It is often said that Adam Smith, despite his general belief in Laissez-faire, made an exception for education. That is not entirely true. In the course of a lengthy and interesting discussion, Smith argues both that education is a legitimate government function, at least in some societies, and that it is a function which governments perform very badly. His conclusion is that while it is legitimate for government to subsidize education, it may be more prudent to leave education entirely private.
My purpose in this essay is to argue that Smith's conjecture was correct. While government schooling, free and compulsory, is at present nearly universal in developed societies, the case for it is unconvincing. There are arguments for government provision of schooling, as there are arguments for government provision of any good or service, but the arguments in favor are weaker, and the arguments against stronger, than the corresponding arguments for other goods and services that we routinely leave to the private market.
The arguments in favor of government involvement in schooling can be roughly divided into four groups: Externality arguments, information arguments, capital market failure arguments, and egalitarian arguments. I will deal with them in that order.
The most common arguments for government schooling involve the claim that it produces large positive externalities, that by schooling my children I greatly benefit society as a whole, and that it is therefore inappropriate to leave either the decision of how to school them or the cost of doing so entirely to me. On further analysis, this claim divides into three variants, one wrong and two dubious.
The simplest version is the one that is wrong. It is said that since education increases human productivity, by educating my child I increase the wealth of the whole society, making all of us better off. One obvious problem with this argument is that, if correct, it applies to a lot of things other than education. Physical capital also increases productivity; does it follow that all investments ought to be subsidized? Better transportation allows workers to spend more time working and less time commuting; should we subsidize the production of cars? The argument suggests that everything worth doing ought to be subsidized-leaving us with the puzzle of what we are to tax in order to raise the money for the subsidies.
What is wrong with this argument is that it misses is the way in which the price system already allocates "social benefits" to those who produce them. Building a factory may increase the wealth of my society-but most (in the limit of perfect competition, all) of the increase goes to the investors whose capital paid for the factory. If I use a car instead of a bus to commute, the savings in time is added either to my leisure or my income. If education makes me a more productive worker, my income will be higher as a result. That is why top law schools are able to sell schooling to willing customers at a price of about twenty thousand dollars a year.
Schooling-like a new car-produces non-market benefits as well. But these too go mostly to the student, enabled by education to appreciate more of the riches of the culture he lives in. There may be effects on other people as well, but they are typically small compared to the benefits to the student, and their sign is not always clear. When my child becomes an expert in Shakespeare and quantum mechanics one result may be to enlighten and entertain her friends, but another may be to make them feel stupid. In just the same way, the beauty of my new car may produce the pleasures of aesthetic appreciation or the pains of envy in those who watch me drive it down the street. To base the design of our institutions for schooling on the uncertain effect on such third parties rather than the direct effect on the schooled makes no more sense than to base the design of cars on their value to everyone except the owner.
There is, however, at least one important respect in which my investment in education-or a factory-does produce substantial external benefits. Even if my income fully reflects my productivity, as it will tend to do in a market economy, not all of my income goes to me. Some of it goes to the tax collector. It follows that some investments, in factories or in people, may not get made even though they are worth making, because the share of the benefit that goes to the investor is not enough to pay the cost of the investment. This inefficient failure to make some worthwhile investments is one form of what economists call "excess burden"-the cost of taxation above and beyond the amount collected.
There is a problem in trying to solve this particular inefficiency by subsidizing investments. In order to pay a subsidy one must collect a tax-and the additional tax increases excess burden at the same time that the subsidy reduces it. Excess burden is an argument against taxation, not for subsidy.
Another version of the externality arguments locates the externality not in the increased economic productivity of educated people but in their increased virtue. Both religious and utilitarian variants of this justification for government schooling were popular in the nineteenth century. Conservatives wanted to use publicly controlled education to teach the masses religious virtue. Many utilitarians, including Bentham himself, believed that while freedom was a good thing in most contexts, it was necessary first to teach people how to use their freedom-which is to say, to teach them utilitarianism. A form of this argument which still remains popular holds that uneducated people are particularly likely to become criminals, justifying government schooling as a form of crime control. While I have not yet heard anyone argue that government schooling is needed to make the public ecologically responsible, to properly train the crew of spaceship earth, it seems the obvious next step in the evolution of the argument-considering what is actually being taught to elementary school students in the more up to date government schools.
