So far, I have generally assumed that if there is a possibility for a trade--if I am willing to sell something at a price at which you are willing to buy it--the trade occurs. I have ignored both the problems of finding a trading partner and negotiating a trade and the associated transaction costs.
Barter vs Money
The simplest form of trade is barter; I trade goods that I have and you want for goods that you have and I want. This raises a problem. I must find a trading partner who has what I want and wants what I have: what economists call a double coincidence of wants. In a simple society in which there are only a few goods being traded, this may not be a serious problem; but in a complicated society such as ours, it is. If I want to buy a car, I first look in the classified ads to find someone who is selling the kind of car I want, then call him up and ask him if he wants to be taught economics in exchange for his car. This drastically reduces the number of potential trading partners.
The solution is the development of money--some good that almost everyone is willing to accept in exchange. Money usually starts out as some good (gold, cloth, cattle--the word "pecuniary" comes from the Latin word for cattle) valued for its own uses; people are willing to accept it even if they do not intend to consume it, because they know they can later exchange it for something else. In a money economy, I find one person who wants what I have, sell it to him, and then use the money to buy what I want from someone else.
The advantage of money is obvious; the disadvantage is that you cannot eat it or wear it (exception: wadmal, wool cloth used as money in medieval Iceland). If markets are thin--if there are few people buying or selling--the individual who chooses to hold a stock of money may find that he cannot easily exchange it for what he needs when he needs it.
Thin markets cause two different problems for someone who wants to buy or sell. The first is that there may be nobody who wants what he is selling today or is selling what he wants to buy; the mere process of locating a trading partner may be expensive and time consuming. The second is that if he does find a trading partner, he becomes part of a bilateral monopoly--one buyer, one seller. Bilateral monopoly, for reasons discussed in an earlier chapter, can lead to substantial transaction costs: time and energy spent haggling over the price, and deals that do not get made because of a breakdown in bargaining.
In a society in which markets are thin and the number of traded commodities is small enough so that the double coincidence problem is not too serious, individuals may find it more convenient to hold wealth in the form of goods rather than money. This was probably the situation in early medieval Europe. Coins existed and were used in exchange; but barter was, for several centuries, more common.
A Market We All Know and Love
In order to understand the difficulties of barter, it is useful to consider the large-scale barter market of which you are all part--the marriage/dating/sex market. The reason this is a barter market is that if I am going out with or married to you, you are necessarily going out with or married to me. I must find a woman whom I want and who wants me--the double coincidence of wants.
We observe, in this market, large search costs, long search times, lots of frustrated and/or lonely people of both sexes--in other words, a market where traders have a hard time getting together, due largely to the high transaction costs of barter.
In Chapter 1, I pointed out that even if every individual in a group behaves rationally, the result may be undesirable--for every individual. This happens when one person's actions impose costs or benefits on others. The examples I gave in Chapter 1 involved students cutting across the lawn and fighters running away in battle, shooting their weapons without aiming them, or not shooting at all. In such situations, the rationality of the individual does not imply that the group acts as if it were rational.
The rest of this chapter will be devoted to a discussion of situations of this sort. I will start with a number of specific examples and then go on to explain the two general categories under which many such problems are usually classed in economics: public goods and externalities. I will end by discussing the special problems associated with imperfect information.
Good for Each May Not Be Good for All: Some Examples
I will give three examples of conflicts between the individual rationality of the members of a group and their welfare. Two--the first and the last--are situations that should be familiar to every reader over the age of 17. The other is a widely discussed public policy issue with which I hope most of you have had no personal experience.
To Vote or Not to Vote? In deciding whether to vote in the next election, one should consider both costs and benefits. The costs are fairly obvious: a certain amount of time standing in line and additional time spent studying issues and candidates in order to decide how to vote. The benefits are of two sorts: those that do not depend on the effect of your vote on the election, and those that do. An example of the first sort might be your feeling of having done your civic duty or your pleasure at voting against a candidate you particularly dislike.
The second sort of benefit comes from the effect of your vote on the outcome of the election. In evaluating such benefits, you should consider two questions: how important it is that the right candidate win and how likely it is that your vote will affect the outcome. In most large elections, the probability that your vote will affect the outcome is very small; in a presidential election, it is well under one in a million. Unless getting the right person elected is immensely valuable to you--so valuable that you are willing to bear the costs of voting in exchange for one chance in a million of influencing the outcome--the effect of your vote on the election is not a good reason for voting unless you expect the election to be extraordinarily close. If you vote anyway--because you enjoy voting or because you believe that good citizens vote or because you like being part of a history-making event reported on nationwide television--the minuscule effect of your vote on the election gives you very little incentive to be sure you are voting for the best candidate.
The usual response to arguments of this sort is either "You are saying people should be selfish" or "What if everyone did that?" The answer to the first is that I have not assumed that you are selfish in any conventional sense of the word. I assume you are concerned with costs and benefits, but I include as a benefit the achieving of whatever objectives you happen to have. Obviously individuals have objectives that are not selfish in any narrow sense--they value the welfare of their children, their friends, and (to a lesser degree) people they do not even know. One reason you might put a high value on electing the right candidate is the belief that doing so will benefit not only yourself but hundreds of millions of other people. If you were so altruistic as to give the same weight to the welfare of every other person as to your own, then the benefit of electing the right candidate would be hundreds of millions of times as great as the direct benefit to you. That might be a sufficient reason to spend an hour or two voting, even if you realized that all you were buying was one chance in a million of influencing the outcome of the election. Casual observation suggests that few people are that altruistic.
The question "What if everybody acted like that?" can be answered in two ways. The first is to point out that if enough people refrained from voting, the remaining voters would each have a substantial chance of influencing the outcome of the election, and it would then pay them to vote. The equilibrium would be a situation in which the (say) ten thousand most concerned citizens voted.
The second answer to the question "What if everybody acted like that?" is to point out that the question implicitly assumes that true beliefs must have desirable consequences--and therefore that beliefs with undesirable consequences must be false. There is no reason why this must always be so. Perhaps it is true both that sensible people will not vote and that if everyone acts on that principle the consequences will be bad. If so, it might be wise for me not to tell you that sensible people do not vote, but that does not make it untrue. A statement may be both true and dangerous. The previous sentence is such a statement--since it provides ammunition for those who wish to argue against free speech.
The apparent paradox--that if everyone correctly perceives how to act in his own interest and does so, everyone may be worse off as a result--comes from the fact that different people have different objectives. Suppose there are a hundred of us, each of whom can individually choose action A or action B. My taking action A gives me $10 and costs the rest of you a total of $20. Your taking action A gives you $10 and costs the rest of us, including me, a total of $20. As long as we act separately, it is in the interest of each of us to take action A--making us all worse off than if we had all taken action B. The problem is that I only control my action--and I am better off taking A than B. This, of course, is the problem we encountered long ago in the discussion of why soldiers run away.
A simple and striking example of such a situation is the prisoner's dilemma discussed back in Chapter 11. Joe and Mike, the two accused criminals, would both be better off if they both kept silent. But if Mike confesses, the D.A. will have the evidence needed to convict Joe--and will punish him for his silence with a stiff sentence. So if Mike is going to confess, Joe had better confess too. If Mike stays silent and Joe confesses, the D.A. will express his gratitude by letting Joe off with a token sentence. So if Mike is not going to confess, Joe is better off confessing. Whatever Mike does, Joe is better off confessing, and similarly for Mike. They both confess, and both get worse sentences than if they had both kept silent.
