Class Notes for Lecture 4: Public Goods Instructor: Taejong Kim

Public Goods dened

A public good is a commodity that is nonrival in consumption. In contrast, with private goods, individual consumers rival each other for consumption. If I eat my apple, the apple will be gone and you cannot consume it, too, and vice versa. A classic textbook example of a public good is a lighthouse. If my ship benets from the service of a lighthouse, it does not diminish the benet that your ship may enjoy from the same lighthouse. If a good is perfectly nonrival in consumption, then we call the good a pure public good. Examples of pure public goods include national security and provision of law and order. You may recall that these two topped Adam Smiths list of three essential functions to be provided by government. Examples of impure public goods include beach area, public library, and highways. Impure public goods are nonrival in consumption if there are a relatively small number of consumers, but as the number of consumers increases, congestion leads to a decrease in benets enjoyed by individual consumers. Note that a public good does not mean a commodity provided by the public sector. Public goods may be provided by the private sector. For instance, we have parks and other public amenities donated by private citizens. The public sector may also provide private goods. Public education is a case in point. Seats in a public school are private in the sense that a seat in a class taken by a student cannot be taken by another.

You may note that several textbooks in public nance, including our Rosen, list two characteristics to dene a public good: nonrivalness and nonexcludability. Of course, my denition above specied nonrivalness, but not nonexcludability, to dene public goods. In this treatment, I followed some leading public nance economists who believe nonrivalness to be the essential characteristic of public goods. Let me explain why. For most public goods (featuring nonrivalness in consumption), it is true that it is hard, if not impossible, to exclude the non-paying members of the community from enjoying the benets nonetheless. It would be dicult indeed to exclude some individuals from enjoying the benets from well-established law and order, for instance. Some of the eciency problems we often associate with public goods, such as the problem of freeridership, arise from nonexcludability. New technologies may arise to make a certain good, so far nonexcludable, excludable. Cable channels are available only for the fee-paying households, while public broadcasting is to be enjoyed by everyone, whether paying or not. New electronic technologies may allow only fee-paying vehicles into a highway or a busy section in downtown. These new technologies will block free riders from using the services. Suppose a nonrival good has turned into an excludable good by new technologies. Does this mean that eciency problems will disappear? No. As long as nonrivalness remains, the marginal cost of letting one more consumer benet from the service remains zero (or low for impure public goods). If a feecharging entrepreneur turns away consumers whose willingness to pay is positive but smaller than the set fee, this means loss in social surplus, an inecient out-

come. It is in this sense that I consider nonrivalness as the dening characteristic, but not nonexcludability. Consider creation and distribution of knowledge for instance. Knowledge costs money to create, but, once it is created, its distribution costs little. Some knowledge may be nonrival in consumption. In this case, the marginal cost of letting in one more consumer into the knowledge is zero. Eciency requires then free distribution of knowledge. Blocking access to knowledge for non-payers would be socially inecient. Of course, we have a dilemma here. If knowledge is to be distributed freely, then incentives to create knowledge in the rst place may be weakened. A patent system is a response to this dilemma. By guaranteeing a temporary monopoly, patents encourage creation of knowledge. At the expiry of patents, the knowledge will be made freely available to the public.1

Ecient Provision of Public Goods

Due to nonrivalness, it is important to note, consumers are restricted to consuming the same amount of a public good.2 Adam and Eve do not have to eat the same number of apples, but if they are going to enjoy reworks, they have
1 A precursor to the modern patent system was the British governments eort to encourage development of a precision clock, or chronometer, in the seventeenth century. Navigation in those days critically relied on accurate measurement of time, a matter of life and death to sailors. The government promised a substantial sum of prize money to the rst inventor to meet the pre-specied quality standards. When the invention nally arrived, the government distributed its copies to the ships in the British eet at a low cost.

