Strategy-proof social choice on multiple singlepeaked domains and preferences for parties

Size: px
Start display at page:

Download "Strategy-proof social choice on multiple singlepeaked domains and preferences for parties"

Transcription

1 Strategy-proof social choice on multiple singlepeaked domains and preferences for parties Alexander Reffgen August 26, 2011 Abstract The concept of single-peaked domains is generalized to multiple single-peaked domains, where the set of alternatives is equipped with several underlying orderings with respect to which a preference can be single-peaked. We provide a complete characterization of the strategy-proof social choice functions on multiple single-peaked domains. We show also in the framework of a spatial voting model for party elections that multiple single-peaked domains are appropriate to represent preferences over parties. Keywords: Single-peaked preferences, Strategy-proofness, Party preferences, Efficiency, Degree of manipulability. The author would like to thank John Weymark, Hans Peters, Dolors Berga, Lars-Gunnar Svensson, Tommy Andersson, Håkan Holm, and Philipp Wichardt for very helpful comments. Financial support from The Jan Wallander and Tom Hedelius Foundation is gratefully acknowledged. Alexander Reffgen. Department of Economics, Lund University, Box 7082, SE Lund, Sweden. Fax: alexander.reffgen@nek.lu.se 1

2 1 Introduction The central concern in the theory of strategy-proof social choice is to investigate the conditions under which it is possible to construct voting procedures that are incentive compatible in the sense that voters can never gain from strategic misrepresentation of their preferences. In general, this problem can not be solved in a satisfactory way: if there are three or more eligible alternatives, then by the Gibbard-Satterthwaite theorem (Gibbard 1973; Satterthwaite 1975) only the dictatorial voting procedures are strategy-proof. A crucial prerequisite for this result is, however, the unrestricted domain assumption, which means that voters are allowed to hold any conceivable preference. At first glance, this is a plausible assumption because the exact shape of a voter s preference is private information, and therefore it is strictly speaking impossible to classify some preference for certain as infeasible. But at the same time, accepting the unrestricted domain assumption also means not taking into account the fact that preferences do not come up in an arbitrary way, but are normally the result of a deliberate mental process, and it is therefore likely that the domain of preferences exhibits some structure (see, for instance, Sen (1970, p. 165)). This objection has inspired a large number of studies that consider restricted preference domains with different structures and investigate which voting procedures are strategy-proof on these domains. Thereby, it turned out that some domains admit large classes of non-dictatorial strategy-proof voting procedures (e.g., Moulin (1980) and Barberà et al. (1991)), while other domains lead back to the same conclusion as in the Gibbard-Satterthwaite theorem (e.g., Aswal (2003)). Among all restricted preference domains, the domain of single-peaked preferences, introduced by Black (1948), has been especially influential. For single-peaked domains, the structure of preferences comes up as a consequence of the nature of the available alternatives. A necessary prerequisite for single-peaked preferences is that the available alternatives have a common characteristic that makes it possible to order these alternatives in an objective way in a line from, say, the left to the right. Given this ordering, a voter forms his preference by first choosing a most preferred alternative, and the further he then moves away from his top to either the left or the right, the less preferred are the alternatives. Single-peaked preferences are reasonable in a wide range of applications. To begin with, it is likely that preferences are single-peaked when the value of a quantitative variable must be chosen, e.g., the height of a particular tax rate, the wage rate of labor, or the rate of CO2 emission (examples of this type were studied in Black (1948)). Further, when the alternatives are qualitative, there may also be a 2

3 natural ordinal ordering of the alternatives which then again induces single-peaked preferences; e.g., this is the case when the voting concerns what civil rights to assign to a certain group in the society, and the alternatives are ordered on a scale from no rights via basic democratic rights such as freedom of speech and franchise to full rights including membership in social security and pension system. More generally, in many political decisions the available alternatives can be ordered on an ideological left right scale, and therefore single-peaked preferences also play an important role in political economy; for example, an extensive social security system and economic regulations are traditionally considered as left-wing politics, while free markets and a broad protection of private property are considered as rightwing politics. A particularly important example of political votings are of course elections involving parties or candidates for official positions, and also in this context it is common to assume a left right ordering and single-peaked preferences (see, for example, the seminal work by Downs (1957) on electoral competition). A crucial assumption for single-peaked preference domains is, however, that all voters apply the same underlying ordering of the available alternatives when they form their single-peaked preferences. This is almost undisputable in the case of quantitative variables, and it is also reasonable in many cases of qualitative variables with a clear ordinal ordering. But when it comes to political alternatives, there is often no equally convincing objective ordering, and hence the assumption of one common underlying ordering accepted by all voters may be unjustified. In particular for the case of political parties, there is empirical evidence casting doubt on the hypothesis that voters preferences are single-peaked with respect to a common ordering of the parties: in several studies performed in the U.S. and Germany, it turns out that for any given left right ordering of the available parties there is normally a large proportion of voters whose preferences are not single-peaked with respect to that ordering (e.g., Niemi and Wright (1987), Feld and Grofman (1988), Pappi and Eckstein (1998)). Instead, it appears to be more likely that voters have different perceptions of how to place political alternatives on a left right scale, but given their personal understanding of the ordering of the alternatives, they still form single-peaked preferences. To describe this preference structure formally, we introduce in this article the concept of a multiple single-peaked preference domain. By this, we mean the union of several (ordinary) single-peaked domains which are constructed with respect to different underlying orderings of the alternatives. Multiple single-peaked domains cover a wide range of domains: at the one extreme, when there is only one underlying ordering, there are of course the usual single-peaked domains, and, at 3

4 the other extreme, the unrestricted domain is also a multiple single-peaked domain. The main purpose of this article is to provide a complete characterization of the strategy-proof voting procedures on multiple single-peaked domains. For singlepeaked domains, Moulin (1980) has shown that a voting procedure is strategy-proof if and only if it is a generalized median voter scheme. Since multiple single-peaked domains consist of several single-peaked domains, a strategy-proof voting procedure on a multiple single-peaked domain must be such that the restriction of the voting procedure to each of the single-peaked domains that belongs to the multiple domain is a generalized median voter scheme. However, a voting procedure that is a generalized median voter scheme on one single-peaked domain is not necessarily a generalized median voter scheme on another single-peaked domain, because the two domains are based on different orderings of the available alternatives, and therefore the voting procedure is not necessarily strategy-proof on the second domain either. Thus, to obtain a strategy-proof voting procedure on the multiple single-peaked domain, we must find a coordination condition to ensure that a voting procedure, which is a generalized median voter scheme on one single-peaked domain belonging to the multiple domain, is also a generalized median voter scheme on any other single-peaked domain that belongs to the multiple domain. In Theorem 1, we show that the following condition is necessary and sufficient for a voting procedure on a multiple single-peaked domain to be strategy-proof: When we compare the different orderings that generate the single-peaked domains belonging to the multiple domain, we can divide the set of alternatives into a left and a right component where all orderings coincide, and a middle component where the orderings differ. A strategy-proof voting procedure must, to begin with, be dictatorial on the middle component. On the left component, the voting procedure applies first a generalized median voter scheme for all voters except the middle component dictator, and, comparing the resulting alternative with the top alternative of the dictator, the voting procedure then chooses the alternative that is closest to the middle component. A similar procedure applies to the right component. Intuitively, the class of strategy-proof voting procedures on a multiple single-peaked domain thus falls in between the class of all generalized median voter schemes, which are the strategy-proof rules on a traditional single-peaked domain, and the class of dictatorial rules, which are the only rules that are strategy-proof on the unrestricted domain. So far, we have argued somewhat loosely that traditional single-peaked domains are not sufficient to represent preferences over parties, but that these preferences belong rather to a multiple single-peaked domain. In the second part of this article, 4

5 we present therefore a more thorough motivation for the use of multiple singlepeaked domains to represent preferences over parties. A first thought on the nature of parties already reveals that parties differ in one important respect from most of the other examples of single-peaked domains mentioned above: In general, parties are not single-decision alternatives, but can be considered as bundles of alternatives from different political dimensions, such as economic policy or social and civil rights policy, which they would implement if elected (see, e.g., Austen-Smith (1996)). To derive the structure of preferences over parties from the nature of parties as bundles of alternatives, we construct a spatial voting model in which parties are represented as elements in a multi-dimensional set of political alternatives. Imposing a weak local structure on voters preferences over the multi-dimensional set of political alternatives, we show then that voters preferences over parties constitute a multiple single-peaked domain (Proposition 1), where the exact number and shape of the underlying orderings depend intimately on how the parties are located in the multi-dimensional set. This article is organized as follows: Section 2 introduces the social choice model. Section 3 provides a complete characterization of the strategy-proof voting procedures on multiple single-peaked domains. In Section 4, we show that multiple single-peaked domains are appropriate to represent preferences over parties. Section 5 considers the trade-off between strategy-proofness, anonymity, and efficiency on multiple single-peaked domains. All proofs are collected in the appendix. 2 The social choice model and some important results from the literature 2.1 The basic model The basic formal framework of this study is as follows: Let N = {1,...,n} be a finite society of n individuals, who consider a finite set A of m social alternatives. The individuals have complete, transitive, and asymmetric 1 preferences over the alternatives in A, and the preference of individual i is denoted by P i. Given P i, the kth ranked alternative (1 k m) in P i is denoted by r k (P i ), which means that #{a A; ap i r k (P i )} = k 1. The set of all complete, transitive, and asymmetric 1 A preference P on A is complete if for all a,a A, a a, either apa or a Pa; P is transitive if apa and a Pa implies apa for all a,a,a A; finally, P is asymmetric if, for all a,a A, apa implies that a Pa does not hold. 5

