When Self-Interest is Self-Defeating:The Public Goods Experiment as aTeaching Tool

Robert G. Nelson and Richard O. Beil, Jr.*

Abstract

This simple classroom experiment demonstrates many of the behavioral phenomenaassociated with the voluntary provision of a public good. The mechanics of the game are explainedin detail and complete instructions are provided, as well as suggestions for follow-up lectures.Influences such as anonymous voting, persuasion, returns to free-riding, and duration of associationcan be explored in connection with concepts of incentives, individual rationality and group welfare.A number of variations and extensions can be used to incorporate prisoners’ dilemmas, incentivecompatible mechanisms, negative externalities, and Coasian bargaining.

Key Words: teaching, experimental economics, public goods

When it comes to the free rider problem,the standard textbook exposition on voluntaryfinancing of public goods can seem bothparadoxical and pedantic. Consider one suchexposition (representative of many) from Stiglitz’stextbook The Economics of the Public Sector:

“But since every individualbelieves that he would benefitfrom the services providedregardless of whether hecontributed to the service, he hasno incentive to pay for theservices voluntarily. That is whyindividuals must be forced tosupport these goods throughtaxation,” (p. 120)

Such statements are paradoxical becausealthough individuals are supposed to have no

incentive to contribute we do in fact observe manyhighly successful examples of voluntarycontributions to public goods, Such statements arepedantic, or at least overly general, in that whiletaxation is usually the recommended solution, manyorganizations do not have such authority, and thereare other mechanisms that can be used to elicitcontributions without coercion.

In introducing the subject to students wemay argue that universal free-riding can bepredicted from our simple theory of self-interested,“rational” choice--a theory that served us well whenwe lectured on markets for excludable, privategoods, But now we encounter a class of goods forwhich self-interest is both rational and self-defeatingat the same time! From the perspective of theexperimental economist or game theorist theseissues are simply different facets of the principle

that “institutions matter’’--it is not so much a case of

*Robert G, Nelson is an assistantprofessor in the Department of Agricultural Economics and Rural Sociology, andRichard O. Beil, Jr. is an associate professor in the Department of Economics at Auburn University, Alabama. Thisarticle is similar to a chapter in a book to be published by Richard D. Irwin, Inc. entitled Illustrating Econom k

Principles with Classroom Experiments: A Teachers Guide by Robert G. Nelson and Richard O. Beil, Jr. Permissionis granted by the publisher to use this material.

J Agr and Applied Econ 26 (2), December, 1994:580-590Copyright 1993 Southern Agricultural Economics Association

J Agr and Applied Econ,, December, 1994 581

different goods, or market failure, or innate greed,as it is that the institution (or set of rules) that wetypically use to represent the market for publicgoods tends to favor non-cooperation.

Still, simplistic and inaccuraterepresentations abound. In their review of theinstructional literature on free-riding, Asch andGigliotti comment that the remarkably uniformtreatment of free-riding in economics textbooks“misses the opportunity to show students howeconomic analysis may come to grips with a set ofapparently inconvenient facts and with the notuncommon opinion that economists are ethicalnihilists, if not worse,” (p, 34) Indeed this notuncommon opinion has led to recurrent accusationsthat we indoctrinate our students to be selfish.These charges range from provocative titles like“Does Studying Economics Inhibit Cooperation?”and “Economists Free Ride, Does Anyone Else?” inprofessional journals (Frank et al.; Marwell andAmes) to recent disparaging remarks in the popularpress such as “Economists are not merely dismal, itappears, but selfish and uncooperative, as well”(Anon., p. 71),

Rather than defend or perpetuate thesestereotypes and prejudices, we can transform theseincongruities into a “teachable moment” through asimple classroom experiment that is easy to conduct,fim to play, and rich in empirical content. Just aslaboratory exercises are an integral part of thecurricula of most scientific disciplines, agriculturaleconomics can also benefit from the selective use oflaboratory exercises in experimental economics, Inthis paper we describe in detail a prototypical one-class-period experiment and then discussmodifications and variations of the proceduresuitable for an extended set of follow-up lectures onpublic goods,

Background and Rationale for the Instructor

The conditions of the classroom laboratoryare not intended to mimic the real world. Quite theopposite, the laborato~ environment is speciallydesigned to give the experimenter control over asmany variables as possible, In this experiment theinstructor has control over several key factors thatare now widely accepted to affect the extent of free-riding, but that are not articulated in the simpletheory of rational choice. The public good is “pure”

in that it is nonrival in consumption andnonexclusionary in its benefits. Moreover, everyindividual shares equally in the pecuniary benefits.The instructor is in the singular position ofcontrolling each student’s marginal valuation of thegood, as well as the group optimum, The instructorcan also vary the group size, period of association,incentive structure, and conditions of anonymity andcommunication.

