Concerned citizens have two responses to situations that require sacrifice for the greater good, such as social dilemmas or provision of aid for the needy. One is voluntary sacrifice. The other is to take political action, in order to change the rules so that others will sacrifice in the same way. For a somewhat selfish and rational utilitarian, under specified assumptions, I show that political action is sometimes worthwhile and superior to voluntarism. This situation is more likely to obtain when the actor is moderately selfish (as opposed to being totally selfish or unselfish), and when: cost of political action is low; cost of cooperation is high; the situation involves aid for the needy and the proportion of potential beneficiaries is large; variability in willingness to cooperate is low; some people are already cooperating, but not too many; or the benefit/cost ratio of contributing increases with the number of contributors.
Jonathan Baron is Professor of Psychology, University of Pennsylvania. His research interests include utilitarian theory and psychological difficulties in applying that theory, particularly: (a) problems in the quantitative measurement of utilities, and (b) moral intuitions at odds with utilitarianism. His books include Thinking and deciding (Cambridge, 1994) and Judgment Misguided (Oxford, 1998).
Suppose you discover that your favorite breakfast cereal is produced in a way that pollutes the environment. Another cereal is identical but costs more because the manufacturer is careful not to pollute. You are an environmentalist, and you want to take some action. Two things occur to you. One is to switch to the more expensive cereal, a personal sacrifice. The other is to write your representatives and ask them to ban the method of production used by the manufacturer of the first cereal. The writing itself is also a personal sacrifice, of smaller magnitude, let us suppose, than switching cereal. If the method is banned, the price of the cereal will, of course, rise for everyone. You are not concerned enough to make both sacrifices. Which should you do?
More generally, people who are morally concerned can make two kinds of efforts. They can make individual voluntary sacrifices for the greater good, or they can try to bring about social change so that they and others like them are forced to make the same sacrifices. This general conflict between voluntary and political action arises in two main kinds of situations, social dilemmas and aid for the needy.
This article analyzes social dilemmas and aid - primarily social dilemmas - from a modified utilitarian perspective. Its purpose is to define the class of situations in which political action to bring about social change is more cost-effective than individual self-sacrifice. A third possibility is that people can try to persuade other people to make voluntary sacrifices. I shall consider this to be a special case of political action, with effects that are both smaller (because fewer people are involved) and more likely to occur. More generally, people can endorse moral norms, partly for the broad purpose of inducing others to behave more consistently with these norms (Baron, 1996; Gibbard, 1990). The question of when such norm-endorsement is rational thus bears on the rationality of supporting moral norms in general.
The assumptions are utilitarian in the sense that I am asking about the possible motives for rational action, and I am assuming that these motives must arise from the achievement of goals (Baron, 1996). These goals may be divided into two sorts, selfish and altruistic. Altruistic goals are those that depend on the goals of others: they are achieved to the extent to which the goals of others are achieved. I am at first putting aside other goals that arise from moral principles, such as goals concerning distributional fairness or personal participation, because these other goals may themselves be affected by the kind of analysis I am attempting here (Baron, 1996). If people are persuaded by utilitarian arguments, for example, their goals concerning fairness may change. Later in the paper, however, I do consider some of these other goals.
The modification of utilitarianism thus involves an allowance for an increased weight on self-interest. This modification stems from the assumption that both self-interested and altruistic goals are more fundamental than other goals but, otherwise, neither altruism nor self-interest is more fundamental than the other. The question here is about the role of basic motivations in establish other principles of conduct, so we must accept the balance of altruism and self-interest that exists. The resulting assumption is equivalent to the sort of ``weighted utilitarianism'' discussed by many writers (e.g., Baron, 1986; Bennett, 1981, p. 78; Hare, 1981, ch. 11; Singer, 1979, p.\ 181). This assumption has serious problems as a normative standard (Baron, 1993, pp. 89-91 , or 1994, p. 482), but it does capture the critical assumption needed here, namely, that these two kinds of goals are fundamental.
For present purposes I would think of family loyalty as part of self-interest. Otherwise, I assume that altruism is impartial within the relevant group. This is not a crucial assumption, but rather a simplifying one.
Although my assumptions are utilitarian, most of the considerations discussed are relevant to any moral analysis, since all analyses consider consequences to be relevant. Insofar as people are influenced by expected consequences, the analysis makes predictions as well, and I mention some of these. Also, while the primary focus is normative (i.e., concerned with the standards by which we evaluate outcomes), I try to provide prescriptively useful guidelines (practical advice that could improve the normative evaluation) for real behavior. (See Baron, 1994, for discussion of the normative-prescriptive distinction.)
