much, say, park space is optimal is no trivial matter in the real world, although it is easy enough when you are given the utility functions in a textbook exercise.) But whatever that optimal quantity is or is believed to be, Schmidtz's mechanism can produce it. So I do not think this concern weakens Schmidtz's argument, although it relates to our conclusions below. For the time being, let's recognize that Schmidtz has not given us a mechanism to reveal degrees of demand, perfect knowledge of which would allow optimal investment in a divisible public good. An assurance contract with unanimity clauses is not a perfect demand revelation scheme; it only answers whether people prefer contributing some dollar figure c over not contributing given that the good won't be produced if they fail to contribute c.

My third objection comes from what I believe many economists would say when presented with the result that an assurance contract with a unanimity clause allows voluntary provision of a public good. They would say, " Fine. But the government can produce the same thing and eliminate all the costs of contracting." Governmental provision, they would say, lowers transactions costs. I believe Schmidtz might respond to this by saying that he has shown that the public goods argument is not sufficient to justify the state. If one added the premise that the government can produce a particular collective good at a lower transaction cost than can voluntary means (and if the unanimity condition is met), then that would seem to justify (teleologically) state provision of that good. A teleological justification for the institution of the state would require more descriptive premises, including ones that would preclude the state from going beyond its narrow public goods mandate.

So if the claim that governmental provision of some public good(s) involves lower transaction costs than voluntary provision is true then governmental provision would indeed be teleologically justified. (Schmidtz has shown that without such a claim the public goods argument for the state fails.) I think Schmidtz would have to admit that such a claim, were it true, would provide grounds for a teleological justification of the state, although he does not discuss it. And so we must ask: Is it true that governmental provision is less costly than voluntary provision?

Economists do a lot of hand waving on this question. A system that allows a government to produce public goods is costly. Resources are spent on voting (demand revelation schemes are costly!), seeking office, lobbying for the policy change, collecting taxes, and so forth. Furthermore, if a referendum on a public goods issue fails a unanimity test, we lose all confidence that the public good in question is unanimously desired, and so we certainly have to go back and justify a more demanding moral premise for the public goods argument. If we use representatives to decide such issues, we not only lose confidence about unanimity but also open the door to rent-seeking activities. That is, people will lobby, under the auspices of public goods, for private goods that benefit them alone but for which others must pay. A major cost, then, of a governmental provision of public goods is that the government will expand into activities well beyond those allowed it by its public goods mandate. These governmental transactions costs must be weighed against the voluntary transactions costs of writing up an assurance contract and collecting signatures. It is not clear to me that collecting signatures is any more costly than a referendum; if cost- effective, an entrepreneur could set up signing booths analogous to voting booths to get the vast bulk of the necessary signatures. And there are far fewer possible negative spin-offs with voluntary provision. But this empirical question requires an empirical answer, a research task we voluntarists must take on.

Source of the Disagreement
In an important way the divisible goods issue and the transaction costs issue exemplify the critical premises underlying the disagreements between those who favor voluntary provision of collective goods and those favoring state provision. Those premises concern what is assumed to be known about people's demands for goods. If we have perfect knowledge of people's demands, then state provision might lower transaction costs and might yield the optimal quantity if the good is divisible. But is this a realistic premise on which to base a criticism of voluntary provision? No. The truth is that we don't know what quantities of which collective goods people really want. Let us consider how Schmidtz's mechanism handles some important informational issues relating to public goods.

The first issue is determining that everyone (or nearly everyone [see note 4]) desires the good in question. The public goods argument rests on the assertion of unanimity, but policy must rest on fact. How could we know whether everyone (or nearly everyone) wants a good? Schmidtz's mechanism certainly gives us that, as witnessed by the signatures. Aside from referenda that clearly link potential cost to benefit, voting mechanisms do not give us a clear picture of the demands of individuals.