This experiment illustrates a behavioralresponse (free-riding) that reduces the efficiency ofan allocation mechanism (private funding of apublic good) even though that mechanism has thepotential to make all the participants better off andnone worse off. It demonstrates a case where the“invisible hand” of rational, utility maximizing self-interest can fail to serve the interests of society.Furthermore, students often do not fully appreciatethe difficulties in organizing for collective action,This classroom experiment is an entertaining way toreveal some of these complexities of humanbehavior,

The Mechanics of the Game

A significant advantage of this experimentis that it can be conducted in one class period andstudents will gain an immediate appreciation of thedynamics of collective action. However, theinstructor may need additional time to recordearnings and provide summary statistics.

It is important for students to understandthe nature of the payoff table early if they are tomake deliberate decisions rather than randomchoices, and this may take a few rounds of practice.However, assuming the instructions are clear, itshould not be necessa~ to explain any theory to thestudents in order for them to start making choices inthe game, Before conducting the experiment, wealso try to avoid describing the fhnding objectivewith specific examples like lighthouses or nationaldefense so as not to bias the marginal valuation (theinstructions just call it a “Public Good”).

To set up this experiment the instructorgives each student an endowment--$0.50 or itsequivalent in bonus points--at the beginning of eachround. Each student must then decide whether togive the endowment to fund a public good, or keepthe endowment, They communicate this decision by

582 Nelson and Bed, Jr When Se~-In~erestis Se~-Defea(mg: The Public Goods Experiment as a Teaching Tool

writing their name and their vote (to “give” or“keep”) on a slip of paper and handing it to theinstructor. The instructor counts the number of“give” slips and announces this number to the class,Students then look uptheir individual earnings forthat round from the table in their instructions, Inour example, if five people (or more) give, theneveryone makes as much as (or more than) theiroriginal $0.50 endowment. The more people whogive, the more everyone makes, But a Keeperalways makes $0.50 more than a Giver.

In contrast to the stylized facts of rationalchoice, but just like the real world, the mostfrequently observed outcome of this experiment isthat some proportion of the class gives in eachround and another keeps. Complete cooperation isexceptional, but wholesale free-riding isextraordinary. As students anticipate the end of theexperiment (which is usually the end of the classperiod) there are often dramatically fewer Giversand more “Keepers. If allowed, some interestingattempts by spokespersons to stimulate cooperationmay develop in intermediate rounds but, with nomechanism to monitor vohng and penalizedefectors, full cooperation tends to be short-hved.

Although “dollars and cents” are used indiscussing the instructions and payoffs, bonus orparticipation points toward the course grade can alsobe used effectively to motivate students to play thegame in earnest. For example, we might offer onepoint added to the midterm grade for every $10earned in the experiment, Obviously, the results ofpublic goods experiments are sensitive to the returnsto free-riding. If there is no payoff to free-ridingthen we should expect to see higher levels ofcooperation. This is an importmt real-worldvariable that may be discussed in a subsequent classconcerning the conditions under which one wouldfind cooperation more or less effechve,

Table 1 in the Instructions for the Student(Appendix) would need to be extended for classeslarger than 25, but the general rule M simple:multiply the number of “gives” by a constant to getthe “gives” payoff (we used a constant of 0,10 inthe table); add the endowment to this to get the“keeps” payoff, Small classes (n<5) would need apayoff table where n givers would earn at least asmuch as theu endowment. Very large classes couldbe divided into teams, This introduces the element

of group dynamics in declslon-making, an elementnot found in the classical model of individualchoice, but one that corresponds to many real-worldinstitutions.