The main point is that, given this modification, the personal cost of political action is often lower than the cost of voluntarism. In particular, this happens when people are somewhat selfish, that is, when they are neither perfect utilitarians (who care no more about themselves than about others) nor perfect self-interest maximizers. Therefore, an individual with a limited budget for self-sacrifice can sometimes do more expected good - taking into account the actor's extra self-interest - through political action than through voluntarism. This is not true for every issue, of course. It depends on the details of the situation, in ways I shall sketch. So, ultimately, the utilitarian do-gooder should adopt some mix of political action and voluntarism. Sometimes these are even hard to separate, when voluntary action is public and has the effect of dramatizing a political cause.
This argument answers a criticism of those who advocate some sort of coerced sacrifice for the greater good, such as a tax increase. The critic can say, ``If you think we should increase taxes, why don't you pay more yourself? Nobody is stopping you.'' The advocate may be tempted to answer that one person paying more is a drop in the bucket. The reply is that advocacy, too, is a drop in the bucket, because one person's activism is so unlikely to have any effect. I believe that this contrast between the size of the individual contribution and the size of the problem is generally off the point. The advocate's true reply is a little less noble. It is that advocacy requires less self-sacrifice than paying more taxes, but it could do just as much good. In other words, the benefit-to-cost ratio is sometimes higher for advocacy.
The argument is inspired by the arguments of several others. Heckathorn (1989) carried out a similar analysis. The main difference between the present analysis and his is that considers the altruistic benefits of cooperation to the cooperator to be independent of the number of others who benefit from cooperation. In essence, he assumes a kind of altruistic motivation that is dependent on action rather than on consequences. His conclusions are still similar to those I draw here, which is that sometimes a kind of apparent hypocrisy is best, i.e., supporting collective action but, meanwhile, free riding. (In a later analysis, Heckathorn [1996] extends this to several other situations, which differ from that describe here largely in that they all involve repeated interactions of small numbers of participants, who are members of a larger group. The analysis described here applies to large groups.)
Brennan and Lomasky (1993) argue for an ``expressive'' account of voting. In essence, the fact that many people vote cannot be explained in terms of rational maximization of self-interest, because the expected effect of voting on the voter is minuscule, taking into account the low-probability of any effect at all. Thus, people must vote for other reasons, and Brennan and Lomasky suggest that they vote largely to express their moral opinions.
Although this suggested expressive motivation is moral, it is not necessarily consequentialist according to Brennan and Lomasky. Because they do not assume consequentialist motivation, they have difficulty explaining the prevalence of two-party systems (pp.\ 121-123): it would seem that expressive considerations alone would lead to as many parties as there were views to express. If the moral motive to vote were, however, consequentialist and impartially altruistic, two-party systems would be easier to explain. Under such an account, many people really do vote in part because they have some chance of affecting the outcome, so they do not want to waste their vote on parties that have no chance of winning. The small chance of affecting the outcome is great enough to justify voting, if people take into account the large magnitude of the potential effect on others , even though it is not great enough if people consider narrow self-interest alone. From here on, then, I take the position of such consequentialist altruists, on the assumption that enough of them exist to matter. These people care about the magnitude of the effect that they might have, not just the opportunity to express their opinion.
Another possible explanation of voting, and activism in general, is that people are irrational. They think that political activism is in their self-interest, because they miscalculate. People tend to treat very low probabilities as if they were higher (Tversky and Kahneman, 1992), and, just as they buy lottery tickets because they ``might win,'' they vote because they ``might determine the outcome,'' which is true even if the probability is minuscule.
One empirical test of this question compares two situations in which people express their opinion about an issue, one in which they are not affected by their expression (low cost) and one in which they are affected (high cost). According to the expressive view of Brennan and Lomasky or the altruist consequentialist view I have just suggested, we ought to be able to find situations in which people would express different opinions in these two situations. Eichenberger and Oberholzer-Gee (1996) found conditions under which subjects would say, when asked in a survey, that roughly equal division between two people was fair, but, when they were asked to make the division, they took more for themselves. The survey was relatively low cost. The same argument may apply to surveys of willingness to pay, when these are discrepant from real behavior (e.g., Seip & Strand, 1991). Eichenberger and Oberholzer-Gee discussed the discrepancy between what people say and what they do as if it were a kind of inconsistency, but I hope to show here that it is not.
