Arguments for government and against private alternatives, both in general and in specific areas, are frequently based on the existence of market failure. In the first part of this essay I briefly sketch the idea of market failure and show that it provides problems for, hence arguments against, both private markets and the political alternatives. In the second part I discuss some of the imperfect ways in which private markets deal with such problems in the small, providing public goods or internalizing externalities within a framework of enforced property rights. The third part describes how that framework itself might be privately provided, examines difficulties that might arise due to forms of market failure within a system of private law and law enforcement, and considers how those problems might be dealt with. In the conclusion I argue that a society without a government, market in the large, could be stable under some but not all circumstances, and if stable could be expected to produce more attractive outcomes than a society with a government for reasons closely related to the reasons to expect the market in the small to work better than its political alternatives.
It is sometime in the 12th century, somewhere in Europe, and I am one of a line of men with spears, on foot, facing another bunch of men--on horseback with spears‹moving rapidly in our direction. I make a rapid cost benefit calculation. If we all stand, we might break their charge. If we run, we die. I should stand.
The mistake I have just made is the word ³we². I only control me, and I am only one spearman out of several thousand. If everybody else stands and I run, my running has little effect on whether their charge is stopped--and I won¹t be one of the men who dies stopping it. If everybody else runs and I stand, I die. So whether the rest of the line is going to run or stand, I should run. Everybody else in the line makes the same calculation. We all run and most of us die.
Welcome to the dark side of rationality.
This is an example of market failure‹a situation where each individual correctly chooses the action that best accomplishes his objectives, yet the result is worse, in terms of those same objectives, than if everyone had done something else. More familiar examples include the Prisoner¹s Dilemma, a situation where each of two criminals is better off confessing even though both would be better off if neither confessed, and air pollution in circumstances where it is in each person¹s interest to pollute but we would all be better off if none of us did so.
Central to these examples is the fact that my decision provides costs or benefits for other people. In deciding what to do, I take account of the effect on me. I correctly conclude that I am better off running than standing, ignoring the cost my running imposes on my comrades. They correctly conclude that they are better off running, ignoring the cost imposed on me. I gain by my decision but lose more by theirs, and similarly, mutatis mutandis, for them. We each decide correctly and are all worse off as a result.
This is clearly a failure of some sort‹but the examples I have given have nothing to do with markets, so why is it called ³market failure?² A likely answer is that the concept was developed in the context of neo-classical economics. Economists generally assume that individuals are rational, that they take the actions which best serve their objectives. That suggests that if we simply let each person do what he wants, the outcome should be attractive for everyone, a suggestion that can be converted into a formal proof, an efficiency theorem showing that, under some set of simplifying assumptions, the outcome of individual choice in a market system cannot be improved even by a wise and benevolent central planner.
In economic theory, market failure provides the exception to that conclusion‹an exception that may arguably swallow the rule. Where one person¹s acts impose costs or benefits on others that he has no reason to take account of, individual rationality cannot be expected to lead to group rationality, so there are opportunities for a wise and benevolent central planner‹perhaps also for a real world government‹to intervene in ways that make everyone better off. So economists are used to viewing the various forms of market failure they have analyzed‹the public good problem, externalities, adverse selection‹as reasons why free markets, laissez-faire, sometimes fails, hence arguments for government intervention.
They are half right. Market failure is a reason why free markets sometimes fail. But it is also a reason why the alternatives to free markets, the political mechanisms proposed for correcting those failures, fail. In order for government intervention to improve on the market outcome, it is not enough that there is something government could do that would give a better outcome. There must also be a reason to expect government to do it. Putting the point in the language of economics, the incentives of the relevant political actors have to be such that it is in their interest to act in ways that result in the improved outcome.
To see why this is unlikely, consider the simplest argument for why democracy works‹what I like to think of as the civics class model. In that model, politicians act in the voters¹ interest because if they do not the voters will vote them out at the next election.