As demonstrated in the “Sample Results”section, the instructor may allow talking amongclass members as an effort to persuade moremembers to be Givers, but anonymity in actualvoting should be preserved. Since many peopleperceive the experiment as dealing with scnsltivemoral issues such as fairness, cooperation, fidelityand solidarity, wc strive to keep the ballotingprocess confidential. The results of theseexperiments are quite sensitwe to whetherparticipants feel that strict anonymity can and willbe maintained. In the terminology of game theory,the potential for “monitoring” is a significantvariable in determining the extent of free-riding,

In many economics experiments talkingamong subjects is prohibited in order to inhibitstrategic signaling and collusion, Relaxation of thisconvention is relatively harmless m this experimentand may be used to illustrate the role of “socialentrepreneurs” (McCaleb and Wagner) m devisingtactics to facilitate cooperation, The instructor maywish to discourage talking in early rounds and thenallow it in later rounds, The number of Giversusually increases with appeals from self-appointedadvocates of cooperation whose exhortations can bequite engaging. Such rules of communication areyet another variable in determining the effectivenessof cooperative action,

A sustained high level of contributions tothe public good 1s unlikely when using the payoffVdbleprovided here. This is predictable mainlybecause of the substantial returns to free-riding thatare built into the table. Another destabilizingelement is introduced when students know the gameis going to end with the class period; they tend todisplay more self-interest in later rounds.

Sample Results

Figure 1 shows the results of an experimentin a class of 16 students. The experiment wasconducted before the topics of public goods andfree-riding were introduced in lecture. The sessionoccupied an entire 50-minute class period. Talkingwas not allowed in the first four rounds, but by the

J Agr, and App[led Econ, December, 1994

Table 1. PayoffTablefor the PublicGoodsGame

If this may pimplechoose “give” You will get this much for cheasing:

o

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

“give”

0.00

0.10

0.20

0.30

0.40

0.50

0.60

0.70

0.80

0.90

1.00

1.10

1.20

1.30

1.40

1.50

1.60

1.70

1.80

1.90

2.00

2.10

2.20

2.30

2.40

2.50

fourth round some students were exhibitingconsiderable frustration at not being able tocommunicate with other students (presumably toexplain the payoff table and bring attention to thebenefits of cooperation). At this point the instructoroffered to relax the rules and allow the students totalk things over. The next round showed asubstantial increase in contributions, but this erodedin Round 6. The students conversed again and inRound 7 they achieved 100 percent of the maximumcontribution rate. By this time the end of the classperiod was imminent and further attempts to sustainthe coalition were progressively less successful untilthe last round was announced and the game ended

“keep”

0.50

0.60

0.70

0.80

0.90

1.00

1.10

1.20

1.30

1.40

1.50

1.60

1.70

1.80

1.90

2.00

2.10

2.20

2.30

2.40

2,50

2,60

2.70

2.80

2.90

3.00

583

with only two Givers. About one “dollar” perperson per round was earned on the average overthe ten rounds, at a dollar-to-bonus point conversionrate of 10:1,

A small fraction of students were free-riders in every round (except the seventh). A fewconsistently contributed throughout the experiment.Most alternated between “giving” and “keeping.”Ironically, a sentiment commonly expressed by thisgroup was “I tried to be cooperative but found itdidn’t pay because others were too greedy!”

584 Nelson and Bei[, Jr: When Se~-In!eres( is SeljDefeating, The Pubkc Goods Experiment us a Teaching Tool

Figure 1. Contributions to the Public Good, by Round; 16 Players

Typical of persu:lsiveefforts

““””””7”””””

{

k. . . . . . . . . . . . . . .

1 3 5 7 9 11

Round Number

Debriefing: What the Experiment Demonstrates

After several rounds of play, students haveaccumulated tangible experience in the dynamics offree-riding behavior. The economic principlesunderlying this phenomenon can be the topic ofsubsequent lectures. Our exposition is somethinglike the following.

Free-riding can lead to a condition that issometimes described as “market failure” wherebythe private market fails to efficiently supply a goodor service for which there is ample, though perhapslatent, demand. Because free-riders do not revealtheir true demand for public goods, it is oftenargued that public financing is the only way toprovide such goods, with the allocation mechanismsof choice being majority rule and fixed tax shares.Nevertheless, in the real world we often seeexamples of private financing by voluntarycontributions that result in the provision of publicgoods such as college athletics programs, churches,art exhibits, and public broadcasting, An important

distinction is that in the real world we do not knoweach person’s marginal valuation of the public good.In the case where some people perceive theirmarginal benefit to be less than the marginal costimposed, public financing by taxation results in theantithesis of free-riding, a phenomenon wrylydescribed by Gordon Tullock as “coerced carrying”(McCaleb and Wagner).