Altruistic behavior, such as giving to charity or engaging in political action, has been analyzed as having two components, one based on consequences for others and one based on the psychic benefits to the actor, the ``warm glow'' that results from doing something to help others (Andreoni, 1990; Margolis, 1982). The psychic benefits are thought to be largely independent of the the number of others helped or the amount by which they are helped. (Heckathorn, 1989, implicitly assumes this too, as noted.) The present analysis assumes that altruistic behavior is somewhat dependent on numbers and amounts, and it deals with that part. Psychic benefits that do not depend on consequences are counted here as simple reductions in the cost of action itself, even to the point where cost becomes negative. These psychic benefits, however, are still probably parasitic on the belief that the action in question does some good. They might even be sensitive to the belief that the action taken is the most efficient way of helping others. (People might fail to think about magnitudes unless they are reminded to do so.) These psychic benefits might therefore be sensitive to the kinds of arguments I am making here.
The analysis applies to two situation, both of which involve a conflict between the self-interest of individuals and the greater good, social dilemmas and aid to the needy. In both cases, some people sacrifice and some people benefit, and the total utility benefits are greater than the total sacrifices. The situations differ in that, in aid, the contributors and beneficiaries are different people, and, in social dilemmas, people are essentially identical in their roles as beneficiaries and potential contributors. From a utilitarian perspective, the distribution of utility does not matter, only the total, so the situations are closely parallel.
A social dilemma, of the sort at issue here, is a situation in which each of several people is faced with one option that is better for the individual and another option that is, on the whole, better for everyone. A classic case is that of fishers. It is in each fisher's interest to fish as much as possible, yet, if all of them do this, the fish stocks become depleted, and the fishers are all out of work. Other environmental problems take this form, in fact, any situation in which each of many agents can create the same kind of externality, such as pollution. It is in each person's or company's interest to pollute, to release carbon dioxide, not to recycle, etc., but the greater good is served by some other option. The same goes for voluntary compliance with the law, including the law of property: it is in each of our interests to steal if we can get away with it, but the more who try to do this the worse it is for everyone on the whole. And the same goes for working. It is in each worker's interest to shirk, but ....
The option that is best on the whole is called cooperation. The option that is best for the self is called defection. We have a variety of social norms that urge cooperation. We exhort each other to cooperate in a variety of ways. We gossip about defectors (shirkers, thieves). Some of us consider it a sign of virtue to use public transportation, recycle voluntarily, avoid buying products that cause pollution, and so on.
An alternative to voluntary cooperation is coercion. We change the rules so as to punish defectors. We fire shirkers (or spurn them, or don't promote them), we pass laws against pollution, we increase enforcement of property rights, or we increase the tax on gas guzzlers. Coercion involves a second-order social dilemma: we must cooperate in supporting it in various ways, such as voting for it, paying the taxes to support it, or doing our part to enforce it. Enforcement need not involve government or a single higher authority; it may be dispersed among some subset of those involved, or all of them (Ostrom, 1990). But even when enforcement is dispersed, the second-order social dilemma must still be solved. People must commit themselves to supporting the system. I use the term ``coercion'' without intending the usual negative connotations; indeed, because I favor it in some circumstances, I am taking the term of its opponents.
The choice between voluntary cooperation and coercion has an analog in the case of charity. The choice is between voluntary contributions to help those in need and coerced transfers made by government. (Coate, 1995, discusses this choice in a different context.)
In general, the mechanisms by which utility increases is different in the two cases. Social dilemmas arise because a single option (relative to the alternative) helps the people who do it but hurts everyone else, by the nature of the option itself. Utility increases if people are induced to avoid this option. In the case of aid, the benefit results from declining marginal utility. The same goods do more good in the hands of the beneficiaries than in the hands of the contributors.
The distinction between aid and social dilemmas is somewhat arbitrary. Really the two situations are ends of a continuum. When we analyze something as a social dilemma, we assume that those who make the decision to cooperate or defect are those who benefit from the cooperation of others. When we analyze something as aid, we assume that the givers and the receivers are different people. In real social dilemmas, some give more than others and some benefit more than others. In real aid situations, givers and receivers can sometimes switch places over time (e.g., tax contributions for medical assistance for those too poor to pay taxes at all).
Suppose that the status-quo is defection in a social dilemma. Let us calculate and compare the net benefit of cooperation with the net benefit of political action, which, if successful, would force everyone to cooperate. All benefits and costs will be in terms of utility. The benefits and costs affect N people, including the actor. Except for the actor, everyone is equally affected by the actor's decision.
I assume that outcomes for the self are given more weight than outcomes for others. The weight of the self is W, which is greater than 1. The fact that W > 1 means that the actor is not a perfect utilitarian. For a perfect utilitarian, W=1. So the analysis presented here is not entirely a moral one. It is a hybrid of an analysis based on morality and one based on self-interest. If we assume that a person's self-interest includes the utilities of other people to some extent, then it can be understood entirely in terms of self-interest. The analysis is addressed simply to people whose concern for others takes a utilitarian form but, otherwise, weigh themselves more than others. When I speak of ``utility,'' I mean utility weighted in this way. In sum, effects on each other person are weighted by N-1, the number of others, and effects on the self are weighted by W.