Plea Bargaining: A Real-World Prisoner's Dilemma. A plea bargain is an arrangement by which a prosecutor, instead of trying a defendant on a charge of, say, first-degree murder, allows the defendant to plead guilty to a lesser charge, such as second-degree murder or manslaughter. It is widely criticized as a way of letting criminals off lightly. In fact, it seems likely that the existence of plea bargaining results in criminals being punished more severely rather than less. If plea bargaining were abolished--as some people suggest it should be--the result might well be to reduce the sentence received by the average criminal.
How can this be? Surely a criminal will only plead guilty to the lesser charge if doing so is in his interest--which means that a certain conviction on the less serious charge is preferable, for him, to whatever he believes the chance is of being convicted on the more serious charge. True. But the chance of a conviction depends on what resources, of money and time, the prosecution spends on that particular case--which in turn depends on how many other cases had to go to trial and how many were settled by plea bargaining.
Suppose there are 100 cases per year, and the district attorney has a budget of $100,000. He can only spend $1,000 on each case, with the result that 50 percent of the criminals are acquitted. With plea bargaining, the D.A. concentrates his resources on the ten criminals who refuse to accept the bargain he offers. He spends $10,000 prosecuting each of them and gets a conviction rate of 90 percent. Each criminal deciding whether to accept the D.A.'s offer knows that, if he refuses, he has about a 90 percent chance of being convicted--so he accepts any offer that he prefers to a 90 percent chance of conviction. On average, all the criminals, both the ones who accept the bargain and the ones who do not, are worse off--more severely punished--than if the D. A. prosecuted all of them on the more severe charge and convicted half. Each individual criminal benefits by accepting the D.A.'s offer--but by doing so, he frees resources that the D.A. can then use against another criminal, raising the average conviction rate. The higher conviction rate makes criminals willing to accept worse bargains. All of the criminals would be better off if none of them accepted the D.A.'s offer, but each is better off accepting. This is the prisoner's dilemma in real life.
Why Traffic Jams. This is a situation in which each individual takes the action that is in his individual interest; they are all, as a result, worse off than if they had acted differently. A more familiar example of such a situation occurs twice a day, five days a week, about two blocks from where I used to live. The time is rush hour; the scene is the intersection of Wilshire Boulevard and Westwood Avenue in Los Angeles, said to be the busiest intersection in the world. As the light on Wilshire goes green, ten lanes of traffic surge forward. As it turns yellow, a last few cars try to make it across. Since Wilshire is packed with cars, they fail and end up in the intersection, blocking the cars on Westwood, which now have a green light. Gradually the cars in the intersection make it across, allowing the traffic on Westwood to surge forward--just as the light changes, trapping another batch of cars in the intersection.
If drivers on both streets refrained from entering the intersection until there was clearly enough room for them on the far side, the jam would not occur. Traffic would flow faster, and they would all get where they are going sooner. Yet each individual driver is behaving rationally. My aggressive driving on Wilshire benefits me (I may make it across before the light changes, and at worst I will get far enough into the intersection not to be blocked by cars going the other way at the next stage of the jam) and harms drivers on Westwood. Your aggressive driving on Westwood benefits you and harms drivers (possibly including me) on Wilshire. The harm is much larger than the benefit, so on net we are all worse off. But I receive all of the benefit and none of the harm from the particular decision that I control. I am correctly choosing the action that best achieves my objectives--but if we each made a mistake and drove less aggressively, we would all be better off.
My point, in this and the previous examples, is not that rationality implies selfishness. That is a parody of economics. Drivers may value other people's time as well as (although probably not as much as) their own. In Chapter 21, we will discuss the economics of altruism--the behavior of people who value the happiness of other people. If drivers value the welfare of other drivers, rationality may prevent the jam instead of causing it.
The point--which to some readers may seem paradoxical--is that rational behavior by every individual in a group may sometimes lead to an outcome that is undesirable in terms of precisely the same objectives (getting home earlier in this case or getting a light sentence or surviving a battle in some of the other cases we have discussed) that each individual's rational behavior is correctly calculated to achieve. Such situations often involve what economists call public goods or externalities, two concepts that we will now discuss.
There are a number of different, closely related definitions of a public good. I prefer to define it as "a good such that, if it is produced at all, the producer cannot control who gets it." The public-good problem arises because the producer of a public good cannot, like the producer of an ordinary ("private") good, tell the consumer that he can only have it if he pays for it; the consumer knows that if it is produced at all, the producer has no control over who gets it.
One example of a public good is a radio broadcast; if it is made at all, anyone who owns a radio and lives in the right area can receive it. This example demonstrates several important things about public goods. The first is that whether or not a good is public depends on the nature of the good. It is not that the producer should not control who gets it but that he cannot; or, at least, he can control who gets it, if at all, only at a prohibitively high cost (hiring detectives to creep around people's houses and arrest them if they are listening to the broadcast without having paid for it). While the publicness of a good may be affected by the legal system (whether it is legal to listen to a broadcast without the broadcaster's permission), it is mostly just a fact of nature; even if it were legal to forbid unauthorized listening, the law would be prohibitively expensive to enforce.
A second important thing to note about a public good is that it is not defined as a good produced by the government. In this country, radio broadcasts are mostly private; they are still public goods. Many of the things government does produce, such as mail delivery, are private goods; the government can and does refuse to deliver your letter if it does not have a stamp on it. The fact that a good is public presents a problem to a private producer--the problem of how to get paid for producing it--but the problem is not necessarily an insoluble one, as the example of a radio broadcast illustrates.
Private Production of Public Goods. There are a number of ways in which the problem of producing public goods privately may be solved. One, which works best if the size of the public (the group of people who will receive the good if it is produced) is small, is a unanimous contract. The producer gets all the members of the public together, tells them how much he wants each to pay toward the cost of producing the good, and announces that unless each agrees to chip in if everyone else does, the good will not be produced.
Assume that they believe him. Consider the logic of the situation from the standpoint of a single member of the group deciding whether he should agree to chip in. He reasons as follows:
Either someone else is going to refuse, in which case the deal falls through, I get my money back, and my agreement costs me nothing, or else everyone else is going to agree. If everyone else agrees and I refuse, I do not have to pay for the public good, but I also do not get it. So as long as the good is worth more to me than my share of the cost, I ought to agree.
The same argument applies to everyone, so if the public good is worth more to the consumers than it costs to produce, the entrepreneur should be able to divide up the cost in such a way that each individual finds it in his interest to agree.
One difficulty with this is that if the public is large, it may be hard to organize a unanimous contract. One solution is to find a privileged minority: a subgroup of the public that is small enough so that its members can form their own unanimous contract and that receives enough benefit from the public good so that its members can be persuaded to bear the whole cost. When I mow my front lawn, I am acting as a privileged minority (of one); the mowed lawn makes the neighborhood more attractive, benefiting everyone, but I receive enough of the benefit to be willing to pay the whole cost.
Consider how this might work in the case of one of the largest public goods in our society and one of the most difficult to produce privately: national defense. Suppose the inhabitants of Hawaii believe that there is a 10 percent chance of a nuclear strike against their island next year. If the strike occurs, the island will be wiped out. The inhabitants can flee the island before the attack, so the cost will be distributed roughly in proportion to the value of the land they own. Table 18-1 is an (entirely imaginary) listing of how land ownership is divided on the island and how much each owner would pay, if necessary, to prevent the attack.