6 preferences over A is called the unrestricted domain over A, and it is denoted by Σ A. The set of all admissible preferences, referred to as the preference domain, is denoted by Ω, and Ω is normally a strict subset of Σ A. If P Ω and B A, then the restriction of P to B is that preference P B on B which satisfies bp B b if and only if bpb for all b,b B. The domain of preferences obtained when every P Ω is restricted to B is denoted by Ω B = {P B ; P Ω}. The preferences of the n individuals in the society are collected in a (preference) profile (P 1,...,P n ) Ω n. In order to focus on individual i s preference, a profile (P 1,...,P n ) can be rewritten as (P i,p i ), where P i Ω n 1 thus denotes the profile of all individuals except individual i. A social choice function (SCF) is a mapping f : Ω n A that assigns to every profile (P 1,...,P n ) Ω n a unique social choice a A. For a given SCF f, the set of all a A that can be attained by f is called the range of f and is denoted by R f, i.e., R f = {a A; a = f (P 1,...,P n ) for some (P 1,...,P n ) Ω n }. If R f = A, then f is surjective. An SCF f : Ω n A has the tops-only property if the choice of f depends only on the top alternatives in every profile, i.e., f (P 1,...,P n ) = f (P 1,...,P n) for all (P 1,...,P n ) Ω n and (P 1,...,P n) Ω n such that r 1 (P i ) = r 1 (P i ) for all i N. Further, an SCF f : Ω n A is manipulable at the profile (P 1,...,P n ) Ω n if there is some i N and some P i Ω such that f (P i,p i)p i f (P i,p i ). If f cannot be manipulated at any admissible profile, then f is strategy-proof. Finally, f : Ω n A is dictatorial on B A if there is some i N, called the dictator on B, such that f (P i,p i ) = r 1 (P i ) for all (P i,p i ) Ω n with r 1 (P i ) B. If f is dictatorial on A, then f is shortly called dictatorial. In this basic model, the reference point in the theory on strategy-proof social choice is the well-known Gibbard-Satterthwaite theorem (Gibbard 1973; Satterthwaite 1975), which states the general incompatibility of strategy-proofness and nondictatorship. The Gibbard-Satterthwaite theorem. Let A be a finite set of at least three alternatives. A surjective SCF f : Σ n A A is strategy-proof if and only if f is dictatorial. 2.2 Social choice on single-peaked domains Given the basic model above, we now introduce a structure for the preference domain Ω which generalizes the structure of single-peaked preferences. To this end, we recall first the definition of single-peaked preferences, starting with the concept of an underlying ordering of A: An ordering S of A is a complete, transitive, and asymmetric binary relation on A, and if asb for a,b A, we say that a is to 6

7 the left of b, or equivalently, that b is to the right of a. The set of all orderings of A is denoted by Σ A, exactly as for preferences. For a given ordering S Σ A, denote by ρ k (S) the kth alternative in A from the left (1 k m), i.e., ρ k (S) is the unique alternative in A that satisfies #{a A; asρ k (S)} = k 1. The total ordering ρ 1 (S)Sρ 2 (S)S... Sρ m (S) of A by S is in short represented as a vector S = [ρ 1 (S) ρ 2 (S)... ρ m (S)]. If S Σ A, define the inverse ordering S 1 of S by setting as 1 b if and only if bsa for all a,b A. Further, for B A, define the maximum max S (B) of B with respect to S to be the unique alternative b B which satisfies bs b for all b B \ { b}. Similarly, the minimum min S (B) of B with respect to S is that alternative b B which satisfies bsb for all b B \ {b}. In particular, note that min S (A) = ρ 1 (S) and max S (A) = ρ m (S). To every S Σ A, we associate a ternary betweenness relation B S on A by setting B S (a,b,c) for a,b,c A if and only if either asbsc or csbsa, and if B S (a,b,c) we say that b is between a and c. Further, for a,b A with asb or a = b, the set [a,b] S {x A; x {a,b} or B S (a,x,b)} is called the (closed) interval of a and b with respect to S; we also define the two half-closed intervals ]a,b] S [a,b] S \{a} and [a,b[ S [a,b] S \{b}. Finally, the projection of x A on an interval [a,b] S with respect to S, denoted by π S [a,b] S (x), is defined as follows: π S [a,b] S (x) = x if x [a,b] S ; π S [a,b] S (x) = a if xsa; and π S [a,b] S (x) = b if bsx. Given some ordering S Σ A, a preference P Σ A is single-peaked with respect to S if B S (r 1 (P),a,b) implies apb for all a,b A. Equivalently, P is single-peaked with respect to S if B S (a,b,c) implies bpa or bpc for all triples a,b,c A. The first of these two definitions gives a global characterization of single-peaked preferences: if the alternatives in A are set out in a line, then a utility function representing P is upward sloping towards the top of P, and downward sloping afterwards. 2 The second definition can instead be seen as a local characterization: it considers one triple at a time and requires that the alternative that is in the middle according to S must not be ranked worst in the triple by P. 3 It is easily checked that these two definitions are equivalent in the given framework where A is one-dimensional. If the set of alternatives is multi-dimensional, however, then this equivalence no longer holds, which is explained in Section 4. Finally, the set of all P Σ A that are single-peaked with respect to some given ordering S Σ A is called the single-peaked domain with respect to S and denoted by Ω S. When preferences belong to a single-peaked domain, then strategy-proofness 2 In this way, single-peaked preferences were introduced in Black (1948). 3 This requirement is a special case of the more general condition of value-restricted preferences introduced in Sen (1966). 7

8 and non-dictatorship do no longer exclude each other, but instead there is a large class of strategy-proof and non-dictatorial SCFs. The following result by Moulin (1980), which is the reference point in the theory of strategy-proof social choice on single-peaked domains, provides a complete characterization of the strategy-proof tops-only SCFs on single-peaked domains. Moulin (1980, Proposition 3) An SCF f : Ω n S A is strategy-proof and has the tops-only property if and only if there exists a family {a T } T N of fixed alternatives from A such that S f (P 1,...,P n ) = min { max S {a T,r 1 (P j )} } for all (P 1,...,P n ) Ω n S. (2.1) T N j T The minmax SCFs in (2.1) are known as generalized median voter schemes in the literature. These SCFs can be interpreted as follows. Every coalition T N of individuals proposes one alternative from A using a max rule, and the scheme then selects the smallest of these proposed alternatives. Thereby, the family {a T } T N of fixed alternatives can be considered as preset votes for all T N, and a coalition T can only deviate from a T if some i T moves the vote to the right. In this sense, a T delimits the possibilities of coalition T to influence the location of the social choice, and, in particular, the more a T is to the right in A, the less power has T to influence f. For a concrete example, consider a dictatorial, and hence strategy-proof SCF f : Ω n S A. To represent f as a generalized median voter scheme, let î N be the dictator for f. If a {î} = ρ 1 (S) and a T = ρ m (S) for all T N with T {î}, then it is easily checked that the righ-hand side in (2.1) reduces to r 1 (Pî). Choosing {a T } T N in this way, all power is thus assigned to the singleton coalition {î}, and no power to any other coalition. From a normative perspective, an important subclass of the generalized median voter schemes consists of the schemes that treat all individuals equally and that hence are called anonymous. Formally, an SCF f : Ω n A is anonymous if f (P 1,...,P n ) = f (P σ(1),...,p σ(n) ) for all permutations σ of N and all profiles (P 1,...,P n ) Ω n. A generalized median voter scheme is obviously anonymous if a T = a T whenever #T = #T. A well-known example of an anonymous strategyproof SCF on a single-peaked domain Ω S is the median rule, which assigns to every profile (P 1,...,P n ) Ω n S the median of the n top alternatives r 1(P i ), i N, in the profile. 4 In fact, Moulin (1980) showed that every anonymous strategy-proof SCF 4 The median med(x 1,...,x n ) of n alternatives x 1,...,x n is the alternative that ends up in the middle when these alternatives are ordered according to S, i.e., if x = med(x 1,...,x n ), then #{x i ; x i 8