Table 1. Decomposition of effects of voluntary or political action into utility components for others and self.
| Direct effect | Direct effect on | Indirect effect on | |
| on N-1 others | self (weighted by W) | self (weighted by W) | |
| Voluntary | (N-1)B | W(B-C) | |
| Political | P(N-1)(NB-C) | PW(B-C) - WA | P(N-1)WB |
| with P=1/N | [(N-1)/N](NB-C) | W(B-C)/N - WA | [(N-1)/N]WB |
Table 1 shows the analysis of action into three components, direct effects on others (first column), direct effects on the self (second column), and indirect effects on the self (third column). The indirect effects occur when political action succeeds in making others cooperate, so that the actor benefits from the cooperation of others. All benefits and costs are positive numbers, so costs must be subtracted.
Voluntary cooperation (first row of Table 1) has a cost C to the actor and a benefit B to each of the N-1 others, as well as to the actor. By the nature of social dilemmas, B < C, but NB > C (i.e., cooperation increases total utility). Thus the total impersonal utility of a cooperative act is NB-C, although the utility to the cooperator is only B-C, which is negative. This assumes that B and C are independent of N and of the number of other cooperators, which I shall call M. (Later I shall examine some violations of these assumptions.) Personal utility for cooperating is thus (N-1)B + W(B-C). The second term is W times the net benefit of cooperation, which is negative. A sufficiently high W will make it better not to act; in particular, cooperation is worthwhile when WC < (N+W-1)B.
Political action (second row of Table 1) in contrast to cooperation, has a fixed cost A. In the simplest case, political action is voting, so that it can either succeed or fail at electing a candidate who will force cooperation. If the action succeeds, then everyone will cooperate, so that benefit to others will be N[(N-M)B-C], where M is the number who would cooperate without coercion. Here, N-M is the number of new cooperators, so (N-M)B-C is the benefit that each person gets from each other person's cooperation, including that of the actor (hence N rather than N-1). In the table, and henceforth unless noted otherwise, I assume that M=0. This means that the actor, among others, will not cooperate unless forced to do so, but will advocate a regime that will force her to act. (This is the more interesting case, since the actor is voting to force herself to do what she would not do otherwise.) In sum, the benefit of success for each person is NB-C. In column 1 of Table 1, this is multiplied by N-1 because N-1 other people benefit in this way.
Action has a probability of success, P. In other cases, P could be interpreted as a dilution factor resulting from the fact that each person's influence is small. In sum, the term P(N-1)(NB-C) is analyzed as follows: P is the dilution factor; N-1 is the number of other people affected by successful action; N is the number whose action is compelled by success; and the C is included within the scope of the first N-1 because they are all compelled to cooperate.
The effect on the self comes from two sources. One (column two of Table 1) is the direct effect of one's own action. It includes the fixed cost A plus the extra cost that will be imposed if the action is successful, W(B-C), but now weighted by P (unlike the case of social dilemmas). The other source (column three) is the indirect effect through others' actions, if the original action succeeds. Others will then be forced to cooperate, and this will benefit the actor. There are N-1 others, so this effect is substantial if it occurs, but it must also be weighed by P.
So utility for political action is the sum of the terms in the second row of Table 1. Political action is worthwhile when this is greater than zero. Rearranging terms, action is personally worthwhile when WA < P(N+W-1)(NB-C). It is sometimes worthwhile to take political action but not to cooperate spontaneously. The opposite could also happen.
In real life, P depends on N. The more voters, the less chance of any voter being decisive. Likewise, other political action is diluted when N is greater. Let us make a simplifying assumption that P=1/N. This represents the fact that, with higher N, the probability of having an effect is lower, or the dilution of one's political effort is greater. This situation is represented in the third row of Table 1. Action will be worthwhile when WA < [(N+W-1)/N](NB-C), or, approximately for sufficiently large N, when WA < NB-C.
The interesting result here concerns C. As N increases, the significance of C in the second column diminishes to essentially nothing because it is multiplied by P. Thus, when C is relatively high (which, of course, means that NB is higher still), the relative attractiveness of political action over voluntary cooperation increases. The benefits also increase, as shown in the first column of Table 1, because each person benefits from the cooperation of every other person. In voluntary cooperation, by contrast, C is part of the fixed cost of action. It cannot be avoided.
We can think of A as the cost of ``expression'' of one's political attitude. For some, this cost may be nonexistent, since they may have moral feelings that make them want to express their views in the political arena. It is rational to express the view that everyone should be forced to cooperate even when one is not willing - because of extra attention to self-interest - to cooperate spontaneously.
