[Reprinted by permission of Kluwer Academic Publishers. This text is based on the version on my hard disk, so may vary in detail from the published chapter.]
In our society, probably in any society, goods are produced and allocated in several different ways. One very common pattern is production for and sale on the market. Butter, computers, and plumbers' services, to take three examples out of a multitude, are produced by individuals and firms, acting in what they perceive to be their own interest, and sold, usually for money, to those who wish to consume them. While this is a common way in which goods are produced and allocated in our society, it is not the only way, nor is it even clear that it represents a larger part of the total economy than alternative ways.
What are the alternatives? The two most important are household production--the way in which children are reared, homes cleaned, clothes washed, and most meals cooked--and political production. While household production represents a substantial fraction of the economy, and perhaps even of total medical services (parents serving as nurses for their sick children, grown children taking care of aging parents, and the like), its role is not at the moment a subject of much controversy and will be ignored in this essay. The two alternatives I will be concerned with are production and allocation on the market and production and allocation by government. The essential question I will try to answer is whether one form of production should be preferred, and if so which.
The meaning of market provision of medical services is fairly clear, although the form may vary; individual physicians, group practices, private hospitals, payment by individuals or by insurance companies, are all possible market arrangements. The meaning of governmental provision is a little less clear. For the purposes of this essay I include three different sorts of government intervention under the general category of government provision and allocation--production, payment, and regulation. An obvious example of governmental production of medical services is a state hospital, or, on a larger scale, the British National Health System. Government payment would include systems such as medicare, where the service is produced on the market but paid for, in whole or in part, by government. Regulation includes such widely accepted activities as licensing of physicians and control by the FDA over the introduction of new drugs.
While it is convenient to speak of government as if it were an independent being, it is also highly misleading. As should become clear in Part II of this chapter, I regard government not as some outside force accomplishing its own good or evil objectives but as a shorthand description for a mechanism--a set of rules--under which individuals interact in order to achieve individual objectives. From this standpoint, calling the alternatives "market" and "government" is itself somewhat misleading, since government can be viewed as merely a second and different market, a political market in which exchanges occur and decisions are made under a different set of rules than in the economic market. The question then is whether it is better for medical services to be produced entirely in the private market--what I mean by medicine being a commodity--or entirely in the public market, or by some combination of the two.
In trying to answer this question I will start, in Part I, by discussing the relation between economic analysis and normative conclusions. Doing so will involve an unavoidably lengthy discussion of what economists mean by efficiency and what, if anything, it has to do with conclusions about what should happen. I will go on, in Part II, to discuss what economics can tell us about the relative efficiency of the economic and political market. In Part III I discuss, first, some general criticisms of the economic approach, and then several arguments that suggest that medical care is, in some fundamental way, a different sort of good, so that the ordinary economic arguments used to discuss how butter or automobiles can best be produced are irrelevant to the production of medical care. In Part IV I discuss arguments that treat medical care as an ordinary good but claim that it, like some other goods, has special characteristics that justify producing it partly or entirely through the political market. In Part V I will attempt to summarize my conclusions.
The purpose of this essay is to discuss, from the viewpoint of an economist, the arguments against use--or at least, against exclusive use--of the market in health care. This section deals with the relation between the philosophical question of what is desirable and the economic question of what is efficient. Part II will be concerned with the economic and philosophical question of whether we have any basis for judging whether the efficiency of particular market arrangements is likely to be increased or decreased by governmental interventions.
Economic efficiency is not a self-evident goal; most people who clearly understand what it means would agree that there are some circumstances in which an inefficient outcome is preferable to an efficient one. Arguments based on efficiency therefor depend on further moral arguments. I will spend much of this paper--almost all of Parts II and IV-- discussing the case for and against the market in terms of economic efficiency; the purpose of this part of the essay is to explain why.
Before doing so it seems only fair to describe, at least briefly, my own position on moral philosophy, in order to help the reader make proper allowances for the biases of the author. This is particularly important since, for reasons shortly to be explained, my views may not be obvious from the arguments I will be using.
The position in moral philosophy that I find least unsatisfactory is that there exist natural rights, that they can be described in terms of entitlements, and that to be entitled to something is not the same thing as to deserve it. Seen from this position, entitlements may not be a complete statement of what I ought to do, but they are a complete statement of my claims against others and theirs against me--what each of us may properly compel the other to do. My position is in this regard similar to that defended by Robert Nozick in Anarchy, State and Utopia.
The rules of original entitlement and transfer that I find plausible correspond fairly closely to the laws of a pure free market society. In the course of this article I will argue that those rules have other attractive features--that the same institutions I consider morally attractive also approximate a utilitarian optimum more nearly than any alternative I know. One should perhaps be suspicious of such a convenient coincidence. One explanation of it is that I am, consciously or unconsciously, distorting and selecting economic arguments in order to justify as efficient institutions I support for other reasons. A second is that I am, consciously or unconsciously, distorting my moral philosophy to justify institutions that I support because they are efficient. A third--and the most interesting--is that the "coincidence" reflects some underlying connection between natural rights and utilitarian arguments.
The conventional definition of economic efficiency, due to Pareto, is that a Pareto-improvement is a change that benefits someone and injures no one and a situation is efficient if it cannot be Pareto-improved. The rule for judgment that this seems to suggest--reject all inefficient outcomes--is less value-free than it at first appears. I therefore prefer to use a slightly different approach, due to Marshall. I define an improvement as a change such that the total benefit to the gainers, measured by the sum of the numbers of dollars each would, if necessary, pay for the benefit, is larger than the total loss to the losers, similarly measured. I call this a Marshall improvement. I define a situation as efficient if it cannot be Marshall-improved.
While my definition of an improvement is not equivalent to the conventional definition, my definition of efficient is; a situation that is Pareto efficient is also Marshall efficient, and vice versa.[1] To see why, we must look a little more carefully at the word "can" in the definition of efficient; what does it mean to say that a situation "can be improved?"
A determinist would argue that nothing can be except what is; in this sense whatever occurs must be efficient.[2] More generally, how strong or weak a requirement efficiency is depends on how wide a range of alternatives we think of as possible. In practice, economists have defined efficiency using "can be improved" to mean "could be improved by a central planner who had complete economic knowledge and complete control over individual behavior." This represents a sort of outer bound on the outcomes that could actually be produced by varying economic arrangements. It is hard to see how any economic institutions could produce outcomes that could not be produced by such a "bureaucrat god" but easy to imagine that the bureaucrat god might be able to produce outcomes that could not be produced by any real world economic institutions--since we do not have gods available to run our economy. So if an arrangement is efficient there is no institutional change that can improve it; if it is inefficient there may be one.
This very broad definition of "can be improved" is the reason why the best available institutions may be inefficient. It is also the reason why Pareto-efficient and Marshall-efficient are equivalent. The proof goes as follows:
Any Pareto improvement is a Marshall improvement; since there is at least one gainer and no losers, gains must be larger than losses. Hence a situation that can be Pareto improved can also be Marshall improved. Hence a situation that is Marshall efficient (cannot be Marshall improved) is also Pareto efficient.
Suppose there were a situation that was Pareto efficient (could not be Pareto improved) but not Marshall efficient. There would then be a possible Marshall improvement--a change that would benefit the gainers by more, measured in dollars, than it would injure the losers. A bureaucrat god could make that change and simultaneously transfer from gainers to losers a sum larger than the losses and less than the gains, taxing each gainer an amount less than his gain and giving each loser an amount greater than his loss. The combination of the Marshall improvement plus the transfer would be a Pareto improvement. But a situation that can be Pareto improved is not Pareto efficient--which contradicts the original assumption. Hence any situation that is Pareto efficient is also Marshall efficient. Hence the two definitions of efficiency are equivalent.[3]
If the two definitions of efficiency are equivalent, why have I bothered to introduce and use the unconventional one? Because the conventional definition, as it is commonly used, is a way of making interpersonal utility comparisons while pretending not to; Marshall's approach makes the same comparisons but is honest about what it is doing.
What justification can there be for making interpersonal utility comparisons in a way that, by comparing gains and losses as measured in dollars, implicitly assumes that the utility of a dollar is the same to everyone? Marshall's answer was that for most economic questions it does not much matter how you weight utilities. Most issues involve large and diverse groups of gainers and losers; unless there is some systematic tendency for one group to have a higher marginal utility of income than another, the differences among individuals average out, so if the utility gain to the gainers is larger than the loss to the losers when measured in dollars, it is probably also larger measured in utiles. Utility measured in dollars is observable, since we can observe how much people are willing to pay to achieve their objectives; utility measured in utiles is not. We use the former as a proxy for the latter. Seen in this way, judgments of efficiency are really approximate judgments about utility. "O1 is more efficient that O2" means "going from O2 to O1 is a Marshall improvement" means "utility is (probably) higher in O1 than in O2." I will make a careful distinction between efficiency and utility only in those special cases where there is a reason to expect the approximation to be a bad one.
My justification for discussing the case for or against the market in medical care in terms of economic efficiency rests on three propositions. The first is that more is known about economics than about moral philosophy, so we are more likely to reach true conclusions and be able to convince others of them through the former than through the latter. The second is that most real world moral philosophies overlap considerably, with utilitarian considerations an important part of the overlap. While only utilitarians will claim that utility is all that matters, and even utilitarians will not claim that efficiency as defined by economists is all that matters, most people believe that utility, and efficiency as an approximation thereof, are among the things that matter. The third proposition is that the difference in efficiency among different institutions is large--large enough to affect the conclusions, with regard to those institutions, of many who are not utilitarians. I suspect, although I cannot prove, that market arrangements are so much superior to any workable alternative that most people, including most non-utilitarians, would prefer their consequences to those of any alternative.[4] If I am right, then political disagreement is fundamentally a disagreement about the economic question of what consequences different institutions produce, not the ethical question of what consequences we prefer. This is in part an economic opinion about the efficiency of the market, and in part an economic opinion about the range of workable alternatives.
I believe that utilitarian comparisons among the outcomes of different institutions are neither irrelevant nor conclusive; I view the economic concept of efficiency as the most practical way of making such comparisons. I will therefore try to use economics to analyze the arguments against market health care, mostly in terms of economic efficiency.