9 f : Ω n S A that satisfies the tops-only property can be represented as the median of the n top alternatives of the individuals in N and n + 1 fixed alternatives in A. The characterization of the strategy-proof SCFs in (2.1) is based on the assumption that all individuals in the society apply the same ordering S when they form their single-peaked preferences. As discussed in the introduction, this may be a reasonable assumption in many situations, but in other cases it is more likely that the individuals form their single-peaked preferences along different orderings of A. This leads to the following central definition of this article. Definition 1. Let S = {S 1,...,S q } be a family of q orderings S τ Σ A, 1 τ q. The multiple single-peaked domain Ω S with respect to S is Ω S q τ=1 Ω S τ. Further, Ω S is called non-trivial if there exist S,S S such that S S and S 1 S. A multiple single-peaked domain consists of a number of ordinary single-peaked domains which are constructed with respect to different orderings of A. Thus, if P Ω S, then P is single-peaked with respect to at least one S S. Note that if Ω S is not non-trivial, then either #S = 1 or S = {S,S 1 } for some S Σ A, and in both cases, Ω S is an ordinary single-peaked domain. This is obvious if #S = 1, and if S = {S,S 1 }, this follows from the fact that B S (a,b,c) for a,b,c A if and only if B S 1(a,b,c), which means Ω S = Ω S 1. The class of multiple single-peaked domains covers a wide range of domains, which is illustrated in the following example. Example 1. At the one extreme, if #S = 1, the ordinary single-peaked domains turn out as a special case of multiple single-peaked domains. At the other extreme, also the unrestricted domain is a special case of multiple single-peaked domains which is obtained for instance if S consists of all possible orderings of A, i.e., if S = Σ A. For a non-trivial example of a multiple single-peaked domain in between an ordinary single-peaked domain and the unrestricted domain, consider the set A = {a,b,c,d}. Suppose that the individuals in the society consider unanimously a to be the leftmost and d to be the rightmost alternative, but that there is disagreement on how to order the two middle alternatives b and c. Then, an individual may either use the ordering S 1 = [a b c d ] or S 2 = [a c b d ] to form his single-peaked preference, and the preference domain is Ω {S1,S 2 }. For a slightly more complicated example, again let A = {a,b,c,d}, and suppose now that there is consensus in the society that A can be divided into a left block consisting of a and b, and a right block [ρ 1 (S), x] S } n/2 and #{x i ; x i [ x,ρ m (S)] S } n/2. Further, if n is odd, then the median rule can be represented as a generalized median voter scheme with a T = ρ 1 (S) if #T = (n+1)/2 and a T = ρ m (S) of #T (n + 1)/2. 9

10 consisting of c and d, but there is no agreement on how to order these two blocks internally. In this case, the collection S generating the domain Ω S consists of the four orderings S 1 = [a b c d ], S 2 = [b a c d ], S 3 = [a b d c], and S 4 = [b a d c]. As a matter of fact, it turns out that Ω S = Σ A. Remark 1. An alternative way to interpret multiple single-peaked domains is as follows. If S is the underlying ordering of an ordinary single-peaked domain Ω S, then the associated betweenness relation B S is complete in the sense that for every triple of distinct alternatives from A, there is always exactly one alternative that is between the other two, and this alternative cannot be ranked worst in the triple. On the contrary, a multiple single-peaked domain can no longer be associated with a complete betweenness relations. For example, if Ω S is generated by S 1 = [a b c d ] and S 2 = [a c b d ], then b is unambiguously between a and d, but for the three alternatives a, b, and c, there is no unique middle alternative and every alternative can be ranked worst in the triple. Allowing for several underlying orderings can thus be seen as a relaxation of the assumption of a complete betweenness relation. In this context, we want to mention that an important alternative generalization of single-peaked domains, as derived from an abstract betweenness relation, has been considered in Nehring and Puppe (2007b). The deep analysis in that article is too different from our approach to motivate a detailed comparison of the two models, but it should nevertheless be pointed out that they are logically independent. 5 3 Strategy-proof social choice on multiple single-peaked domains We now provide a complete characterization of the strategy-proof SCFs on multiple single-peaked domains. It turns out that given a collection S = {S 1,...,S q } of orderings of A, the class of strategy-proof SCFs on Ω S depends on how similar the orderings in S are. In particular, to what extent the orderings in S coincide at their left and right ends will be of importance. To make this precise, we introduce 5 For the interested reader, it can be mentioned that this logical independence can be established easily by means of examples. To begin with, for many of the examples of rich single-peaked domains considered in Nehring and Puppe (2007b) it is almost obvious that they cannot be represented as multiple single-peaked domains. For a simple converse example, it is straightforward to check that the multiple single-peaked domain Ω S generated by the two orderings S 1 = [a b c d ] and S 2 = [b a d c] does not satisfy the closure condition in Theorem 1 in Nehring and Puppe, and therefore, cannot be represented as a rich single-peaked domain in the sense of Nehring and Puppe. 10

11 first a way to decompose the set of alternatives A into three components, namely a left component A L S, a middle component AM S, and a right component AR S, such that A = A L S AM S AR S, and all orderings in S coincide on the left and on the right component, while there is some disagreement on how to order the alternatives in the middle component. Thereby, we must take into account the fact that if S and S are inverse orderings to each other, then the single-peaked domains Ω S and Ω S are actually identical because B S (a,b,c) if and only if B S (a,b,c) for all a,b,c A. When we analyze the similarity of the orderings in S, we must therefore direct these orderings in a similar way, because the class of strategy-proof SCFs on Ω S depends of course only on dissimilarities among the orderings in S that enlarge the preference domain Ω S, but not on dissimilarities that depend on different directions of the orderings, and that hence do not have any impact on Ω S. We now present the formal construction of the decomposition of A: (1) Consider first the case when there exist two orderings S,S S such that { ρ1 (S),ρ m (S) } { ρ 1 (S ),ρ m (S ) } =. (3.1) This means that none of the endpoints of S and S coincide, even if we take the inverse of one of the two orderings. Then there cannot be some common left or right component for all orderings in S, and therefore we set A L S = AR S = and A M S = A. (2) Consider now the case when there exist no two orderings in S for which (3.1) occurs. Then, replacing some of the orderings in S with their inverse orderings if necessary, we can assume that ρ 1 (S) = ρ 1 (S ) for all S,S S. If now all orderings in S are identical, set A L S = A and AM S = AR S =. Otherwise, there exists a unique integer l with 1 l < m such that ρ k (S) = ρ k (S ) for all k with 1 k l and all S,S S, and ρ l+1 (S) ρ l+1 (S ) for some S,S S. Then, take some S S and set A L S = l k=1 {ρ k (S)}, and note that A L S does not depend on the particular choice of S. We distinguish two cases: (i) If the orderings in S do not have a common right endpoint, i.e., if ρ m (S) ρ m (S ) for some S,S S, set A R S = and AM S = A \ AL S. (ii) If (i) is not the case, then all orderings in S have a common right endpoint, i.e., ρ m (S) = ρ m (S ) for all S,S S. Then there exists a unique integer r with l < r m such that ρ k (S) = ρ k (S ) for all k with r k m and all S,S S, but ρ r 1 (S) ρ r 1 (S ) for some S,S S. Take some S S and set A R S = m k= r {ρ k (S)}, and note that A R S does not depend on the particular choice of S. Finally, set A M S = A \ (AL S AR S ). 11