The relative utility of cooperation and political action can also be compared under the simplifying assumption that P=1/N. It follows from Table 1 (or from the stated conditions) that political action should be preferred to spontaneous cooperation when [(N-1)/N](W-1)C > WA. As N increases, this approaches (W-1)C > WA. Again we see that C weighs against spontaneously cooperation.
Of course, it is perfectly possible for a person to engage in both spontaneous cooperation and political action for the same social dilemma. However, in some situations the personal costs of each option come out of the same source (time or money), and the benefits to others do not decline noticeably with the amount of the individual contribution, within the range under consideration. In these situations, it is better to choose the most cost-effective option and put as much into it as possible, rather than to divide one's effort between the two options. On the other hand, if one option draws on time and the other on money, and if the disutility of the contribution (C or A) is marginally increasing with the amount of the contribution, then it could be reasonable to put some effort into both options.
The case of aid is roughly the same except that the contributors and beneficiaries are different. Those who pay the cost are different from those who get the benefit. From a utilitarian point of view, however, this does not matter much, for we can simply aggregate the costs and benefits. Typically, in cases of monetary aid, the beneficiaries can benefit much more than the donors from the same amount of money, so total utility is increased by transferring money from donors to beneficiaries. But this situation is formally analogous to one in which the donors and beneficiaries are the same people, and the increase in utility comes from the externalities, the effects of each person's behavior on everyone else, rather than from differing utilities for money. In both social dilemmas and aid, total utility increases as a result of individual sacrifice. Because of this formal analogy, all the arguments of the last section would thus apply, for example, to a population consisting of N contributors (including the actor) and N beneficiaries, different from the contributors.
It is possible that P is reduced. It might make sense to replace the simplifying assumption that P=1/N with the assumption that P=1/2N. Political action would thus lose half of its appeal. (Also, considering the whole population of 2N, the average cost per person and the average benefit per person is about half of what it would be in the social-dilemma case, but this is simply a matter of scale and does not affect the conclusion just reached.)
As a general conclusion, this is highly suspect. In real cases of aid, two situations are typical. In one, the contributors are the only ones who ``vote,'' i.e., who participate in political action. Examples include foreign aid and benefits for children. In this kind of situation, P would still be 1/N (still assuming two equal groups of size N). In the other situation, the beneficiaries can act politically, but they tend to favor the aid, and the contributors tend to oppose it. With two large blocs of votes precommitted, a relatively small number of swing voters can drive the decision.
I will shortly examine the effect of changing the relative sizes of the populations of contributors and beneficiaries. The general point should be clear, however. There are cases in which political action in favor of aid has a higher expected benefit/cost ratio than simply giving the aid.
Let us examine how various parameters affect the relative balance of voluntarism vs. political action. Unless I state otherwise, I shall use social dilemmas as the primary example and shall assume that the argument extends to aid.
The assumption that P=1/N is reasonable when P is taken to be a dilution factor. The positions of real legislators and government officials are reasonably seen as lying along a continuum. For example, they decide how much money will be spent on a given program, or how stringent a certain regulation will be. In making these decisions, they often listen to politically active citizens. It is reasonable to suppose that each citizen's voice is less influential, the more citizens who speak out. To a first approximation, it is also reasonable to suppose that the number of citizens who speak out is proportional to the population. (If anything, participation rates seem to be slightly higher for larger populations. People are a bit more likely to vote in national elections than in local elections, although these results are confounded with the fact that national elections typically include local elections, so the trip to the polls is more worthwhile.) Thus, the effect of any given citizen's action on the legislator's position on the continuum is diluted by a factor of 1/N. This is true even for actions that few people engage in, such as making campaign contributions, if we assume that the number of people engaging in such actions is a fixed proportion of N: each individual's expected effect is still inversely proportional to N.
In such cases of inverse proportionality, we can ask about the effect of increasing N on the utility of voluntary and political and on their relative utility. It is easy to see from Table 1 that the net benefit of both options increases with N. Moreover, once N is reasonably large, the difference between the two options is essentially unaffected by N, as I argued earlier.
The greater benefit with increasing N implies that, when other things are equal, one should act on behalf of a larger group rather than a smaller group.
Matters get more complicated when we consider voting. In order to consider seriously the effect of population size, we must drop the simplifying assumption that P=1/N. If, for example, each of the other voters (of which the number is even) has exactly a .5 chance of voting my way, my chance of being the decisive voter is inversely proportional to ÖN, rather than inversely proportional to N as I have assumed so far. It is easy to show that this changes the conclusion of the last paragraph. As N increases, the relative advantage of political action over spontaneous cooperation increases. Of course, the the probabilty of being decisive, and hence the potential benefit of voting as a political action, declines quickly as the expected probability of voting my way moves away from exactly .5 in either direction.