Value of Land
Value of Defense
United Fruite Co.
Maxwell House Coffee
In this case, the story has a happy ending. Suppose, however, that the total cost of the defense is $149 million. It is still worth having--the top five landowners alone value it at more than its cost--but it will be very hard to get. If the entrepreneur asks the Big 5 to each put up the same proportion of the value of their land (just under 10 percent), Howard Johnson will refuse. Unfortunately for Hawaii, the Howard Johnson firm is run by an optimist who believes the chance of an attack is only 5 percent and therefore is willing to pay only 5 percent of his land value to protect against the attack.
If the information on Table 18-1 were a matter of public knowledge, agreement could still be reached, with Howard Johnson contributing at half the rate of the other four. The problem is that the other contributors are likely to view Howard Johnson's optimism as a bargaining ploy, a way to get them to pay more than their share of the cost. If there is no simple rule for dividing up the cost of defense, agreement on who pays what may well be impossible.
The larger the number of people whose agreement is needed and the less obvious it is how much each values what he is getting, the harder it will be to get agreement. If the public good is cheap--if defense costs only $40 million--the problem is soluble; the entrepreneur can either leave Howard Johnson out of the contract or else charge everyone 5 percent of land value and still raise enough money. But if the cost of the public good is a large fraction of the benefit it produces and if the benefit is spread among many people, raising the money is a serious and perhaps insoluble problem.
In the example discussed, the concentration of land ownership in Hawaii greatly simplified the situation. The Big 5 were a privileged minority; they received a large fraction of the total benefit, so the entrepreneur could, with luck, raise the money he needed from them while ignoring the large number of small holders. The term "privileged minority," which is commonly used in this way, has always struck me as somewhat strange, since the minority has the "privilege" of paying for what all the other members of the public get for free.
Unanimous contracts are one solution to the problem of producing a public good. Another solution is to convert the public good temporarily into a private good. Suppose the public good is flood control; building a dam will reduce floods in the valley below, increasing the value of farm land there. One way to pay for the dam is for the entrepreneur to buy up as much as possible of the land in the valley (or buy options on the land at its current price), build the dam, then sell the land back (or sell the options back to the owners). Since the new flood protection makes the land worth more than when he bought it, he should be able to get a higher price than he paid, for either the land or the options.
Another ingenious solution, which would never have occurred to me if I had not seen it in operation, is to combine two public goods and give away the package. The first public good has a positive cost of production and a positive value to the customer; the second has a negative cost of production and a negative value to the customer. The package has zero or negative cost of production and positive value to the consumer.
This is how radio and television broadcasts are produced; the first good is the program and the second the commercial. Commercials have a negative cost of production from the standpoint of the broadcaster; he gets paid by the sponsor to broadcast them. Since there is usually no convenient way to listen to the program without hearing the commercials, the listener must choose to accept or reject a package deal--program plus commercial. If the net value of the package is positive to him, he will accept it. If the net cost (cost of operating the station minus payment from the sponsor) is negative, if advertising revenues more than cover operating expenses, the broadcaster can and will stay in business.
An interesting example of the public-good problem, and several interesting solutions, occur in the computer industry. A $300 computer program can be copied onto a $3 floppy disk. Programs can be protected against copying, but this is inconvenient for the user, who would like at least one backup copy in case his original gets damaged and who may also find it convenient to copy several of the programs he has purchased onto one disk. Even if programs are protected, someone with a reasonable amount of expertise can frequently "break" the protection--figure out how to copy them. There are even programs on the market designed to copy copy-protected programs. In one case, a program capable of copying other copy-protected programs was copy-protected against itself; a second company sold a program to copy it!
If you cannot effectively copy-protect a program, selling it to one person means, in effect, giving it to everyone. The program is then a public good and figuring out how to make money producing it is a public-good problem. Firms that produce and sell software have come up with a number of ingenious solutions. One of them is bundling. You sell a computer along with a bundle of programs designed to run on that particular computer; in effect you charge for the programs in the price of the computer. Anyone can copy the programs--but to use them, he has to buy the computer. Another kind of bundling is to sell a package consisting of a program plus service: a voice on the other end of a telephone to answer questions about how to make the program work. The seller keeps track of who bought the program and only gives help to registered owners. A third kind of bundling is exemplified by the way in which I "sell" the computer programs that go with this book. A professor who adopts the book is given a free copy of the programs and permission to make copies for his students. I get paid for my work writing the programs in increased sales of the book. I hope.
As these examples suggest, there are a variety of ways in which public goods can be privately produced. Each of these may succeed, under some circumstances, in producing some quantity of a public good. None of them can be relied on to lead to an efficient level of production in the strong sense in which we have been using the term--an outcome so good that it could not be improved by a bureaucrat-god. Typically, the private producer of a public good succeeds in collecting only part of the additional value of each unit of the good produced. He produces up to the point where what he gets for an additional unit (an additional hour of broadcasting, or an additional dollar spent making the program better) is equal to what it costs him. That is a lower level of output than the efficient point where marginal cost to the producer equals marginal value to the consumer.
To see more clearly the sense in which private production of public goods is inefficient, consider some of our examples. In the Hawaiian defense case, Hawaii was worth defending as long as the cost was less than $245 million, since that was the total value of the defense to all the inhabitants put together. If the cost of the defense happened to be only $40 million, private arrangements might produce it, which is the efficient outcome. If the cost were $235 million, it is unlikely that the defense would be produced; since it still costs less than its value, a bureaucrat-god who ordered Hawaii defended would be producing a net benefit. So if the cost of defending Hawaii is $235 million, private production results in an inefficient outcome. Hawaii is worth defending--and is undefended. The private production of public goods is inefficient in the sense of sometimes leading to an inefficient outcome--failing to produce a good that is worth producing.
We have assumed that there are only two possible amounts of defense: none or enough. Whether or not that is plausible in the case of defense, the equivalent assumption is obviously wrong for radio broadcasts or computer programs; in each case, the manufacturer decides how much he will spend and what quality of product he will produce. The efficient outcome is one in which he makes all quality improvements that are worth more to the consumers than they cost him to make. But from his standpoint, improvements are worth making only if they increase his revenue by at least as much as they cost. Since he will be able to collect only part of the value he produces, there may be improvements worth making that he does not find it in his interest to make; so here again the outcome could be improved by a bureaucrat-god. The good may be produced, but it is generally underproduced: An increase in quality, number of hours of broadcasting, or some other dimension would result in net benefits. So private production of public goods is generally inefficient in the technical sense in which I have been using the word.
The Efficient Quantity of a Public Good. While we have talked about producing the efficient outcome, we have not yet discussed how, in principle, one would find out what it is. Figure 18-1 shows the answer to that question, for a very small public. D1, D2, D3 are the demand curves for radio broadcasting of three listeners. Each shows how much broadcasting a listener would buy if it were an ordinary private good--how many hours per day he would pay for as a function of the price per hour. The figure assumes that number of hours per day of broadcasting is the only relevant quality variable, the only way in which the broadcaster can affect the value to his "customers" of what he produces. MC shows the marginal cost curve faced by the broadcaster--how much each additional hour per day of broadcasting costs him.
Calculating the efficient quantity of a public good. MV shows the total marginal value to the three customers--the vertical sum of their demand curves. The efficient quantity Q* is where MV=MC.