12 The triple (A L S,AM S,AR S ) is called the maximal common decomposition of A with respect to S. The sets A L S, AM S, and AR S are pairwise disjoint, and AL S AM S AR S = A. The maximal decomposition of A is unique up to a possible inversion in the sense that if (A L S,AM S,AR S ) and (ĀL S,ĀM S,ĀR S ) are two maximal common decompositions of A, then A M S = ĀM S and either AL S = ĀL S and AR S = ĀR S, or AL S = ĀR S and AR S = ĀL S. Note also that all orderings in S, some of them if necessary replaced by their inverse orderings, order the alternatives in A L S and AR S in exactly the same way, and if al A L S, am A M S, and ar A R S, then al Sa M Sa R for all S S. We now consider the maximal common decomposition in two concrete examples. Example 2. Consider the set A = {x 1,x 2,x 3,x 4,x 5 }, and suppose that A is equipped with the two orderings S 1 = [x 1 x 2 x 3 x 4 x 5 ] and S 2 = [x 2 x 1 x 3 x 5 x 4 ]. In this case, the sets of endpoints {ρ 1 (S 1 ),ρ 5 (S 1 )} and {ρ 1 (S 2 ),ρ 5 (S 2 )} are disjoint, which means that the maximal common decomposition of A with respect to S = {S 1,S 2 } is (A L S,AM S,AR S ) = (,A, ). Example 3. As a second example, now let A = {x 1,...,x 7 } be a set of seven alternatives, and suppose that S = {S 1,S 2,S 3 } consists of the following three orderings: S 1 = [x 1 x 2 x 3 x 4 x 5 x 6 x 7 ], S 2 = [x 7 x 5 x 6 x 4 x 3 x 2 x 1 ], S 3 = [x 1 x 2 x 3 x 5 x 4 x 6 x 7 ]. Here, the sets of endpoints {ρ 1 (S 1 ),ρ 7 (S 1 )}, {ρ 1 (S 2 ),ρ 7 (S 2 )}, and {ρ 1 (S 3 ),ρ 7 (S 3 )} coincide for all three orderings, and hence both A L S and AR S are non-empty. To find the maximal common decomposition of A with respect to S, replace first S 2 by its inverse ordering S2 1 = [x 1 x 2 x 3 x 4 x 6 x 5 x 7 ] to make the three orderings have the same left endpoint. Then it is easily checked that A L S = {x 1,x 2,x 3 }, A M S = {x 4,x 5,x 6 }, and A R S = {x 7}. Using the maximal common decomposition of A with respect to S, we can now give a complete characterization of all strategy-proof surjective SCFs on Ω S. Theorem 1. Let A be a finite set of m alternatives, and suppose that Ω S is a non-trivial multiple single-peaked domain over A. Further let (A L S,AM S,AR S ) be a maximal common decomposition of A with respect to S. Then a surjective SCF f : Ω n S A is strategy-proof if and only if π S ( [r 1 (Pî),a] GL (P S î ) ) if r 1 (Pî) Ā L f (Pî,P î ) = r 1 (Pî) if r 1 (Pî) Ā M (3.2) π[a,r S ( 1 (Pî)] GR (P S î ) ) if r 1 (Pî) Ā R 12

13 for all (Pî,P î ) Ω n S, where (1) î N, (2) S S, (3) ĀM = [a,a] S for some a A L S {ρ 1(S)} and a A R S {ρ m(s)}, and further Ā L = A L S \ ĀM and Ā R = A R S \ ĀM, and (4) G L : Ω n 1 S A and G R : Ω n 1 S A are generalized median voter schemes, defined for all individuals except individual î, with ρ 1 (S) R GL and ρ m (S) R GR, i.e., there are two families {a G L T } T N\{î} and {a G R T } T N\{î} of fixed alternatives in A with a G L T = ρ 1(S) for some T N \ {î} and a G R = ρ m (S) such that G L (P î ) = min S T N\{î} and G R (P î ) = min S T N\{î} { max S {a G L j T T,r 1(P j )} } for all P î Ω n 1 S, { max S {a G R j T T,r 1(P j )} } (3.3) for all P î Ω n 1 S. Theorem 1 characterizes the strategy-proof surjective SCFs on a multiple singlepeaked domain as follows. To begin with, A is divided into three parts Ā L, Ā M, and Ā R, which are intervals with respect to some ordering S S. The condition Ā M = [a,a] S, where a A L S {ρ 1(S)} and a A R S {ρ m(s)}, means that the left endpoint a of Ā M belongs to A L S if AL S is non-empty, and the right endpoint a belongs to A R S if AR S is non-empty; if AL S is empty, then a equals ρ 1(S), i.e., the left endpoint of A M S with respect to S, and if AR S is empty, a equals the right endpoint ρ m(s) of A M S. Thus, ĀM always contains A M S, and, if possible, ĀM continues to at least one alternative in A L S and AR S. Given ĀM, the set Ā L (Ā R ) is that part of A L S (AR S ) which is not covered by Ā M. If f now is a strategy-proof surjective SCF, then some individual î must be a dictator on Ā M. Further, if individual î reports a top alternative r 1 (Pî) that does not belong to Ā M, say r 1 (Pî) Ā L, then the choice of f is determined by first applying a generalized median voter scheme G L to all individuals with the exception of individual î, and the resulting alternative G L (P î ) is then projected on the interval between r 1 (Pî) and a; in particular, if G L (P î ) is between r 1 (Pî) and a, then G L (P î ) is the final outcome. A similar procedure applies when r 1 (Pî) belongs to the right part Ā R. Thus, individual î has no dictatorial power on Ā L and Ā R, but he has nevertheless large influence on the location of the social choice even if his top alternative belongs either to Ā L or Ā R. Finally, the two conditions ρ 1 (S) R GL and ρ m (S) R GR, which are equivalent to a G L T = ρ 1(S) for some T N \ {î} and = ρ m (S), are necessary and sufficient for f to be surjective on Ā L and Ā R. The a G R following example illustrates how Theorem 1 works in two different cases. Example 4. Consider again the two maximal common decompositions from Examples 2 and 3. In Example 2, we found (A L S,AM S,AR S ) = (,A, ), so in this case only the dictatorial SCFs f : Ω n S A are strategy-proof and surjective. Thus, combining the two single-peaked domains Ω S1 and Ω S2 turns the possibility result for singlepeaked domains depicted in Moulin (1980) into a dictatorial result. 13

14 In Example 3, both A L S = {x 1,x 2,x 3 } and A R S = {x 7} are non-empty. Since the dictatorial part Ā M of a strategy-proof SCF f : Ω n S A must cover AM S = {x 4,x 5,x 6 } and continue into A L S and AR S if possible, f must be dictatorial at least on {x 3,x 4,x 5,x 6,x 7 }. In particular, the non-dictatorial right part Ā R is empty here. A strategy-proof and not completely dictatorial SCF f : Ω n S A can now for example be constructed as follows: Let some î N be a dictator on {x 3,...,x 7 }. If r 1 (Pî) {x 1,x 2 }, calculate the median med(p î ) of the top alternatives in P î Ω n 1 S (assume here for simplicity that n 1 is odd so that med(p î ) is well-defined); if this median belongs to A L S, choose that alternative of r 1(Pî) and med(p î ) which is closest to x 3, and if med(p î ) / A L S, choose x 3. Remark 2. Compared with the characterization of strategy-proof SCFs in Moulin (1980), the important contribution of Theorem 1 is of course that it applies for a more general class of preference domains. The assumptions in these two results differ, however, slightly in one more respect. While Moulin (1980) assumes that the SCF in addition to strategy-proofness also satisfies the tops-only property, Theorem 1 requires strategy-proofness and surjectivity. A priori, surjectivity is a weak normative requirement guaranteeing voters sovereignty, while the tops-only property is a strong limitation on which information on voters preferences may be used by the SCF. However, in the proof of Theorem 1, we show that on multiple singlepeaked domains, strategy-proofness and surjectivity actually imply the tops-only property. Note also that Theorem 1 does not apply to traditional single-peaked domains, i.e., when S contains only one ordering. The SCFs in (3.2) are of course strategyproof even if #S = 1, but the characterization is no longer complete in this case because (3.2) cannot represent for example the anonymous strategy-proof SCFs which exist on single-peaked domains, e.g., the median rule. Instead, if S = {S}, the complete characterization of all surjective strategy-proof SCFs f : Ω n S A is given by Moulin s characterization in (2.1) with the additional requirement that a T = ρ 1 (S) for some T N and a = ρ m (S) to ensure surjectivity. Remark 3. As a special case, Theorem 1 covers the Gibbard-Satterthwaite theorem because if S contains all possible orderings of A, then Ω n S equals the unrestricted domain Σ A and (A L S,AM S,AR S ) = (,A, ), so by (3.2), a strategy-proof surjective SCF on Σ A must be dictatorial. Furthermore, from the perspective of the literature on maximal domains admitting non-dictatorial strategy-proof SCFs (see, e.g., Barberà (2011, p. 817)), Theorem 1 shows how the possibility result of anonymous 14