A real-world agricultural example relates togeneric advertising of a commodity like beef, butter,or broccoli. By definition, any increase in thedemand for a generic product benefits all producersof that product. The benefits of advertising cannotbe captured privately the way they can if theproduct has a private label, brdnd, or trademark.Thus, generic advertising is often fhnded by a“check-off” program that automatically chargesproducers some percentage of their input costs orsales revenues. Some of these programs have arefund provision whereby a producer can ask thatthe amount charged under the check-off be refunded

J Agr. and Apphed Econ, December, 1994 585

in full or in part. Producers that exercise thisrefund provision could be chamcterized as free-riders,

Students often suggest that free-ridingcould be eliminated if everyone were required tocontribute to the Public Good. In fact, most genericpromotion programs have responded to perceivedfree-riding in this very way, terminating their refundprovisions when refired levels reached disturbingproportions. For the Honey Board this occurredwhen refi.mds reached 13 percent; for the NationalPotato Promotion Board, 18 percent; for the CottonBoard, 35 percent; and for the National Egg Board,45 percent (Forker and Ward pp. 92 and 108).However, not all organizations are vested with theauthority to tax and, for those that are, coercedcarrying can be an unwelcome consequence.

Students also complain that their mainfrustration with the game is that they are notallowed to know how other students had voted,Members of international cartels like OPEC voice asimilar complaint when they cannot effectivelymonitor cartel members’ actions and penalize themfor departures from the monopoly price. Indeed, theeconomics of public goods is similar in many waysto that of oligopolies.

A homework question on “What do youthink were important factors determiningcontributions in each round?” can generate somepenetrating insight and novel extensions. Forexample, one student commented that theexperiment was “like a prediction game” in that hisdecisions were based on anticipations of how manyothers would give--a fair description of the processof rational expectations formation. Another studentremarked that persuasive efforts were likeadvertising, and we noted that they do display someof the same “wearing out” effects in subsequentperiods, Responses to the question “WhatImprovements can you suggest to increase voluntarycontributions?” produced such refinements as “askfor a show of hands of intentions to give beforeeach round” and “put the givers’ slips into a lotterypool, draw names at random, and give a bonus tothe winners, ”

While we have used the Public Goodsgame primarily in lower-division undergraduate

classes in marketing and macroeconomics (with bothmajors and non-majors) it can be adapted tointermediate, advanced and graduate classes, Suchextended expositions could use game theory toformalize the notion of free-riding as a Nashequilibrium that is Pareto-dominated by thenonequilibrium outcome (full contributions),Students could be challenged to design an allocationmechanism that is non-dictatorial, Pareto-optimal,and strategy-proof (Ledyard). Majority rule satisfiesthese conditions, but only if there are no more thantwo possible allocations, For three or moreallocations an impossibility theorem has establishedthe nonexistence of such an optimal mechanism(Satterthwaite). Although no dominant alternativeinstitution has been developed, numerous rules formaking committee decisions have evolved, andsome unconventional mechanisms have beenproposed such as the Wicksell-Lindahl tax scheme,Thompson’s D-Process, and the Groves-Ledyardmechanism (Feldman).

Variations and Extensions of the Game

The research literature of experimentaleconomics conlams many extensions andelabortions on the theme of public goods provision,Most of these investigations are beyond the scope ofthe basic principles under discussion here, or areoriented toward research rather than teachingapplications, However, there are several variationsthat may prove instructive in the undergraduate orgraduate classroom setting.

Random Endpoint

One simple variation is to make theduration of the game more uncertain, Thepropensity for cooperation to deteriorate is not aspredictable if players do not know when the gamewill end, especially if it is perceived to continuewell beyond one class per]od in the future. A gamewith a random endpoint can be used to examine th]seffect on the sustainability of cooperation. Therandom endpoint can be determined by rolling apair of dice or picking a card from a shuffled deck.When a certain number or card appears, the gameends, Although not common in real-worldapplications, endpoint anticipation has been noted asan Important treatment variable influencing the

586 Nelson und Bed, Jr When Se~-in/ere.yt is Se~-Dejeatmg l?-te Subllc Goods Experunent as a Zkuchmg Tool

degree of cooperwvc action In researchexpcrlmcrrts; the “one-shot” game IS an extremecase (Andreoni).