As usual, the efficient solution is to produce where MV=MC--to keep increasing the number of hours as long as the value of an additional hour to the listeners is at least as great as its cost of production. We know from Chapter 4 that each demand curve is also a marginal value curve. Each extra hour of broadcasting benefits all three customers; its marginal value is the sum of its marginal value to Customer 1, its marginal value to Customer 2, and its marginal value to Customer 3. So the total MV curve is the vertical sum of the MV curves for the customers, each of which equals the corresponding demand curve. The result is shown on the figure. Q* is the efficient quantity.
Public Production of Public Goods. One obvious solution to the public-good problem is to have the government produce the good and pay for it out of taxes. This may or may not be an improvement on imperfect private production. The problem is that the mechanism by which we try to make the government act in our interest--voting--itself involves the private production of a public good. As I pointed out earlier in this chapter, when you spend time and energy deciding which candidate best serves the general interest and then voting accordingly, most of the benefit of your expenditure goes to other people. You are producing a public good: a vote for the better candidate. That is a very hard public good to produce privately, since the public is a very large one: the whole population of the country. Hence it is underproduced--very much underproduced. The underproduction of that public good means that people do not find it in their interest to spend much effort deciding who is the best candidate--which in turn means that democracy does not work very well, so we cannot rely on the government to act in our interest.
If we cannot rely on the government to act in our interest, we cannot rely on it to produce the efficient quantity of public goods. Just as with a government agency regulating a natural monopoly, the administrators controlling the public production of a public good may find that their own private interest, or the political interest of the administration that appointed them, does not lead them to maximize economic welfare.
Even if the government wishes to produce the efficient amount of a public good, it faces problems similar to the problems of regulators trying to satisfy the second efficiency condition. In order to decide how much to produce, the government must know how much potential consumers value the good. In an ordinary market, the producer measures the demand curve by offering his product at some price and seeing how many he sells. The producer of a public good cannot do that, since he cannot control who gets the good, so the government must find some indirect way of estimating demand. Individuals who want the public good have an incentive, if asked, to overstate how much they want it--which means that a public opinion poll may produce a very poor estimate of demand.
In dealing with the public-good problem, just as in dealing with the closely related problem of natural monopoly, we are faced with a choice among different imperfect ways of solving the problem, some private and some governmental. None of the alternatives can be expected to generate an efficient result. As I pointed out earlier, the fact that something is inefficient means that it could be improved by a bureaucrat-god. That does not necessarily mean that it can be improved by us, since we do not have any bureaucrat-gods available.
As you may have realized by now, public-good problems of one sort or another are very common--indeed many common problems, both public and private, can be viewed as public-good problems. One example is the problem of getting anything accomplished in a meeting. Most of us like attention: When we are in a meeting and happen to have the floor, we take the opportunity not only to say what we have to say about the issue on hand but also to show how clever, witty, and wise we are. This imposes a cost on other people (unless we really are witty and wise); if there are sixty people in the room, every minute I speak costs a person-hour of listener time. Brevity, in this case, is a public good--and underproduced.
At the beginning of this section, I mentioned that different economists use slightly different definitions of a public good. The definition I have used emphasizes non-excludability: the inability of the producer to control which consumers get the good. The other characteristic usually associated with a public good is that one person's use does not reduce the amount available for someone else. A different way of stating this is to say that the marginal cost of producing the good is zero on the margin of how many people get it, although there may still be a cost to producing more on the margin of how much of it they get. Something that is a public good in only this sense (it has zero marginal cost, but the producer can control who gets it) is simply a natural monopoly with MC = 0. Since the problems associated with natural monopoly have already been discussed, I prefer to concentrate on the inability of the producer to control who consumes the good, which seems to me to be the essential characteristic of public goods responsible for the special problems associated with them.
The long-winded speaker is underproducing the public good of brevity. Another, and equivalent, way of describing the situation is to say that he is overproducing his speech. The problem can be described either as underproduction due to the public-good problem or as overproduction due to the existence of an externality.
An externality is a net cost or benefit that my action imposes on you. Familiar examples--in addition to the cost of listening to me talk too long in a meeting--are pollution (a negative externality--a cost) and scientific progress as a result of theoretical research (a positive externality--a benefit). Externalities are all around us: When I paint my house or mow my lawn, I confer positive externalities on my neighbors; when you smoke in a restaurant or play loud music in the dorm at 1:00 a.m., you confer negative externalities on yours.
The problem with externalities is that since you, rationally enough, do not take them into account in deciding whether or not to smoke or play the music, you may do so even when the total cost (including the cost to your neighbors) is greater than the total benefit. Similarly, I may fail to mow my lawn this week because the benefit to me is less than the cost, even though the total benefit (including the benefit to my neighbors) is more.
As you can see by these examples, "externalities" and "public goods" are really different ways of describing the same problems. A positive externality is a public good; a negative externality is a "negative" public good and refraining from producing it is a positive public good. In some cases, it may be easier to look at the problem one way, in some cases the other--but it is the same problem.
Figure 18-2 is a graphical analysis of the inefficiency due to an externality. D is the demand curve for steel. S is the industry supply curve for the competitive industry that produces steel. The industry produces a quantity Qs at a price Ps.
In addition to the costs that the industry pays for its inputs, there is another cost to producing steel: pollution. For every ton of steel it produces, the industry also produces a negative externality of $10. So the true marginal cost of a ton of steel is $10 above the marginal private cost, the cost to the industry, which is what determines the industry's supply curve. S' is what the supply curve would be if the industry included in its calculations the cost of the pollution it produced. The efficient level of output is where marginal cost equals marginal value--where S' intersects D at a quantity of Qs'. From the standpoint of efficiency, the situation is exactly as if the supply curve were S' but the industry, for some reason, produced Qs. The resulting inefficiency is the colored area A on the figure. The society as a whole--producers, consumers, and victims of pollution--is that much worse off than if the firms produced the efficient quantity Qs'.
So far we have assumed that the only way of reducing pollution is to reduce the amount of steel produced. There may be other alternatives. By filtering its smokestacks or using low sulfur coal, the firm may be able to eliminate a dollar's worth of pollution at a cost of less than a dollar.
Supply and demand for steel. (Qs, Ps) is the uncontrolled equilibrium. S' is what the supply curve would be if the steel firms included in their cost calculations the cost ($10/ton) that their pollution imposes on others. S'' is what it would be if the firms included pollution cost, but reduced it by purchasing the efficient level of pollution control, as shown on Figure 18-3.
Figure 18-3 shows that possibility. For simplicity, I assume that the cost for a given reduction of pollution per ton is proportional to the amount of steel the firm is producing. TC is the total cost function for producing pollution control. It shows how many dollars must be spent on pollution control per ton of steel produced in order to reduce pollution per ton to any particular level. MC is the corresponding marginal cost function, showing the cost of the additional pollution control required to eliminate an additional dollar's worth of pollution.
As we already know, the efficient quantity of output occurs where marginal cost equals marginal value. The value of eliminating $5 worth of pollution is $5; marginal value is $1 per dollar. The steel firm should keep increasing its expenditure on pollution control until the last dollar buys exactly a dollar's worth of pollution control. The efficient amount of pollution per ton is at Qp, where MC crosses MV.
Cost curves for controlling pollution. TC shows total cost/ton of steel for reducing pollution as a function of the amount of pollution produced. MC is the corresponding marginal cost, MV the marginal value of pollution control.