The thesis has two versions-education and indoctrination. The first assumes that crime and sin are the result of ignorance rather than rational choice. The evidence for this thesis is far from clear. As a general rule, criminals seem to exhibit rational behavior in their crimes-little old ladies, for example, get mugged a lot more often than football players. Criminals who have been caught and imprisoned frequently return to a life of crime-although that experience surely teaches them more about the consequences of their actions than they are likely to learn in any school. And, of course, even if ignorance is one source of crime, the argument depends on the assumption that government schools are better at dissipating ignorance than private ones. As we will see, both theory and history provide reasons to doubt that.
The indoctrination version of the argument may make somewhat more sense. In a private system, children will be taught what their parents want them to know. In a government system, children will be taught what the state wants them to know. So the government system provides an opportunity for the state to indoctrinate children in beliefs that it is not in their interest, or their parents' interest, for them to hold. Insofar as some virtues require one to act against one's own interest-for instance, by not stealing something even when nobody is watching-that is an opportunity to indoctrinate children in virtue.
One good reply to this argument was made by William Godwin, who, in 1796, expressed his hope "that mankind will never have to learn so important a lesson through so corrupt a channel." To put the argument in more modern language, government schooling does indeed provide the state with an opportunity to indoctrinate children-but there is no good reason to believe that it will be in the interest of the state to indoctrinate them in beliefs that it is in the interest of the rest of us for them to hold. Many modern societies have strong legal rules designed to keep the state from controlling what people believe-the first amendment to the U.S. constitution being a notable example. It seems odd to combine them with a set of institutions justified as doing the precise opposite.
In an interesting recent article, John Lott explores the question of why schooling is controlled by the state in modern societies. His conclusion is that government schooling is a mechanism by which the state lowers the cost of controlling the population. Part of his evidence is the organization of modern government school systems-in particular the almost complete absence of systems where parents choose the school and funding is proportional to number of students, an arrangement which would put pressure on the school to teach what the parents, rather than the state, wanted. Part is a statistical analysis of data for a large number of nations, designed to explore the relation between government schooling and other characteristics of government.
One final version of the externality argument is the claim that my education provides benefits to others because it makes me a more rational voter. While the argument is logically correct, its implications are limited. It is perhaps best understood as an argument for subsidy, not control. It is in my private interest to have a correct understanding of the world around me, and such an understanding will make me more able to evaluate government policy as well as more able to make private decisions. The only argument for government control is that it can force me to learn more about issues relevant to voting, instead of issues relevant to private choice. The problem with this is that the agency that does the controlling has its own interest with regard to how I vote-which brings us back to the indoctrination argument.
A second problem with the argument is that it implicitly assumes that different voters have the same interest, so that my rational voting benefits you as well as me. For some issues this is no doubt true. But other issues-many of them in a modern state, unfortunately-involve attempts by one group to benefit itself at the expense of others. In such situations, your rational voting may well make me worse off. Subsidizing education in how to use the political system in one's own interest becomes the political equivalent of subsidizing an arms race, and equally unproductive.
A final problem is that the argument works only if government run or government subsidized schools actually educate better than private schools. If the costs of government control more than cancel the benefits of government subsidy, the advantages of educating students well provide no argument for having the state educate them badly.
Externality arguments, not only for government schooling but for many other issues as well, often make the mistake of adding up only externalities with one sign-positive in the case of schooling, negative in discussions of population or global warming-while ignoring externalities with the opposite sign. The result may be misleading, since it is the net externality that provides an argument for government involvement. If my action benefits one person by a dollar and injures someone else by two dollars, that is an argument against subsidy, not for it.
What negative externalities might result from schooling? One I have just mentioned-you may use your improved education to more effectively pressure the government to benefit you at my expense. A similar possibility exists for private transfers. Ignorance may perhaps produce crime-but education produces more competent criminals.
Another possibility is that schooling may produce negative externalities because it is used in the competitive pursuit of status. Consumption bundles of physical goods and services are not the only thing that individuals care about. If one reason I wish more schooling for myself or my children is so that I or they will have more income or more degrees than my neighbor or his children, and if my neighbor has similar tastes, then the gains of each come at the other's expense.