2 An individuals consumption of a public good thus generates positive externalities to be enjoyed by fellow consumers. A public good is a special example of a positive externality where consumers are restricted to consuming the same amount of the good.

to enjoy the same number of rockets in the Garden of Eden. It is useful to recall the eciency results of a competitive market equilibrium for a private good, say g leaves, using the concept of market demand. In the market for g leaves, market demand curve for a private good is obtained by horizontal summation of individual demand curves. Thus, if Adams demand is one g leaf at the price of 5 dollars, and Eves three g leaves at the same price, market demand at the given price is three g leaves (4 = 1 + 3). Market equilibrium is determined at the intersection of the market demand and supply curves. For convenience, assume the price of an apple is $1. Then at the equilibrium, Adam and Eve face the same price of g leaves to which they equate their MRS (MRS of g leaves in terms of apples for Adam = MRS for Eve =
Pf Pa

= price of g leaves). The equilibrium price is also equal to marginal

cost of g leaves through prot maximization of competitive producers, which is also equal to the MRT of g leaves for apples. Thus the market equilibrium satises the condition required for eciency. Eciency in the provision of public goods requires that the sum of marginal valuation on the last unit be just equal to the marginal cost. Assuming again for convenience that the price of an apple is $1, this means that the sum of MRS for rockets be equal to the marginal cost, which is equal to the MRT of rockets for apples. In other words, the sum of marginal subjective valuations of individuals for an additional rocket should be equal to the marginal cost of having one more rocket provided.

r,a r,a M RSAdam + M RSEve = M RT r,a

(1)

Graphically, collective demand curve for a public good may be found by vertical summation of individual demand curves for the public good. Ecient amount of public good provision is then found at the intersection of the collective demand curve and the supply curve (= marginal cost curve for rockets). Note that public goods represent a special case of positive externalities. One might consider the market price as the marginal private cost of an individual consumer. The individuals own MB represents his marginal private benet. The other individuals marginal benets are then marginal external benets arising from the given individuals consumption. The individual has no reason to take into account external benets, and thus is likely to decide to consume at the level that equates his marginal cost (=price of rockets) and his marginal private benet, which is sub-optimal from the societys point of view.

Free Rider Problem

In fact, the outcome may be worse in the presence of public goods than the analogy to positive externalities suggests. For private goods, we can rely on the market to provide an ecient amount. Can we also rely on the market for ecient provision of a public good? The answer is no. Adam has an incentive to lie about his preferences. So does Eve. Adam may falsely claim that reworks mean nothing to him. If he can get Eve to foot the entire bill, he can still enjoy the show and yet have more money to spend on apples and g leaves (private goods). This incentive to let other people pay while you enjoy the benets is known as the free rider problem. Hence, not only that the market supply is likely to fall short of the ecient amount, but

also that the supply may well fall to naught. Consider the following results based on lab experiments. Marwell and Ames (1981): lab experiment involving a public good and a private good. 10 subjects are given 10 dollars each. Each individual is a given two investment options, a private account and a public account. The rate of return guaranteed with a private account is 100 Isaac, McCue, and Plott (1985): Sadly, when IMP let subjects go through several rounds of MA games, they found that contribution to the public pot decreases as experience accumulates.

Can the government do better?

The question then is whether the government can do better. Even the most ardent free-market advocates agree that national defense and domestic security (why are they public goods?) are proper functions of the government. Whether the government can provide an ecient amount of public goods is a dierent question. Just as the private market nds it dicult or impossible to let consumers reveal their true preferences, it will be also dicult for the government to gure out the ecient level of public good provision.

4.1
3

Preference Revelation Mechanisms

Private provision of public goods generally results in insucient supply. If only the government can induce individuals to reveal their true preferences, it
3 This

subsection borrows heavily from Varians Microeconomic Analysis, third ed., chapter

23.

should be able to provide an optimal amount of public good. There are several mechanisms designed for the purpose. For illustration, we present the GrovesClarke mechanism. Suppose there are only two individuals i = 1, 2 (or Adam and Eve) for illustration. Consider the case of a discrete public good. Thus the amount to be provided is either zero or one unit. Let ri be individual is reservation price and si his cost share of the public good. Since the public good costs c to provide, si c is the total amount of money that individual i must pay if the public good is to be provided. Then vi = ri sic is individual is net value for the public good. It is ecient to provide the public good if and only if v1 + v2 > 0 or r1 + r2 > s1 c + s2 c = c. (This is in line with the familiar condition for eciency that sum of marginal benets is equal to the marginal cost of public good provision.) One mechanism that we might use is simply to ask each individual to report his or her net value and provide the public good if the sum of these reported values is positive. The trouble with such a mechanism is that it provides incentives to lie. For instance, anyone with a positive net value will have an incentive to (falsely) report positive innity for his net value, and anyone with a negative net value will have an incentive to report negative innity (or any large negative number). Groves-Clarke mechanism solves this problem by adding a simple twist. 1. Each individual reports a bid for the public good, bi . This may or may not be his true net value vi . 2. The public good is provided if b1 + b2 > 0, and it is not provided if the