If steel firms install the efficient level of pollution control, they will spend $2.50/ton on pollution control and produce $4.50/ton worth of pollution. The cost of producing steel, including the cost to the producers of controlling pollution and the cost to everyone else of the pollution they do not control, is $7/ton higher than the cost to the firms of producing steel with no pollution control. The corresponding supply curve is S'' on Figure 18-2. The efficient quantity of steel is Qs''.
What is the efficiency loss from producing Qs without pollution control instead of Qs'' with pollution control? Producing Qs without pollution control instead of Qs' without pollution control costs area A. Producing Qs' without pollution control instead of Qs'' with pollution control raises the supply curve from S'' to S' and moves quantity from Qs'' to Qs', so it costs the shaded area B, the resulting change in total surplus. Going from S' and Qs to S' and Qs' saves area A; going from there to S'' and Qs'' saves an additional area B; so the net savings in moving from the initial inefficient outcome to the final efficient one is A+B. That is the inefficiency of the uncontrolled outcome, compared to the outcome that would be chosen by a bureaucrat-god.
Efficient Pollution and How to Get It: The Public Solution. The textbook solution to externalities is to impose the cost on, or give the benefit to, the producer. If I am benefiting others by scientific research, subsidize me; if I am polluting the air, charge me an effluent fee of so many dollars per cubic foot of pollution emitted, corresponding to the costs that my pollution imposes on others. I will continue to pollute only if the net value of what I am doing is more than the damage done--in which case, pollution is efficient. If each steel firm must pay an effluent fee of $1 for each dollar's worth of pollution it produces, the supply curve for steel will shift to S'' and the quantity produced will be Qs''. The industry will produce an efficient amount of steel--and an efficient amount of pollution.
"Pollution" is a loaded word. To be in favor of pollution sounds like being in favor of evil; the phrase "an efficient level of pollution," lifted from a book like this one, would be fine ammunition for a speech on the inhumanity of economics--and economists.
If you find the idea that some amount of pollution is desirable a shocking one, consider that carbon dioxide is commonly regarded as a pollutant, and the only way you can stop producing it is to stop breathing. This is an extreme case, but it makes an important point--that the real issue is whether, in any particular case, the costs of pollution are greater than the costs of not polluting.
While there is, in this sense, an efficient level of pollution, it is not clear how to get that level. The problem with using effluent fees to control externalities is the same as the problem with government provision of public goods; it depends on the government finding it in its interest to act in the interest of the public and knowing how to do so. Just as in previous cases, "knowing how" includes somehow estimating the value of something to people by some method other than offering it at a price and seeing whether they take it. The result of the governmental solution may be better or worse than the alternatives of either accepting the overproduction of negative externalities and the underproduction of positive ones, or dealing with the problem in some imperfect private way.
Private Solutions. How might one control externalities privately? One (real-world) solution is a proprietary community. A developer builds a housing development and sells the houses with the requirement that the buyer must join the neighborhood association. The neighborhood association either takes care of lawns, painting, and other things that affect the general appearance of the community or requires the owners to do so. A friend of mine who lived in such a community could not change the color of his front door without his neighbors' permission.
This sounds rather like government regulation masquerading as a private contract, but there are two important differences. It is in the private interest of the developer to set up the best possible rules, in order to maximize the price for which he can sell the houses. And nobody is forced to purchase a house and membership from that developer; if the package is not at least as attractive as any alternative, the customer can and will go elsewhere.
There is another private solution that applies to the case where "You" and "I" are not two people but two firms--merger. If a factory and a resort are both on the same lake and the factory's pollution is ruining the resort's business, one solution is for the two firms to join. After the resort buys out the factory, or vice versa, the combined firm will be trying to maximize the combined income. If controlling the factory's effluent increases the resort's income by more than it costs the factory, it will pay the merged firm to control the effluent. The externality is no longer external.
One way of looking at firms is precisely as ways of controlling such problems. As I pointed out back in Chapter 7, one could imagine an economy of tiny firms, perhaps with only one person in each, coordinating their activities through the market. One reason we do not do things that way is that, when many firms are jointly producing a single product, decisions by each one affect all the others. If I am doing a crucial part of the job and make a mistake that delays it for six months, I am imposing large costs on the other firms--which I may not be able to compensate them for. By combining all of us into one firm, that sort of externality is internalized.
The disadvantage of doing it that way is that we introduce a new kind of externality. Now that I am an employee instead of an independent business, the cost of my sleeping on the job is borne by everyone else. So a firm must monitor its employees in ways in which it does not have to monitor other firms. The efficient size of firm is then determined by the balance between problems associated with coordinating a lot of small firms and problems associated with running one large firm.
Another solution to externality problems is the definition and enforcement of property rights in whatever is affected by the externality. It is in one sense a governmental solution, since property rights are defined by courts and legislatures, and in another sense a private solution, since once property rights are defined it is the market and not the government that decides what happens. An example is the case of British trout streams. Trout streams in Britain are private property. Each stream is owned by someone--frequently the local fishing club. An industrial polluter dumping effluent into such a stream is guilty of trespass, just as if he dumped it on someone's lawn. If he believes the stream is more valuable as a place to dump his effluent than as a trout stream, it is up to him to buy it. If he believes (and the fishing club does not) that his effluent will not hurt the trout, he can buy the stream and then--if he is right--rent the fishing rights back to the previous owners.
As this example suggests, what is or is not an externality depends in part on how property rights are defined. When I produce an automobile, I am producing something of value to you. It is not an externality because I can control whether you get it and will refuse to give it to you unless you pay me for it. Some externality problems arise because property rights are not defined when they should be: If land were not property, my fertilizing it or planting a crop would confer positive externalities on whoever later came by and harvested my crop. Under those circumstances, crops would not be planted. Other problems arise because there is no way of defining property rights that does not lead to externalities in one direction or another. If I have to get your permission to play my stereo when you want to sleep, I can no longer impose an externality on you--but your decision to go to sleep when I want to play my stereo imposes an externality on me! If only two people are involved, they may be able to work out an efficient arrangement by mutual negotiation--but air pollution in Los Angeles affects several million people. Just as in the case of producing a public good, the problems of negotiating a unanimous contract become larger the larger the number of people involved.
One way of looking at this is to say that all public-good/externality problems are really transaction-cost problems. If bargaining were costless, then the problems leading to inefficiency could always be solved. As long as there was some change that would produce net benefits, someone could put together a deal that would divide up the gain in such a way as to benefit all concerned. This argument has a name--it is called the Coase Theorem (after economist Ronald Coase). Looked at in this way, the interesting question is always "What are the transaction costs that prevent the efficient outcome from being reached?"
Joint Causation, or Why Not Evacuate Los Angeles?
Half of Coase's contribution to understanding externalities was the observation that the problem would vanish if bargaining between the affected parties were costless; the problem could thus be seen as the result not of externalities but of transaction costs. The other half was the observation that the traditional analysis of externalities contained a fundamental error.
So far we have followed the pre-Coasian analysis in treating an externality as a cost imposed by one person on another. That is not quite right. As Coase pointed out, the typical externality is a cost jointly produced by the actions of both parties. There would be no pollution problem in Los Angeles if there were no pollution, but there would also be no problem, even if there were lots of pollution, if nobody tried to live and breath in Los Angeles.
If evacuating Los Angeles does not strike you as a very satisfactory solution to the problem of smog, consider some more plausible examples. The military owns bomb ranges: pieces of land used to test bombs, artillery shells, and the like. If you happen to be camping in one, the dropping of a three hundred pound bomb next to your tent imposes serious externalities. It seems more natural to solve the problem by removing the campers than by removing the bombs.