I conclude that externality arguments provide little independent support for government schooling. At most they suggest that private schooling ought to receive some subsidy-and even that conclusion is an uncertain one, given both the weaknesses of the arguments for the existence of net positive externalities and the difficulty of separating subsidy from control.
Another argument is that government schooling is necessary because parents, being themselves inadequately educated, are incompetent to choose schooling for their children. As John Stuart Mill put it, "The uncultivated cannot be competent judges of cultivation." This argument concedes that government schools will teach what the state wants children to learn instead of what their parents want them to learn, but views that as an advantage of the government system.
This argument seems to justify at most one generation of government schooling. Once we educate the first generation, they should then be competent to choose an education for their children. The U.S. and Britain have now had universal government schooling for at least five or six generations. If it has done a good job of educating students it should now be unnecessary, and if it has done a bad job perhaps we should try something else.
A further problem with the argument is that most of what the government schools actually teach-or, too often, fail to teach-is well within the comprehension of virtually all parents. Insofar as the main business of the schools is to teach children the basic skills needed to function in our society, the children's parents are usually competent to judge how good a job is being done. Even a parent who cannot read can still tell whether his child can. And, while a few educational issues may go beyond the parents' competence to judge, parents qua parents, like parents qua taxpayers, have the option of making use of other people's expert opinion. The crucial difference between the two roles is that a parent deciding what school his child shall go to has a far stronger incentive to form as accurate an opinion as possible than does a parent deciding how to vote.
Parents have one other advantage over educational administrators-a flood of detailed free information. By observing their children, and by listening to them, parents can learn a great deal about how well they are being schooled. As West put it, describing the situation in England in the 19th century, "Parents were their own inspectors and, compared with official ones, they were not only much more numerous but exercised continuous rather than periodic check."
Parental preferences have often clashed with "expert educational opinion," but it has not always been the parents who turned out to be in the wrong. Thus in Scotland, around 1800, parents "Increasingly resisted traditional parochial school emphasis on classical languages and Religion ... . Parents complained that their children did not get their due in the school `By not having been teached writing.' " Modern examples might include the controversies associated with the shift away from phonics and towards the look-see approach to teaching literacy and the introduction of the "new math" somewhat later-both arguably among the causes of the massive decline in the output of the American school system from 1960 to 1980. Parents have to live with the results of educational experiments; the educators can always go on to a new generation of experimental subjects. As Adam Smith put it:
"Were there no public institutions for education, no system, no science would be taught for which there was not some demand; or which the circumstances of the times did not render it either necessary, or convenient, or at least fashionable, to learn. A private teacher could never find his account in teaching, either an exploded and antiquated system of a science acknowledged to be useful, or a science universally believed to be a mere useless and pedantic heap of sophistry and nonsense. Such systems, such sciences, can subsist no where, but in those incorporated societies for education whose prosperity and revenue are in a great measure independent of their reputation, and altogether independent of their industry."
In a striking passage, E.G. West hints that much of the support for teaching children what they ought to know instead of what their parents want them to know, in the past and presumably today, depends on each expert assuming that it is his version of what children ought to know that will win out:
"The French Physiocrats wanted a national system of education because they could use it to propagate their new found knowledge of the `secrets' of the workings of the economy. ...For the nineteenth century cleric, the `ignorance' which led to crime was primarily the ignorance of the teaching of his particular church. For the utilitarian the crucial issue was ignorance of the laws of the state or in other words the want of knowledge and effective warning of the pain that would inevitably follow from certain actions. For Malthus it was the ignorance of his population principle which mattered most. Public education for him was needed to suppress the `sophistries' of persons such as Condorcet. The latter happened to be the successful instigator of French state education, and undoubtedly intended it to instruct according to his conception of truth." West (1975) p. 123.
The special problems of investing in human capital are sometimes offered as an argument for government intervention in schooling. If I wish to borrow money to pay for a profitable investment in building a factory, I can offer the factory as collateral. If I wish to make a profitable investment in my own education, I have no similar option. Under the present legal rules of the U.S. and most advanced countries, I can acquire the education and then wipe out the debt by declaring bankruptcy. So profitable investments in human capital may fail to be made if the human in question cannot finance them himself.