15 strategy-proof social choice in Moulin (1980) turns gradually into the dictatorial result of the Gibbard-Satterthwaite theorem when the preference domain is extended successively from a single-peaked domain to the unrestricted domain. Remark 4. Theorem 1 characterizes the strategy-proof SCFs on multiple singlepeaked domains under the additional requirement of surjectivity. But Theorem 1 can also be used to derive the functional form of an arbitrary strategy-proof SCF on Ω S, i.e., also when range restrictions are allowed. To see this, suppose that f : Ω n S A is a strategy-proof SCF with range R f = B A. Note first that f (P 1,...,P n ) must only depend on how the preferences in (P 1,...,P n ) rank the alternatives in B because if P i,p i Ω S are such that P i B = P i B, then P i and P i rank f (P i,p i ) and f (P i,p i) in the same way for all P i ΩS n 1, so strategy-proofness requires f (P i,p i ) = f (P i,p i). Hence, it is possible to construct a well-defined SCF ˆf : (Ω S ) n B B by assigning to ( P 1,..., P n ) (Ω S ) n B the outcome f ( P 1,..., P n ) = f (P 1,...,P n ), where (P 1,...,P n ) Ω n S is some profile with P i B = P i for all i N. Then, ˆf is (1) defined on a multiple single-peaked domain because (Ω S ) B = Ω (S B ), (2) strategy-proof because f is strategy-proof, and (3) surjective by construction. Then, ˆf can be characterized by Theorem 1, using the maximal common decomposition of B with respect to S B. In a final step, the functional form of f is obtained via the relationship f (P 1,...,P n ) = ˆf (P 1 B,...,P n B ) for all (P 1,...,P n ) Ω n S. Remark 5. The ordering S S used in the construction of Ā M, the projections in (3.2), and the minmax rules in (3.3) can be replaced by any other ordering S S, holding everything else in Theorem 1 fixed, because such a replacement will only affect the representation of f, but not f itself. The reason for this is that the functional form of f depends only on S on those parts of A where all orderings in S coincide. This is obvious in (3.2), and in (3.3), the choice of for example G L is only of importance for f if G L (P î ) Ā L, and if this is the case, then G L (P î ) will not change if S is replaced by some other ordering S S since all orderings in S coincide on Ā L (a similar statement applies for G R ). However, replacing S by some other S S may change the choices of G L and G R on A M S, where the orderings in S differ. Note also that neither G L nor G R are in general strategy-proof, because a generalized median voter scheme with respect to S is not necessarily a generalized median voter scheme with respect to some other S S, and will hence be manipulable when preferences from Ω S are admitted. Remark 6. A similar characterization of strategy-proof SCFs as a combination of dictatorial rules and generalized median voter schemes is obtained in Schummer 15

16 and Vohra (2002). Their result differs, however, both motivationally and formally from Theorem 1: Schummer and Vohra consider the problem of choosing a strategyproof location on a network, which is represented as a continuous graph in a Euclidean space, when the individuals have quadratic single-peaked preferences, i.e., the ranking of an alternative depends only on the Euclidean distance to the top alternative. In this setup, Schummer and Vohra show that a strategy-proof SCF must be dictatorial on the circular parts and a generalized median voter scheme on the cycle-free parts of the graph. 4 Preferences over parties In this section, we show in the framework of a spatial voting model that preferences over parties can be represented by multiple single-peaked domains. Thereby, our starting point is the assertion that the political sphere normally covers several political dimensions such as economic policy, civil rights policy, security policy, etc., and most political parties take a position in every dimension. Identifying every party with its positions in the different political dimensions, parties can be considered as multi-dimensional alternatives. In addition, the political system usually has the following two properties which are crucial for the structure of party preferences. First, in every political dimension, the possible positions a party can take can be categorized by an ideological left right scale. For example, for the political dimension of socioeconomic policy, equalizing redistribution of income is a typical leftist position, while restrictive redistribution in favor of protection of private property is a typical rightist position. Similarly, concerning economic policy, an extensive regulation of markets would be classified as leftist and minimal regulatory policy as rightist. Based on these left right scales, it is reasonable that preferences are single-peaked within every political dimension, provided that all other dimensions are kept fixed. Moreover, the ideological left right scale can be used in some cases also to order combinations of positions from different dimensions. For example, the combination (high income redistribution, high market regulation) is certainly to the left of the combination (no income redistribution, no market regulation). On the other hand, the two combinations (high income redistribution, no market regulation) and (no income redistribution, high market regulation) cannot be compared unambiguously on a left right scale. For the structure of preferences, we assume that whenever a collection of policy combinations can be ordered unambiguously on a left right scale, then preferences are single-peaked on this collection. Second, 16

17 only a small proportion of all possible policy combinations is normally represented by parties, i.e., the party system is sparse in the multi-dimensional set of all possible political alternatives. This means that preferences over policy combinations are not identical to preferences over parties, but the latter are obtained by restricting the former to the prevailing party system. We now formalize the reasoning from the preceding paragraph. Thereby, we restrict ourselves to the case when there are exactly two political dimensions along which parties locate their positions; this will be sufficient to gain insight into the structure of preferences over parties, while adding further dimensions would only make the analysis more technical, but lead to the same qualitative results. So let X 1 = {x 11,...,x 1m1 } and X 2 = {x 21,...,x 2m2 } be two finite sets of m 1 respectively m 2 alternatives, which are referred to as the two political dimensions. The sets X 1 and X 2 are equipped with left right orderings S X1 and S X2, respectively. The set of political alternatives is the product set X = X 1 X 2. The two orderings S X1 and S X2 induce a (partial) ordering S X on X by (x 1,x 2 )S X (y 1,y 2 ) def [x 1 S X1 y 1 and x 2 S X2 y 2 ] or [x 1 S X1 y 1 and x 2 = y 2 ] or [x 1 = y 1 and x 2 S X2 y 2 ]. This means that x = (x 1,x 2 ) is to the left of y = (y 1,y 2 ) according to S X if at least one coordinate of x is to the left of the corresponding coordinate of y, and the other coordinate of x either coincides with or is also to the left of the corresponding coordinate of y. Note that S X is transitive and asymmetric, but not complete in all non-trivial cases (i.e., when m 1 2 and m 2 2), because if x 1 S X1 y 1 (where x 1,y 1 X 1 ) and x 2 S X2 y 2 (where x 2,y 2 X 2 ), then neither (x 1,y 2 )S X (x 2,y 1 ) nor (x 2,y 1 )S X (x 1,y 2 ). Denote the betweenness relation associated with S X by B X, i.e., for all x,y,z X, we have B X (x,y,z) if and only if xs X ys X z or zs X ys X x. Definition 2. A complete, transitive, and asymmetric preference P on X is singlepeaked with respect to B X if B X (x,y,z) implies ypx or ypz for all x,y,z X. The domain of all P Σ X that are single-peaked with respect to B X is called the singlepeaked domain with respect to B X, and it is denoted by Γ X. The domain Γ X consists of preferences with a variety of different shapes, which is illustrated by the following example. Example 5. Let X 1 = {x 1,x 2,x 3 } and X 2 = {y 1,y 2,y 3 }, and assume that X 1 is ordered by S X1 = [x 1 x 2 x 3 ], while X 2 is equipped with the ordering S X2 = [y 1 y 2 y 3 ]. 17

18 Consider the following three preferences on X = X 1 X 2, where the number corresponding to the pair (x i,y j ) indicates the rank r k (P) of (x i,y j ) in P: P 1 x 1 x 2 x 3 y y y P 2 x 1 x 2 x 3 y y y P 3 x 1 x 2 x 3 y y y It is straightforward to check that these three preferences are single-peaked with respect to B X. Moreover, the shape of these preferences can be interpreted in different ways: Preference P 1 can be represented by the utility function u(x i,y j ) = 3i j for (x i,y j ) X, which means that P 1 can be thought of as being represented by a utility plane that attains its maximum on X at (x 1,y 1 ) from which it slopes downward in both directions towards (x 3,y 3 ). Preference P 2 has its top at the center (x 2,y 2 ) X and can be represented by a hat-shaped utility function on X. On the contrary, preference P 3 has rather a saddle shape since P 3 is -shaped on the line from (x 1,y 1 ) to (x 3,y 3 ), but -shaped from (x 1,y 3 ) to (x 3,y 1 ). A party system on X is a subset A X. For a given party system A, the domain Γ X induces a domain Γ A (Γ X ) A of preferences over parties. Further, S X induces a left right ordering (S X ) A on A. In general, (S X ) A will not be complete, but by the Szpilrajn extension theorem, there is always at least one complete ordering S Σ A that is compatible with (S X ) A. 6 The set S A = {S Σ A ; S is compatible with (S X ) A } is thus non-empty, and S A is called the set of all party orderings of A. Example 6. Suppose that X 1 = {x 1,x 2,x 3,x 4 } and X 2 = {y 1,y 2,y 3,y 4 } are two political dimensions, equipped with the left right orderings S X1 = [x 1 x 2 x 3 x 4 ] and S X2 = [y 1 y 2 y 3 y 4 ], and X = X 1 X 2 is the set of political alternatives. If the party system on X is A = {(x 1,y 1 ),(x 2,y 2 ),(x 3,y 3 ),(x 4,y 4 )}, then (x 1,y 1 )S X (x 2,y 2 )S X (x 3,y 3 )S X (x 4,y 4 ), and hence the restriction (S X ) A of S X to A is complete in this case. This means that (S X ) A is the unique party ordering of A, i.e., S A = {(S X ) A }. 6 If B is a set of alternatives, and S is a transitive and asymmetric, but not necessarily complete ordering of B, then a complete, transitive, and asymmetric ordering S of B is compatible with S if bsb implies b Sb for all b,b B. The Szpilrajn extension theorem (Szpilrajn, 1930) states that for every transitive and asymmetric ordering S of B there exists at least one complete ordering S of B compatible with S. 18