Another example is airplane noise, which can be a considerable problem for people who live near large airports. One approach to the problem is to modify planes to make them quieter, close the airport when people are asleep, and instruct pilots to begin their descent as near the airport as possible. An alternative is to soundproof the houses near the airport. Another alternative is not to have anyone living near the airport: keep the land empty, use it for a water reservoir, or fill it with noisy factories where no one will notice the minor disturbance produced by a 747 two hundred feet over the roof.
It is not immediately obvious which of these alternatives provides the most efficient way of dealing with airport noise. If we try to solve the problem by the equivalent of an effluent fee (more generally described as a Pigouvian tax, after A.C. Pigou, the inventor of the traditional analysis of externalities), we may never find out. Charging the airlines for the cost of the noise they produce give them an incentive to reduce noise, but that may be the wrong solution--it might be less costly to soundproof the houses or pay their occupants to move out.
The problem is that the cost is jointly produced by the actions of both parties. If we do nothing, the cost is entirely born by one party (the homeowners in our example) so the other has no incentive to reduce it--even if he can do so at the lower cost. If we impose a Pigouvian tax on the "polluter," the "victim" may find that his best tactic is to do nothing--even if he is the one who can solve the problem at the lower cost. If, as a third alternative, we let the victim sue the polluter, the victim has no incentive at all to avoid (or reduce) the cost--whatever he loses he gets back in damage payments. Any of the alternatives might or might not give the efficient outcome, depending on whether it happens to impose the externality on the party who can avoid it at the lowest cost. If the efficient solution requires actions by both parties--soundproofing plus some noise reduction, for example--none of the alternatives may be able to produce it.
What lessons can we learn from this depressing tangle? The first is that the traditional analysis of externalities, and the associated solution of Pigouvian taxes, applies only to the special case where we already know which party is the least cost avoider of the problem--that emission controls for automobiles in Southern California cost less than evacuating that end of the state. The second is that in the more general situation, where we do not know who can solve the problem at lowest cost, the best solution may be to fall back on Coase's other idea: negotiations between the parties. If the airlines are liable for damage produced by noise pollution, they may choose to pay people living near the airport to soundproof their houses. They may even choose to buy the houses, tear them down, and rent out the land to people who want to build noisy factories. If the airlines are not liable for damages, it may be in the interest of the local homeowners to offer to pay the cost of noise reduction if that is cheaper than soundproofing. So the best solution to such problems may be for the legal system to clearly define who has the right to do what and then permit the affected individuals to bargain among themselves.
In defining the initial rights--in deciding, for instance, whether the airlines have the right to make noise or must buy that right from the homeowners--one should consider the transaction costs of getting from each possible definition to each possible solution. If there are 10,000 homeowners living near the airport, raising money to pay the airlines to keep down their noise will be a public good for a public of 10,000; so it will almost certainly not be produced, even if it is worth producing. If the airlines have the right to make noise and not pay damages, they will continue producing noise whether or not it is efficient--homeowners will put up with the sound, soundproof, or sell out. If the airline is permitted to make the noise but must pay damages to affected homeowners, the airline can negotiate separately with each homeowner, buying or soundproofing some houses and paying damages on the rest if that is cheaper than modifying the planes--which should ultimately lead to the efficient solution.
Suppose, as another alternative, that each homeowner has an absolute right to be free from noise. In that case it does the airline no good to soundproof houses or buy them unless all 10,000 are included. The result is a holdout problem. Any one homeowner can try to get the airline to pay him the entire savings from soundproofing the houses instead of the planes, by threatening to withhold his consent. With 10,000 homeowners, every one of whom must agree, the deal is unlikely to go through--even if it is the lowest cost solution to the problem.
In this particular case, the best solution may be a legal rule permitting homeowners to collect damages but not to forbid the noise. That allows whichever of the three solutions turns out to be most efficient to occur with either no transaction (the airline reduces its noise) or a relatively simple and inexpensive one (the airline deals separately with the homeowners who are willing, and pays damages to the holdouts). This solution depends, however, on the damage done by the noise being something a court can measure. One can imagine many cases where that would not be the case, and where a different rule might be more likely to lead to an efficient outcome.
Externalities can be eliminated by a contractual arrangement, as when two firms merge or when I agree, in exchange for a payment, not to do something that injures you. Externalities can also be created by contract. One example that I will discuss a little later is the case of insurance. By purchasing fire insurance, I create an externality: If I am careless with matches, part of the cost will be borne by the insurance company instead of by me. A second example is the case of sharecropping.
Sharecropping means that a farmer pays, instead of rent, a fixed percentage of his crop to the owner of the land he farms. It seems an odd and inefficient arrangement. If I must pay half of my crop to my landlord, it only pays me to make investments of labor or capital if the payoff is at least twice the cost. I have, by contract, created an externality of 50 percent.
This raises an obvious puzzle. Sharecropping is a common arrangement, appearing in many different societies at different times in history. If it is inefficient, why does it exist?
One way of answering the question is to consider the alternatives. There are two obvious ones. The landlord could hire the farmer to work his land, paying him a fixed wage, or the farmer could pay a fixed rent to the landlord and keep all of the crop.
Converting the sharecropper into an employee is hardly a solution; instead of collecting half the return from additional inputs of labor he collects none of it. Switching from sharecropping to renting may be a solution, but it has some problems. For one thing, farm output may vary unpredictably from year to year. If the farmer has agreed to pay a fixed rent, he does very well in good years but may starve to death in bad ones--the rent may be more than the full value of his crop.
Seen from this standpoint, sharecropping is, like insurance, a device for spreading risk. The landlord and the farmer divide the risk, instead of the farmer taking all of it. If the random factors affect different pieces of land differently, a landlord who owns several pieces of land can expect random effects to average out, just as they do for an insurance company. When there is lots of rain he gets very little from tenants farming low-lying areas, which flood, but lots from tenants farming hilltops that are usually too dry to grow much. Just as with insurance, the two parties pay a price in inefficiency due to externalities in order to get a benefit in risk spreading.
One way of reducing that price is for the landlord to monitor the farmer--just as he would do if the farmer were an employee. If he concludes that the farmer is not working hard enough there is nothing the landlord can do this year, but he can find another sharecropper next year. Sharecroppers require more monitoring than tenants but less than employees, since they get at least part of the output they produce.
Another explanation for sharecropping, at least in some societies, may be that the landlord is also contributing inputs: experience, administration, perhaps capital. If so, giving him a fraction of the output reduces the farmer's incentive but increases the landlord's. In this case as in many others, there may be no efficient contract--no contract that does as well as rule by a bureaucrat-god. Just as in choosing firm size, or controlling externalities that are jointly caused, or picking a rule for product liability, choosing the optimal contract involves tradeoffs among different imperfect ways of coordinating individuals whose actions are interdependent.
Suppose something I do imposes both positive and negative externalities, and by some coincidence they are exactly equal. I will, as always, treat the external costs and benefits as if they were zero--and in this case, I will be right. Since on net all of the costs and benefits caused by my action are borne by me, I will make the efficient decision as to whether or not to do it.