Economics is not limited to arguments about economic efficiency; indeed, some economists would argue that such "welfare economics" is too much involved with issues of value to be a real part of the science. From my viewpoint, efficiency is simply one useful way of summing up positive results--results about the consequences of particular institutions. It provides a convenient compromise between the questions people ask economists and the questions economists are equipped to answer. It is not, however, the only thing economics can say about the outcome of institutions, hence not the only contribution that economics can make to disputes as to what institutions are best. I shall, to take one example, feel free to use economic arguments not only to discuss whether government interference with the market has a cost in decreased efficiency, but also whether it produces any benefit in increased equality.
Economic efficiency is a strong requirement for the outcome of any real world system of institutions, since an outcome is efficient only if it could not be improved by a bureaucrat god--a benevolent despot with perfect information and unlimited power over individual actions. While it may be seen as an upper bound on how well an economic system can work, one might think that using that bound to judge real systems is as appropriate as judging race cars by their ability to achieve their upper bound--the speed of light.
Surprisingly enough, it is possible to prove that a market system that meets certain assumptions is efficient in this strong sense. There are several sets of assumptions that will do; two of them may be roughly stated as follows:
I. Perfect Knowledge: Individual producers know the cost of all alternative ways of producing and the market price for what they produce; individual consumers know the price and the value to them of goods they consume.
II. Private Property: Property rights are defined and costlessly enforced and can be costlessly transferred for all scarce goods.
Either
III. No Transaction Costs: Any transaction mutually advantageous to the parties involved can be arranged at no cost.
Or
IV. Perfect Competition: Every producer and consumer is so small a part of the market that the quantity he produces or consumes does not affect the market.
V. Private Goods: Every producer can control the use of the goods he produces.
These assumptions are chosen to eliminate market failures traditionally associated with the terms public goods, externalities, imperfect competition, and imperfect or asymmetric information. Assumption III substitutes for assumptions IV and V because of the Coase Theorem; under conditions of zero transaction costs all of the inefficiencies associated with imperfect competition, externalities, and public goods can be eliminated by appropriate bargains among the affected parties.[5]
Of course, no real world economy satisfies either assumptions I, II, and III or I, II, IV and V exactly. This fact is frequently used to advocate government intervention in the market, on the grounds that it can improve the inefficient outcome due to one or another sort of market failure. The difficulty with such arguments is that there is no adequate theory of government behavior that implies that government would choose to do the right things--that its intervention would make things better rather than worse.
It is one thing to show that there is something government could do that would improve on the outcome of the unregulated market; it is an entirely different, and much more difficult, thing to show that what government would do, given the power, would improve on that outcome. That would require a theory of governmental behavior comparable in power and precision to the theory of market behavior from which the original efficiency theorem, and the inefficiencies due to failures of its assumptions, were derived. No widely accepted theory of that sort exists, and much of the large and growing literature that attempts to produce such a theory seems to suggest that government intervention is more likely to worsen than to improve market outcomes.[6]
Even if one drops the traditional argument that every deviation of the market outcome from perfect efficiency justifies a countervailing intervention by the state, one can still argue that since real markets do not meet the requirements of the efficiency theorem we can expect them to be inefficient and do not know by how much, that since there is no efficiency theorem for the political system we have no way of knowing how inefficient it is, hence we have no way of guessing whether political intervention in the market system will lead to more or less efficiency. A similar argument on a grander scale suggests that since we have an efficiency theorem for capitalist economies, whose assumptions real capitalist economies do not meet, and no efficiency theorem at all for socialist economies (or other alternative systems), we again have no basis for predicting which is superior. By defining a perfect outcome that we cannot achieve, we seem to have made it impossible to choose among imperfect outcomes. The best appears to be the enemy of the good.
This argument is, I believe, wrong for at least three reasons. The first is that, although we do not have an economic theory of the political process as well worked out and broadly accepted as the theory of private markets, we do have enough of such a theory to have some idea of where and why the political market is likely to produce less efficient outcomes than the private market. Second, if the situation really were that we had no theoretical basis at all for predicting how efficient the alternatives to the market would be, we would have some reason to expect the market to be (imperfectly) efficient, no reason to expect its alternatives to be, and thus a (rebuttable) presumption that the market is more efficient than its alternatives. Third, whatever the theoretical situation may be, there exists a large and growing body of empirical studies of the effects of government regulation, most of which cast serious doubt on the idea that intervention in the market is likely to lead to improvement.
Conventional economics provides us with an analysis of the economic market, including some understanding of why and to what extent it fails to be efficient. In order to compare the outcomes to be expected from the economic market with those to be expected from the political market, we need some way of analyzing the latter. There exists a body of economics called public choice theory that attempts to analyze the political system by using the same approach with which ordinary economics analyzes the private market. While no single version of public choice theory is as well worked out and widely accepted as is the conventional price theory used to analyze ordinary private markets, public choice theory as it now exists provides economists with some limited ability to predict where and why political alternatives to the market will or will not be efficient. We do not have an adequate theory, but we have something more than no theory at all.
Public choice theory is simply economics applied to a market with peculiar property rights. Just as in the economic analysis of an ordinary market, individuals are assumed to rationally pursue their separate objectives; just as in that analysis, one may first make and later drop simplifying assumptions such as perfect information or zero transaction costs. But the property rights on the public market include the right of individuals to vote for representatives, of representatives, acting through the appropriate procedures, to make laws, of various government officials to enforce the laws, of judges to interpret them, and so on.
Ordinary economics is greatly simplified if we treat firms as if they were imaginary individuals trying to maximize their profits; in this way we reduce General Motors from several hundred thousand individuals to one. There is some cost to the simplification, since it ignores the conflicts of interest within the firm among managers, employees, and stockholders. So far no alternative simplification seems to work as well, so economists continue to analyze an economy of profit maximizing firms, except when the particular problem being considered hinges on intra-firm interactions--as it does, most obviously, in the theory of the firm.
One of the ways in which different public choice theories differ is in what they take to be the equivalent of the firm on the political market, and what it is assumed to maximize. Downs, one of the founders of public choice, took his unit to be the political party.[7] Niskanen took it to be the individual government bureau.[8] Other analyses take interest groups or politicians as the "firms," or eliminate firms entirely and consider individual voters. For the purposes of the following brief sketch, I will consider the firms on the political market to be elected politicians (or, equivalently, the political organizations of which they are a part), and limit my discussion to the market for legislation. This is a simplification of the political market, and only one of several possible simplifications, but it provides a convenient way of sketching the theory.
Consider, then, the market for legislation. Individuals perceive that they will be benefitted or harmed by various laws. They therefor offer payments to politicians for supporting some laws and opposing others. The payments may take the form of promises to vote, of cash payments to be used to finance future election campaigns, or of (concealed) contributions to the politician's income.[9] The politician is seeking to maximize his long run income (plus non-pecuniary benefits, one of which may be "national welfare"), subject to the constraint that he can only sell his support for as long as he can keep getting reelected.
Is the outcome of this market efficient? That depends on additional assumptions. If all transaction costs are zero, the answer is yes. As long as there is any potential change in legislation--any law to be passed or repealed--that confers net benefits, there is some possible agreement that will produce the change while benefitting all those whose cooperation is required. The Coase theorem applies to political as well as private markets.
Even with transaction costs, we would still expect efficient outcomes as long as the amount individuals or groups are willing to pay for any piece of legislation is proportional to the benefit they receive from it, with the constant of proportionality the same for all. In that case, any law whose passage conferred net benefits would be profitable to pass--more would be offered to support it (by those benefitted) than to oppose it (by those injured).
Alternatively, we might expect an efficient outcome if we assume that all voters are perfectly informed. In that case campaign contributions are useless, since no voter can be persuaded to vote against his own interest, and we have civics class democracy, with the candidate who best represents the public interest getting elected.[10]
The important question, however, is not whether the political market works under conditions of zero transaction costs and perfect information; under those assumptions, the private market is also perfectly efficient. The interesting question is how badly each system breaks down when the assumptions are relaxed.
Consider a political market with realistic transaction costs. A legislator proposes a bill that inefficiently transfers income from one interest group to another; it imposes costs of $10 each on a thousand individuals and gives benefits of $500 each to ten individuals. What will be bid for and against the law?
The total cost to the losers is $10,000, but the amount they will be willing to offer to a politician to oppose the law is very much less than that. Why? Because of the public good problem.[11] Any individual who contributes to a campaign fund to defeat the bill is providing a public good for all the members of the group. The same arguments that imply underproduction of public goods by the market apply here. Just as in that case, the larger the public the lower the fraction of the value of the good that can be raised to pay for it.
The benefit provided to the winners is also a public good, but it goes to a much smaller public--ten individuals instead of a thousand. A small public can more easily organize, through conditional contracts ("I will contribute if and only if you do") to fund a public good. Even though the benefit to the small group is smaller than the cost to the large one, the amount the small group is able to offer politicians to support the bill will be more than the amount the large group will offer to oppose it.
The effect is reinforced by a second consideration--information costs. I now drop the assumption of perfect information; instead, I assume that information about the effect of legislation on any individual can be obtained, but only at some cost in time and money. For the individual who suspects that the bill may injure him by $10, it is not worth paying much to obtain the information. Not only is his possible loss small, but the effect on the probability that the bill will pass of any actions he would consider taking is also small. The member of the dispersed interest chooses (rationally) to be worse informed than the member of the concentrated interest.
If we abandon the particular example and consider the extreme case of an American presidential election, the argument becomes even clearer. Suppose the average election is won by 2.5 million votes. We may simplify the analysis by replacing the actual probability distribution for the outcome by a uniform probability distribution from -5 million to +5 million. The probability that the election will be a tie, hence that one additional vote can decide it, is then one in ten million. So the return to an individual who figures out who is the right candidate and votes accordingly, instead of voting at random, is an increase by one chance in ten million of the probability that the right candidate will be elected. Unless the voter has an extraordinarily high value for electing the right candidate, it is not worth paying very much in order to increase the probability of that outcome by one in ten million, so he does not pay the information costs necessary to decide for whom he should vote. This is rational ignorance. It is rational to be ignorant if the cost of information is greater than its value.