The problem lies in the incentives not of the politicians but of the voters. In order to punish politicians for doing bad things, voters have to know that they are doing them. Politicians rarely run as bad guys or introduce bills to Congress entitled ³A Program to Make Farmers Richer and City Folk Poorer.² In order to figure out both what a politician is doing and whether he should be doing it, the voter must spend substantial amounts of time and effort studying the issues and the politician¹s voting behavior. In doing so, he is producing a public good‹better laws‹for a very large public; he himself collects only a tiny fraction of any benefit. Seen from the other side, he is bearing a large cost for a trivial gain‹an increase of perhaps one chance in a million in the probability that the right politician will get elected. Spearmen facing that logic run, firms pollute, and voters remain ignorant‹rationally ignorant. So far as achieving their objectives is concerned, since they know their vote has almost no effect on the outcome, it makes more sense to choose how they vote on other grounds‹which candidate is more handsome, more articulate, more popular with their friends and neighbors. That fits my observations of how voters behave. When I ask my students if they know the name of their congressmen, about half of them say they do. It is hard to keep track of what a politician is doing if you don¹t know his name.
As a second example, consider a less idealistic theory for why democratic government might work. A congressman proposes a bill that will benefit some interest groups and harm others. People on both sides of the issue offer campaign contributions, illegal bribes, endorsements, and other goods and services of value to politicians in exchange for voting their way. The amount they are willing to spend getting their way depends on how important it is to them. If the gainers gain more than the losers lose, the gainers are willing to spend more, lobby harder, with the result that the bill passes‹and should.
This argument too runs into the public good problem. How much an interest group is willing to spend to get its way depends on how important the issue is to that group, but that is not all it depends on. From the standpoint of the members of an interest group, contributing to the group¹s political efforts is the production of a public good‹where the public is not the whole population but the members of the interest group. If the auto industry gets a tariff passed all the firms will benefit, not just the ones that contributed to the campaign funds of the politicians who passed it. If opponents block the tariff, all consumers of autos will benefit, not just the ones who contributed to the campaign against the tariff.
One of the things public good theory tells us is that it is harder to produce a public good for a very large public than for a very small public. A concentrated interest group‹the auto interest group, say, which consists mostly of a handful of firms, one large union, and a few Michigan politicians‹can raise a substantial fraction of the value to its members of legislation they support in order to support the legislation. But consider a dispersed interest group such as those injured by auto tariffs, mostly consumers of autos and producers of export goods. It is a large and very dispersed group‹lots of people, each of whom loses only a little. Each individual member has little incentive to spend his time and effort opposing the tariff, when the result will be a tiny reduction in the probability that the tariff will pass‹a benefit received mostly by the other members of the group.
Two conclusions follow, both confirmed by real world experience. The first is that tariffs will get passed, even though they do net damage‹hurt the losers by more than they help the winners. The second is that the opposition to tariffs will come not from those who bear most of the loss but from those who bear a concentrated loss‹not auto consumers and export producers but foreign car dealers.
Generalizing these examples to the more general political market, we conclude that there is no reason to expect individual rationality in that market to lead to group rationality. In private markets, most of the time, an individual who makes a decision bears most, although not all, of the resulting costs, and receives most of the resulting benefits. In political markets that is rarely true. So we should expect that the market failure that results from A taking an action most of whose costs or benefits are born by B, C, and D should be the exception in the private market, the rule in the political market. It follows that shifting control over human activities from the private market to the political market is likely to increase the problems associated with market failure, not decrease them.
A market failure is also a profit opportunity. If the result of individuals acting rationally in their own interest is to make them worse off than if they acted in some other way, it follows that an entrepreneur who could somehow move them to the better outcome would produce a net benefit‹some of which, with luck, he could pocket. Hence in a market society there is an incentive for private parties to find ways around the inefficiencies due to market failure.