sum is negative. 3. Each individual receives a side payment equal to the (sum of) the other bid(s), if the public good is provided. (If the other bid is positive, the individual receives it; if negative, he must pay the amount.) Given the scheme above, it is easy to show that truth-telling or bi = vi is the dominant strategy. For individual i = 1, the payos are v1 + b2 if b1 + b2 > 0, and 0 if b1 + b2 < 0. We have two cases to consider. In the rst case, suppose v1 + b2 > 0. In this case, individual 1 will be better o with the public good provided with b1 + b2 > 0. He can make sure this situation occurs by simply telling the truth, that is, by reporting b1 = v1 . In the second case, suppose v1 + b2 < 0. In this case, individual 1 will be better o with the public good not provided with b1 + b2 < 0. He can make sure this situation occurs again by simply telling the truth, that is, by reporting b1 = v1 . In other words, truth telling is the dominant strategy for individual 1. We can similarly demonstrate that truth-telling is also the dominant strategy for individual 2. So rational individuals (clever enough to understand the payo structure and gure out the dominant strategy) will reveal their true preferences. The government then can use the information to decide optimally whether to provide the public good or not. (Exercise) Two roommates, Lisa and Bart, are considering whether to buy owers to put in the vase in their room. Flowers cost 4 dollars. Lisa and Bart are going to evenly divide the cost in case they buy owers. Lisas reservation price for

owers is 5 dollars, and Barts reservation price 1 dollar. 1. What is the ecient thing to do, buy or not? 2. Consider Groves-Clarke mechanism. What is the payo structure from Lisas point of view under the mechanism? 3. Show that Lisa cannot do better than bid her true net value of 3 dollars. In other words, Lisas dominant strategy is to tell the truth. 4. Show that the mechanism fails to achieve a balanced budget. More specifically, calculate the total amount of side payments required by the mechanism. This theoretical result is interesting, but the question remains whether this kind of sophisticated mechanism can be used in practice. Individuals may fail to understand the mechanism. They may not care to spend time to determine their own MRS schedules. Balancing the budget is almost always a problem with these mechanisms. Finally, it may be too costly to gather and process the required information when many individuals are involved.

4.2

Lindahl equilibrium and unanimity rule

Assume for convenience that a unit price of the public good is $1. Allow the shares of the price borne by Adam and Eve to be dierent. SA is Adams share, and SE Eves. They are called Lindahl prices. (Chapter 6) Lindahl proposes to use the unanimity decision rule to reach the ecient level of public good provision. Let a government ocial announce an arbitrary pair of Lindahl prices and let Adam and Eve announce their respective quantities of public 9

goods desired. If the quantities are identical, then the process stops. Otherwise, the process will continue until the two parties reach a unanimous decision. This process should be able to attain eciency, since the unanimity requires the equality between the sum of two individuals MRS and the price (MRT). This process may prove impracticable, however. First, people may vote strategically, not revealing their true preferences. Second, even if they vote sincerely, the process may take too long before reaching a decision, if at all, if there are many individuals involved.

4.3

What else?

Under some conditions, it has been shown that majority voting will lead to the outcome most preferred by the median voter. Yet, there is no guarantee that the outcome will be ecient. In addition, majority voting suers from the voting paradox, if preferences are not single-peaked. (See chapter 6.) There is no political mechanism found so far that leads to ecient provision of public goods under reasonably general conditions. Microeconomic theory about public goods undoubtedly paints a pessimistic picture. This is alarming, when one considers that there is an important class of public goods whose adequate provision can prove critical for successful governance of society, namely collective action. See Mancur Olsons The Logic in the reading list. One may presume that if some action represents collective good for all the members of a group or society (monitoring of, and responsible participation in, the political process), then the appropriate action will be taken. According to Olson, this premise is wrong, as collective action is a public good:

10

let someone else provide the necessary action, while I will just enjoy the benet. Do we have to be pessimistic? Not necessarily. It may be dicult, or nearly impossible, to reach the ecient level of public good provision, as the discussion above suggests. However, we have many important examples from the real world whereby communities have succeeded in maintaining a reasonably satisfactory regime for public good provision. See The Lighthouse in Economics, for instance, for an example. Ostrom also provides several interesting examples, whereby communities around the world have been able to provide the required collective action for successful management of common property resources. Communities are usually better equipped to deal with provision of public goods compared to markets, as they are able to apply extra disciplinary measures to free riders, and also better able to collect and process information on members true preferences.

11