19 On the other hand, if A = {(x 1,y 1 ),(x 2,y 3 ),(x 3,y 2 ),(x 4,y 4 )}, then (x 1,y 1 )S X (x 2,y 3 ), (x 1,y 1 )S X (x 3,y 2 ), (x 1,y 1 )S X (x 4,y 4 ), (x 2,y 3 )S X (x 4,y 4 ) and (x 3,y 2 )S X (x 4,y 4 ), but (x 2,y 3 ) and (x 3,y 2 ) cannot be ordered by S X. Thus, (S X ) A is incomplete in this case. There are two complete orderings of A that are compatible with (S X ) A, namely S = [(x 1,y 1 ) (x 2,y 3 ) (x 3,y 2 ) (x 4,y 4 )] and S = [(x 1,y 1 ) (x 3,y 2 ) (x 2,y 3 ) (x 4,y 4 )], and the set of party orderings of A is S A = {S,S }. The structure of the domain of preferences over parties is described by the following result. Proposition 1. Γ A = Ω SA. The main message of Proposition 1 is that if voters preferences over political alternatives are single-peaked with respect to B X, then the ordinary single-peaked domains are in general too narrow to represent preferences over parties, but the domain of preferences over parties is instead a multiple single-peaked domain. Furthermore, the orderings needed to generate the multiple single-peaked domain are given by S A. We conclude this section by comparing our results with another approach to multi-dimensional single-peaked preferences, which has been very influential in the literature. Definition 2 can be considered as a generalization of single-peaked preferences to a multi-dimensional setup based on a local characterization of singlepeakedness in the spirit of Sen (1966), since only one triple from X is considered at a time. In one dimension, single-peakedness can equivalently be described by the property that alternatives become less preferred the further one moves away in either direction from the top alternative of a preference, which is the definition used in Black (1948). A generalization of this global characterization of singlepeakedness to multi-dimensional sets of alternatives has been introduced in Barberà et al. (1993). Using a geometric betweenness relation 7, in contrast to the ideological betweenness relation in our approach, Barberà et al. (1993) define a preference P Σ X to be generalized single-peaked if xpy whenever x is between r 1 (P) and y. Denote the domain of all P Σ X which are single-peaked in this sense by ˆΓ X. 7 Formally, the betweenness relation in Barberà et al. (1993) is defined as follows: Assuming that X 1 and X 2 are integer intervals, Barberà et al. define first the L 1 -norm of x = (x 1,x 2 ) X 1 X 2 by x = x 1 + x 2. Given the L 1 -norm, y X is between x X and z X if x z = x y + y z. Geometrically, this means that y belongs to the minimal box containing x and z. 19

MAXIMAL POSSIBILITY AND MINIMAL DICTATORIAL COVERS OF DOMAINS

MAXIMAL POSSIBILITY AND MINIMAL DICTATORIAL COVERS OF DOMAINS MAXIMAL POSSIBILITY AND MINIMAL DICTATORIAL COVERS OF DOMAINS Gopakumar Achuthankutty 1 and Souvik Roy 1 1 Economic Research Unit, Indian Statistical Institute, Kolkata Abstract In line with the works

More information

On the Strategy-proof Social Choice of Fixed-sized Subsets

On the Strategy-proof Social Choice of Fixed-sized Subsets Nationalekonomiska institutionen MASTER S THESIS, 30 ECTS On the Strategy-proof Social Choice of Fixed-sized Subsets AUTHOR: ALEXANDER REFFGEN SUPERVISOR: LARS-GUNNAR SVENSSON SEPTEMBER, 2006 Contents

More information

Coalitionally strategyproof functions depend only on the most-preferred alternatives.

Coalitionally strategyproof functions depend only on the most-preferred alternatives. Coalitionally strategyproof functions depend only on the most-preferred alternatives. H. Reiju Mihara reiju@ec.kagawa-u.ac.jp Economics, Kagawa University, Takamatsu, 760-8523, Japan April, 1999 [Social

More information

The Structure of Strategy-Proof Social Choice:

The Structure of Strategy-Proof Social Choice: The Structure of Strategy-Proof Social Choice: General Characterization and Possibility Results on Median Spaces * Klaus Nehring Department of Economics, UC Davis, Davis, CA 95616, U.S.A. and Institute

More information

A Characterization of Single-Peaked Preferences via Random Social Choice Functions

A Characterization of Single-Peaked Preferences via Random Social Choice Functions A Characterization of Single-Peaked Preferences via Random Social Choice Functions Shurojit Chatterji, Arunava Sen and Huaxia Zeng September 2014 Paper No. 13-2014 ANY OPINIONS EXPRESSED ARE THOSE OF THE

More information

On Domains That Admit Well-behaved Strategy-proof Social Choice Functions

On Domains That Admit Well-behaved Strategy-proof Social Choice Functions On Domains That Admit Well-behaved Strategy-proof Social Choice Functions Shurojit Chatterji, Remzi Sanver and Arunava Sen May 2010 Paper No. 07-2010 ANY OPINIONS EXPRESSED ARE THOSE OF THE AUTHOR(S) AND

More information

The Structure of Strategy-Proof Social Choice

The Structure of Strategy-Proof Social Choice The Structure of Strategy-Proof Social Choice Part II: Non-Dictatorship, Anonymity and Neutrality * Klaus Nehring Department of Economics, University of California at Davis Davis, CA 95616, U.S.A. kdnehring@ucdavis.edu

More information

Strategy-Proofness on the Condorcet Domain

Strategy-Proofness on the Condorcet Domain College of William and Mary W&M ScholarWorks Undergraduate Honors Theses Theses, Dissertations, & Master Projects 5-2008 Strategy-Proofness on the Condorcet Domain Lauren Nicole Merrill College of William

More information

DICTATORIAL DOMAINS. Navin Aswal University of Minnesota, Minneapolis, USA Shurojit Chatterji Indian Statistical Institute, New Delhi, India and

DICTATORIAL DOMAINS. Navin Aswal University of Minnesota, Minneapolis, USA Shurojit Chatterji Indian Statistical Institute, New Delhi, India and DICTATORIAL DOMAINS Navin Aswal University of Minnesota, Minneapolis, USA Shurojit Chatterji Indian Statistical Institute, New Delhi, India and Arunava Sen Indian Statistical Institute, New Delhi, India

More information

Non-Manipulable Domains for the Borda Count

Non-Manipulable Domains for the Borda Count Non-Manipulable Domains for the Borda Count Martin Barbie, Clemens Puppe * Department of Economics, University of Karlsruhe D 76128 Karlsruhe, Germany and Attila Tasnádi ** Department of Mathematics, Budapest

More information

Resource Allocation via the Median Rule: Theory and Simulations in the Discrete Case

Resource Allocation via the Median Rule: Theory and Simulations in the Discrete Case Resource Allocation via the Median Rule: Theory and Simulations in the Discrete Case Klaus Nehring Clemens Puppe January 2017 **** Preliminary Version ***** Not to be quoted without permission from the

More information

6.207/14.15: Networks Lecture 24: Decisions in Groups

6.207/14.15: Networks Lecture 24: Decisions in Groups 6.207/14.15: Networks Lecture 24: Decisions in Groups Daron Acemoglu and Asu Ozdaglar MIT December 9, 2009 1 Introduction Outline Group and collective choices Arrow s Impossibility Theorem Gibbard-Satterthwaite

More information

Hans Peters, Souvik Roy, Ton Storcken. Manipulation under k-approval scoring rules RM/08/056. JEL code: D71, D72

Hans Peters, Souvik Roy, Ton Storcken. Manipulation under k-approval scoring rules RM/08/056. JEL code: D71, D72 Hans Peters, Souvik Roy, Ton Storcken Manipulation under k-approval scoring rules RM/08/056 JEL code: D71, D72 Maastricht research school of Economics of TEchnology and ORganizations Universiteit Maastricht

More information

W. ALLEN WALLIS Institute of POLITICAL ECONOMY

W. ALLEN WALLIS Institute of POLITICAL ECONOMY Strategy-Proofness and Single-Crossing Alejandro Saporiti Working Paper No. 48 October 2007 W. ALLEN WALLIS Institute of POLITICAL ECONOMY UNIVERSITY OF ROCHESTER Strategy-Proofness and Single-Crossing

More information

A General Impossibility Result on Strategy-Proof Social Choice Hyperfunctions

A General Impossibility Result on Strategy-Proof Social Choice Hyperfunctions A General Impossibility Result on Strategy-Proof Social Choice Hyperfunctions Selçuk Özyurt and M. Remzi Sanver May 22, 2008 Abstract A social choice hyperfunction picks a non-empty set of alternatives

More information

INTEGER PROGRAMMING AND ARROVIAN SOCIAL WELFARE FUNCTIONS

INTEGER PROGRAMMING AND ARROVIAN SOCIAL WELFARE FUNCTIONS MATHEMATICS OF OPERATIONS RESEARCH Vol. 28, No. 2, May 2003, pp. 309 326 Printed in U.S.A. INTEGER PROGRAMMING AND ARROVIAN SOCIAL WELFARE FUNCTIONS JAY SETHURAMAN, TEO CHUNG PIAW, and RAKESH V. VOHRA

More information

Social Choice. Jan-Michael van Linthoudt

Social Choice. Jan-Michael van Linthoudt Social Choice Jan-Michael van Linthoudt Summer term 2017 Version: March 15, 2018 CONTENTS Remarks 1 0 Introduction 2 1 The Case of 2 Alternatives 3 1.1 Examples for social choice rules............................