One would think it an unlikely coincidence for positive and negative externalities to precisely cancel; but there is an important situation, called a pecuniary externality, in which that is exactly what happens. Whenever I decide to produce more or less of some good, to enter or leave some profession, to change my consumption pattern, or in almost any other way to alter my market behavior, one result is to slightly shift some supply or demand curve and so to change some price; this affects all other buyers and sellers of the good whose price has changed. In a competitive market, the change in price due to one person's actions is tiny--but in calculating the size of the effect, one must multiply the small change in price by the large quantity of goods for which the price has changed--the entire market. When, for example, I decide to become the million and first physician, the effect of my decision in driving down the wages of each existing physician is tiny, but it must be multiplied by a million physicians. The product is not necessarily negligible.
It appears that there can be no economic action without important externalities. But these are precisely the sort of externality that can be ignored. When price falls by a penny, what is lost by a seller is gained by a buyer; the loss to the physicians is a gain to their patients. The result is a pecuniary externality. My decision to enter a profession, to buy or to sell goods, may have more than a negligible effect on others through its effect on the price of goods or services they buy or sell, but that effect imposes neither net costs nor net benefits, so ignoring it does not produce an inefficient outcome.
Religious Radio: An Application of Public-good Theory
Whenever I spend much time listening to a variety of stations on the radio, I am struck by how many of them are religious. One could take this as evidence that America is a very religious country--except that the popularity of religion on the airwaves does not seem to be matched elsewhere. If I go to a newsstand or a bookstore, I see relatively few religious newspapers, magazines, or books--far fewer, as a percentage of the total, than radio programs.
There is a simple explanation for this discrepancy. Publishers can control who gets their publications; broadcasters cannot control who listens to their broadcasts. Broadcasters, unlike publishers, are producing a public good and depend on some solution to the problem of producing a public good privately in order to stay in business.
Commercials are one solution to that problem; religion is another. The people who listen to religious broadcasters presumably believe in the religion. For most of them, that means that they believe in the existence of a god who rewards virtue and punishes vice. If, as many radio preachers claim, donating money to their programs is a virtuous act, then the program is no longer a pure public good. The preacher may not know which listeners help pay for the show and which do not, but God knows. One of the benefits produced by the program is an increased chance of a heavenly reward; you are more likely to get that benefit if you pay for it. Thus religion provides a solution to the public-good problem.
Nothing in the analysis depends on whether the particular religion is or is not true; what matters is only that the listeners believe it is true and act accordingly. The result is that religious broadcasters have an advantage over secular broadcasters. Both produce programs that their listeners value, but the religious broadcaster is better able to get the listener to pay for them. The religious publisher has no corresponding advantage over the secular publisher. So religion is more common on the air than in print.
Long ago and far away--in Chapter 1, to be precise--I pointed out an ambiguity in the definition of "rational." In some contexts a rational individual was one who made the right decision, the decision he would have made if he knew all of the relevant facts, in other contexts a rational individual was one who made the right decision about what facts to learn and then the best possible decision in the context of what he knew. I suggested that the latter definition is appropriate in situations where an essential part of the problem is the cost of getting and using information.
It is tempting to argue that information costs are simply one of the costs of producing and consuming goods, and so can be included in our analysis just like any other costs. In some situations that argument is correct. But, as we will see in this part of the chapter, information costs are frequently associated with problems that lead to market failure.
Information as a Public Good
One cost of buying goods is the cost of acquiring information about what to buy. This may be one reason firms are as large as they are; brand names represent a sort of informational capital. There may be a better deal available from an unknown producer, but the cost of determining that it is a better deal may be greater than the savings. Not only do you know that the brand-name product has been of good quality in the past, you also believe that the producer has an incentive to maintain the quality so as not to destroy the value of his brand name.
Why do we rely on brand names instead of buying information about the quality of goods from someone who specializes in producing such information? To some extent, we do buy information: by reading Consumer Reports, Car and Driver, or Handgun Tests and by taking economics courses. Yet much of the information we use we produce for ourselves--probably a much larger fraction than of most other things we consume. Since we do not have the time to become experts on everything we buy, we end up depending on brand names and other indirect (and very imperfect) ways of evaluating quality.
Why do we produce so much information for ourselves? Why is information a particularly hard good to produce and sell on the market?
The problem is that it is hard to protect the property rights of a producer of information. If I sell you a car, you can resell it only by giving up its use yourself. If I sell you a fact, you can both use that fact and make it available to all your friends and neighbors. This makes it difficult for those who produce facts to sell them for their full value. It is the same problem that I earlier discussed in the case of computer programs--which can be thought of as a kind of information. Information is in large part a public good; because it is a public good, it is underproduced.
One solution to this problem is provided by large brand-name retailers such as Sears. Sears does not produce what it sells, but it does select it. You may buy any particular product only once every year or two, which makes it hard to judge which producer is best. But you buy something from Sears much more often, so it is easier for you to judge that Sears (or one of its competitors) "on average" gives you good value for your money. Sears is in the business of learning which brands of the products it buys represent good value for the money and selling them to you under its brand name, thus implicitly selling you the information. By not telling you who really makes the product, it prevents you from reselling the information--to a friend who would then buy the same brand at a discount store. All you can tell your friend is to buy from Sears--which is fine with Sears.
Consider a situation where information is not merely imperfect but asymmetrical. The market for used cars may be a good example. The best way of finding out whether a car is a lemon is to drive it for a year or two. The seller of a used car has done so; potential buyers have not. While they can, at some cost, have the car examined by a mechanic, that may or may not be sufficient.
Suppose, to simplify our analysis, that there are only two kinds of cars: good cars and lemons. There are also two kinds of people: sellers and buyers. Each seller has a car, which he is interested in selling if he can get a reasonable price. Half have good cars; half have lemons. Each buyer would like to buy a car--if he can get it for a reasonable price. Sellers know what kind of car they have; buyers do not.
Both buyers and sellers prefer good cars to lemons. Sellers value lemons at $2,000 and good cars at $4,000--at any price above that they are willing to sell. Buyers value lemons at $2,500 and good cars at $5,000--at any lower price they are willing to buy. It appears that all of the cars should sell--lemons for between $2,000 and $2,500, good cars between $4,000 and $5,000.
There is a problem. Buyers cannot, at a reasonable cost, tell whether a car is a lemon. The sellers know, but have no way of conveying the information, since it is obviously in the interest of every seller to claim that his car is a good one. So each buyer is buying a gamble--some probability of getting a good car and some probability of getting a lemon.
It looks as though the probabilities are 50-50, since half the cars are lemons. If so, and if the buyers are risk-neutral, they will offer no more than the average of the values of the two kinds of cars, which is $3,750. At that price, owners of lemons will be glad to sell, but owners of good cars will not.
The buyers can work out the logic of the preceding paragraph for themselves. While a car offered for sale has a 50 percent chance of being good, a car that is actually sold is certain to be a lemon, since owners of good cars will refuse the best offer buyers are willing to make. Buyers take that fact into account, and reduce their offers accordingly. All of the cars are worth more to the buyers than the sellers, but only the lemons get sold. That is an inefficient outcome. In more complicated situations, with a range of qualities of cars, the result may be even worse; in some cases only the single worst car gets sold.
One obvious solution is for sellers with good cars to offer a guarantee--perhaps a guarantee to buy it back a year later for purchase price minus a year's rental if the buyer decides the car is a lemon. One problem with this solution is that the condition of the car a year hence depends on a lot of things other than its condition today, including how it is treated by its new owner.