Moreover, if I am unsure of this expected probability, the probability of being decisive is not proportional to ÖN. Margolis (1983) has argued that it is approximately proportional to 1/N. I am thus inclined to take the initial assumption of P=1/N as a more accurate representation of political action even when voting is involved. Moreover, modern legislators are, it seems, exquisitely sensitive to public opinion. Because they must put together coalitions of voters with concerns about different questions, they pay attention to the sizes of majorities that favor one issue or the other. Even in single-issue referendums, the proportion of voters choosing each side has an effect, beyond whether the referendum wins or loses, on the putting forward of similar referenda and on their terms. Given the existence of modern opinion polling, most people can guess the outcome of elections with considerable accuracy, and most of the time they know that the vote will not be very close. In particular, the outcome is almost never close enough so that the probability of being decisive plays a substantial role in voting, as opposed to the probability of influencing the course of events by being counted on one side or the other. I cannot prove this, of course. But this argument also supports the 1/N assumption that I have made.
If the actor is a perfect utilitarian, W is 1, and the utilities of voluntarism and political action are, respectively, NB-C and PN(NB-C). Both options are always worthwhile, because we have assumed that the action in question is beneficial on the whole, which implies that NB is greater than C. Moreover, if P=1/N, the two options are equally worthwhile.
If the actor is perfectly selfish and rational, W is infinity, and neither option is worthwhile. (It would make more sense here to divide all terms by W, so that the terms dealing with effects on others would disappear. This is allowed because the units of utility are arbitrary.) We have assumed that, in both cases, the personal costs are greater than the benefits.
Under the assumption that P=1/N, political action can be worthwhile, and more worthwhile than voluntarism, only if W is neither 0 nor infinity. Earlier we concluded that political action is more worthwhile than spontaneous cooperation (for large N) when (W-1)C > WA. As W increases, the less it matters that 1 is subtracted from it, so the higher the relative personal utility of political action. In sum, political action is sometimes preferable because we are somewhat selfish, but not completely selfish (or else we would not act in either way). The advantage of political action is that it has little effect, so we can avoid paying the full cost of its success unless it actually succeeds. If W were no greater than 1, political action would never be worthwhile. This is the main point of the present analysis.
In the case of aid, I assumed that the number of contributors was equal to the number of beneficiaries. What happens when this changes? All sorts of situations are possible, and I can only approximate them here. For example, a progressive income taxation scheme is a kind of aid, in which the degree of contribution vs. benefit depends on income. This situation is approximated by the assumption of equal sizes that I made earlier. In foreign aid, the potential beneficiaries are typically more numerous than the contributors, although real foreign aid does not reach this many people. In the case of social welfare, the number of beneficiaries is typically smaller than the number of contributors. Moreover, as I noted, aid situations differ in the group that defines the policy in cases of political action. In foreign aid, it is the contributors alone. In other situations, the beneficiaries can also participate. Call the proportion of beneficiaries R.
Earlier, I considered the effect of R on P, concluding that there was little systematic to be said. R has another effect, however. When it is low, i.e., when the number of contributors is large relative to the number of beneficiaries, each beneficiary gets more benefit relative to the cost borne by each contributor. The opposite is true when R is high.
If utility is completely transferable from person to person this will have no effect. However, the entire rationale for aid denies this assumption. We have no reason to give aid at all unless the goods being transferred have more utility for the beneficiaries than for the contributors, so that total utility increases. Moreover, from a utilitarian point of view, it is senseless to assume that the utility per unit of the good is constant with the amount of goods being transferred, for both groups. Such an assumption would imply that the contributors should donate all of their goods to the beneficiaries. We must assume, instead, that utility of the goods is marginally declining with the amount of goods, at least for the beneficiaries. Thus, Coate (1995) assumed that utility is marginally declining for the beneficiaries but constant for the contributors.
If we make this assumption, then the optimal size of the benefit B does not depend on R. From a utilitarian point of view, it should be set so that the marginal utility of the good to each beneficiary equals the (fixed) marginal utility of the good to each contributor. Once this is done, though, the proportion of beneficiaries will affect C, the size of the contribution required. In particular, C=BR/(1-R). If B is fixed, C decreases as R decreases. As C decreases, the cost of voluntarism decreases, and the contributors ought to be more willing to contribute voluntarily. Moreover C plays a larger role in the utility of voluntarism than in the utility of political action, because, in the latter case, C is borne only with probability P and the main determinant of personal disutility is A, the cost of action itself. So, as R decreases, the cost of political action stays about the same but that of voluntarism decreases. Conversely, when R increases, political action becomes relatively more beneficial compared to its nearly fixed cost.