More information

A MAXIMAL DOMAIN FOR STRATEGY-PROOF AND NO-VETOER RULES IN THE MULTI-OBJECT CHOICE MODEL

A MAXIMAL DOMAIN FOR STRATEGY-PROOF AND NO-VETOER RULES IN THE MULTI-OBJECT CHOICE MODEL Discussion Paper No. 809 A MAXIMAL DOMAIN FOR STRATEGY-PROOF AND NO-VETOER RULES IN THE MULTI-OBJECT CHOICE MODEL Kantaro Hatsumi Dolors Berga Shigehiro Serizawa April 2011 The Institute of Social and

More information

Tighter Bounds for Facility Games

Tighter Bounds for Facility Games Tighter Bounds for Facility Games Pinyan Lu 1, Yajun Wang 1, and Yuan Zhou 1 Microsoft Research Asia {pinyanl, yajunw}@microsoft.com Carnegie Mellon University yuanzhou@cs.cmu.edu Abstract. In one dimensional

More information

Finite Dictatorships and Infinite Democracies

Finite Dictatorships and Infinite Democracies Finite Dictatorships and Infinite Democracies Iian B. Smythe October 20, 2015 Abstract Does there exist a reasonable method of voting that when presented with three or more alternatives avoids the undue

More information

The Gibbard random dictatorship theorem: a generalization and a new proof

The Gibbard random dictatorship theorem: a generalization and a new proof SERIEs (2011) 2:515 527 DOI 101007/s13209-011-0041-z ORIGINAL ARTICLE The Gibbard random dictatorship theorem: a generalization and a new proof Arunava Sen Received: 24 November 2010 / Accepted: 10 January

More information

Political Economy of Institutions and Development. Lectures 2 and 3: Static Voting Models

Political Economy of Institutions and Development. Lectures 2 and 3: Static Voting Models 14.773 Political Economy of Institutions and Development. Lectures 2 and 3: Static Voting Models Daron Acemoglu MIT February 7 and 12, 2013. Daron Acemoglu (MIT) Political Economy Lectures 2 and 3 February

More information

Arrow s Theorem as a Corollary

Arrow s Theorem as a Corollary Arrow s Theorem as a Corollary Klaus Nehring University of California, Davis October 2002 Abstract Arrow s Impossibility Theorem is derived from a general theorem on social aggregation in property spaces.

More information

Social Choice and Mechanism Design - Part I.2. Part I.2: Social Choice Theory Summer Term 2011

Social Choice and Mechanism Design - Part I.2. Part I.2: Social Choice Theory Summer Term 2011 Social Choice and Mechanism Design Part I.2: Social Choice Theory Summer Term 2011 Alexander Westkamp April 2011 Introduction Two concerns regarding our previous approach to collective decision making:

More information

Economic Core, Fair Allocations, and Social Choice Theory

Economic Core, Fair Allocations, and Social Choice Theory Chapter 9 Nathan Smooha Economic Core, Fair Allocations, and Social Choice Theory 9.1 Introduction In this chapter, we briefly discuss some topics in the framework of general equilibrium theory, namely

More information

Strategy-proof allocation of indivisible goods

Strategy-proof allocation of indivisible goods Soc Choice Welfare (1999) 16: 557±567 Strategy-proof allocation of indivisible goods Lars-Gunnar Svensson Department of Economics, Lund University, P.O. Box 7082, SE-220 07 of Lund, Sweden (e-mail: lars-gunnar.svensson@nek.lu.se)

More information

arxiv: v2 [math.co] 14 Apr 2011

arxiv: v2 [math.co] 14 Apr 2011 Complete Characterization of Functions Satisfying the Conditions of Arrow s Theorem Elchanan Mossel and Omer Tamuz arxiv:0910.2465v2 [math.co] 14 Apr 2011 April 15, 2011 Abstract Arrow s theorem implies

More information

Strategic Manipulability without Resoluteness or Shared Beliefs: Gibbard-Satterthwaite Generalized

Strategic Manipulability without Resoluteness or Shared Beliefs: Gibbard-Satterthwaite Generalized Strategic Manipulability without Resoluteness or Shared Beliefs: Gibbard-Satterthwaite Generalized Christian Geist Project: Modern Classics in Social Choice Theory Institute for Logic, Language and Computation

More information

Approximation Algorithms and Mechanism Design for Minimax Approval Voting

Approximation Algorithms and Mechanism Design for Minimax Approval Voting Approximation Algorithms and Mechanism Design for Minimax Approval Voting Ioannis Caragiannis RACTI & Department of Computer Engineering and Informatics University of Patras, Greece caragian@ceid.upatras.gr

More information

Parameterization of Strategy-Proof Mechanisms in the Obnoxious Facility Game

Parameterization of Strategy-Proof Mechanisms in the Obnoxious Facility Game Journal of Graph Algorithms and Applications http://jgaa.info/ vol. 2, no. 3, pp. 247 263 (207) DOI: 0.755/jgaa.0045 Parameterization of Strategy-Proof Mechanisms in the Obnoxious Facility Game Morito

More information

Antonio Quesada Universidad de Murcia. Abstract

Antonio Quesada Universidad de Murcia. Abstract From social choice functions to dictatorial social welfare functions Antonio Quesada Universidad de Murcia Abstract A procedure to construct a social welfare function from a social choice function is suggested

More information

14.770: Introduction to Political Economy Lectures 1 and 2: Collective Choice and Voting

14.770: Introduction to Political Economy Lectures 1 and 2: Collective Choice and Voting 14.770: Introduction to Political Economy Lectures 1 and 2: Collective Choice and Voting Daron Acemoglu MIT September 6 and 11, 2017. Daron Acemoglu (MIT) Political Economy Lectures 1 and 2 September 6

More information

Comparing impossibility theorems

Comparing impossibility theorems Comparing impossibility theorems Randy Calvert, for Pol Sci 507 Spr 2017 All references to A-S & B are to Austen-Smith and Banks (1999). Basic notation X set of alternatives X set of all nonempty subsets

More information

Arrow s Paradox. Prerna Nadathur. January 1, 2010

Arrow s Paradox. Prerna Nadathur. January 1, 2010 Arrow s Paradox Prerna Nadathur January 1, 2010 Abstract In this paper, we examine the problem of a ranked voting system and introduce Kenneth Arrow s impossibility theorem (1951). We provide a proof sketch

More information

Ordered Value Restriction

Ordered Value Restriction Ordered Value Restriction Salvador Barberà Bernardo Moreno Univ. Autònoma de Barcelona and Univ. de Málaga and centra March 1, 2006 Abstract In this paper, we restrict the pro les of preferences to be

More information

THE UNIVERSITY OF KANSAS WORKING PAPERS SERIES IN THEORETICAL AND APPLIED ECONOMICS

THE UNIVERSITY OF KANSAS WORKING PAPERS SERIES IN THEORETICAL AND APPLIED ECONOMICS THE UNIVERSITY OF KANSAS WORKING PAPERS SERIES IN THEORETICAL AND APPLIED ECONOMICS AN EFFICIENCY CHARACTERIZATION OF PLURALITY SOCIAL CHOICE ON SIMPLE PREFERENCE DOMAINS Biung-Ghi Ju University of Kansas

More information

Michel Le Breton and John A. Weymark

Michel Le Breton and John A. Weymark ARROVIAN SOCIAL CHOICE THEORY ON ECONOMIC DOMAINS by Michel Le Breton and John A. Weymark Working Paper No. 02-W06R April 2002 Revised September 2003 DEPARTMENT OF ECONOMICS VANDERBILT UNIVERSITY NASHVILLE,

More information

Approximation Algorithms and Mechanism Design for Minimax Approval Voting 1

Approximation Algorithms and Mechanism Design for Minimax Approval Voting 1 Approximation Algorithms and Mechanism Design for Minimax Approval Voting 1 Ioannis Caragiannis, Dimitris Kalaitzis, and Evangelos Markakis Abstract We consider approval voting elections in which each

More information

UNIVERSITY OF KANSAS WORKING PAPERS SERIES IN THEORETICAL AND APPLIED ECONOMICS

UNIVERSITY OF KANSAS WORKING PAPERS SERIES IN THEORETICAL AND APPLIED ECONOMICS UNIVERSITY OF KANSAS WORKING PAPERS SERIES IN THEORETICAL AND APPLIED ECONOMICS Strategy-Proofness versus Efficiency in Exchange Economies: General Domain Properties and Applications Biung-Ghi Ju Paper

More information

Political Economy of Institutions and Development: Problem Set 1. Due Date: Thursday, February 23, in class.