So far I have discussed only one characteristic of a group--its size. It is useful to think of the terms "concentrated" and "dispersed" as proxies for the set of characteristics that determine how easily a group can fund a public good for that group; the number of individuals in the group is only one of those characteristics.
Consider, for example, a tariff on automobiles. It benefits hundreds of thousands of people--stockholders in auto companies, auto workers, property owners in Detroit, and so forth. But General Motors, Ford, Chrysler, American Motors, and the UAW are organizations that already exist to serve the interests of large parts of that large group of people. For many purposes one can consider all of the stockholders and most of the workers as "being" five individuals--a group small enough to organize effectively. The beneficiaries of auto tariffs are a much more concentrated interest than a mere count of their numbers would suggest. That may explain why such tariffs exist, even though they are inefficient--the costs they impose on consumers of automobiles and American producers of export goods (both dispersed interests) are larger than the benefits to the producers of automobiles.
The reason the public good problem leads to inefficiency on ordinary private markets is that the amount a group can raise to buy a public good benefitting that group is less than the total value of the good to the members of the group, hence some public goods that are worth more than they cost to produce fail to get produced, which is inefficient. The problem on public markets is that both costs and benefits are only fractionally represented on the market, due to the public good problem--and if the weights are different, as they almost always will be, laws that impose net costs may be passed and laws that impose net benefits may not be. This again is inefficient.
To compare the efficiency of the two systems--the private and the political market--we must specify the degree of publicness of the good and the nature of the publics involved. If, for example, we consider the private production of a pure public good for which the public is the entire population of the country, the political alternative may be an attractive one. The private market will produce the good only if total benefit times the weighting factor relevant to that public (the percent of the benefit that can be raised to pay for the good) is larger than the cost of producing the good. The public market will produce the good only if total benefit times the weighting factor (the same as before) is larger than total cost times a weighting factor representing the fraction of the tax money saved by not producing the good that taxpayers (also a public with a public good problem) can raise to oppose its production. Since the weighting of cost as well as of benefit is (much) less than one, the chance that weighted benefit is greater than weighted cost (so that the good will be produced politically) is greater than the chance that weighted benefit is greater than unweighted cost (so that it will be produced privately). The public good is more likely to be produced publicly than privately.
This is not entirely a good thing. If the weighting factor on cost is smaller than on benefit--if taxpayers are a more dispersed interest than beneficiaries, as they would be if the good benefitted only a small group--then the good may be produced even if cost is greater than benefit. This too is inefficient, although in the opposite direction--producing a good that should not be produced rather than not producing one that should.
In many cases, although perhaps not in the case of defense, the size of the public and the degree of publicness are quite different on the private and political markets. Consider immunization against contagious disease. Immunization is only partly a public good; if I have myself immunized I receive a private benefit (I am safe from the disease) and confer a benefit on others (they cannot catch it from me). If the public and private parts happen to be of equal value, I will choose to get immunized as long as the total benefit is at least twice the total cost. Private immunization is then only "half public."[12]
Suppose we decide to provide immunization publicly instead of privately. This does not mean that we decide to provide it if and only if it is worth providing--systems that automatically generate right answers are not one of the options available to us. It means rather that we have a system in which what immunizations are given to whom--and who pays for them--are decided politically. The decision is determined by a tug of war between two interests--those benefitted and those paying--both large groups whose political efforts are restricted by the public good problem. The groups would have to be very well balanced indeed in order to generate a more efficient outcome than "produce if benefits are at least twice costs"--the outcome of the private system.
This article is an essay on the role of the market in medical care, not a textbook on public choice theory. I have tried to sketch enough of the theory to show the reader why the efficiency of the public market generally breaks down much faster as we weaken our simplifying assumptions (perfect information and zero transaction costs) than does the efficiency of the private market. On the private market, an individual who pays for information gets the benefit of the information; if I spend time and energy deciding what car to buy, my conclusion will determine what car I get. On the public market, gathering information typically involves producing a public good for a large public; if I gather information on which politician to vote for, the result is a miniscule increase in the probability that I (and everyone else) will get that politician and the legislation he supports. Hence we may expect people to be much better informed about their private than their public decisions--a conclusion that seems consistent with casual observation. Imperfect information, while a problem for both markets, is a much more serious problem for the public market.
Similarly with the set of problems associated with public goods and externalities. The typical good produced on the private market is a mostly private good--an office building that provides $50,000,000 worth of benefit to the individuals renting space in it, plus a $1,000,000 benefit spread among all the inhabitants of the city whose skyline it ornaments. If expensive buildings strike you as a bad example, substitute my front yard (mowed) or your neighbor's phonograph (a negative externality if you have sufficiently different tastes in music--or in when you wish to listen to it). Exceptional cases are pure or almost pure public goods (scientific research of a non-patentable nature, the invention of the supermarket) or goods for which external costs are a large fraction of all costs. In the normal case we have something close to an efficient outcome--the building is constructed as long as benefit is at least 1.02 times cost. Inefficient outcomes are unlikely to occur, and if they do occur the inefficiency is small--the failure to produce something that is worth slightly more than it would cost to produce, or the production of something worth slightly less than it costs to produce. Only in the exceptional cases is inefficiency likely to be a serious problem
On the public market, on the other hand, public goods with large publics and goods whose costs or benefits are mostly external are the rule rather than the exception. When I support a bill to benefit me at your expense, all of the cost of what I am getting is external--imposed on you and ignored in my calculations. When a politician tries to pass a good law, he is attempting to produce a public good with a very large public. He would almost certainly be better off, from his own standpoint, using the same tax money to buy the votes (or bribes) of some concentrated interest instead. Even if the politician has a public spirited desire to pass good laws, he will find that unless he is very rich he cannot afford to--just as a capitalist cannot afford to give his customers what he thinks they should have instead of what they are willing to pay for. Unless the politician has large political assets that his competitors lack, his attempt to do good will result in his defeat by someone who follows a more nearly vote maximizing policy.
The conclusion of this analysis seems to be that the public market is preferable to the private one, at most, only in cases of extreme market failure. As long as the private market is anywhere close to efficient, it is probably preferable to the political alternative.
How good are our reasons for believing that the analysis I have sketched accurately describes the real world? There is some evidence, but it is not conclusive. The observed outcomes of the political system--what industries get tariffs, how professions come to be regulated, and the like--appear to fit the patterns suggested by the analysis, but neither the evidence nor the theory is sufficiently good to make us as confident about the economic theory of government as we are about the economic theory of ordinary markets.[13]
This brings me to my third reason for believing that the market is typically more efficient than its alternatives. There now exists a substantial and growing body of studies of the effects of government regulation. The conclusions suggested by many of these studies are that regulation rarely achieves its stated purpose, that it frequently imposes serious costs, and that it frequently serves as an indirect way of transferring income from one group to another.
Consider, for example, Linneman's study of the effects of minimum wage legislation.[14] The minimum wage is often defended as a way of helping low income workers by raising their wages. Linneman obtained, for a large sample of individuals, data on employment, wages, and a variety of characteristics (such as age, education, and race) that might be expected to affect employment. Using data from before and after a large increase in the minimum wage rate, he estimated the effect of the increase on the wage rate and employment of the individuals in his sample. He concluded that the effect had been to lower the earnings of the subminimum population--the workers who would have made less than the new minimum wage--and increase the earnings of high wage and union workers. In effect, the increase priced unskilled labor out of the market, increasing the demand for skilled labor.
Consider, as a second example, Peltzman's study of the effects of the Kefauver amendments to the Pure Food and Drug Act.[15] The stated purpose of the amendments was to force drug companies to demonstrate the effectiveness of "New Chemical Entities" before putting them on the market; the argument was that because of imperfect information, consumers paid unnecessarily to buy, and drug companies to produce, new drugs that were no better than the old drugs. Peltzman showed that the effect of the amendment was to reduce the rate of introduction of NCE's by more than half without any detectable improvement in their average quality; he estimated that the effect was equivalent to imposing a ten percent sales tax on all consumption of drugs.
The point of these examples is not merely to show that the best laid plans of mice, men, and legislators gang aft agley. It is rather to suggest that insofar as we have evidence on the effect of intervention, the evidence is that the intervention imposes net costs, and does so without achieving the objectives--sometimes involving efficiency and sometimes not--for which it purports to exist. It is further to suggest that the observed pattern is more consistent with rational behavior aimed at objectives other than those announced than it is with the other obvious explanation--repeated error. In the case of the minimum wage, for instance, it is worth noting that the congressmen supporting it have generally represented high wage, not low wage, states and industries.
In summary, then, the conclusion of this section of the paper is that both economic theory and empirical studies provide a basis for preferring market to political arrangements except, perhaps, in cases of extreme market failure. The conclusion does not have the rigor of a mathematical theorem--but neither, so far as I know, does any conclusion about the real world, even in the most precise of sciences. In addition to the uncertainty associated with any attempt to apply a theoretical construct to a real-world situation, the conclusion has the additional uncertainty associated with a body of theory--public choice--that is still developing and many of whose elements are still matters of dispute.
In Part I of this paper, I attempted to deal with general philosophical arguments concerning the appropriateness of economic efficiency as a criterion for judging alternative arrangements; in Part II I tried to show how one might judge the relative efficiency of the market and its alternatives. In Part IV I will try to apply that analysis to the question of how medical care should be produced and distributed.
Before doing so, there are several potential objections to this approach that I would like to try to deal with. Some are objections to the economic approach in general, some objections to its application to medical care. There is little point in my subjecting you to a lengthy economic analysis of the efficiency of providing medical care in various ways, without first making some attempt to persuade you that economic analysis is a legitimate way of learning things and that its conclusions are relevant to questions of the provision of medical care.
I will start by discussing three objections to the economic approach that are, I think, widely held. They are, first, that economic analysis depends on the unrealistic assumption of individual rationality, second, that economic efficiency fails to take account of the fact that money, which it uses as its measure of value, is of different value to different people, and third that it ignores the infinite value of life relative to money or material goods, and the fact that medical care, like food, is necessary for life. I will go on to discuss the claim that medical care has some sort of special moral status, making it a "priority good" or implying that people have a "right to medical care."