How important this argument is depends on whether the unit of analysis is the individual or the family. If it is the family, then the argument applies to only a small fraction of the population. Most families can pay the cost of schooling their children out of current income. Indeed, most families do pay the cost of schooling their children out of current income-in the form of taxes to support government schools. Such expenditures might be harder for those with large families and low incomes than they are now, and easier for those with small families and high incomes. On the other hand, there is evidence that private schools provide a given level of education at a considerably lower cost than government schools. If so, most parents would face a lower burden under a completely private system. The market failure argument would then apply only to a small fraction of families at the bottom of the income distribution.
So far as that part of the population is concerned, several points are worth noting. The first is that it makes very little sense to construct a government school system for everyone in order to subsidize investments in human capital for a few percent of the population. The second is that the present system does a very bad job of educating just those people who would have the hardest time educating themselves, which casts some doubt on the idea that it is, for them, an improvement on a purely private system. The third is that the evidence of the nineteenth century suggests that even quite poor people are able to provide their children at least a minimal education. British workers of the early nineteenth century were very much poorer than the inhabitants of America's inner cities at present. Yet the evidence suggests that most were able, without government help, to buy enough education for their children to provide at least minimal literacy-more than many inner city children get now.
Additional issues arise if we consider the problem from the standpoint of the child rather than the family. Most families can afford to pay for schooling their children, but very few children can afford to pay for schooling themselves. A private system depends, for almost all children, on parents caring enough about the welfare of their children to be willing to pay the cost of their education.
Most parents, in most societies, do care for the welfare of their children. In part this may be explained by altruism, itself explainable on evolutionary grounds, and in part by the desire of parents to have children capable of supporting them in their old age. These incentives are not perfect-there are parents who sacrifice the welfare of their children to their own welfare. But the alternative to allowing parents to make decisions for their children is not, as a general rule, having the decisions made by the children-five year olds lack not only income, but information and political power as well. The alternative to having a child's parents make decisions for him is having other adults-school administrators, politicians, voters-make those decisions. Parents may not always be altruistic towards their children, but a child's parents are, of all adults, the ones most likely to be. The argument against letting the parents make the decision is an even stronger argument against letting anyone else make it instead.
Here again, the empirical evidence is striking. Under circumstances of poverty difficult for most of us to imagine, British parents of the early 19th century managed to send almost all of their children to school-not for as long as our children go to school, but for long enough to acquire at least minimal skills. In this country a century later, immigrant parents routinely sacrificed themselves to promote the education of their children. We have yet to see any similar level of altruism on the part of those who control the government schools-say a teacher strike aimed at lowering teacher wages in order to leave more money to pay for books.
A final, and powerful, argument against an entirely private system of schooling is that it promotes and perpetuates inequality. Wealthier parents will spend more on their children, making those children in turn better educated, more successful, and wealthier. This effect is increased by the fact that family background is itself a strong predictor of school performance, even with equal levels of expenditure. In order to give a child from a poor and badly educated family as a good an education as a child from a rich and well educated family, it would, on average, be necessary to spend substantially more on the former.
There are at least two possible replies to that argument. One is that our objective ought to be education, not equality. If shifting to an entirely private system improves the education of the bottom half of the income distribution a little and the education of the top half a lot, both groups are better off. Pursuing that line of argument would take me farther afield than I intend to go in this essay.
A second reply is that while a completely private system would indeed result in unequal educational accomplishment, so does our present government system-and it is far from obvious which leads to more inequality. At present, the quality of government schools varies enormously and non-randomly from place to place. One reason is that high income suburbs, on average, can and do spend more on their schools than low income inner cities, although in the U.S. this difference has probably decreased in recent years as a result of legal pressures. A second reason is that the children of affluent and well educated parents are, on the whole, easier to educate and to be educated with than the children of the inner city poor. A third may well be that affluent suburbanites are better than the inner city poor at getting political institutions to act in their interest.
The first two effects would still exist in an entirely private system, but several factors might reduce the inequality they now produce. A private system would be less rigidly geographical than the present government system. Poor parents with bright children who were willing to sacrifice for them, as many have been in the past, would have the option of sending them to better schools instead of being limited to the school district where they happened to live. Such arrangements are technically possible in a government system as well, and occasionally permitted, but not often-perhaps because they transfer power from the schooling bureaucracy to parents.