The second issue is determining that such and such a way is the best way to provide benefits people desire. How do we know, for instance, whether it is better to spray a large area with pesticide against mosquitos or just wear mosquito repellant? This is in part a question of whether access to the good should be limited or open. (see note 5) Fortunately, Schmidtz has implicitly given us a way to find out which goods are really public goods. That is, if a project can succeed using the assurance contract with unanimity, then by god it's a public good. The project is satisfying a demand people have and doing so in a way such that they know no better way. With governmental provision, on the other hand, we have no such assurance. The fact that the government supplies some benefit at zero marginal cost often precludes people from seeking lower cost ways to provide the same benefit, as they have to keep paying for what the government is doing anyway. (see note 6) And importantly, not much of what the government does can be thought of as providing public goods. Most of it is simply wealth transfers, either in the context of actual private goods or of restrictions on trade to benefit a particular class of producers. (see note7)

The third issue gets back to optimal production of a public good, which we discussed above as a possible area of objection to voluntary provision. With governmental provision, we have no confidence that what we get is an optimal level. For example, do we have an optimal level of national defense? In theory the optimality issue is decided by adding up people's marginal benefit of an additional unit of defense (whatever a " unit" of defense means) and producing defense to the point where the marginal social benefit equals the marginal cost. I may have missed it, but I do not remember Congress going through this calculation or anything like it. In fact, the voting models political economists use to explain Congressional behavior suggest that the optimal outcome would not in general be reached. (For a good discussion of how the government hides costs and magnifies benefits through " fiscal illusion," see James Buchanan's " ÔLa Scienza Delle Finanze': The Italian Tradition in Fiscal Theory," in Fiscal Theory and Political Economy [Chapel Hill: University of North Carolina Press, 1960, pp. 24-74].)

Of course, suggesting that one government does not produce an optimal amount of a particular alleged public good does not mean that we can't imagine some demand revelation scheme that does approximate such optimality. We would need to compare such schemes to voluntary schemes like Schmidtz's. While Schmidtz does not suggest this, I believe clever entrepreneurs might approach optimal provision -- and placate Sugden -- by repeated use of the assurance contract. For example, everyone signs on for a 100-acre park, and then everyone signs on for 20 more acres, and then there are non-contributors at another 20 acres, etc., with a convergence on some optimally sized park.

And so in the real world, entrepreneurs must seek ways to encourage people to reveal their true demands. A procedure that invokes consent and discourages free riding such as Schmidtz's has a lot going for it when it comes to our world of demand uncertainty. Specifically, Schmidtzian outcomes pass a market test. Perhaps the " transaction costs" in a voluntary scheme are necessary for us to be confident that there really is a demand for a prospective good at a particular price. (Giovanni Montemartini shows how political entrepreneurs work to generate free- riding -- rather than reduce it -- in " The Fundamental Principles of a Pure Theory of Public Finance," in Classics in the Theory of Public Finance, by R.A. Musgrave and A.T. Peacock [London: Macmillan, 1958].)

Criticisms of voluntary provision almost always ride on the high horse of better-than-realistic information. Schmidtz has put voluntary provision on the same size horse as sits the public goods argument and found that the joust is a draw. My claim is that, under real world assumptions, we can extend Schmidtz's analysis to show that voluntary provision wins the match.

Reciprocity and the Feedback Theory of Morality
In Chapter 7, " The Morality of a Cooperative Society," Schmidtz covers a lot of ground and makes many interesting points. Space limitations allow me to discuss only my favorite. This point concerns reciprocity in social interaction: " A reciprocator will cooperate if and only if his partner is a cooperator." (p.138) Schmidtz wants to bring this notion of reciprocity into the discussion of moral norms. After invoking John Rawls' definition of society as a cooperative venture for mutual advantage, he states,

My first premise ... is that, all other things being equal, one candidate for the status of fundamental moral norm is superior to another along a critically important dimension if it is more conducive to the success of this cooperative venture. Superiority along this dimension may not always be decisive, but it surely counts for something, especially in a discussion of the moral norms pertaining to public goods production. (p.141)
Because moral agents choose moral norms independently, the first premise demands that we look at how a suggested moral norm fares in the world, regardless of whatever other agents choose as moral norms. " Because we choose norms for ourselves, the norms that actually promote cooperation often differ from the norms that would promote cooperation if we were choosing norms for everyone." (p.141, emphasis in original) Thus, tit-for- tat is a better norm under this reciprocity standard than is nonstrategic altruism, because tit-for-tat helps ensure cooperation whereas nonstrategic altruism encourages free riding. Schmidtz wants us to ask the following question of moral norms: " Is this norm conducive to a cooperative society's success, given how people will react both to it and to agents guided by it?" (p.153) He calls this the " feedback theory" of morality. Schmidtz convinces me that, when we judge norms, we should consider how they fare in a world of alternative norms. Much as Axelrod did with iterated game strategies, we should mentally " run" proposed norms against each other to test which norms lend themselves most to cooperation.