The central assumption of economics is rationality--that people have objectives and tend to choose the correct way of achieving them. While the assumption can be modified to deal with information costs, individuals are still assumed to make the correct decision, in an uncertain environment, about how much information to buy.
The use of the term "rationality" to describe this central economic assumption is somewhat deceptive, since it suggests that people find the correct way to achieve their objectives by rational analysis--using formal logic to deduce conclusions from assumptions, analyzing evidence, and so forth. No such assumption about how people find the correct means to achieve their ends is necessary.
One can imagine a variety of other explanations for rational behavior. To take a trivial example, most of our objectives require that we eat occasionally, so as not to die of hunger (exception--if my objective is to be fertilizer). Whether or not people have deduced this fact by logical analysis, those who do not choose to eat are not around to have their behavior analyzed by economists. More generally, evolution may produce people (and other animals) who act rationally without knowing why. The same result may be produced by a process of trial and error. If you walk to work every day you may by experiment find the shortest route, even if you do not know enough geometry to calculate it. "Rationality" does not mean a particular way of thinking but a tendency to get the right answer, and it may be the result of many things other than thinking.
Half of the assumption is that people tend to reach correct conclusions; the other half is that people have objectives. In order to do much with economics, one must strengthen this part of the assumption somewhat by assuming that people have reasonably simple objectives. The reason for this additional assumption is that if one has no idea at all about what people's objectives are, it is impossible to make any prediction about what they will do. Any behavior, however peculiar, can be explained by assuming that that behavior was itself the person's objective (why did I stand on my head on the table while holding a burning thousand dollar bill between my toes? Because I wanted to stand on my head on the table while holding a burning thousand dollar bill between my toes).
This element in economic theory may be partly responsible for the idea that economists assume that "all anyone is interested in is money." Put in that way the assertion is wrong; economists usually assume that people desire money only as a means to other objectives. What is true is that although economics can, in principle, take account of the full richness of human objectives, it is necessary for many practical purposes to assume away all save the most obvious--the consumption of goods, leisure, security, and the like.
Economics is based on the assumption that people have reasonably simple objectives and choose the correct means to achieve them. Both halves of the assumption are false; people sometimes have very complicated objectives and they sometimes make mistakes. Why then is the assumption useful?
Suppose we know someone's objective, and also know that half the time he correctly figures out how to achieve it and half the time he acts at random. Since there is usually only one right way of doing things (or perhaps a few) but very many wrong ways, the rational behavior can be predicted but the irrational behavior cannot. If we predict his behavior on the assumption that he is always rational we will be right half the time; if we assume he is irrational we will almost never be right, since we still have to guess which irrational thing he will do. We are better off assuming he is rational and recognizing that we will sometimes be wrong. To put the argument more generally, the tendency to be rational is the consistent and hence predictable element in human behavior. The only alternative to assuming rationality (other than giving up and concluding that human behavior cannot be understood and predicted) would be a theory of irrational behavior, a theory that told us not only that someone would not always do the rational thing but also which particular irrational thing he would do. So far as I know, no satisfactory theory of that sort exists. As I argued in Part II of the essay, it takes a theory to beat a theory; until some better alternative is found, rationality is the best we have.
One possible response to this is that although rationality may give us the best available theory, a theory built on so weak a foundation is not very useful; perhaps we are better off depending on "common sense" answers instead.
This sounds plausible, but when it comes to analyzing a market--a complicated interacting system--"common sense" turns out in practice to mean a poorly thought out, inconsistent, and untested theory. If analyzing such a system were easy, economics would be easy too. Any good economist can provide a collection of horrible examples from casual conversation and his daily newspaper.
A second set of objections to the market and the economic approach is based on the claim that neither gives proper consideration to the implications of income inequality. This involves two different arguments, one of which I in part agree with. The first is that the market--and the criterion of efficiency according to which its outcome is often judged--measures individual values in dollars, not in intensity of feeling; since some people have fewer dollars than others, their desires receive less attention, even if they are as strong as, or stronger than, the desires of the more fortunate. The second argument is that economic analysis, and the economic defense of market outcomes, is put in terms of choice, but that choice is irrelevant to the poor; how can one say that an individual chooses not to have that which he cannot afford?
The first argument can be restated by going back to my original defense of the efficiency criterion as a proxy for the utilitarian objective of maximizing total welfare. I conceded there that the measure of improvement used in defining efficiency (explicitly in the Marshallian approach, implicitly in the more convention Paretian approach) compares costs and benefits to different individuals measured by how much money those individuals would spend, if necessary, to get the benefit or avoid the cost. This is equivalent to doing an interpersonal utility comparison on the assumption that the marginal utility of a dollar is the same for everyone.
I defended that way of defining efficiency, from a utilitarian standpoint,[16] by arguing that the particular rule used to weight utilities did not matter very much. Most of the decisions an economist is interested in affect large and heterogeneous groups of people. Unless there is some reason to expect that a particular weighting rule favors the losers more than the gainers, or vice versa, differences among individuals can be expected to average out when we consider the effect on the whole group.
There are some cases in which we do have good reason to expect that the choice of weighting rule will differently affect the different groups. Most of us believe that people with high incomes have, on average, a lower marginal utility of income than people with low incomes. If so, there will be a systematic discrepancy between efficiency and utility when we are considering decisions that have differential impact on high and low income people. One obvious case is the decision of whether or not to provide free medical care (or housing, or food, or money) to the poor at the expense of the rest of us. If we evaluate the transfer in terms of efficiency, we conclude that what the poor get is worth at most its cost to them (if it were worth more, they would have bought it) while the cost to us of paying for it is at least its cost (more if there is some cost to collecting taxes), so the transfer provides benefits that are at most equal to the cost and almost certainly less. If we evaluate it in terms of utility, we conclude that even if a hundred dollars worth of medical care is worth only ninety dollars to the poor recipient, that may well represent more utility than the hundred and ten dollars that it costs the not-poor taxpayer.
This is the traditional utilitarian argument for redistribution. Three things are worth noting about it. The first is that the assumption that marginal utility of income is correlated inversely with income, while plausible, is not necessarily true; under some circumstances one would expect the opposite. The second is that the argument has no special relevance to medical costs, except to the extent that differing medical bills may be an important cause of inequalities in the marginal utility of income among individuals; it is a general argument for redistribution from the rich to the poor. The third is that while the argument suggests that we should prefer outcomes that are biased towards the poor, everything else being equal, it does not tell us that we should favor political over market processes. That conclusion depends on some further argument to show that the outcome of political processes is likely to modify the market outcome towards equality and not away from it--that the poor can be expected to do better on the political market than on the economic market.
I begin with the first point. The standard argument for redistribution begins with the claim that, for a given individual, the marginal utility of income declines as income increases. This seems plausible in terms of introspection, observation (of behavior under uncertainty) and theory, although one can construct counterexamples. The next step is to claim that, absent information about differences in individual utility functions, we must treat each individual utility function as a random draw from the same population, so declining marginal utility of income applies not only to the same individual with different incomes but (on average) to different individuals with different incomes. It follows that the poor have, on average, a higher marginal utility for income than the rich.
This argument assumes that there is no causal connection running from utility function to income. That is plausible enough if income is determined exogenously--by inheritance, or lottery, or some other chance mechanism. It is quite implausible if differing income is the result of differing effort. An individual who greatly values the things that money buys will be more willing than others to give up other goods, such as leisure, in order to get income, so he will, on average, end up with a higher income. If instead of assuming that different individuals have the same utility function but different incomes we instead assumed identical opportunities but different utility functions, we would expect income and marginal utility of income to be positively correlated. So our opinion about the sign of the relationship will depend very much on our opinion about the origin of observed inequalities in income.
This brings me to my second point. The cost of medical services represents a large, and to a considerable extent random, subtraction from income. Even if we believed that people with low incomes are, on average, poor because they do not value money, rather than that they value money because they are poor, we should still make an exception for those who are poor because of medical expenses. In that case at least the traditional argument seems to hold rigorously. The income difference is the result of a random exogenous force, so differences in utility functions should be unrelated to differences in income net of medical expenses, so those whose consumption of other goods is low because of their medical expenses should have a high marginal utility of income. It follows that the utilitarian case for transferring income to those who are impoverished by high medical expenses is stronger than the conventional utilitarian case for transferring income to the poor.[17]
Is this a conclusive argument, within the utilitarian framework, for political intervention to alter the outcome of market provision of medical services? That depends on the costs of such intervention--if large enough, they might outweigh the gains. This argument is usually put in terms of the costs resulting from the distortion of incentives introduced by both tax and subsidy.
While this is a legitimate argument for limiting the size of income transfers to the poor, whether for medical or other purposes, it gives us no reason to expect that the optimal transfer would be zero. Elsewhere I have made a different argument that does suggest that conclusion--essentially that giving government the power to transfer income sets off an expensive free-for-all, with individuals expending considerable effort to ensure that they will be the beneficiaries of transfers instead of the ones who pay for them.[18]
There is, however, another important problem with political redistribution. It is usually assumed without discussion that having government intervene to redistribute in favor of the poor, while leaving most of the rest of the system unaffected, is one of the available alternatives. This is the same sort of assumption made by those who argue that wherever the market outcome is inefficient, government should step in; in both cases, government is treated as a deus ex machina, introduced to produce whatever outcome we have decided is desirable. But in discussing equality, just as in discussing efficiency, it is not enough to show that there is something government could do that would improve on the outcome of the market. The question is whether what government will do will be an improvement.[19]
Public choice theory does not give any clear answer as to whether government, given the power to redistribute, is likely to increase or decrease inequality, to distribute to the poor or from them. On the one hand, votes are more evenly distributed than income, which should tend to make the political market more egalitarian than the private market. On the other hand, many of the characteristics that give groups and individuals political influence are closely related to income. Education reduces information costs, labor skills differentiate their possessors into (relatively) concentrated interest groups, stockholders have their interest represented by well organized firms and skilled (hence highly paid) workers by well organized unions, and so forth. So far as theory is concerned, it is difficult to predict whether the political system is more likely to transfer money down the income ladder or up.