Consider one example of the public good problem‹radio and television broadcasts. By producing and broadcasting an entertaining program, I provide a benefit to everyone who listens to it. Since I cannot control who listens to it I cannot, as in the case of ordinary production, collect my share of that benefit by charging for it. The public in question is a large and disorganized one so it is clear, on theoretical grounds, that programs cannot be privately produced.
Yet they are. Some clever person thought up the idea of combining a public good with positive production cost and positive value with a public good of negative cost and negative value and giving away the package‹a program plus advertisements. As long as the net value is greater than zero and the net cost less than zero, people listen to the program and the broadcaster covers his costs.
There are a lot of other ways in which such problems get solved‹imperfectly but adequately. Consider any case in which two firms interact in such a way that decisions of one have large costs and benefits for the other, of a sort not easily controlled by contract. One way of internalizing the externalities is for the firms to merge. Indeed, one way of looking at the theory of the firm is to consider the size of the firm as a balance between the advantages of getting mutually related activities within a single organization with a single bottom line and the disadvantages of organizational diseconomies of scale‹too many layers of administration between the president and the factory floor.
For a final example out of many I might give, consider the problem of producing ideas and information. One solution is intellectual property law, but that solution is difficult to apply in areas such as basic research or business methods where defining just what is owned and who is infringing can become a very difficult problems. It is hard to imagine how a patent on the laws of physics or the idea of the supermarket could be defined and enforced. In the context of software, an additional problem is enforcement cost. When any individual customer can copy a four hundred dollar program onto a one dollar CDR and pass it on to a friend, it becomes hard for the producer of the program to enforce his copyright.
Solutions to this problem in the context of basic research are discussed at some length by Kealey. Knowledge of current cutting edge research‹Kealey¹s field is biology‹is of considerable value, and it is not the sort of knowledge easily summed up on a one page memo. In practice, the knowledge is largely restricted to the people doing the research, both because they are the ones who can understand each other¹s work and because they are the ones that the other researchers want to talk to. That makes such researchers valuable employees and consultants for firms and universities. While the researchers are unlikely to internalize the entire value of the information they produce, they may internalize enough so that the resulting income, along with nonpecuniary rewards of their work, make their research worth doing for them, and subsidizing for firms and universities. Kealey¹s conclusion, looking at several different fields where government subsidies went from near zero to very substantial, was that there was no observable effect on the rate of progress in the field. One might interpret that as evidence that the cost of misallocation of resources through the political mechanism‹diverting smart people into whatever field looked good in the popular imagination at the moment‹at least balanced the benefit of the additional money.
A similar pattern of incentives can be observed in the case of Open Source software and has been discussed at some length by Eric Raymond. Programmers claim no rights over others use of the code they contribute‹the one restriction is that any program derived from an open source program must itself be open source. Why then is it in the interest of programmers to spend their time and effort contributing to an open source project?
Part of the answer is nonpecuniary returns‹status from doing work that others use and know is yours, satisfaction from helping to produce something worth producing. Part of it, as in the case of scientific research, is the real benefit to the programmer of being part of the relevant community. A contributor to an open source project who encounters a problem with the software that he cannot deal with or finds it lacking some feature that he cannot easily add has immediate access to the other contributors, some one of whom may be well positioned to solve the problem‹and happy to do favors for someone who is contributing to the project and may next week do a similar favor for them.
That explains how programmers get useful services from their work, but not how they pay their bills. Part of the answer to that is that most programming is not done to be sold but to be used‹customized software for a particular firm. By basing its software on open source code, a firm not only saves a lot of programming time, it also provides itself with access to a pool of experienced programmers familiar with the code. That makes familiarity with an open source project and access to the associated community‹advantages that programmers contributing to the project already possess‹valuable assets for a programmer who wishes to be an employee or consultant of such a firm.
These examples are suggestive, not exhaustive. They do not imply that the problem of market failure does not exist. But they do suggest that the problem can easily be overestimated, due to the failure to consider the many ways in which ingenious individuals can work around the inefficiencies due to market failure‹and find it in their interest to do so.