Insofar as people are sensitive to these considerations, R would affect the social choice between reliance on voluntarism vs.\ government-coerced transfers. Voluntarism would be used more often for small classes of beneficiaries.
This situation also creates opportunities for redefining classes of beneficiaries so that they are larger, making political action worthwhile. There would thus be more political support for ``aid for the disabled'' than for aid for people with a particular disability. Public programs would deal with broader categories than those used by private charities. This is a testable prediction.
I assumed earlier that the number of other cooperators, M, was zero and that the cost and benefit of cooperation did not depend on it. Even under the latter assumption, increases in M decrease the expected benefit to others of political action, which is P(N-1)[(N-M)B-C]. Political action is most useful, then, when others are not already cooperating. On the other hand, when the number of cooperators is extremely low, the probability of success of political action is probably also low. Political action might be most cost-effective when some substantial number of people are already cooperating in the absence of coercion. The recent passage of compulsory recycling laws in the U.S. is a good example of this kind of situation.
In real life, the cost and benefit of cooperation often depends on the number of other cooperators. Recycling laws decreased the cost of cooperation substantially, both for trash services and individuals. Curbside recycling is easier than taking trash to the dump, and the efficiencies of scale increase the profitability of recycling for trash services. Public radio and television provide a contrasting example, in which the benefit of cooperation declines once a station has enough money to do its basic job.
Political action is more effective, compared to voluntarism, in those cases, like recycling, in which the benefit/cost ratio increases with the number of cooperators. It is likely to insure the high levels of cooperation needed for this kind of success. Schelling (1978) and Elster (1989) provide many examples of the great variety of cases in which costs and benefits depend on the number of contributors.
One case of special interest is that of ``step-level public goods provision,'' in which a public good is provided only if the number of cooperators is larger than some number, and otherwise is not provided at all. This is close to the case of voting analyzed here. Voting, however, is closer still to the money-back guarantee condition of Dawes, Orbell, Simmons, and van de Kragt (1986). In that situation, people get their contribution returned if the number of cooperators is insufficient to provide the good. This is analogous to the condition I have assumed for political action, in which the personal cost C-B is not paid unless the action is successful. However, I have assumed that there is some cost, A, in ``offering'' the contribution. The point here, however, is that political participation is itself a kind of cooperation in a second-order social dilemma.
When faced with questions about political support of coercive action vs. voluntarism, many people are inclined to say that political action is better because, if it succeeds, everyone will contribute, and, without that, each contribution will be wasted. For example, Baron (1997) presented subjects with a dilemma about overfishing. Subjects were asked whether they would cut back their fishing voluntarily and whether they would support a rule to make everyone cut back. Many who would not cut back themselves but who would support the rule said that their cutting back would have no noticeable effect. They did not realize that this was probably also true of their political action. They tended to think of political action as making a broad contract with others, of the form, ``I'll cut back if you do.'' Many subjects also saw voluntary cooperation in these terms.
I have just argued that there are situations in which this kind of argument makes sense, such as recycling. An individual's contribution becomes more effective when others are also contributing. The overfishing example I used, however, was carefully constructed so that this was not true. I suspect that people tend to think of contingent cooperation - I'll do it if you will - as a basic heuristic, which they apply to many situations whether it is useful or not. In this sense, people are sometimes non-utilitarian (Baron, 1994).
A related motive for political action is support of fairness in contributing. People want sacrifices to be fairly distributed. Those willing to contribute do not want to be suckers, allowing others to free ride on their contributions or avoid their responsibility. Political action to force contributions from all potential contributors insures fairness in this sense, and it also can reduce the contribution required from each contributor, when there is some limit.
Utilitarian theory can justify this fairness motive to some extent in some cases. If the disutility of contributions is marginally increasing with the size of the contribution for each contributor, for example, then it is for more contibutors to make smaller contributions each than for a few contributors to bear the whole burden. (Philanthropy by rich people is an obvious exception, to this, since they can more easily bear the cost.) In other cases, however, the magnitude of people's concern with fairness goes beyond what utilitarianism can justify.
Coercive programs require enforcement. When the costs of enforcement are high, voluntarism becomes more attractive, other things being equal. (If these costs are voluntary, they are included in C, perhaps as part of another social dilemma. If they are compulsory, they are subtracted from B, as in the case of political action.)
In general, we might expect enforcement costs to be high when many people do not want to participate. Coercive arrangements are almost never completely coercive (Ostrom, 1990). Rather, they simply lower the benefit of defection (C), thus moving people from defection to voluntary cooperation. B-C is still typically negative.
