Preferential treatment

It is misleading to think decisions reached by voting must reflect the characteristics of underlying individual opinions. It is well known, for instance, that consistency of individual voters' preferences is no guarantee of collective consistency in decisions reached by majority vote. Populations can contain majorities in favour of jointly incompatible propositions, even though no individual believes in all these inconsistent propositions together. Majorities can, for example, prefer marginal public funds to be spent on health rather than education, on education rather than the police and on the police rather than health, even though no one holds to all three of these incompatible preferences. The problem of democratic policy choice in such circumstances are obvious. Kenneth Arrow's famous general possibility result demonstrated the seriousness of the issue by establishing the apparent impossibility of designing any acceptable rule capable of aggregating individual preferences into a consistent social ordering in all circumstances.

This volume, edited by Arrow and two other distinguished contributors to the associated literature, Amartya Sen and Kotaro Suzumura, contains part of the proceedings of a conference on latest developments in the formal analysis of social choice. Closely tied to issues of collective consistency are issues of manipulability. Governments know individual preferences only insofar as voters choose to reveal them, and particular ways of aggregating individual preferences can give voters little reason to express them accurately. The case for "tactical voting" under the UK system of plurality voting in parliamentary elections provides a familiar example in which it can serve someone's interests better to declare a preference rather than the truth. Though sometimes labelled "insincere" voting, this behaviour is typically quite sincere in its attempt to serve the voter's best interests under a system recognised not to reflect preferences well. The problem can be shown to be far more pervasive than this, to the extent that there is no social choice rule satisfying certain appealing conditions that never gives any voter an incentive to disguise his or her true preferences.

Must these problems always be a source of concern? Rules for aggregating preferences can be designed in ways that guarantee consistent outcomes and give no incentive to misdeclare preferences if those preferences are known to be sufficiently homogeneous in certain respects. (Of course, this does not mean that manipulability problems cannot still exist under other voting rules, such as plurality voting, even with such well-behaved preferences.) This has been known for some time but Salvador Barbera provides a useful summary of recent research on manipulability that deepens understanding of this point. The existence of mechanisms that implement social choice rules under specific notions of equilibrium is something about which research has given more reason to be hopeful. Bhaskar Dutta's interesting paper points such research in an interesting direction by seriously considering the practicality of such mechanisms. Much recent attention has been focused on the problem of ensuring that social choice rules protect individual rights. Majority voting offers little safeguard against abuse of individual rights - the danger of majorities exploiting minorities is widely recognised. Sen, more dramatically, has a famous example that suggests that not even unanimity voting can protect individual rights if individuals care more about intervening in others' affairs than about protecting their own spheres of action, for reasons of censoriousness, mischief or whatever. One response to this is to question the implicit formulation of rights underlying the argument. Perhaps what matters for whether rights are respected is not how outcomes of social choice relate to individual preferences but whether social institutions allow individuals to act in ways such as to protect their rights if they so choose. Peter Hammond and Prasanta Pattanaik explore the promising suggestion that rights should be thought of formally in terms of restrictions on availability of strategies in social games. As Hammond points out: "Social systems are game forms ... from which social states or consequences emerge as a result of individuals' strategic decisions." In this, as in several other contributions, the value of the game-theoretic perspective on social choice questions is clearly brought out.

Charles Blackorby, Walter Bossert and David Donaldson consider the extension of social choice ideas to questions regarding population size. The issues raised here are tricky. How, if at all, should choice of population numbers relate to the hypothetical preferences of potential people? One superficially appealing suggestion holds that it is better for an extra person to be alive if they would prefer to be so. Combining this, however, with a preparedness to weigh up one potential person's welfare against another's seems to draw one ineluctably towards what Derek Parfit has elsewhere called the "repugnant conclusion". No matter how happy were to be the lives of a potential future community, it would seem always to be acceptable to drive its members into poverty if this were to allow the creation of enough extra people barely happy to be alive. Basing judgement on the average well-being of populations denies the suggested principle but seems to imply the equally repellent conclusion that it would in some sense be better if those less happy than average were not to have existed. Blackorby, Bossert and Donaldson consider the foundations of criteria for choice in this area. Here, as in other papers, one sees the potential for conflict between seemingly attractive principles and the benefit of rigorous thought about issues that invite easy but questionable responses.

Ian Preston is lecturer in economics, University College London.