Another advantage of the private system, from the standpoint of poor parents, is that parents could control what they got without having to acquire political power-which poor people, as a rule, have very little of. Subject to the limits of their income, poor people have the same economic power as anyone else-the ability to choose whom they buy from.
A final advantage is that a private system might actually provide poor children with some education. Under our present system, the largest determinant of educational output is family background. One explanation of that is that parents are a major part of their children's environment and thus a major source of their education. But a second explanation may be that our schools do not do a very good job of teaching, making children more dependant than they need be on the education they get from those around them. If so, poor children, who are in more need than rich children of things they cannot get from their parents, might well benefit more from a general improvement in the schools.
In many areas of human activity there are two histories-the popular history, mostly mythological, and the real history. In education, quite a lot of the real history has been provided by E. G. West. In examining the history of the rise of government schooling in Britain and the U.S., he has established several important points which go far to refute the popular idea that mass education can exist only through the intervention of the state. They are:
1. Schooling expenditure in Britain represented about the same fraction of national income prior to government intervention and compulsory schooling laws as it did after both were introduced.
2. Prior to government involvement, almost all children were going to school. The opposite claim, widely made in Britain by the supporters of government involvement, was based on fairly simple statistical errors. The most common was to calculate how many children should be in school by picking an arbitrary and unrealistic number of years of schooling and using it to calculate how many children would be in school if all children went to school for that number of years. The ratio of the number of children actually in school to the calculated number was then treated as if it was the fraction of children who went to school. In practice, as West shows, more direct evidence suggests that almost all children in the period just before the beginning of government involvement (c. 1830) went to school for at least a few years. The discrepancy between actual and calculated attendance mainly reflected actual school attendance for fewer years than assumed in the calculation.
A particularly striking example of this fallacy was an unfavorable comparison of the British private system to the Prussian state system, made by the Manchester Statistical Society in 1834. The authors assumed that British students attended school for ten years, used that assumption to calculate that just under two thirds of the children in Manchester attended school, and contrasted that to the (claimed) hundred percent attendance rate of the Prussian system. The Prussian system, however, provided for only seven years of schooling-so even if the claim that every child got the full seven years was true, the average years of schooling per child were about the same in the two systems (7 in Prussia, about 6.5 in Manchester). The Statistical Society offered no evidence that the British number represented two thirds of the students attending school for ten years each, and later evidence made it clear that it did not. The actual number who never attended school seems, from slightly later studies, to have been between one and three percent.
3. Attempts to measure educational output in the form of literacy, using both a variety of studies at particular times and a crude measure (percentage of grooms who signed their names when they got married) that is available over a long time period, show no significant effect of government intervention. So far as one can tell by the (very imperfect) evidence, literacy was already rising rapidly prior to the beginning of government subsidy. Most of the measured increase in literacy had already occurred by the time a nationwide system of government schools and compulsory attendance was established.
4. The eventual expansion of the government school system was in large part the result of efforts by the people running it, plausibly explained by their own self-interest. Its main effect was to replace, not to supplement, the pre-existing private system.
I have been considering two alternatives-government and private schooling. Another alternative, in some ways intermediate between the two, is for the state to provide a fixed amount per pupil per year, which may be used to buy schooling from any of a variety of private providers. How well does such a system deal with the problems we have discussed?
A voucher system solves some of the problems associated with market failure on the human capital market. Families that are too poor to pay to send their children to school will be able to use the voucher to pay for schooling. Parents who do not care enough for their children to be willing to pay for their schooling will be able to use vouchers to provide schooling for their children at no cost to themselves. A voucher system might also reduce educational inequality, relative to both government and private systems. It would not, however, eliminate inequality, both because parents would be free to supplement the voucher and because the parents themselves are a major input to the child's education.
It is not clear whether a voucher system solves any of the other problems raised by a purely private system. It is, for example, a poor tool for solving inefficiencies associated with positive externalities-supposing that one believes such externalities exist and are substantial. With a voucher, the cost to the parents of an additional dollar of schooling is zero up to the amount of the voucher and one dollar above it. That means that parents who would in any case spend more than the voucher will buy the same amount of schooling with a voucher as with private schooling. If there are substantial net positive externalities, that amount will be inefficiently low. Parents who would have spent less than the amount of the voucher will now spend the full amount-which might buy more or less than the efficient amount of schooling. A better way of dealing with such externalities would be for the state to pay a percentage of school expenses corresponding to the percentage of net benefits that went to people other than the student and his family. A voucher makes sense, from this standpoint, only if the optimal educational expenditure is known and is about the same for all families-which seems implausible.