The point applies to the political market as well as private ones. There too, inefficiencies due to market failure provide opportunities for enterprising individuals to rearrange the outcome and gain by doing so. Arguably that is why government does not work even worse than it does. One relevant mechanism was hinted at by the title of a news story in the Harvard Crimson that I remember from when I was an undergraduate, reporting on a talk by a prominent political scientist: ³Banfield Favors Corruption.² When the incentives generated by the political marketplace are sufficiently perverse, illegal market transactions may be the best way of dealing with the problem.
The argument for favoring private over political markets is not that market failure is always insoluble in one but not in the other. It is that the conditions leading to market failure are the rule in political markets, the exception in private markets.
It is tempting to think that we could get the best of both worlds by permitting the political market to intervene only where the private market fails badly. But someone has to decide what situations fit that criterion‹and it is in the political market that the boundaries of political control are set. The range of possible arguments for the existence of market failure is broad enough so that intervention can be justified almost anywhere‹if there is enough to be gained by justifying it.
My examples so far have been within the familiar structure of legal rules created and enforced by a government. In this section I give a brief sketch of how private markets might replace government in its most fundamental activities, including the creation and enforcement of legal rules, and then discuss problems such a system might raise and how they might be dealt with.
Imagine a society without a government to provide law and law enforcement. Individuals wish to protect their rights and settle their disputes, so private firms arise to produce those services. Each individual is the customer of a rights enforcement agency that provides him the service of enforcing his legal rights against others. Each pair of such agencies contracts with an arbitration firm, a private court that settles disputes between their customers, and agrees to abide by its decisions.
There is no government over the agencies to force them to abide by their contracts. Instead there is the discipline of repeat dealings. A firm that reneges on its arbitration agreement when the decision goes against it will find other firms unwilling to contract with it for arbitration. Violent conflict is more expensive and risky than arbitration, so a firm that can only enforce its clients¹ rights by violence is at a severe market disadvantage compared to firms that abide by mutual arbitration agreements. So we can expect an equilibrium in which such agreements are made and followed.
The legal rules and the structure of rights they embody in this society are created not on the political but on the private market. Part of the product that an arbitration firm sells to the enforcement agencies that are its customers is the set of legal rules it applies in deciding cases. 
As in ordinary private markets, the result is a tendency towards an efficient product‹in this case, a set of legal rules that maximizes the welfare of the people living under it. To see why, consider first a case where some change in legal rules would, on average, benefit the customers of both agencies. The better the service they provide to their customers, the more willing customers will be to pay for the service, so it is in the interest of both agencies to persuade the arbitration firm to alter its rules accordingly‹or if it does not, to shift to one that does. In the case where the change produces benefits for customers of one agency and (smaller) costs for customers of the other, it is still in the interest of both to agree on the change‹accompanied by an appropriate side payment from the one agency to the other, or changes in other rules that favor the second firm.
This is one advantage of the institutions I have described over those we are familiar with. For reasons discussed above, the creation of legal rules in the political market has only a very weak tendency to generate efficient rules, since the efforts that go to supporting or opposing legal changes are proportioned not only to the size of their effects on different interest groups but also to the degree to which each interest group is able to solve its internal public good problem‹loosely speaking, to how concentrated or dispersed each interest group is.
A second advantage of private law is that it avoids the public good problem associated with rationally ignorant voting. What determines the legal rules of the private market for law is individual choice‹the same mechanism that determines the characteristics of ordinary private goods. The individual consumer who decides that agency A has, on the whole, chosen a better set of legal rules for its customers than agency B is free to switch agencies, just as a consumer is free to decide to buy a different brand of car. He pays the cost of his research‹and gets the benefit. We expect rational individuals to be better informed about their market choices than about their political choices‹and in this system, law is chosen on the market.
These are examples of my earlier point about market failure on private and political markets. On the political market each is jointly choosing for all. Externalities and public goods are routine features of that market, private goods exceptions, hence market failure is the norm. The market failures I have just discussed occur naturally on the political market‹and are naturally absent in the alternative private market.