The evidence is also unclear. There are a variety of well publicized programs to help the poor at the expense of the not poor--food stamps, welfare, and the like. There are also a considerable number of government programs to help the not-poor at (on average) the expense of the poor--state subsidies to higher education being a notable example. Social Security, by a large margin the biggest income transfer program in our society, has ambiguous effects; a poor individual who works and collects for the same number of years as a rich individual gets a somewhat better deal, but poorer individuals, on average, start work earlier and die earlier, hence pay for more years and collect for fewer--which may more than balance the progressive element in the payment and benefit schedules.[20]
The greatest weakness in the (utilitarian) argument for government intervention, in the market for medical care or for anything else, is that it is an argument for an outcome. It simply assumes that government intervention will produce that outcome. If the assumption is correct, if government intervention results in redistribution down the income ladder with relatively small costs, then the argument is a correct one. If government intervention results in large costs, as I have argued elsewhere, or if the direction of the redistribution is ambiguous or even perverse, so that there is no gain in equality to balance the loss in efficiency, then the argument is wrong.
So far I have been discussing the traditional utilitarian argument for government intervention to redistribute to the poor. There is a second, and to my mind less defensible, set of arguments sometimes employed against the market--especially in medical services and other "necessities." This is the claim that poor people cannot really be said to choose poor medical care, or poor nutrition, or whatever, since they cannot afford anything else. Hence, it is argued, such goods should be provided to the poor, whether or not they are able or willing to pay for them.
Insofar as this is simply a repeat of the argument for redistribution from differing marginal utility of income, I have already discussed it. What is disturbing about it is that the argument seems, in practice, to be used to justify programs to provide "necessities" to the poor, whether or not such programs are actually redistributive--and often enough they are not.[21] Such programs, in effect, force the poor to spend their own money on what someone else has decided they need. It seems odd to argue that the poor family cannot afford to pay for adequate medical care (because there will not be enough money left for food?) and should therefore be compelled to pay for adequate medical care (and starve?). Yet this is what a program of governmentally funded medical care implies, unless it also redistributes. And if one is going to redistribute, there seems no obvious reason why the subsidy to the poor should take the form of medical care. Why not give money, and let the poor family decide for itself what it is most important to spend it on? The claim that the rich choose among their wants while the poor "must" buy their "necessities" may be effective rhetoric, but what it actually says is that choice is less relevant to large values than to small ones, which seems peculiar.
One popular and persuasive criticism of both market provision of health care and the economic analysis thereof is the claim that both embody a fundamental error--the failure to realize that life is infinitely more valuable than money. Medical care, it is argued, is not merely a want but a need. Without medical care we may--some certainly will--die, and without life all other values are meaningless. Hence neither willingness nor ability to pay is relevant to who should or should not get medical care; the only proper criterion is who needs it.
This argument embodies several errors. The first, and least important, is the confusion between money as a thing of value and money as a measure of value. While money is convenient in defining economic efficiency, it is not essential--any tradable commodity will do. The real comparison is not between life and money but between life and other things people value--leisure, consumption goods, education for their children, housing, et multa caetera.
But the question still remains: Is not life, and hence medical care as (sometimes) a necessity for life, infinitely more important than other values?
Not if we accept the behavior of ourselves and others as evidence about our values. Anyone who both smokes and believes the conclusions of the surgeon general's report is deliberately trading life--an increased probability of dying of heart disease or lung cancer--for the pleasure of smoking. Anyone who spends time and money on anything that does not increase his life expectancy while there remains some unexploited opportunity for an expenditure that does--an extra visit to the doctor, a marginal improvement in nutrition, a slightly safer car--is demonstrating that life, while it may be valuable, is not infinitely valuable in comparison to other things.
Perhaps we do trade life for lesser values but should not. I know no way to prove what we should do, but the sort of life that would be implied by that prescription does not seem very attractive. Not only does the assumption that life is infinitely valuable imply that we should take no avoidable risks--no sky diving, no skiing, no skin diving--it also implies a society wholly devoted to achieving a single goal. It is all very well to say that we need food, or air, or housing, or medical care, but how much do we need? In the case of medical care, at least, the level of expenditure per capita at which additional care produces additional returns in life expectancy is surely far above our per capita income. If so, a society that treated life as infinitely valuable would have no resources left to spend on anything else--including the things that, for many of us, make life worth living.
Another argument that could be made for the infinite value of life is based on the economist's principle of revealed preference. Few of us would agree to be hung tomorrow in exchange for a payment of a million dollars, or even ten billion. Does not that imply that our value for life is very high and perhaps infinite?
No. It proves that money is of no use to a corpse. What is special about the situation being considered is not the high value of what is being bought but the low value of what is being paid with.
Consider the difference between an offer to buy all of someone's life immediately and offers to buy half of the remaining years, or payments for some specified probability of death. In both cases, many more people would accept than with the first offer--and those who would sell all their life for some price would sell part of it for a much lower price. This is not merely speculation; studies of wage differentials in hazardous professions generate value of life estimates well below a million dollars. The reason for the almost discontinuous change as we move from offering to buy all of a life to offering to buy a fraction of it is not the reduction by a factor of several in the amount of life we are buying but the increase, by a much larger factor, in the amount of life available in which to enjoy the money.[22 ]
While I have very little intellectual sympathy with the claim that life is infinitely more important than other goods, I have a good deal of emotional sympathy with it--it appeals more to my moral intuition than to the ideas with which I try to make sense out of both my moral and economic beliefs. To show why it does so, and why I nonetheless reject it, I will discuss a case in which the claim seems at first very plausible.
Suppose there is some individual who requires--and does not get--a ten million dollar operation to save his life. Further suppose that ten million dollars is precisely the sum spent, during a year, by all the people in the U.S. in order to have mint flavor in their toothpaste. Surely this is a monstrous outcome--a man losing his life in order that others can have the trivial pleasure of mint flavor in their toothpaste.
The conclusion seems unavoidable, but I believe it is wrong. The problem, I think, is a fault in my (and I presume your) moral intuition--our inability to multiply by large numbers. To most of us, a number such as two hundred million has only a vague meaning--we have no intuition for how much the importance of a trivial pleasure is increased when it is multiplied by two hundred million. In contemplating the situation I have described, we end up comparing the value of one life to the value of a trivial pleasure to one person, or perhaps a few. Seen that way, the answer appears obvious.
In trying to test my intuition, I find it useful to remove the other people from the problem and convert numbers into probabilities. Suppose I know that by eliminating mint flavor from my toothpaste I can avoid a one in 200 million chance of my own death. That is, in some sense, an equivalent problem--at least if we specify, in the first situation, that nobody knows whether or not he will be the one in need of medical treatment.
Put this way, the answer is far from obvious; this time it is the tiny probability that my intuition finds it difficult to deal with. In order to help it out, I imagine that I get to make the same choice two hundred times, each time eliminating some minor pleasure. It seems far from clear that giving up two hundred minor pleasures--mint flavored toothpaste, the availability of my favorite flavor of ice cream, lying in bed an extra minute each morning, rereading a favorite book one more time--in order to eliminate a one in a million chance of dying is a good deal.
As a further crutch to my intuition, I try converting one chance in two hundred million of losing all of my life into a certainty of losing one two hundred millionth of my life; while there is no obvious reason why I must be risk neutral with regard to years of life, it seems the natural first approximation. Assuming that I have fifty more years to live, one two hundred millionth of my life is about eight seconds. That is not an obviously exorbitant price for mint flavor in my toothpaste.
A variety of different arguments are used by philosophers to defend the proposition that medical care has a special moral status and that choices involving it cannot be properly made on the same basis as most other choices. I have neither the space nor the competence to explore all of these claims, but I will briefly discuss three--the attempt to derive a right to medical care from a right to life, the claim that medical care involves objective needs and hence that the correct amount of medical care is not an economic issue, and the claim that medical care has a special status because of its connection with the ability to make choices in the future.
It is sometimes claimed that individuals have a self-evident right to life, hence to that necessary to life, hence to medical care. One difficulty with this argument is that it proves too much. Medical care is not the only thing whose consumption affects life expectancy. Nutrition, clothing, housing, education--a very large fraction of all desirable things have some effect on how long one lives. If a right to life means a right to anything that increases life expectancy, then we each have a right to most of what everyone else possesses--since many of those goods, like medical care, continue to increase life expectancy at levels of consumption well beyond their current average.
One possible reply would be that there is a fundamental difference between denying someone a kidney machine, hence killing him with certainty, and denying him the additional amenities that would increase his life expectancy by a few days. I find this argument unconvincing. A kidney machine does not, after all, prevent death--it only postpones it. Nothing, so far as we know, prevents death. It is hard to see any profound difference between giving someone an additional five years with certainty and increasing by ten percent his chance of an additional fifty years.
My own view is that to talk of a right to life in this sense is already a mistake, even before it is translated into a right to medical care. A right to have life is by its nature contradictory. The concept of a right to life makes sense as my right to have other people not kill me. It does not make sense as a blank check against the rest of the human race for anything that extends my life.
A second argument for the special status of medical care holds that economic considerations are appropriate for choices that involve differing tastes, but inappropriate for choices where the right answer is a matter of objective fact. Thus, it could be claimed, one may properly decide who gets a piece of land by who is willing to offer more for it, but one should not decide who wins a lawsuit, or which scientific theory is true, by how much each party is willing to offer the judge. In the first case, the question being decided is one of value; in the second, it is one of truth.
The problem with this argument is that decisions about medical care, like most decisions human beings make, involve issues of both fact and value. When I buy a steak at the grocery store, my decision depends both on what I think the characteristics of the steak are and how I value them. Precisely the same is true of medical care. The expert can provide information about the chance that an operation will save the patient's life, or about possible side effects of a drug. But the information does not by itself compel a conclusion. The patient may care or not care about particular side effects, value his life more or less in comparison to monetary or non-monetary costs of the operation, and so on.
A final argument goes roughly as follows. It is said that there is something peculiarly important--essentially human--about the ability to make choices. One consequence of lack of medical care may be a drastic reduction in the number of available alternatives--a cripple cannot become an athlete, to take a particularly sharp example. Hence, it is said, while poor people may not have any general right to be given money, they do have a right to be provided with medical care. It is claimed that this argument justifies medical vouchers--payments to the poor that can only be used for medical expenses.