Vanah[e Cono-ibullons

Another insuwmonal mochiicamm allowsthe student to rmkc varlablc comnmmons to thepubhc good instead of’ the aU-or-rlothlng cholccdescrtiied above. Whale this varwtuon IS morereprcsemauvc of real-world sltuauons, It reduces Ihenumber of” rounds that cm De played in a classpcrlod because mslcad 01”just courumg “gives” theinstructor now has to sum the dil fcrem comnbutl onson all the slips of paper beiore the value of thepubhc good can be announced. Typically theinstructor adds some percentage LOthe sum ot thecomt”ibutlons to deterrnrnc the value of’ the publlcgood.

This varlanon has been adapted forclassroom usc by Brock [or repeated play amongteams of students, and by Sutock for a one-shotgame whe~c students contribute their own money.Brock suggests a homework exercmc to derive themarket demand cum’e for the pubhc good byvertically summmg the willingness to pay of’ theindividual teams.

Pmotler’s I)ilerrwtra

13rock also describes a game-tneoreticmtcrprctauon of’ the public goods game that Nconstructed as a Prlsoncr’s IMernrna, Since ourvcrwon uses all-ol-ntrthmg contr-lbutlons it fits thisirueqmetatlon beucr than the variable conmbuuonsversion, For a game with 20 players, the Prwtmcr’sIMemma payoff rrmtrlx that corresponds to table 11s shown in figure 2. The larger type tmhcates“your payofl:” whale the smaller type mchcatcs“every othcrplayer’s payoff” assummg they all dotile same thing.

The game-theoreuc Cxposltlon 1sparmxtlarly well-stuted to dlustmte why tree-ridingIS a Nash equihbrmm even though IuILcooperationis Pareto-dommant.

f+’ovlslon POI}12LS

[swat, Schmidt~, and Walker describe amodificauon In which cormibuuons to the pub~icgood can be increased substantially. In this designthe pubhc good is provided only if totalcontrlbuuons reach some pre-specdied thresholdcalled a “provmon point”; otherwise cormbuhonsarc rerurncd under a money-back guarantee. In theall-or-r~oth]rlg case, the provlslon point can beconsuucced to render a Nash equlhbnurn where fidlcontrmtmon now Pareto-dornmates the free-riderequdlbnum. Again using an example with 20players and table 1, the payoff matrix correspondingto a full-conmbtmon provlslon point of 20 with amcmcy-track gwaramce is shown in [igurc 3.

Any prowslon point less than the fuliconmbunons case illusuatcd m figure 3 [cads to adlf ferem set O(~nccnuves to free-rldc. For wmmplc,with 20 players and a prowwon point of 19, figure4 shows that “kcepmg” is the best strategy (i[everyone else can be counted on to “give”).

Figure 5 illustrates the results from a classof 44 students where the first six rounds followedthe rules described above for the prototype gdrneand the last SIXrounds used various prowsion pointswith the money-back guarantee. The prowsionpom~s used in this game were suggested by classmcrrlbers. Round 7 had a prowslon point set at 30and met this condmon with 35 gwcrs. Rounds 8, 9,and 10 fell short of their prowsion points and therefund opuon was excrcmd: round 8 required 35givers but got only 28; round 9 reqtured 35 gm’ersbut got only 25; round 10 required 43 givers but gotonly 41. Rounds 11 and 12 attained thclr prowslonpomls, which were both set at full contrlbuhons(44). Students apparently carnc to realize that anyplovlslorl point less than full contributions t>mesthesame free-rider problems as the basic game.