Rather than going on to a more extensive discussion of the logic and structure of the private market for law and law enforcement, I next consider problems that might be expected to arise with such a system and might outweigh its benefits.
In the system I have described, enforcement agencies serve much the same function as police forces in our system. This raises an obvious problem‹the risk that a group of such agencies may have sufficient force to overcome the others and establish a government, perhaps a worse government than we now have.
Such a project faces several difficulties. One is that, if there are many agencies, customers threatened by an agency that wishes to convert them to subjects can hire another to defend them. To put the point differently, what I am describing is a cartel and cartels are hard to maintain if there are many firms and easy entry to the industry. Thus one important issue will be how many agencies there are; the answer will depend upon economies of scale in the rights enforcement industry.
A second difficulty, also faced also by a tyrannical government of the present sort, is that governments do not have all of the relevant resources. Individual citizens control themselves and may have access to weapons, communication equipment, reputational and information assets of various sorts, and so some ability to use force in their own defense.
Finally, the practicality of converting a society from market anarchy to tyranny depends on a set of important but hard to define ideological factors. Consider the corresponding change in our society. Our military and police forces control most of the weaponry and are not particularly well paid. Why do they not seize power and revise our political system in their favor? Presumably the answer has to do with what sorts of actions they regard as appropriate and expect potential allies and opponents in such a move to regard as appropriate. Similar constraints would exist in the society I have described.
Earlier I mentioned the issue of economies of scale in the enforcement industry. If a firm with a third of the market can produce a better service at a lower cost than any smaller firm, we will end up with at most three firms‹making a cartel agreement in favor of them and against us a likely outcome. If a firm of optimal size serves only one percent of the market, such an outcome is unlikely.
So far as the ordinary business of rights enforcement is concerned, the evidence of existing police forces suggests that economies of scale do not go very far‹big city forces do not seem to provide better services at lower cost than smaller forces, although that judgment is complicated by the fact that big cities and small towns differ in lots of ways relevant to the cost of preventing crime. It is also complicated by the fact that most of what we observe are geographical monopolies. It is possible that a firm with 80% of the customers in a given area can outcompete a firm with only 20%, even if police forces in large areas cannot outcompete those in small.
A more interesting complication comes from the fact that the agencies are producing two related products. One is the enforcement of legal rules. The other, via the arbitration firms, is the set of rules being enforced. I have described how bargaining between agencies would change legal rules in the direction of efficiency but I have left out one important element of the problem. I have ignored the distributional outcome, the background starting point from which the agencies bargain. If the customers of one firm prefer rule A to rule A¹ by a million dollars and those of another prefer A¹ to A by two million, we expect them to end up with A¹‹but does the bargaining involve the second group offering to pay the first more than a million to get A¹, and having the offer accepted, or the first group offering the second less than a million to get A and having the offer refused? Where does the bargaining start?
The pessimistic answer is that the distributional outcome arises form the underlying threat game‹the options each party has if bargaining breaks down and conflicts must be settled by violence. If so, we would expect firms to get better results for the customers the better they were at interfirm violence. Economies of scale then depend both on economies in the business of enforcing rights and on economies in the business of threatening other firms with violence‹and evidence from current institutions suggests that economies of scale in the latter activity may exist to a considerably larger size than in the former. That brings us to the nightmare scenario that some critics of private law imagine, with the big fish eating the small and private order dissolving into civil war and eventual tyranny.
The optimistic answer is that once a stable system is established the recourse to violence is no longer a credible threat, hence the ability to use violence is no longer an important asset. If two firms get into a mini-war, both lose‹because both now have higher costs for producing a lower quality service than all the other firms that are peacefully settling their disagreements by arbitration. The distributional outcome, the background state from which firms bargain, is determined by history not threats‹the dead hand of the past providing the Schelling points of the bargaining game which both parties fall back on if no offer to change the rules can be agreed on. That view is supported by the extraordinary stability of national boundaries, also, presumably, the outcome of a mutual threat game. They do not shift a mile one way or another every time one of the two countries expands its army by a division or launches a new battleship. If this view is correct, it will be economies of scale in police services, not warfare, that determine the equilibrium size and number of rights enforcement firms.