Under such conditions of unwillingness and strong sanctions, we might expect that voluntarism would not do much better. People would refrain from voluntary cooperation for the same reason they would require strong sanctions. But there is one situation in which voluntarism might work well even though coercion would not. This the situation in which individuals vary greatly in their judgment of B. For example, it might make sense to rely on voluntary contributions for such purposes as providing abortions for the poor. Some people strongly oppose the use of their tax money to pay for any abortions, to the point at which the cost of making them pay their taxes increases noticeably. In such cases, B could be negative for some people.
For a somewhat selfish and rational utilitarian, political action is sometimes worthwhile and superior to voluntarism. This situation is more likely to obtain when:
The cost of political action can still be positive. This cost may be reduced by some feeling of personal satisfaction that is independent of consequences, but this reduction is not required. The benefits of participation can be completely proportional to the expected consequences, in an actor motivated by consequentialist concerns for others.
These conclusions are, of course, limited to the modified utilitarian assumptions adopted. Utilitarian and non-utilitarian moral theories, however, differ mainly in the extent to which other factors are relevant. Most of the considerations raised here are relevant in any moral system.
The conclusion is also relevant, as noted, to the rationality of endorsing moral norms in general. Such endorsement is analogous to political action. Moreover, the benefit of norm endorsement increases with the size of the group. The present argument thus supports the endorsement of norms for the largest group possible.
Baron, J. (1986). Tradeoffs among reasons for action. Journal for the Theory of Social Behavior, 16, 173-195.
Baron, J. (1993). Morality and rational choice. Dordrecht: Kluwer.
Baron, J. (1994). Thinking and deciding (2nd ed.). New York: Cambridge University Press.
Baron, J. (1997). The illusion of morality as self-interest: a reason to cooperate in social dilemmas. Psychological Science.
Baron, J. (1996). Norm-endorsement util itarianism and the nature of utility. Economics and Philosophy, 12, 165-182.
Bennett, J. (1981). Morality and consequences. In S. M.\ McMurrin (Ed.), The Tanner Lectures on human values (Vol. 2, pp. 45-116). Salt Lake City: University of Utah Press.
Brennan, G., & Lomasky, L. (1993). Democracy and decision: The pure theory of electoral politics. Cambridge: Cambridge University Press.
Coate, S. (1995). Altruism, the Samaritan's dilemma, and government transfer policy. American Economic Review, 85, 46-57.
Dawes, R. M., Orbell, J. M., Simmons, R. T., & van de Kragt, A. J. C. (1986). Organizing groups for collective action. American Political Science Review, 80, 1171-1185.
Eichenberger, R., & Oberholzer-Gee, F. (1996). Rational moralists: The role of fairness in democratic economic politics. Institute for Empirical Economic Research, University of Zurich.
Elster, J. (1989). The cement of society: A study of social order. Cambridge: Cambridge University Press.
Gibbard, A. (1990). Wise choices, apt feelings: A theory of normative judgment. Cambridge, MA: Harvard University Press.
Hare, R. M. (1981). Moral thinking: Its levels, method and point. Oxford: Oxford University Press (Clarendon Press).
Heckathorn, D. D. (1989). Collective action and the second-order free-rider problem. Rationality and Society, 1, 78-100.
Heckathorn, D. D. (1996). The dynamics and dilemmas of collective action. American Sociological Review, 61, 250-277.
Margolis, H. (1982). Selfishness, altruism, and rationality: A theory of social choice. New York: Cambridge University Press.
Margolis, H. (1983). The Banzhaf fallacy. American Journal of Political Science, , 321-326
Ostrom, E. (1990). Governing the commons: The evolution of institutions for collective action. New York: Cambridge University Press.
Schelling, T. C. (1978). Micromotives and macrobehavior. New York: Norton.
Seip, K. & Strand, J. (1991). Willingness to pay for environmental goods in Norway: A contingent valuation study with real payment. Manuscript, SAF Center for Applied Research, Department of Economics, University of Oslo.
Singer, P. (1979). Practical ethics. Cambridge University Press.
Tversky, A., & Kahneman, D. (1992). Advances in prospect theory: cumulative representations of uncertainty. Journal of Risk and Uncertainty, 5, 297-323.
1This research was supported by
N.S.F. grant SBR95-20288. I thank William C. Hale, Julie
Irwin, and the reviewers for comments. Send correspondence to Jonathan
Baron, Department of Psychology, University of Pennsylvania, 3815
Walnut St., Philadelphia, PA 19104-6196, or (e-mail)
baron@psych.upenn.edu.