The problem I have just been describing is known, in the context of insurance markets, as adverse selection. Consider health or life insurance. The customer has information about himself that the insurance company cannot easily obtain: how carefully he drives, what medical problems he has had in the past, whether he is planning to take up hang gliding, skydiving, or motorcycle racing in the near future. The more likely a potential customer is to collect on his insurance the greater its value to him--and its cost to the insurance company. If the customer knows he is a bad risk and the insurance company does not, insurance is a good deal--for the customer.
The good risk would be happy to buy insurance at a price reflecting the low probability that he will get sick or die next year, but the insurance company will not offer it to him at that price, since the insurance company does not know he is a good risk. The result is that bad risks are more likely to buy insurance than good risks. Insurance companies, knowing that, must adjust their rates accordingly--the very fact that someone buys insurance is evidence that he is a bad risk and should therefore be charged a high price. The higher price results in even fewer good risks buying insurance--resulting in an even higher price. The equilibrium result may well be that many good risks are priced out of the market, even though there is a price at which they would be willing to buy insurance and the insurance companies would gain by selling it to them. Just as with automobiles, one can even construct a situation where only the worst risks end up insured, everyone else having been driven out of the market. Again we have an inefficient outcome.
Insurance companies try to control this problem in a variety of ways, including medical checkups for new customers and provisions in insurance contracts denying payment to people who say they have no dangerous hobbies and then die when their parachutes fail to open two miles up. A less obvious solution is selling insurance to groups. If all employees of a factory are covered by the same insurance, the insurance company is getting a random assortment of good and bad risks. The good risks get a worse deal than the bad, but since they still get insured the insurance rates reflect the risk of insuring an average employee rather than an average bad risk. If insuring everyone is the efficient outcome, the group policy produces an efficient allocation of insurance, plus a redistribution of income from the good risks, who are paying more than their insurance costs to produce, to the bad risks who are paying less.
One argument in favor of universal, governmentally provided health insurance is that it is a group policy carried to its ultimate extreme--everyone is in the group. It thus eliminates the problem of adverse selection (except, perhaps, for people with health problems who decide to immigrate in order to take advantage of the program). Whether the net effect is an improvement depends on how well the government can and does deal with other problems of providing insurance.
It may have occurred to you that there is another potential inefficiency associated with insurance. Most of the things we insure against are at least partly under our own control. That is true not only of my health and the chance of my house burning down, but even of losses from "acts of God" such as floods or tornadoes. I cannot control the flood, but I can control the loss--by deciding where to live and what precautions to take.
Whether or not I am insured, I take those precautions, and only those precautions, that save me more than they cost me. Once I have bought fire insurance, part of the cost of being careless with matches and part of the benefit of installing a sprinkler system have been transferred to the insurance company; the cost to me is no longer the entire cost, so the result is no longer efficient. If a sprinkler system costs $1,000 and produces a benefit of $800 to me in reduced risk of being burned alive and another $600 to the insurance company in reduced probability of having to replace my house, it is worth buying--but not to me.
So people who are insured will take less than the efficient level of precaution. This problem is known as moral hazard. It is an inefficiency resulting from an externality; once I am insured someone else bears some of the cost of my actions.
Insurance companies try to control moral hazard just as they try to control adverse selection. One way is by specifying, so far as possible, the precautions that the insured will take--requiring a factory to install and maintain a sprinkler system as a condition of providing fire insurance. Another is co-insurance--insuring for only part of the value, in order to make sure that the customer has at least a substantial stake in preventing the risk that is insured against. If, in my previous example, the house was insured for only half its value, the sprinkler system would be worth more to me than it cost, so I would buy it. If, at the opposite extreme, the insurance company makes the mistake of insuring a building for more than it is worth, the probability of a fire may become very high indeed.
In thinking about market failure, it is often tempting to interpret the problem in terms of fairness rather than efficiency. Externalities are then seen as wrong because they are unfair, because one person is suffering and another gaining, and public goods as a problem because some consumers get what others pay for.
That is a mistake. Consider the situation of a hundred identical individuals polluting and breathing the same air. On net there is no unfairness--everyone gains by being able to pollute and loses by being polluted. Yet because each person bears only 1 percent of his pollution, each pollutes at far above the efficient level and all are, as a result, worse off. This is precisely analogous to the effect of the potato subsidy discussed in Chapter 3; everyone gets back in subsidy as much as he pays in taxes yet ends up worse off, not because he is poorer but because he is buying too many potatoes.
The same is true for the other kinds of market failure. The ultimate problem with public goods is not that one person pays for what someone else gets but that nobody pays and nobody gets, even though the good is worth more than it would cost to produce. The major cost of adverse selection is not that some people buy lemons or write life insurance policies on skydivers. The major cost is that cars are not sold, even though they are worth selling, and people do not get insured, even though they are worth insuring.
1. Describe two public-good problems that you have yourself observed and in some way been involved with in the past year and discuss how they might be dealt with; you should not use any that are discussed in the chapter.
2. In ordinary markets, supply and demand are balanced by price. Given that our customs prohibit, in most social contexts, cash payments as part of a date (or a marriage), what sorts of "prices" balance those markets in the United States at present? If supply and demand on the dating/sex/marriage market are not balanced (quantity supplied is not equal to quantity demanded: more men want to go out or have sex or get married than women, or vice versa), what mechanisms ration out the insufficient supply (decide which men get women, or vice versa)? What prices balance supply and demand for similar markets in other countries or have done so at other times?
3. How would the style of dating and marriage change if a war substantially reduced the ratio of men to women? How would it change if a lot of men migrated to the United States, substantially raising the ratio of men to women?
4. "Heterosexual men are traditionally hostile to homosexual men. If they correctly considered their own interests, their attitude would be just the opposite." Discuss.
5. "The public-good problem is both an argument for government intervention in the market and an argument against government intervention in the market." Explain.
6. Students frequently argue that grades should be deemphasized or abolished. The same students start the first class of the quarter by asking me about my grading policy--and continue throughout the course to exhibit a keen interest in what will or will not be on the final exam. Is their behavior inconsistent?
7. A tape recorder can copy a recording of a concert onto a cassette just as a computer can copy a program onto a disk. Why is the problem of pirating (making copies without paying royalties) less serious in the case of tapes than in the case of programs?
8. In my experience, FM radio is less religious than AM; you may wish to check that conclusion for yourself. Can you suggest any reasons why? (I am not sure I know the answer to this one.)
9. Last year Bryan and Brian occupied separate apartments; each consumed 400 gallons per month of hot water. This year they are sharing a larger apartment. To their surprise, they find they are consuming 1,000 gallons per month. Explain.
10. One of my students cannot possibly take the midterm at the scheduled time. I am afraid that if I give it to him early, he might talk about it to other students, giving them an unfair advantage, and that if I give it to him late, other students might talk to him about it, giving him an unfair advantage. Given the problems associated with property in information, which problem do you think is more likely to arise? Discuss. Does it depend on whether the students believe that I grade on a curve?
11. The following table shows, for three different goods (produced by three different firms), total cost of production (as a function of quantity produced), total external cost imposed by producing the goods, and their total value to the consumer. Assume the manufacturer can sell the goods at their value, which is the same for all three goods ($10/unit). He must pay the cost of production but does not have to pay for the external cost. Fractional units cannot be produced; output can be 0,1,2,3, ... but not 2 1/2.
a. How much does each firm choose to produce?
b. In which, if any, cases is the outcome efficient?
c. In the inefficient cases, how large would the net gain be if the firm was forced to produce the efficient level of output instead of the profit-maximizing level?