Unformatted text preview:

Social Choice Josh Cohen 1. The Voting ParadoxThe puzzles in the area of social choice begin with the paradox of voting. So suppose we as a group are trying to decide when the course should meet: 9, 10, or 11. Suppose that choosing rationally (whether for a group or person) is choosing the best alternative from a consistently ordered set of feasible alternatives, and that the group ranking is fixed by majority support: A is ranked higher than B by the group iff A is ranked higher than B by a majority of group members. Assume finally that consistency is a matter of transitivity of “better than”: that if A is better than B and B is better than C, then A is better than C (we will see later that there are ways to weaken this notion). Assume that we divide into three groups of equal size: Best Second Worst I 9 10 11 II 11 9 10 III 10 11 9 Lets ask how the group ranks the three possibilities, using majority rule as the procedure for group ranking: 9 beats 10, 10 beats 11, and 11 beats 9. Group rationality (i.e. transitivity) requires that 9 beat 11. But 11 is a majority rule winner over 9. In fact matters are worse: putting aside the failure of transitivity, notice that each alternative loses to another alternative. Thus we have not just a failureof transitivity, but further a failure of acyclicity, which is an even more minimal condition of rational decision: Acyclicity: If x1Px2Px3...xn-1Pxn, then not xnPx1 A violation of acyclicity means that the feasible set contains no best element. If rationality means choosing the best from the feasible, then there is no basis here for rational choice. One might wonder whether there is something very special about this case. The answer appears to be: No. Assume we have at least three people and a set of alternatives that vary continuously on two dimensions (e.g., guns and butter; spending on environmental regulation and economic growth). Two things turn out to be true: Using majority rule as the basis of the collective ordering, there is virtually never a best element—that is, everything loses to something.1 And, what is worse, the majority rule cycle is global. That is, beginning from any alternative, we can arrive at any other point by finitely many applications of majority choice: including points that are worse for everyone that the original (outside the pareto set). So if I tell you the preferences of the members of a group and that the group will make its decisions by majority rule, this information, taken on its own, will not enable you to narrow the class of outcomes at all. The product of majority-rule procedures will simply a result that reflects features of circumstances (say, the order of decision-making, permissible amendment procedures, institutions of decision-making), but will lack any deeper rationale in 1 Special symmetry conditions are required to prevent this.a rational group decision based (via majority rule) on individual judgments, values, and tastes. 2. Social Choice TheoryFaced with these majority rule troubles, one might wonder if there are special troubles with rational group decisions using majority rule as a method of aggregation. Perhaps there is some alternative to majority rule that enables us to base a group or collective ranking of alternatives on rational individual rankings of those alternatives, where the collective ranking is itself rational. Thus Arrow says (SCIV, 2): “The methods of voting and the market…are methods of amalgamating the tastes of many individuals in the making of social choices. The methods of dictatorship and convention are, or can be, rational in the sense that any individual can be rational in his choices. Can such consistency [rationality] be attributed to collective modes of choice, where the wills of many people are involved?” Social choice theory explores this possibility: of basing rational collective choices on rational individual choices. The field of social choice theory provides a way to formalize these two fundamental ideas: (1) the idea that a group or collective ranking is constructed from rational individual rankings, with the members treated as equals; and (2) the idea that the group acts rationally. We can then study whether the difficulties in the case of majoritarian conception of group preference can be circumvented if we opt for some other account of groupjudgment, while preserving the idea that this judgment must bear some systematic connection to the rankings of alternatives by individual members. Framework I want now to sketch out the main elements of the social choice theory framework. In each case, I will make some comments on interpretation that will figure in later discussion. 1. Alternatives. Individuals face a set of alternatives (policies or outcomes) that they personally rank and over which we are looking to characterize a group ranking. Let the set of alternatives be X={x, y, z,...}. In Arrow, the alternatives ordered by individuals are vectors that represent total social states: complete specifications of all the goods of each person, the resources devoted to each economic activity, the level of government activity, etc.2 2. Individuals. There is a set of individuals I={1,2,...,n}, each of whom has an ordering R (complete, reflexive, transitive) over X: thus, we read xRiy as: foriindividual i, x is at least as good as y; xPiy as x is preferred to y; and xIiy as individual i is indifferent between x and y. We define R in terms of P and I: xRy =df xPy or xIy. Completeness: For all x,y in X, either xRiy or yPix. Reflexivity: for all x in X, xRix Full Transitivity: for all x,y,z in X, if xPy and yPz, then xPz; if xIy and yIz, then xIz; (and similarly for PI/P and IP/P) 2 Arrow, 17.iAn array of such orderings, one for each individual, is called a profile: [R]. Because X is finite, the ordering can be represented by an ordinal utility function. For every person, there is a function ui from the set X of alternatives to the real numbers, such that xPiy iff ui(x)>ui(y). And of course if there is one such function, there an infinitely many others. 3. Social Preference Relation. Let R be the social preference relation: so 'xPy' is to be read as "x is socially better than/preferred to y;" 'xRy' that "x is socially at least as good as y." What “social preference” comes to is not so clear, and two interpretations are available. One is that the social ranking represents a ranking of alternative social states to be used by officials—or by


View Full Document

MIT 17 960 - Social Choice

Download Social Choice
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Social Choice and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Social Choice 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?