There are some obvious problems with this. Just as in the first case discussed, the argument proves too much--many inputs other than medical care affect our future choices. Further, the argument for vouchers appears to contain a simple error of logic--the proposal to increase choice in fact reduces it.
To see why, consider, not the issue of whether poor people with medical needs should be given money, but the question of whether, if they are given money, they should be compelled to spend it on medical care. The claim is that they should. The argument is that our objective is to increase the range of choice available to the recipient, that spending money on an operation does that while spending it on a vacation does not.
There is only one unambiguous sense in which choices can be increased--if the new set of alternatives includes the old, plus at least one additional alternative. In this sense, removing the requirement that the money be spent on medical care obviously increases choice. The individual who must spend the money on medical care is free to choose among alternatives A-M--all the things he can accomplish if healthy. The individual who can choose how to spend the money can also choose A-M, by spending the money on medical care, but has additional choices N-Z. It is hard to see how the former can be said to have a wider range of choice than the latter. Of course, after spending the money, on either a vacation or medical care, the individual has reduced his choices--but that is the usual consequence of choosing.
In ending this part of the essay, I should perhaps apologize to the philosophers among my readers for trespassing on their domain. I will refrain from doing so, on the grounds that many of the arguments propounded by philosophers trespass on territory that properly belongs to economics. Instead, I invite the philosophers to correct my errors in their field, as I attempt to correct theirs in mine.
So far, I have discussed--and tried to establish--the relevance of economics and of economic efficiency to choice in general and to choices associated with medical care in particular. In this part of the essay I will consider a variety of economic arguments that might be used to demonstrate the desirability of governmental involvement in the production and allocation of medical care, discussing the different ways in which market imperfections can be expected to occur in connection with the market for medical care, and the possible interventions in the market that might improve the market outcome. In all cases "imperfection" means "failure to achieve an efficient outcome" and "improve" means "make a change that produces net benefits" in the sense described in Part II.
In each case, the discussion contains four parts. The first is a description of the reason for market failure. The second is a discussion of the (imperfect) market solutions--the ways in which the market, while failing to achieve the efficient outcome, approaches it more closely than a casual consideration of the problem might suggest. The third step is to discuss the ideal governmental intervention, the efficient solution that would be imposed by an all wise and all powerful benevolent despot--a bureaucrat god. The fourth step is to discuss the consequences of actual governmental involvement, and their attractiveness relative to the outcome of the market.
The first and third steps of this process, taken alone, constitute the traditional theory of regulation--the theory that regulation consists of government stepping in to correct market failures. The fourth step is what distinguishes the modern theory of regulation. Instead of asking what an ideal government could do we ask what a government, possessing the authority to intervene in certain ways, will do.
The problems discussed will be grouped according to the source of the market failure: imperfect information, asymmetric information, imperfect competition, and externalities/public good problems.
The assumption of perfect information seems most appropriate for goods that are purchased repeatedly and whose characteristics are easily observed by the user. If I buy packages of fruit from a store, I discover whether the fruit at the bottom is worse than the fruit at the top as soon as I open the package, and I discover whether the fruit tastes as good as it looks as soon as I eat it. Since each individual purchase represents a tiny fraction of my lifetime expenditure on fruit, the cost of shopping around to determine the quality of fruit offered by different stores is small compared to the cost and value associated with consumption of fruit, and can for most purposes be ignored.
Some medical purchases seem to fit this pattern--cold medicines, for example, are used repeatedly, providing the customer an opportunity to determine which ones do noticeably better than others at relieving his symptoms. But the value of a cold medicine depends only in part on how well it relieves symptoms; other relevant considerations are its effect on how fast you recover from the cold and its side effects, if any. Measuring these effects is difficult both because your health depends on many different factors and because side effects may be very long term. So the information on the quality of medicines that we get by consuming them is very imperfect. The same applies to information on the quality of medical services that are consumed regularly--such as the services of a pediatrician patronized by a large family.
For many medical services the situation is far worse. Few of us break our bones often enough to form a competent opinion of the skills of those who set them; still fewer are so unfortunate as to acquire a large sample of the services of oncologists or brain surgeons. The same is true for our experience with medicines used for the rarer and more serious ailments. In these cases we may be poorly informed. If so, our willingness to pay for drugs or services reflects only very approximately their real value to us, hence providers of goods and services may find it in their interest to provide them even when their real benefit is less than their cost, or not to provide them even when it is greater. In either case we have an inefficient outcome.
There are a number of market solutions to problems of this sort. One is to voluntarily shift the decision, and the associated costs and benefits, to some organization better informed than the individual consumer. By buying health insurance, for example, and allowing the insurance company to provide expert advice on what doctors I should patronize or what drugs I should use, I transfer the decision to a firm that has better information and more expertise than I do.
This raises some additional problems associated with insurance, which will be discussed below. It also raises the problem of how to decide which insurance company to trust. One solution is to purchase life and health insurance in the same package--since the seller then has a strong incentive to keep me alive, his incentives are at least roughly the same as my own. Another is to rely on published reports of the performance of insurance companies. While this again imposes information problems on the customer, they are less severe than the original problems--there are fewer insurance companies than doctors, so it is easier to get some measure of their relative performance.
Another solution is for some expert body to certify the quality of drugs or physicians; a familiar example in another field is the Underwriter's Laboratory. In the case of prescription drugs, and to some degree of non-prescription drugs, the customer hires the physician to give him expert advice--and the physician hires the people who produce medical journals. In the case of physicians, group practice provides one form of certification and the acceptance of a physician by a hospital provides another. In a market without governmental licensing of physicians, other forms of certification, by medical associations, insurance companies, or the like, would presumably arise, as they have in other professions.
Another solution is a guarantee. If customers believe that they are ignorant about the effects of drugs and that the drug companies are not, they should strongly prefer drugs produced by companies that assume liability for unexpected side effects--and be willing to pay more for the drugs sold by such companies. Under those circumstances, a company's refusal to guarantee its product is tantamount to an admission that it knows something the customers do not. Similarly, if patients believe that physicians vary widely in quality, they may choose to patronize those who voluntarily make themselves legally responsible for their errors.
There are two sides to the question of guarantees--or more generally, of liability. Given that the court system is costly and imperfect, the cost to a physician of being liable for his mistakes--or what a court decides are mistakes--may be larger than the value to his patients of having him so liable. Under a market system the rule is freedom of contract. The legal system establishes a default rule--caveat emptor or caveat venditor or something in between--and the physician is free to transfer the liability to himself by offering a guarantee or away from himself by requiring patients to waive some or all of their rights to sue in exchange for his treating them. Under such a system the market differential between "guaranteed" and "non-guaranteed" physicians, or between the services of the same physician with or without the guarantee, reflect the cost to the physician of being liable--the cost of additional malpractice insurance, lawyer's fees, damage payments, and the like. The customer is free to decide whether the additional security is worth the additional price. Under our present system liability rules are determined by the courts and waivers are unenforceable. The customer ends up paying for the malpractice insurance whether or not he thinks it is worth the price.
So far I have discussed private solutions to problems of imperfect information. What about governmental solutions? The obvious one is for the government to generate information, leaving the customer free to decide for himself how to make use of it. A familiar example is the labeling of cigarettes; the customer is informed that the government believes they cause cancer, and is free, if he wishes, to reject the conclusion--or to decide that he is willing to pay a price in increased cancer risk in exchange for the pleasure of smoking.
As long as we are dealing with rational--albeit imperfectly informed--consumers, it is better for the government to leave the consumer free to utilize the information it provides as he wishes. Although the government may have superior information about the side effects of a drug or the consequences of smoking, the consumer has superior information about his own values--how much he enjoys smoking, how important side effects are to him, how much he is willing to pay for a superior product or treatment.
In the case of physicians, this is an argument for certification and against licensing. An individual who, knowing that the government believes a practitioner to be insufficiently skilled, still prefers to go to him--perhaps because he is the only one available, is less expensive, or speaks the same language as the customer--is then free to do so. This has the further advantage of allowing the customer to ignore the government's opinion if he concludes that it is probably wrong. If his experience with government generated information is more extensive than his experience with physicians, he may be able to evaluate the former even if not the latter.
If we look not at what the government should do to deal with problems of imperfect information but at what it does do, we find that certification is an uncommon solution for either physicians or drugs. Physicians are licensed, and unlicensed physicians are forbidden to practice. Similarly, although there is some control over the labelling of drugs, the main effect of government intervention is to keep drugs off the market until the FDA has approved them.
One explanation is that present policies assume customers who are not only poorly informed but irrational as well--even when the government tells them what they should do, they refuse to do it. An alternative explanation is that the chief objective of at least some regulation is the welfare not of the patient but of the doctor. Whether or not medical licensing improves the quality of medical care, it surely holds down the number of physicians and so holds up their salaries. There is extensive evidence that the American Medical Association has used medical licensing for this purpose, in some cases supporting requirements, such as U.S. citizenship, that made more sense as ways of restricting entry to the profession than as ways of maintaining quality. An in many cases of licensing--not only of physicians but of barbers, manicurists, egg graders, yacht salesmen, tree surgeons, potato growers, and a host of others--it is clear that the political pressure for licensing came not from the customers but from the profession.
This raises a general issue frequently ignored by those who wish to substitute governmental for private decisions. Even if the government is better informed than the individual, the individual has one great advantage in making decisions about his own welfare--he can be trusted to have his welfare as one of his principal objectives. That is less true of anyone else, including the set of interacting persons called government.
In the case of FDA regulation of drugs, the analysis of what happens and why is less clear than in the case of medical licensing; while it is possible to interpret FDA policy as the enforcement of a cartel agreement on behalf of the producers of existing drugs, it may also be interpreted as a response to public pressure produced by widespread publicity on the hazards of new drugs, in particular due to the Thalidomide case.
Even if the only interest group affecting the legislation is the general public acting on free information (because of rational ignorance, it rarely pays the general public to base its political decisions on anything else), the results may still be undesirable, since free information is often not very accurate. If the FDA licenses a drug that turns out to have disastrous side effects, the result is a front page story and the end of the career of whoever made the decision. If it refuses to license a useful drug, the result is to keep a cure rate from rising--say from 92% to 93%. The total cost may be very large, but it is not very visible, so the FDA may have a strong incentive to be overcautious, possibly with lethal effects.