Limits deftly integrates a clear game- theoretic discussion of how and why collective goods can be supplied voluntarily with a lucid explanation of why that fact undercuts the public goods argument for the state. On the way, Schmidtz gives us a useful categorization of justifications for social institutions and a new way to think about moral norms. The book is peppered with insights, only some of which have I discussed here. (see note 8)

Limits is a short book, and so it is an easy book to hand to someone who is being sloppy with a public goods argument for the state. On the other hand, it gives only a partial criticism of the public goods argument. There is a great deal more that has been said and that still must be said about how collective goods can be supplied voluntarily, how governmental provision is inefficient and misguided, and how a voluntary mechanism for collective goods provision may be more conducive to entrepreneurial innovation than a governmental mechanism would be. (see note 9) Limits takes the public goods argument and shows that the emperor has no clothes. I believe that Schmidtz has given us some useful weapons in the fight to overthrow the emperor.

1 Schmidtz makes nice use of the 2x2 reduced-form payoff matrices often used in game theory. The assurance contract puts zeros for the payoffs in the left column, which makes mutual contributing a dominant strategy except for free-rider effects. I shall not be using the 2x2 matrices, although Schmidtz uses them quite effectively.

2 In the limiting case, if all benefits are excludable, there is no public goods problem. See Harold Demsetz, " The Private Production of Public Goods," in Journal of Law and Economics, 13 (October 1970), pp. 293-306; also in The Theory of Market Failure: A Critical Examination, edited by Tyler Cowen (Fairfax, VA: George Mason University Press, 1988).

3 He might allow that assurance contract entrepreneurs would be expected to figure out how large their output should be to maximize the appeal of their product.

4 We can set aside for a moment the moral hurdle of coercing honest holdouts given governmental provision and smugly enjoy the fact that this is not a moral hurdle faced by voluntary provision.

5 An excellent discussion of equal versus selective access is in Kenneth Goldin, " Equal Access vs. Selective Access: A Critique of Public Goods Theory," in Public Choice 29 (Spring 1977), pp. 53-71; also in Cowen (1988).

6 An additional point to consider is the adaptability of voluntary or governmental provision schemes to innovations and changes in technology. I've already suggested that governmental provision often prevents people from seeking alternatives. But there is also a factor of inertia more common with governments than with private firms. Important in this light is that historically, most collective and quasi-collective goods now produced by governments were initially produced voluntarily. People saw the need for something and solved it before governments decided they'd like to do it. Voluntary mechanisms, not governmental, were the important innovators. There are examples in law (see Bruce L. Benson, The Enterprise of Law: Justice Without The State [San Francisco: Pacific Research Institute for Public Policy, 1990]), education (see Jack High and Jerome Ellig's chapter in Cowen [1988]), lighthouses (see Ronald Coase's chapter in Cowen [1988]), toll roads (see Dan Klein, " The Voluntary Provision of Public Goods? The Turnpike Companies of Early America," in Economic Inquiry 28 [October 1990]), and even welfare (see David Beito's " Mutual Aid for Social Welfare: The Case of American Fraternal Societies," in Critical Review, Vol. 4, No. 4 [Fall 1990]), as well as other areas yet to be explored by classical liberals.

7 We can ask what behaviors are rewarded under governmental versus voluntary provision. Schmidtzian entrepreneurs are rewarded for minimizing free ridership and transaction costs. The agents of the state are rewarded by coercing honest holdouts (making the agents more popular with the genuine demanders) and extending the role of the state into other more questionable areas.

8 Limits covers an interesting array of both arguments and fields of scholarship. Schmidtz deals with ethics, political philosophy, game theory, and experimental economics; one of the strengths of the book is its interdisciplinary approach. Political philosophers will be most comfortable with chapters 1 and 7, economists and political scientists with chapters 4 through 6. Chapters 2 and 3 would, I suspect, be especially appreciated by classical liberals. Specialists in these areas might read only selected chapters, though the book is so well written that most will end up reading the entire thing.

9 I am not criticizing Schmidtz for not saying everything that has been said or could be said about voluntary public goods provision; I am just encouraging the reader to read more. A good book to complement Limits would be Cowen's The Theory of Market Failure: A Critical Examination.

Dan Garrett is a graduate student in Economics at Stanford University.

Copyright 1992 by the Institute for Humane Studies.

Return to Spring 1992 Humane Studies Review menu IHS Publications or IHS homepage