N. R. Miller
1/28/02
Rev. 8/19/02
2nd Rev. 1/27/03
3rd Rev. 1/10/05
SOCIAL DILEMMAS AND RELATED PROBLEMS
OF A GAME-THEORETICAL NATURE
Social Coordination Problems. Suppose that each member of a group is given an array of choices concerning how he or she might behave, and that everyone does better if everyone chooses to behave the same way (or in some other coordinated fashion) than if different people choose to behave in different (or uncoordinated) ways. (One example: which side of the road should we drive on? Another example: there are one conservative and three liberal candidates in an election and liberals constitute 60% of the electorate; how should liberals vote?) This is a social coordination problem. [See the handout on The Social Coordination Game.] Sometimes members of a group can communicate with each other readily and thereby can solve social coordination problems more or less spontaneously. But external cues and/or leadership may be required, especially if there is some conflict of interest among members of the group. In any event, once coordination is achieved, no one has an incentive to change his or her behavior. (Technically, the coordinated behavior is a Nash equilibrium — named for John F. Nash, the subject of the [somewhat fictionalized] movie A Beautiful Mind that came out several years ago.)
Prisoner’s Dilemma Problems. Suppose each of two people must decide to behave in one of two ways, which we call “defect” and “cooperate.” If both cooperate, they are both better off than if they both defect. But if one defects while the other cooperates, the defector is even better off than if they both cooperate, and the cooperator is even worse off than if they both defect. A salient example: two rival nations (e.g., the U.S. and the Soviet Union during the Cold War), each deciding whether to build a new and terrible and expensive new weapon (“defect”) or to refrain from building it (“cooperate”). This is a Prisoner’s Dilemma problem. If the players cannot communicate, it is very unlikely that they can mutually cooperate. If they can communicate, they would readily agree to cooperate, but the agreement probably will not stick, but — because mutual cooperation is not a Nash equilibrium — each player has an incentive to defect from the agreement
Social Dilemma Problems. A Prisoner’s Dilemma may extend to many people, each of whom must choose to cooperate or defect. If everyone cooperates, everyone does well. But if a few people defect while most cooperate, the defectors do even better at the expense the cooperators. If everyone defects, everyone does poorly, but if almost everyone defects while a few cooperate, the cooperators do still worse. (One example: car owners can either install pollution control devices on their cars [“cooperate”] or not [“defect”]. Another example: nations can reduce their greenhouse gas emissions [“cooperate”] or not [“defect”].) This is a Social Dilemma problem. [See the handout on The Social Dilemma Game.]
Collective Action Problems. In many circumstances (but certainly not all, e.g., international arms races), Prisoner’s Dilemma problems and small-scale social dilemma problems can be solved, in that mutual cooperation can be achieved through straightforward collaboration. But if the number of participants is large, or their actions are not readily observable, or communication among them is difficult, or the stakes a very high, or the parties don’t have reason to trust each other, mutual cooperation is very difficult to achieve. The result is a Collective Action Problem [see David Hume’s example below]. Collective action problems typically have the following elements.
Public Goods Problems. An economic good is something everyone wants more of rather than less, other things being equal. A private good is an ordinary economic good (such as food, housing, entertainment, etc.) which can be simultaneously made available to some people (in particular, to those who pay for it) but not to others (in particular, those who don’t pay). A public good, by contrast, is an economic good (such as clean air, flood control, military defense, public health measures, etc.) which, if made available to anyone (in some group or area), must be made equally available to everyone (in that group or area), regardless of whether they have contributed their “fair share” of the cost of providing the good. Thus, if payment is voluntary (as it is for private goods), everyone has an incentive (to try) to free ride on the contributions of others. (In some contexts, this is called rational abstention.) And if few if any people voluntarily contribute, the public good will not be provided at all. Thus, provision of public goods presents large scale social dilemma and collective action problems. When everyone chooses to defect in a social dilemma (e.g., not put pollution control devices on their cars, thereby imposing externalities on others), the result is a public bad (polluted air), which requires collective action to remove.
Transactions Costs Problems. In order for collective action to be undertaken, people must communicate and negotiate with one another. But, especially given a large-scale social dilemma, this is difficult and time-consuming. Such barriers to cooperation are called transactions costs. Transactions costs can also hinder ordinary economic exchange in private goods. For example, lots of (mostly older and better off) people have basements and garages full of useful items that they no longer need and would be glad to sell at a modest price. At the same time there are many other (mostly younger and less well off) people who need such items and would be glad to buy them at a modest price. But the potential buyers and sellers usually can’t find each other or don’t even know of each other’s existence. The novel and useful characteristic of Internet trading sites such as eBay and half.com is that they greatly reduce transactions costs, as organized markets (e.g., farmers’ markets, stock exchanges) likewise do for other kinds of goods. (Hierarchical organizations such as firms and government agencies can also reduce transactions costs.)