Elagnolland McKee relate several instanceswhere Lheprowslon point mcchamsm wds used toobLam voluntmy conmbuuons to fund a publicgood. in 19’/9 a succcssfiul effort was rnddc to hlrca Iobbylst to represent Ihculty welfare before theOregon stale legislature. All Faculty m the statewere asked to contnbutc to the lobbyist’s annualsttlary, with the provuio that comribuhons would berctumcd d’the funding Objective ($30,000) was not

J Agr and Applted Econ, December, 1994 587

Figure 2, Payoff Matrix for 20 Players--Basic Game

ALL OTHERS

give keep

Y give 2.00 0.10

0 2.00 0.60

u keep 2.40 0.50

1.90 0.50

Figure3. Payoff Matrix for20Players--Provision Point (=20) with Money-Back Guarantee

ALL OTHERS

give keepr

Y give 2,00 0.50

0 2.MI 0.50

u keep 0.50 0.50

0.50 0.50L

Figure4. Payoff Matrix for20 Players--Provision Point (=19) with Money-Back Guarantee

ALL OTHERS

gve keep

Y give 2.00 0.50

0 2.(N 0.50

u keep 2.40 0.50

1.90 0.50

588 Nelson and Bed, Jr When Selj.Interes[ lsSelJ~Defeatmg 7he Publlc Goods Experiment asa Teaching 100I

Figure 5. Basic Game (Rounds 1-6) Compared to Prowslon Point Game (Rounds 7-12); 44 Players

Provisionpoint=44

II

““r””..

..”L_/!0 reach .provision

point

. H!....... .

2 4 6 8 10 12 74

Round Number

reached by a given date, The New DemocraticParty in Manitoba, Canada used the approach ontwo occasions to secure additional contributionsfrom larger donors, achieving their targets of$200,000 and $250,000 in 1980 and 1985respectively (Bagnoli and McKee, p. 351),Applying the provision point mechanism to genericadvertising of agricultural commodities couldmitigate the negative effects of coerced carrying andreduce dependence on dictatorial processes such asthose endorsed by Forker and Ward: “The free-riderissue will always exist in some form, but thenegative effects can be mitigated throughnonrefundable mandatory assessments on all of thevolume going into the rndrket area in which thedemand-expansion program is to be conducted.” (p,258)

Negative Externalities

In addition to these variations on thevoluntary contributions game there are severalparallels and extensions that may find a place in theclassroom exposition. So far the experimentsdescribed have focused on free-riding as a

consequence of the non-exclusionary property inpublic goods. The non-rival property implicit inthese games has generated only positivecxterndities. But when anyone can use the goodand it can be depleted, as with an open-accessresource like public grazing land, negativeexternalizes can lead to the “tragedy of’ thecommons”.

Walker, Gardner, and Ostrom investigatedthese effects with experiments using a “common-pool-resource mechanism”. Subjects could investthcw token endowment in either of two markets.One market offered a modest but constant return.The other market appeared to offer a higher returnexcept that as more people invested in it the returnsteadily decreased. Resource overuse caused rentsin the second market to accrue at levels averagingonly 5.7 percent of the optimum, a level of rentdwsipation even lower than predicted by standardmodels of noncooperative behavior.

Market Pdilure problems arising fromnegative externalities can sometlmcs be resolved ifproperty rights are assigned and side payments are

J. Agr. and Applied Econ., December, 1994 589

allowed. Leuthold has adapted a Coasianbargaining experiment for the classroom. Studentsare paired and one is chosen as the “controller” whodetermines the number of units to be produced (say,units of water pollution abatement), The secondplayer earns progressively higher returns for moreunits (because cleaner water is more valuable tohim), while the controller earns progressively less(because he has to pay for cleaning the water), Thepayoff table is constructed so that both will benefitif the controller produces some units in exchangefor part of the second player’s earnings, The twoplayers bargain over the number of units to beproduced and the amount of the side payment.Leuthold reported that students bargained to theoptimal number of units an average of 83 percent ofthe time.

Conclusions

Experiments are not just for researchersanymore. They can be an effective tool in teachingagricultural economics and are especially well-suitedto involving students in the process of discovery.Moreover, student evaluations of the gameconsistently range from favorable to enthusiastic,

The attractive features of the prototypeexperiment described here are that it is easy to setup and quick to execute, and it demonstrateseconomic behavior in a novel way not duplicated bythe classroom lecture or textbook reading. Itprovides students with an empirical foundation--atouchstone--that they can use to understand andappreciate more elaborate concepts of cooperativeinstitutions, resource economics and public finance.

References

Andreoni, James. “Why free ride? Strategies and learning in public goods experiments.” J, Public Econ.37(1988):291-304.

Anonymous. “How do you mean, ‘fair’?“ The Economist 327(May 29, 1993):71.