While vouchers give the wrong pattern of incentives for solving the externality problem, government schools do still worse. A parent who wishes to give his child a thousand dollar education when the government schools are spending only nine hundred dollars per pupil must pay the full cost of sending his child to a private school: a thousand dollar cost for a hundred dollars of additional schooling. If the additional schooling is worth less than the additional cost, the parent leaves his child in the public school, where he gets less schooling than his parent would have bought for him in an entirely private system. So a government system might result in less expenditure on schooling than a completely private system-making the inefficiency associated with the failure to allow for positive externalities worse rather than better.
Whether a voucher answers the arguments of those who believe that parents are incompetent to control their children's schooling, either because they have the wrong objectives or because they have the right objectives but not enough knowledge to achieve them, depends on how much control the state exercises over schools that accept vouchers. This suggests an important disadvantage of vouchers. If the government is paying the piper, it may well choose to call the tune. If it is giving vouchers to pay for education, it will probably want to determine what counts as education. Thus a voucher system, like a government school system, has the potential to be used either to encourage indoctrination or to redirect "educational expenditure" to benefit politically well organized groups such as school teachers and administrators. While one could design a voucher system to minimize such problems, perhaps by permitting schools to qualify if the mean performance of their students on objective exams matched the mean performance of students at government run schools, it is far from clear that such a system could be either passed or maintained.
A second argument against vouchers is that they may encourage wasteful expenditure on schooling. During the two decades when the performance of U.S. schools, measured by objective exams, plumetted, real expenditure per pupil roughly doubled. Under a voucher system, interest groups selling inputs to schooling-textbook publishers, teacher's unions, and the like -have an incentive to lobby to raise the amount of the voucher above the optimal level of school expenditure. While their ability to divert such expenditures to themselves will be limited by quality competition among the schools, an increase in demand for their product will still tend to raise its price.
There are arguments in favor of having government pay for and produce schooling, as there are arguments in favor of having government pay for and produce practically any good or service. I have tried to show that the arguments in the case of schooling are not very strong. There are also arguments against having government produce and pay for any good-the arguments for political failure combined with the general economic argument that private markets tend, at least in some approximation, to produce the optimal output at the minimal cost. In the case of schooling, there are additional and very powerful arguments against government control. One of the most important is its potential use to indoctrinate the population in views that the government, or the schooling bureaucracy, or powerful lobbying groups, wish people to hold.
In this regard, one of the great disadvantages of government schooling is its uniformity. Any education can be viewed as indoctrination from the standpoint of those who do not believe what is being taught. Under a private system, however, there is no single orthodoxy. Different children are taught different things, reflecting the differing preferences of their parents and, to a lesser degree, the beliefs of teachers, textbook authors, and other contributers to the educational process. As adults, the graduates of such schools have the opportunity to correct the deficiencies in their education by interacting with the graduates of other schools who have been taught very different things. Under a government system, there is a serious risk that one official orthodoxy will be taught to all.
A further disadvantage to state education, especially in a diverse society, is that it inevitably involves a state religion. One cannot educate children without talking about issues on which religions differ. The pretence of a religiously neutral education, at least in the U.S., is maintained mainly by the tendency of teachers, like other people, to regard what they believe in as fact and only what other people believe in as religion. A government school system in a diverse society is thus deeply divisive, since it means that some people's children are being indoctrinated with other people's religion.
Many of the disadvantages of government schooling could be eliminated, or at least reduced, by a voucher system. While such a system would be a great improvement over government schooling, there seems little reason to believe that it would be superior to an entirely private system. The great argument against it is that a voucher system must include some definition of what is or is not schooling, in order to determine what can be paid for with the voucher. Imposing such a definition on private schools implies the same sorts of problems of government control that would arise with a government school system, although possibly to a much reduced degree.
I conclude that Adam Smith was correct in his suggestion. Whether or not it is proper to have a government system of schooling, it is prudent not to.
July 7, 1993
This paper was delivered at a regional meeting of the Mont Pelerin Society and later published in Liberty.