This point was illustrated some years ago when the FDA put out a press release confessing to mass murder. That was not, I should add, the way the press release was phrased, nor the way in which it was reported. The release announced that the FDA had approved the use of timolol, a beta-blocker, to prevent recurrences of heart attacks; it was estimated that its use would save between seven and ten thousand lives a year.[23] Since beta-blockers were already widely used outside of the U.S., the FDA was, in effect, confessing that by preventing their use for over a decade it had killed about a hundred thousand people--a sizable cost, even when compared to the benefit produced by the FDA's decision to keep Thalidomide off the market.
Peltzman's study of the effect of the Kefauver amendments concluded that they had imposed large net costs. So far as I know nobody has yet done a comparable study attempting to estimate the net effect, in either dollars or lives, of FDA regulations restricting the introduction of potentially dangerous drugs. The case of beta-blockers suggests that it is not obvious whether there has been a net gain; absent a complete study, it is hard to say much more than that.
Asymmetric information involves a more subtle kind of problem, and different solutions, than imperfect information. It involves situations in which one party to a transaction has information that the other lacks, and there is no (convincing) way to share the information with the other party. A standard example is the used car market.[24] Owners of used cars have information about their quality, based on experience, that potential buyers cannot reproduce. The seller can tell the buyer that a particular car is a cream puff rather than a lemon--but since the seller has an incentive to say that whether or not it is true, the buyer has no reason to believe it.
While this is unfortunate for the buyer who ends up with a lemon, why does it lead to an inefficient outcome? The problem is that a car may fail to be sold even though it is worth more to a potential buyer than it is to its present owner. The buyer, who cannot distinguish good cars from bad, makes an offer based on the average quality of used cars being sold. The seller accepts or rejects the offer based on the actual quality of the particular car, which he knows. In the case of a lemon, the buyer is paying for a better car than the owner is selling, so he is likely to make an offer the owner will accept. In the case of a cream puff the reverse is true; the buyer's offer assumes the car to be average, the seller knows it is better than average, so he may be unwilling to sell at any price the buyer is willing to offer. The average car sold is then worse than the average car offered for sale, since below average cars are more likely to be sold than above average ones.
That fact further depresses the amount the buyer is willing to offer--the relevant average for him is the average of all cars sold, not of all cars offered for sale. The acceptance of his offer will be evidence that the car is below average. In extreme cases, this process may proceed so far that only one car is sold--the worst car on the market, sold at a price correctly reflecting its quality. In more realistic cases, there is a partial market failure--some of the better quality cars fail to be sold even though they are worth more to the potential buyer than to the seller, because the potential buyer's offer does not take account of their actual quality.
In the context of health insurance, the same problem is called adverse selection. Individuals know more about their own health, past and future, than insurance companies can learn. An individual's knowledge of how dangerously he drives, how well he takes care of himself, what medical problems he has been having that he has not yet reported, how willing he is, if he does get sick, to sit in the hospital for as long as possible at the insurance company's expense, and the like, is relevant to how much it costs to insure him and is inaccessible to the insurance company. With regard to such information the company must treat each customer as an average over the population of individuals buying insurance, and set its rates accordingly. This makes insurance more attractive for the customer who knows he is a bad risk--more precisely, a worse risk than the insurance company thinks he is--and less attractive for the customer who knows he is a better risk than the insurance company thinks. So the people choosing to buy insurance represent a biased sample of the population as a whole--a sample biased towards the bad risks. The insurance company will allow for this in setting its rates. That makes insurance even less attractive to the good risks, reducing even further the number of good risks who buy it.
One market solution is a group policy. If an insurance company insures all the employees of a firm together, the sample of insured individuals is only slightly biased towards bad risks, since the existence of the insurance is only a minor factor in determining who chooses to work for that company. This is an imperfect solution, since it means that some people get insured who would not want insurance at a price that correctly reflects their medical circumstances--individuals whose risk aversion is not sufficient to make up for the administrative cost of providing the insurance. It is also imperfect because it applies only to members of suitable groups.
The obvious governmental solution is a group policy for the entire population--national health insurance. This eliminates one of the imperfections of the private solution--its limited applicability. If provided by a bureaucrat god perfectly informed about the appropriate level of coverage and all the associated administrative details, and suitably tailored to the requirements of different customers, it would be more efficient than the private alternative.
If we consider the governmental solution under more realistic assumptions, the case for it becomes far weaker. National health insurance affects many people other than patients--and some, such as physicians and hospitals, are more concentrated and better organized. The arguments of Part II of this paper suggest that the public good problems associated with providing "the right amount of health insurance in the right way" may be much greater than the advantages gained by eliminating adverse selection, especially given that a good deal of adverse selection can be eliminated privately by group plans.
In addition to adverse selection, insurance in general and health insurance in particular involve a second efficiency problem. While it is due to externalities rather than to asymmetric information, this seems the most convenient place to discuss it. The problem is called moral hazard.
Consider an individual deciding whether to stay an extra day in the hospital. The cost of doing so is $200. The value to him, in terms of a slight reduction in the chance of a relapse, is $50. If he is paying his own bills, he goes home. If the insurance company is paying more than 3/4 of the cost, he stays. Since he is buying something whose cost is more than its value, the outcome is inefficient. Put differently, his decision imposes an external cost on the insurance company; since he ignores that cost in his decision he may make an inefficient decision. The same problem occurs whenever the individual either makes decisions affecting benefits he receives and the insurance company pays for, or makes decisions affecting costs he pays and benefits that go, at least in part, to the insurance company. An example of the latter would be an expense--a medical exam not covered by the insurance, or an improvement in nutrition--designed to reduce future medical problems.
There are several ways in which the costs of moral hazard can be reduced. One is coinsurance--if the insured is responsible for part of the bill, he has at least some incentive to keep it down. Another is for the insurance company to make some of the decisions--to pay for only the number of days in hospital that its doctor recommends, for example.
Moral hazard applies to government health insurance just as it does to private health insurance--indeed, it applies to any government program that transfers some of the cost (or benefit) of individual decisions from the individual concerned to others. The advantage of the private system is that insurance will occur only if the gain due to risk sharing (or other advantages) at least balances the cost imposed by moral hazard--otherwise the insurance company will find that there is no price it can charge at which it can both cover its costs and sell insurance. There seems to be no comparable constraint on the political alternatives. In this case the market outcome is as good as the best possible political outcome, and better than any outcome that we would expect the political system to produce.
One of the assumptions that is usually used in proving the efficiency of the market outcome is perfect competition--every consumer and producer is assumed to be a small enough part of the market so that the amount he consumes or produces does not affect the price. To see why this matters, consider the situation of a producer who knows that the higher his rate of output the lower the market price will be. If he produces at a rate of 101 units a month instead of 100 units a month, his revenue will rise by the price of one unit and fall by 100 units times the amount of the resulting price drop. His marginal revenue--the change in revenue due to one more unit of output--will be the sum of the two changes. Since his objective is to maximize profit--revenue minus cost--he will produce up to the point where marginal revenue equals marginal cost. But it is price, not marginal revenue, that measures the value to a consumer of one more unit. Units for which the cost of production is more than marginal cost but less than price will not get produced; this is inefficient, since it means that there are units not produced that would be worth more than it would cost to produce them. Similar problems arise if the assumption of perfect competition is violated for consumers.
Much of the medical market--most obviously the services of general practitioners in large cities--has the characteristics necessary for an almost perfectly competitive market. The main hindrances to competition on that part of the market are the result of government interference--the prohibition on advertising the price of medical services and the restriction on entry to the profession, both enforced by state regulation of who can practice medicine. The same applies to the retailing of medicine; advertising of the prices of prescription drugs has frequently been illegal.
There is an important point here that I have made before and will make again. It is not sufficient for the supporters of intervention in the market to say "we oppose those particular interventions--what we are in favor of are only the interventions that increase efficiency." Unless they have some solid theoretical or empirical grounds for claiming that they know how to achieve that objective, we must take existing regulation as evidence of what government does--and will do--given the opportunity. If regulation has often consisted of the use of government power to drive up the price of some good or service for the benefit of the producers--and it has--we must treat that fact not as a repeated accident but as evidence. That is particularly true if the evidence fits such theory as we have--and it does. Producers are typically a more concentrated interest than consumers.
What about imperfect competition resulting from economies of scale in the medical industry? Examples are drug research, hospitals, and physicians in sparsely populated areas. In such cases there exist policies that a bureaucrat god could follow that would produce improvements. In order to follow the optimal policies, however, the real world regulator must know the cost curves of the regulated firms--how much it costs to produce any level of output, where level includes quality as well as quantity--and the demand curves of consumers. This is difficult in any industry, and there is a good deal of evidence that regulation of monopolies does not and perhaps cannot force them to charge efficient prices.[25] It is particularly difficult in an industry such as medicine, where quality variables are important and hard to measure. Regulation has the additional disadvantage of providing the industry with a tool that may be used to maintain a monopoly position that would otherwise be eliminated by technological change; a classic example is the regulation of canal traffic and trucking by the ICC in order to defend the railroads from competition.
My own conclusion, considering both the theoretical arguments and historical experience, is that regulation of monopolies may never be desirable, and certainly is not in cases that fall substantially short of complete and very long-lived monopoly. The question is one on which there is a long literature, and many economists would take a less extreme position.
The final category of market imperfection that I will discuss is the category of externalities and public goods. I combine them because they are, to a considerable extent, two ways of looking at the same problem.
A public good is a good that, if produced, will be available to all the members of a pre-existing group: the producer cannot control who gets it. A pure public good is a public good for which consumption by one individual does not interfere with consumption by another. The fact that a good happens to be produced by government does not make it a public good--postal service, for example, is a private good that happens to be produced by government--nor does the fact that something is a public good mean that it cannot be produced privately. Radio and television broadcasts are pure public goods, yet both are produced privately. The fact that something is a public good does, however, pose a problem for the producer who wants to be paid for what he produces--if he cannot control who gets it, how can he make consumers pay him? This in turn implies a problem from the standpoint of economic efficiency. If a good is worth more to the consumers than it costs to produce, it should be produced; but it will be produced only if the amount the producer can get paid for producing it is at least as great as his cost of production. Since the amount that can be raised to pay for a public good is generally much less than its value to the consumers, there will be public goods worth producing that do not get produced--which is inefficient.