Asch, Peter and Gary A. Gigliotti, “The Free-Rider Paradox: Theory, Evidence, and Teaching, ” J Econ.Education 22(1991):33-38.

Bagnoli, Mark and Michael McKee, “Voluntary Contribution Games: Efficient Private Provision of PublicGoods,” Econ. Znqui~ 29(1991):351-66.

Brock, John R. “A Public Goods Experiment for the Classroom.” Econ. Inquity 29(1991):395-401.

Feldman, Allan M, Welfare Economics and Social Choice TheoV. Hingharn, Mk Kluwer AcademicPublishers, 1983.

Forker, Olan D. and Ronald W. Ward. Commodity Advertising: the Economics and Measurement ofGeneric Programs. New York: Lexington Books, 1993.

Frank, Robert H., Thomas Gilovich, and Dennis T, Regan, “Does Studying Economics InhibitCooperation?” J. Econ. Perspectives 7(1993):159-17 1.

Isaac, R. Mark, David Schmidtz, and James M. Walker, “The Assurance Problem in a Laboratory Market.”Public Choice 62(1989):2 17-236.

Ledyard, John O, “Incentive Compatibility, ” In The New Palgrave: A Dictionary of Economics, vol. 2, J.Eatwell, M, Milgate, and P, Newman, eds., London: Macmillan Press, 1987.

Leuthold, Jane E. “A Public Goods Experiment for the Classroom.” J. Econ. Education 18(1987):58-65.

590 Nelson and Beil, Jr.: When SeljInteresl is Se~-Defea(ing: The Public Goods Experiment as a Teaching Tool

Marwell, Gerald and Ruth E, Ames. “Economists Free Ride, Does Anyone Else?” J. Public Ecmz.15(1981):295-310,

McCaleb, Thomas S. and Richard E. Wagner. “The Experimental Search for Free Riders: Some Reflectionsand Observations.” Public Choice 47(1985):479-490.

Satterthwaite, Mark A. “Strategy-proof Allocation Mechanisms.” In The New Palgrave: A Dictionary ofEconomics, vol. 4, J. Eatwell, M, Milgate, and P. Newman, eds., London: Macmillan Press, 1987.

Stiglitz, Joseph E. Economics of the Public Sector, 2nd ed. New York: Norton, 1988.

Sulock, Joseph M. “The Free Rider and Voting Paradox ‘Games’. ” J Econ. Educa~ion 21(1990):65-69.

Walker, James M,, Roy Gardner, and Elinor Ostrom, “Rent Dissipation in a Limited-Access Common-PoolResource: Experimental Evidence.” J Environmental Econ. and Management 19(1990):203-11.

Appendix

Instructions for the Student

This is an experiment in the private funding of a Public Good. Individual contributions from themembers of a group are used to fund a good or service that will benefit all the members collectively,regardless of who contributed and who did not.

The experiment will last for several rounds, At the beginning of each round you will each be givena sum of money as an allowance, You can keep this allowance, or you can contribute it toward the PublicGood. To indicate what you wish to do with your allowance, write your name on a slip of paper and thenwrite either “give” if you wish to give your allowance toward the purchase of the Public Good, or write“keep” if you wish to keep your allowance for yourself. Fold the paper in half (so that no one else can seewhat you wrote), and send the paper up to the front of the room. When everyone’s slip of paper has beencollected, the instructor will count the slips of paper that say “give” on them and announce that number.Using the attached table (table 1) you can then figure out how much money you made in that round. Keepa running total of your earnings to hand in at the end of the experiment.

Take some time now to study the attached table. There are three columns on the table. The firstcolumn shows all the possible numbers of people in the class who could choose to “give” in a particularround, The middle column shows how much money each class member would receive depending on howmany “gives” were counted. For example, if 15 people wrote “give” on their slip of paper then each oneof these people would receive $1.50 (reading across the table from row 15). All the other people in theclass (i.e. those who wrote “keep” on their slip of paper) would receive that amount plus the amount of theirallowance (since they chose to keep this). This amount is shown in the third column. Notice that any valuein the third column is always larger than its corresponding value in the second column by $0,50, the valueof the allowance in each round, The key feature of this table is that ~hemore people that “give”the moreeve?yone makes, but those who “keep” always make $0.50 more than lhose who ‘~ive”.

After the experiment ends, hand in your record of total earnings to the instructor,