So far I have been discussing the stability of the system against internal threats. Another concern is stability against external threats‹the defense of an anarcho-capitalist territory against aggression by adjacent states. Defense against governments is a public good with a large public, hence difficult to provide privately.
Difficult but not necessarily impossible. Providing an open source operating system is also a public good with a large public‹yet Linux exists. Tipping cab drivers who do their job well, and so providing incentives to good service, provides a public good for a large public‹and happens. As these two very different examples suggest, there are a variety of social mechanisms by which it may be possible to provide, at some level, public goods even for quite large publics.
Consider the following model for one way in which a large, modern, stateless society might defend itself‹a model variously inspired by Open Source, a Kipling story, and the 18th century militia system underlying the Second Amendment to the U.S. Constitution.
At the bottom level we have the militia, made up of a large number of volunteer units of amateur soldiers. For the volunteers, part of the reward is the same sort of fun they used to get from paintball, video games or mass medieval combat in the Society for Creative Anachronism‹a chance to play soldier. Another part is the satisfaction of feeling that they are doing their part in defending their homes, families, and the society they are a part of.
Military units require more than volunteer manpower. In my imaginary future, many of them are financed by firms. What the firms get out of it is good public relations‹when the liberty parade goes by, on April 15th of each year, their banner is there, carried by a trimly uniformed band of their employee/volunteers. Thus the firm demonstrates‹as firms today try to demonstrate, sometimes at substantial cost‹that they are a good corporate citizen, the sort of people that one ought to buy from and work for.
What the model so far lacks is organization‹ten thousand separate companies of a hundred men each do not an army make. To provide that organization we have a small cadre of full time professional soldiers, funded by charitable donations. In peacetime they organize war games for the militias, define communication standards, recommend weaponry, teach tactical doctrine, provide the professional superstructure for an amateur army. In wartime, if there is a wartime, the cadre is the command structure of the army.
Whether such a system could successfully depend its territory depends on a number of factors. One crucial one is the size of the threat relative to the resources of the ungoverned society. If, as we might expect, a stateless society grows faster and so becomes richer than competing states, it may not take a large fraction of its resources to fund an adequate defense‹perhaps no more than can be provided in the ways just described. The collapse of the Soviet Union considerably increased the chances that a stateless America could defend itself, since what I have just described, or something similar, should be more than adequate against any likely threats from either Canada or Mexico.
Another factor is the system of norms and values in the society. Some stateless societies, such as the Apache, have been militarily formidable despite the lack of a government to fund and coordinate their efforts. Others have not.
One advantage to this particular model for defense, like the historical model on which it is based, is that it provides a protection against the threat of internal tyranny. The cadre, like the professional army of the original U.S. system, is too small to seize power. The amateurs who control most of the military force are ordinary citizens widely distributed through the population.
So far I have been concerned with the stability of a society without government against internal and external threats. Another set of issues arise if we ask how well such a society, if stable, would work‹how nearly the laws it enforced would fit our views of either justice or efficiency.
To a first approximation, the answer is that such a system would generate efficient law, for reasons sketched earlier and discussed in more detail in Friedman (1996,2), legal rules that maximized to welfare of the people to whom they applied. But in this case as in other private markets, efficiency is only the first approximation, and may be prevented by market failure.