Monitoring Problems. A further barrier to resolving social dilemmas is that everyone knows that everyone has an incentive to defect from any negotiated agreement (e.g., an arms control agreement between two nations). If one party defects from the agreement, it can make a huge difference how quickly this defection can be detected by the other party. If defections are quickly detected, the other party to the agreement can quickly retaliate and protect itself by defecting also. More importantly, the prospect of quick detection and retaliation should deter each party from defecting in the first place. But if there is a monitoring problem, i.e., if defection cannot be readily and reliably detected, defection will not be deterred in this way. Rather each side will suspect the other has defected (or is about to) and is therefore likely to defect preemptively itself. The history of the Cold War nicely illustrates the importance of monitoring. Until the mid-1960s, the U.S. and the S.U. engaged a largely unconstrained (and especially dangerous and expensive) arms race because neither side could reliably monitor what the other was doing (especially the U.S. could not readily monitor what the S.U. was doing; the S.U. could get much of the information it needed by subscribing to the New York Times). After reconnaissance (“spy”) satellite technology was developed, the pace of the arms race was substantially damped down.
Credible Commitment Problems. Since a barrier to resolving social dilemmas is that everyone knows everyone has an incentive to defect from any negotiated agreement, it helps greatly if there is some way in which parties to the agreement can make credible (i.e., believable) commitments to do what they have agreed to do. It can be difficult to make a credible commitment if it requires you carry out an action (e.g., a promise or threat) that will be contrary to your immediate interests when the time comes to carry out the commitment. The ability to make credible commitments facilitates many types of mutually advantageous transactions and agreements. (It also facilitates attempts at coercion — see below.) In many ordinary circumstances, you can pretty readily sort out credible from non-credible commitments. If you are beginning a new job with an established employer, it probably won’t greatly concern you that you will have to work for a week or two (or whatever the pay period is) before your employer pays you anything. On the other hand, you would be rightly suspicious of an anonymous stranger who wants to buy your used car and drive off in it right now while promising to mail you a check next week. Notice also that you are willing to take as payment tokens that are intrinsically worthless (e.g., a stack of small paper sheets printed in green with pictures of Alexander Hamilton [cash] or single signed piece of paper [a check]) or have only “virtual” existence [i.e., an electronic fund transfer to your bank account]) only because the government and banks can make credible commitments. Several circumstances make it easier for a party to an agreement to make credible commitments: (i) long-term existence during which a reputation for keeping promises can be acquired (ebay and half.com have devised procedures by which sellers can establish reputations for reliability), (ii) open-ended existence, so that so that such a reputation is valuable, and (iii) an established legal system that gives each party the “right to be sued” in (which sounds like a “right” that you wouldn’t wish for, until you recognize that the “right to be sued” is the “power to make credible commitments”). The ability to make credible commitments makes threats as well as promises, credible and thus also facilitates coercion.
Bargaining Problems. Suppose that, in the absence of an agreement, everyone is badly off and with an agreement everyone can be better off. And suppose transactions costs, monitoring, and credible commitment problems do not stand in the way of making a mutually beneficial agreement. There is still the problem that (almost always) there are many different potential agreements that could be reached, all of which make all parties better off than they would be in the absence of an agreement, but some of these potential agreements are better for some parties and others for other parties. This is the bargaining problem (so named by the same John Nash). Here is a simple example involving just two parties. A company and union are engaged in collective bargaining for a new wage agreement. Failure to reach an agreement would result in a strike or lockout harmful to both parties. The company figures it would be better off paying anything up to $20 an hour than getting no agreement. The union figures that its members would be better off accepting anything above $15 an hour than getting no agreement. Thus there is a bargaining range from $15 to $20 an hour that leave both parties better off than getting no agreement. But they must agree on some specific wage rate within this range and, within the bargaining range, the two parties have directly conflicting interests, which makes it difficult for them to strike a bargain (and may even prevent them from reaching an agreement).
Government as a Solution to Social Dilemma Problems. A basic lesson is that the existence of government (with a legal systems, police, courts, taxing power, etc.) can alleviate many of these problems and provide public goods reasonably efficiently. For example, a governments can make contributions toward the provision of public goods mandatory by imposing taxes.) Here is a classic statement of this conclusion. (Try to translate Hume’s language into the terms introduced in the handout.)