One solution to this problem is to have the good produced by government. This solution, as I pointed out in Part II, raises a second public good problem. Production of good law--or bad law for that matter--is a public good from the standpoint of the group benefitted by the law, so getting the government to do things requires that someone solve a public good problem.
There are also a number of private solutions to the public good problem. One is charity. Another is for the producer to organize a contract among those who will benefit from the public good by which each agrees to contribute only if the others do. Each member knows that either his signature has no effect (if someone else refuses to sign) or it gives him the package "good minus payment" (if everyone else signs). As long as the specified payment is less than the value of the good to the consumer, he signs and the good gets produced.
This solution runs into problems in a world of imperfect information and non-zero transactions costs, especially if the public is large. It pays each individual to pretend the good is of little value to him, or to claim that even though he values it he is too stubborn to agree to pay for it. If he succeeds in convincing the entrepreneur drawing up the contract, his name will be omitted from the list and he will get the public good without having to pay for it. Such bargaining problems imply that for any save a very small public, the amount that can be raised to pay for a public good is a small fraction of its total value--smaller the larger the public. So methods of this sort can be used only for small publics or for goods whose cost is small compared to their value.
There are other ways of producing public goods, including the ingenious solution employed by the broadcast media, but a discussion of all of them would exceed the bounds of this chapter. The important results are that public goods can be produced privately, that the outcome of private production is not in general efficient (some goods that are worth producing do not get produced), and that the problems tend to increase with the size of the group.
An externality is a cost (or benefit) that one individual's actions impose on another. A public good can be described as a positive externality, and the prevention of a negative externality is a public good. As a rule, the term "public good" gets used to describe situations where the purpose of the action is producing the good (radio broadcast) and the term "externality" to describe situations where the action has a separate purpose and the injury or benefit is a side effect. The line between the two terms is a blurred one, and the decision of whether to describe a particular problem in terms of externalities or public goods is to some degree arbitrary.
The reason that externalities lead to inefficient outcomes is straightforward. An individual producer produces a good if and only if the benefit he receives (by selling or consuming it) is at least as great as the cost he pays. As long as he pays all of the associated costs and receives all the associated benefits, this is equivalent to the efficient rule-- "produce if and only if total benefits are at least equal to total costs." In the case of goods produced without externalities on a competitive market, the condition is satisfied; the price he receives equals the marginal value of the good to the consumer, the prices he pays for his inputs equal their marginal cost of production, so cost and revenue on his balance sheets equal the "total social cost" and "total social benefit" produced by his actions.
With externalities, this is no longer true. Since the producer pays only part of the cost (or, in the case of positive externalities, receives only part of the benefit ), the revenue and cost figures that determine his decision of what and how much to produce no longer equal the corresponding total values and total costs. He may choose to produce a good even though total cost (including the external cost) is greater than total benefit, or choose not to produce even though total benefit (including the external benefit) is greater than total cost. The outcome is no longer in general efficient.
How does all of his apply to the market for medical care? As on most markets, some of the goods impose externalities--in the case of medical care, typically positive ones. If I get inoculated against a contagious disease, that reduces the chance that I will infect you--a positive externality. If my drug company discovers a new family of drugs, that provides information useful to other companies. If I spend money on keeping myself healthy, that benefits all those who care for me and would be made unhappy by my illness; it may even benefit people who have never met me, but feel a glow of pleasant pride at knowing that "My country is the healthiest in the world" or "year by year the human race is getting healthier." All of these are properly counted as externalities. Economic theory suggests that the market will underproduce inoculations, drug research, and health because in each case the individual paying the cost receives only part of the benefit. Similarly, if the use of some antibiotics imposes external costs by encouraging the development of resistant strains of bacteria, such antibiotics will be overused, since the individual using them pays only part of the cost.
All of these, however, are what I earlier described as mostly private (or at least, largely private) goods. In each case a large part of the benefit goes to the person who pays for it--I stay healthy because of my inoculation, the drug company makes money off its new drugs, and my own health probably gives more pleasure to me than to even the most altruistic of my friends. If ninety percent of the benefit goes to me, then I will make the wrong decision only in those cases where the cost is more than ninety and less than a hundred percent of the value, so the result, although inefficient, will not be very inefficient.
I argued in Part II that the political market is usually very inefficient, since the political pressures for and against legislation (and other government activities) represent the values to the groups affected, weighted by their widely varying ability to exert political influence in defense of their interests. My conclusion was that the substitution of the political for the private market was justified, if at all, only in cases of extreme market failure on the private market. None of the case I have described involve such extreme failure. I conclude that although the outcome of the market could be improved by a bureaucrat god, it is likely to be worsened by real world regulation.
I believe I have now covered all of the obvious causes of market failure on the market for health care. In most cases the failure implies that a sufficiently wise, powerful, and benevolent authority could improve--in terms of economic efficiency--the outcome of the market. In no case is there any clear reason to believe that assigning additional power to government--as government actually exists--would improve the situation; in many there is reason to believe that doing so would make it worse. In several cases existing problems are the direct result of government interference with the market.
In the course of this essay, I have attempted to make plausible a thesis many readers will find absurd--that health care should be provided entirely on the private market, just as shoes and potato chips are now provided. Obviously I have not, and cannot, answer all imaginable arguments against doing so; I have tried to answer the ones I find most persuasive.
One argument that I have not yet answered--and find very unpersuasive--is the claim that "health is too important to be left to the market." My response would be that the market is, generally speaking, the best set of institutions we know of for producing and distributing things. The more important a good is, the stronger the argument for having it produced by the market.
Both barbers and physicians are licensed; both professions have for decades used licensing to keep their numbers down and their salaries up. Government regulation of barbers makes haircuts more expensive; one result, presumably, is that we have fewer haircuts and longer hair. Government regulation of physicians makes medical care more expensive; one result, presumably, is that we have less medical care and shorter lives. Given the choice of deregulating one profession or the other, I would choose the physicians.
I suggest to those readers who remain entirely unconvinced that they may be making the error of judging a system by the comparison between its outcome and the best outcome that can be described, rather than judging it by a comparison between its outcome and the outcome that would actually be produced by the best alternative system available. If, as seems likely, all possible sets of institutions fall short of producing perfect outcomes, then a policy of comparing observed outcomes to ideal ones will reject any existing system.
It is easy, and satisfying, to pick some unattractive outcome--a poor man, actual or imaginary, turned away from the expensive private hospital that could have cured his disease--and describe it as "intolerable," "unacceptable," or some similar epithet designed to prevent further discussion. This is, however, a game that any number can play. It is equally easy, as I demonstrated earlier in this essay, for the defender of the market to orate about the hundred thousand people who died of heart attacks because the FDA refused to permit American physicians to prescribe beta blockers to American patients. In a large and complicated society, it is likely that any system for producing and allocating medical care--or doing anything else difficult and important--will sometimes produce outcomes that can plausibly be labeled as intolerable.
The question we should ask, and try to answer, is not what outcome would be ideal but what outcome we can expect from each of various alternative sets of institutions, and which, from that limited set of alternatives, we prefer. I have tried to do so. My conclusion is that there is no good reason to expect governmental involvement in the medical market, either the extensive involvement that now exists or the still more extensive involvement that many advocate, to produce desirable results.
*: This essay originated as a comment on a manuscript version of Buchanan (1985). My debt to my target, and to the organizers of the conference at which both papers were given, will be apparent to anyone familiar with Buchanan's work.
1. G. Akerlof, "The Market for `Lemons'," Vol. No. 336 QJE 488-500 (1970).
2. G. Becker, "Competition and Democracy," JLE Oct 1958, 5 pages.
3. Buchanan, Allen, Ethnics, Efficiency and the Market, Roman and Allenheld, 1985.
4. Buchanan, James M. and Tullock, Gordon, The Calculus of Consent. University of Michigan Press, Ann Arbor (1962).
5. R. Coase, "The Problem of Social Cost", JLE Oct. 1960.
6. Downs, Anthony, An Economic Theory of Democracy. New York: Harper and Row, 1957.
7. Eckert and Hilton, "The Jitneys," JLE August 1976.
8. D. Friedman, The Machinery of Freedom, first edition, Harper and Row, 1973, 2nd edition, Open Court, 1989.
9. D. Friedman, "Many, Few, One - Social Harmony and the Shrunken Choice Set," AER March 1988.
10. D. Friedman, "What Does Optimum Population Mean?" in Research in Population Economics, Vol. III, Simon and Lindert, eds. 1981.
11. D. Friedman, "What Is Fair Compensation for Death or Injury?" International Review of Law and Economics, 2, 1982.
12. D. Friedman, Price Theory: An Intermediate Text. South-western Publishers, 1986, 1989.
13. D. Friedman, "Does Altruism Produce Efficient Outcomes? Marshall vs Kaldor." Journal of Legal Studies, 1987 Vol. XVII, January 1988.
14. P. Linneman, "The Economic Impacts of Minimum Wage Laws," JPE June 1982.
15. William A. Niskanen, Bureaucracy and Representative Government, Chicago: Aldine-Atherton, 1971.
16. Mancur Olson, The Logic of Collective Action. Cambridge, Mass.: Harvard University Press, 1965.
17. S. Peltzman, "An Evaluation of Consumer Protection Legislation: 1962 Drug Amendments." JPE September/October 1973.
18. S. Peltzman, "The Effects of Automobile Safety Regulations," JPE August 1975.
19. S. Peltzman, "Toward a More General Theory of Regulation," Vol. XIX (2) JLE 211-240 (August 1976).
20. R. Posner, "Theories of Economic Regulation," Bell Journal, Fall 1974.
21. G. Stigler, "Director's Law of Public Income Redistribution,"Vol. XIII (i) JLE 1-10 (April 1970).
22. G. Stigler, The Citizen and the State, University of Chicago Press: 1975.
23. G. Stigler and C. Friedland, "What Can Regulators Regulate?" Vol. V JLE 1-16 (October 1962).
Back to the list of articles.
Back to my home page.