One problem arises
when the legal rule applying between A and
B has substantial effects on C. Intellectual property law provides one example. By agreeing to respect your copyrights, I increase the incentive for you to write books or create computer programs. One consequence is that there are more books and programs available to be pirated without payment by people who, via their enforcement firm and arbitration agency, have not agreed to respect copyright. My incentive to agree to respect your intellectual property rights understates the real benefit from my doing so. It follows that I will sometimes fail to so agree even when doing so would produce net benefits. We can expect, in a world of privately produced law, a less than optimal level of protection for intellectual property. Similar arguments imply a less than optimal legal protection against pollution. Another form of the same problem might be associated with the deterrence of crime. By paying my protection firm to make strenuous efforts to apprehend and punish those who violate my rights I make crime less profitable‹and the reduction in the number of criminals may benefit you as well.
In this case, however, the result is ambiguous. It is in my interest to not merely pay for protection, but make sure that potential aggressors know I have done so‹to identify myself and my property as protected. We observe such efforts today‹³We prosecute shoplifter² signs in department stores, ³These premises protected by Š² signs on stores. In a world where all protection against crime is private we can expect more such efforts, converting deterrence, in large part, into a private good. So while the deterrence I have paid for may help you by deterring a criminal who does not know which of us is protected, it may also deter criminals from crimes against me, leaving them more time for crimes against you. Thus the sign of the potential externality from my efforts at deterrence is uncertain‹we might get either a suboptimal or superoptimal level.
These examples bring us back to a central point of this essay. Market failure is a real phenomenon on private markets. Hence we cannot expect perfectly efficient outcomes from private markets, whether in law or in other things. But we have more reason to expect them from private markets than from public markets.
Intellectual property law at present is in large part the product of concentrated producer interests‹while the claim that whenever Mickey Mouse is about to go out of copyright the term of protection is extended is no doubt an exaggeration, it is an exaggeration of a real pattern. On theoretical grounds, it is hard to tell what the optimal level of protection would be, given the advantages and disadvantages of treating ideas as property. Whatever it is, there is no reason to expect our present institutions to produce it. As some evidence that they do not, consider the extraordinary stability of the term of U.S. patent protection, roughly speaking from fourteen to twenty-one years, over the past two centuries--a period during which the conditions likely to determine the optimal term, most notably the rate of innovation, have changed drastically.
Similarly for protection against crime. It is possible that a private market would produce a sub or superoptimal level of criminal deterrence. But is hard to see any reason to expect the political system to do better, or as well.
Similarly for defense against foreign nations. The private market may well produce a suboptimal level. But the public market has not always produced an optimal level of defense either, and what it produces often seems to be produced at a considerably superoptimal cost. And a government equipped with an army may face political incentives to use it under circumstances that do not increase the welfare of its subjects.
I have discussed the law provided by the market so far in terms of economic efficiency‹to what degree it maximizes the welfare, the ability to achieve their ends, of the people it applies to. Some readers may reasonably ask, not whether the law will be efficient but whether it will be just.
Unfortunately I have no theory of justice adequate to answer that question. For those readers who are libertarians, however, I can point out the existence of a variety of arguments and evidence for the claim that liberty is efficient, that it permits people to achieve their individual goals better than any alternative set of social arrangements. If so, a society with efficient law should be, on the whole, a free society.
What follows from these arguments is an uncertain conclusion. There are circumstances in which a stateless society would be unstable against internal or external threats, making the creation of such a society an unattractive gamble. There are other circumstances in which such a society should be able to maintain itself.
If a modern stateless society did prove stable, would it be attractive? Here I think the conclusion is clearer, although it is hard to imagine any rigorous proof. There are forms of market failure that would make the outcome of such a society less than perfectly attractive, inferior to what could be produced by a wise, all powerful and benevolent despot. But Superman as philosopher king is not a real world option. The political alternatives to the market for law can be expected to suffer from more frequent and more serious problems of market failure.
The private market will not produce perfectly efficient law, but it is hard to see why the public market will come even close‹and there is little evidence that it does. The economic arguments to show that a tariff injures the country that imposes it, for most countries most of the time, were worked out by David Ricardo almost two hundred years ago. Most countries still have tariffs‹and we know why.
Friedman, David, The Machinery of Freedom. Harper and Row (1973), Arlington House (1978), Open Court (1989).