Two neighbours may agree to drain a meadow, which they possess in common: because it is easy for them to know each other's mind; and each must perceive, that the immediate consequence of his failing in his part, is the abandoning the whole project. But it is very difficult, and indeed impossible, that a thousand persons should agree in any such action; it being difficult for them to concert so complicated a design, and still more difficult for them to execute it; while each seeks a pretext to free himself of the trouble and expense, and would lay the whole burden on others. Political society easily remedies both these inconveniences.... Thus, bridges are built. harbours opened, ramparts raised, canals formed, fleets equipped, and armies disciplined, everywhere, by the care of government, which, though composed of men subject to all human infirmities, becomes, by one of the finest and most subtle inventions imaginable, a composition which is in some measure exempted from all these infirmities. [David Hume, A Treatise of Human Nature, 1740]
A long tradition in political theory (e.g., Thomas Hobbes, John Locke) invites us to imagine the state of nature (or state of anarchy) that would exist in the absence of government and to observe that widespread social coordination and (especially) social dilemma problems would prevail. (Another tradition in international relations theory invites us to consider the international system as an anarchic state of nature, with this important difference: the parties in a Hobbsian/Lockean state of nature [i.e., individual human beings] are essentially equal in relevant, whereas the parties in the international state of nature [i.e., nations] clearly are very unequal in relevant respect.) If collective action problems could be overcome, all individuals would simultaneously enter into a social contract promising to act cooperatively and setting up a monitoring and enforcement mechanism (i.e., a government and legal system) that would enforce the contract.
Even authoritarian or otherwise undemocratic governments (like Thomas Hobbes’ Leviathan) may be preferable to anarchy, though many of the problems discussed here change form and do not disappear entirely when a government is created. A direct democracy is simply not feasible on a large scale (and is difficult to operate successfully even in a quite small community) especially because of transactions costs and information costs [see below] problems. Representative democracy mitigates these problems but directly raises the following new problem.
Delegation problems. The world is very complicated and dealing with it makes a great many demands on all of us, most which we cannot attend to personally. Thus individuals, groups, and organizations (called principals) often delegate some responsibilities to others (called agents). For example, most people with cars that don’t work delegate the task of fixing them to auto mechanics. People with legal problems delegate responsibilities to lawyers and people with health problems delegate to doctors. Business owners delegate responsibilities to employees. Shareholders delegate responsibilities to directors who into turn delegate to CEOs and other managers. In the “social contract” political theory of Thomas Hobbes, John Locke, and others, “the people” delegate power to a government. When a representative government has been established, voters delegate their voice and votes to elected representatives. Within such a government, Congress delegates powers to its committees and, when it passes a law, almost invariably delegates responsibilities to some administrative agency (or to the President); agency heads (or the President) in turn delegate to lower-ranking officials.
This principal-agent relationship is always more or less problematic. Asking someone to do something for you (delegating to an agent) relieves you of the burden carrying out the task yourself, but almost always the agent — even if sincerely trying to carry out your wishes — doesn’t carry out the task exactly as you would like. You can limit such agency loss by giving the agent extremely detailed instructions and monitoring the agent’s actions very closely, but this may be more be more difficult and time-consuming than carrying out the task yourself — that is, this may defeat the rationale for delegating in the first place. Beyond this, the agent has his or her own interests, which are never exactly the same as the principal’s. (We have recently been reminded that CEOs don’t have the same interests as shareholders.) How closely the interests of an agent may be aligned with those of the principal can depend importantly on the particular mode of delegation, and therefore choosing the appropriate mode is a very important matter. (One example: your house needs to be repainted — should you pay the painter by the hour or by the job? Another example: you need legal assistance — should you pay your lawyer a flat fee or a contingency fee?) Principals may have reason to delegate to agents who are personally known them, family members, of the same ethnicity or religion, etc., so that the interests of the principal and agent may be better aligned.
Moreover, some principals are plural entities (“the people,” Congress, voters in an election district, etc.), in which case the relationship between principal and agent may be especially tenuous because (i) the plural principal probably has internal disagreement concerning what the agent should do and (ii) it probably faces internal collective action problems.
Information Cost Problems and Rational Ignorance. The world is extremely complex and confusing, and nobody has more than superficial knowledge of most of parts of it. To become better informed about any particular part of it demands scare time, attention, and energy from people who have many other things to do. Such information cost problems are a basic reason for delegation (and representative democracy generally) and affect many kinds of social interactions. Generally people look for simple “rules of thumb” (e.g., keep on using the same brand of a good, or the same auto repair shop, if it seems to be satisfactory) and other devices for cutting information costs, and thus they usually remain rationally ignorant of many matters that affect them. The incentive to remain rationally ignorant is especially strong with regard to issues pertaining to public goods, because many people have a say in resolving such issues (e.g., have right to vote in an election or referendum) and one’s own say in the matter is necessarily slight.
Asymmetric Information Problems. While nobody has more than superficial knowledge (if that) about most of the world, different people are relatively well informed about different little bits of it. Many social and political interactions take on distinctive features because they involve (i) a few people who are well informed about a particular problem and (ii) many people who are minimally informed (think voters trying to oversee the activities of legislators, or legislators trying to oversee the activities of bureaucrats) — that is, because of asymmetric information problems. Principal-agent relationships almost by definition entail asymmetric information problems, specifically a poorly informed principal trying to get a well-informed agent to act in the principal’s interest. (Think of the relationship between Enron shareholders and managers — or your relationship with your auto mechanic.)
Social Choice Problems. Once a representative government has been established to set up a legal system and a tax-collection mechanism, provide public goods, etc., social choice problems arise dealing with the mode of representation, elections, voting, etc., including the following.
Conformity costs. When purchasing private goods through the market, consumers can pick exactly what they want (within their ability to pay) but, when “purchasing” public goods through voting and political action, everyone necessarily gets the same thing and therefore almost no one gets exactly what they want.
Preference intensity problems. People differ not only in terms in terms of the direction of their preference but also the intensity of their preference (i.e., how strongly they feel about issues).
Collective choice problems. Majority rule provides a straightforward way for collectively choosing between two alternatives, but once the number of alternative exceeds two, all voting systems (including majority rule) have serious problems.
Agenda control problems. It is not practical to vote on all possible alternatives and whoever controls the agenda (decides what alternatives will be voted on and in what order) can importantly influence to the collective choice.
Misrepresentation of preferences. When choices are made collectively, individuals often have incentives to misrepresent their preference. (For example, the company engaged in collective bargaining referred to earlier may try to persuade the union that it prefers a strike to paying more than $15 (as opposed to $20) an hour, and the union can engage in similar tactics.)
Institutional Design Problems. All this implies that it is important to design constitutions and political institutions carefully. In particular, it is important to try to design institutions that will work well (e.g., produce “fair” outcomes) even when participants in the collective choice process are pursuing their own interests and trying to “game the system.” (The authors of the Federalist Papers believed that the “new science of politics” available of the framers of the Constitution allowed them to design such institutions for the new federal government.) Here are some very simple examples of such so-called incentive-compatible institutions for collective choice.
Fair division. Two children probably know how to agree on a procedure to divide a piece of cake fair and square: one cuts and then the other chooses. The greed of the children and the design of the choice procedure interact to produce a fair outcome (and a Nash equilibrium).
Collective choice by averaging. Members of a club must decide to how much money to spend on a particular project. Some members are intensely interested in the project and want to spend a lot, while others don’t much care about the project and want to spend little if any money on it. Members write down their preferred expenditures on ballots and the collective choice will be the average of the reported preferences. But there are two kinds of averages: the mean and the median. The mean is found by adding up the dollar amounts reported on all the ballots and dividing by the number of ballots. This procedure not incentive-compatible. Members who want to spend more than the average preference have an incentive to report on their ballots even larger expenditures than they really want, since this misreporting of their preferences pulls the mean closer to what they truly want. But suppose the ballots are sorted from lowest to highest in reported preferences to determine the median preferred expenditure reported by the ballot right in the middle of this ranking (half want less, half want more). Collective choice using the median is incentive-compatible. Marking high ballots even higher (or low ballots even lower) than you want has no effect on the median, so members with such preference cannot gain by misreporting them. And the voter who casts the median ballot is getting exactly what he or she wants (i.e., experiences zero conformity costs) and certainly has no incentive to misrepresent.
Judicial appointments. When the framers of the Constitution were discussing the mode of appointment for federal judges, Ben Franklin “in a brief and entertaining manner related a Scotch mode, in which the nomination proceeded from the lawyers, who always selected the ablest of the profession in order to get rid of him, and share his practice [among themselves]. It was here he said the interest of the electors to make the best choice, which should always be made the case if possible.” (James Madison’s Notes on the Federal Convention, June 5, 1787) Presumably Franklin’s proposal was whimsical; in any event, the Constitution provides for judicial appointments by the President with the advice and consent of the Senate.