³Private Creation and Enforcement of Law ‹ A Historical Case.² Journal of Legal Studies, (March 1979), pp. 399-415.
³Efficient Institutions for the Private Enforcement of Law.² Journal of Legal Studies, June (1984).
Price Theory: An Intermediate Text. South-Western (1986, 1990).
³A Positive Account of Property Rights,² Social Philosophy and Policy 11 No. 2 (Summer 1994) pp. 1-16.
³Making Sense of English Law Enforcement in the Eighteenth Century,² The University of Chicago Law School Roundtable (Spring/Summer 1995): 475-505.
Hidden Order: The Economics of Everyday Life. Harper-Collins, 1996. German translation published 1999, Japanese translation (date?).
³Anarchy and Efficient Law² in For and Against the State, edited by Jack Sanders and Jan Narveson, 1996 (2).
Law¹s Order: An Economic Account, Princeton University Press, Spring 2000.
Kealey, Terence, The Economic Laws of Scientific Research, New York: St. Martin¹s Press, 1996
Kipling, Rudyard, ³The Army of a Dream,² in Traffics and Discoveries .
Eric Raymond, The Cathedral & the Bazaar. O'Reilly 2001. Part of the material is webbed at:
 I may, of course, care about some other people, although rarely as much as they care about themselves. Part of the effect on them is then also an effect on me, and I take account of it to that extent.
 A discussion and defense of this assumption can be found in Chapter 1 of Friedman (1990) and Friedman (1996). The former is webbed at
 For the purposes of this essay I am not discussing the precise meaning of economic efficiency or its various definitions; interested readers may wish to look at Chapter 15 of either my Hidden Order or my Price Theory. The latter is webbed at:
 The analysis can be found in more detail in chapter 19 of Friedman (1990) and Friedman (1996). The former is webbed at:
 This statement is true, but far from obvious. It is not an assumption but a conclusion from price theory, which tells us that in a competitive market price, whether of consumer goods or of inputs such as labor, equals both marginal cost and marginal value.
 Kealey (1996).
 Eric Raymond, The Cathedral and the Bazaar. Webbed versions of his essays can be found at Š .
 This means both that any such software must be freely copyable without charge and that the source code, the information needed to modify the software, must itself be freely available.
 In this essay I refer to enforcement agencies and arbitration firms not because of some fundamental difference in the organizations but merely as a way of making it easier to distinguish between the two in my presentation.
 There is indirect evidence to support this conclusion in the way in which auto insurance firms in resolving conflicts between their clients. Just as the enforcement agencies face a choice between expensive violence and cheap arbitration, so the insurance companies face a choice between an expensive conflict in the courts and a less expensive informal resolution. Most such conflicts manage to stay out of the courts.
 A much more detailed description of such institutions can be found in Part III of D. Friedman, The Machinery of Freedom.
 The analysis of this market and the forces that end to move it towards an efficient set of legal rules can be found in Friedman (1996).
 We may think of a concentrated interest group as one that can raise a relatively large fraction of the value of a public good for the members of the group in order to pay for it, a dispersed group as one able to raise only a small fraction.
 The discussion here somewhat oversimplifies the situation, since the market for law is constrained by the fact that legal rules apply to a pair of individuals. My rights against you are part of both my legal rules and yours. Implications of that problem are discussed in Friedman (1996,2).
 I am indebted to James Buchanan for pointing out this problem in a perceptive review of my The Machinery of Freedom.
 The idea of Schelling points underlying the equilibrium of the social mutual threat game is discussed in some detail in Friedman (1994).
 ³The Army of a Dream,² in Traffics and Discoveries.
 For a historical example of institutions to make deterrence a private good in a society where criminal prosecution was private, see Friedman (1995).
 This point was discussed at greater length in Friedman (1984), along with other issues having to do with the efficiency of private law enforcement.
 An issue discussed at some length in Chapters 10 and 11 of Friedman (2000).