Political Economy of Institutions and Development: Problem Set 1. Due Date: Thursday, February 23, in class. Political Economy of Institutions and Development: 14.773 Problem Set 1 Due Date: Thursday, February 23, in class. Answer Questions 1-3. handed in. The other two questions are for practice and are not

More information

Unanimity, Pareto optimality and strategy-proofness on connected domains

Unanimity, Pareto optimality and strategy-proofness on connected domains Unanimity, Pareto optimality and strategy-proofness on connected domains Souvik Roy and Ton Storcken y February 2016 Abstract This is a preliminary version please do not quote 1 Extended Abstract This

More information

On the Chacteristic Numbers of Voting Games

On the Chacteristic Numbers of Voting Games On the Chacteristic Numbers of Voting Games MATHIEU MARTIN THEMA, Departments of Economics Université de Cergy Pontoise, 33 Boulevard du Port, 95011 Cergy Pontoise cedex, France. e-mail: mathieu.martin@eco.u-cergy.fr

More information

Strategic Manipulation and Regular Decomposition of Fuzzy Preference Relations

Strategic Manipulation and Regular Decomposition of Fuzzy Preference Relations Strategic Manipulation and Regular Decomposition of Fuzzy Preference Relations Olfa Meddeb, Fouad Ben Abdelaziz, José Rui Figueira September 27, 2007 LARODEC, Institut Supérieur de Gestion, 41, Rue de

More information

Hans Peters, Souvik Roy, Soumyarup Sadhukhan, Ton Storcken

Hans Peters, Souvik Roy, Soumyarup Sadhukhan, Ton Storcken Hans Peters, Souvik Roy, Soumyarup Sadhukhan, Ton Storcken An Extreme Point Characterization of Strategyproof and Unanimous Probabilistic Rules over Binary Restricted Domains RM/16/012 An Extreme Point

More information

Social Choice Theory. Felix Munoz-Garcia School of Economic Sciences Washington State University. EconS Advanced Microeconomics II

Social Choice Theory. Felix Munoz-Garcia School of Economic Sciences Washington State University. EconS Advanced Microeconomics II Social Choice Theory Felix Munoz-Garcia School of Economic Sciences Washington State University EconS 503 - Advanced Microeconomics II Social choice theory MWG, Chapter 21. JR, Chapter 6.2-6.5. Additional

More information

Repeated Downsian Electoral Competition

Repeated Downsian Electoral Competition Repeated Downsian Electoral Competition John Duggan Department of Political Science and Department of Economics University of Rochester Mark Fey Department of Political Science University of Rochester

More information

Dependence and Independence in Social Choice Theory

Dependence and Independence in Social Choice Theory Dependence and Independence in Social Choice Theory Eric Pacuit Department of Philosophy University of Maryland, College Park pacuit.org epacuit@umd.edu March 4, 2014 Eric Pacuit 1 Competing desiderata

More information

Abstract Arrowian Aggregation *

Abstract Arrowian Aggregation * Abstract Arrowian Aggregation * Klaus Nehring Department of Economics, University of California at Davis Davis, CA 95616, U.S.A. kdnehring@ucdavis.edu and Clemens Puppe Department of Economics, University

More information

Coalitional Structure of the Muller-Satterthwaite Theorem

Coalitional Structure of the Muller-Satterthwaite Theorem Coalitional Structure of the Muller-Satterthwaite Theorem Pingzhong Tang and Tuomas Sandholm Computer Science Department Carnegie Mellon University {kenshin,sandholm}@cscmuedu Abstract The Muller-Satterthwaite

More information

A Systematic Approach to the Construction of Non-empty Choice Sets

A Systematic Approach to the Construction of Non-empty Choice Sets A Systematic Approach to the Construction of Non-empty Choice Sets John Duggan Department of Political Science and Department of Economics University of Rochester May 17, 2004 Abstract Suppose a strict

More information

Single-peaked consistency and its complexity

Single-peaked consistency and its complexity Bruno Escoffier, Jérôme Lang, Meltem Öztürk Abstract A common way of dealing with the paradoxes of preference aggregation consists in restricting the domain of admissible preferences. The most well-known

More information

Approval Voting for Committees: Threshold Approaches

Approval Voting for Committees: Threshold Approaches Approval Voting for Committees: Threshold Approaches Peter Fishburn Aleksandar Pekeč WORKING DRAFT PLEASE DO NOT CITE WITHOUT PERMISSION Abstract When electing a committee from the pool of individual candidates,

More information

Fairness and Redistribution: Response

Fairness and Redistribution: Response Fairness and Redistribution: Response By ALBERTO ALESINA, GEORGE-MARIOS ANGELETOS, AND GUIDO COZZI This paper responds to the comment of Di Tella and Dubra (211). We first clarify that the model of Alesina

More information

Quasi-transitive and Suzumura consistent relations

Quasi-transitive and Suzumura consistent relations Quasi-transitive and Suzumura consistent relations Walter Bossert Department of Economics and CIREQ, University of Montréal P.O. Box 6128, Station Downtown, Montréal QC H3C 3J7, Canada FAX: (+1 514) 343

More information

Integer Programming on Domains Containing Inseparable Ordered Pairs

Integer Programming on Domains Containing Inseparable Ordered Pairs Integer Programming on Domains Containing Inseparable Ordered Pairs Francesca Busetto, Giulio Codognato, Simone Tonin August 2012 n. 8/2012 Integer Programming on Domains Containing Inseparable Ordered

More information

Introduction to spatial modeling. (a mostly geometrical presentation)

Introduction to spatial modeling. (a mostly geometrical presentation) Introduction to spatial modeling (a mostly geometrical presentation) Alternatives X = R n e.g. (x 1,x 2,, x n ) X Alternatives are infinite set of policies in n- dimensional Euclidean space Each dimension

More information

arxiv: v1 [cs.gt] 29 Mar 2014

arxiv: v1 [cs.gt] 29 Mar 2014 Testing Top Monotonicity Haris Aziz NICTA and UNSW Australia, Kensington 2033, Australia arxiv:1403.7625v1 [cs.gt] 29 Mar 2014 Abstract Top monotonicity is a relaxation of various well-known domain restrictions

More information

The core of voting games: a partition approach

The core of voting games: a partition approach The core of voting games: a partition approach Aymeric Lardon To cite this version: Aymeric Lardon. The core of voting games: a partition approach. International Game Theory Review, World Scientific Publishing,

More information

Ordinal Bayesian Incentive Compatibility in Restricted Domains

Ordinal Bayesian Incentive Compatibility in Restricted Domains Ordinal Bayesian Incentive Compatibility in Restricted Domains Debasis Mishra October 12, 2015 Abstract We study deterministic voting mechanisms by considering an ordinal notion of Bayesian incentive compatibility

More information

The Importance of the Median Voter

The Importance of the Median Voter The Importance of the Median Voter According to Duncan Black and Anthony Downs V53.0500 NYU 1 Committee Decisions utility 0 100 x 1 x 2 x 3 x 4 x 5 V53.0500 NYU 2 Single-Peakedness Condition The preferences

More information

The Axiomatic Method in Social Choice Theory:

The Axiomatic Method in Social Choice Theory: The Axiomatic Method in Social Choice Theory: Preference Aggregation, Judgment Aggregation, Graph Aggregation Ulle Endriss Institute for Logic, Language and Computation University of Amsterdam Ulle Endriss

More information

Fair Divsion in Theory and Practice

Fair Divsion in Theory and Practice Fair Divsion in Theory and Practice Ron Cytron (Computer Science) Maggie Penn (Political Science) Lecture 6-b: Arrow s Theorem 1 Arrow s Theorem The general question: Given a collection of individuals

More information

1. The Problem. Table 1

1. The Problem. Table 1 1 A Possibility Theorem on Aggregation Over Multiple Interconnected Propositions Christian List 1 forthcoming in Mathematical Social Sciences Abstract. Drawing on the so-called doctrinal paradox, List

More information

Envelope Theorems for Arbitrary Parametrized Choice Sets

Envelope Theorems for Arbitrary Parametrized Choice Sets Envelope Theorems for Arbitrary Parametrized Choice Sets Antoine LOEPER 1 and Paul Milgrom January 2009 (PRELIMINARY) 1 Managerial Economics and Decision Sciences, Kellogg School of Management, Northwestern

More information

Probabilistic Aspects of Voting

Probabilistic Aspects of Voting Probabilistic Aspects of Voting UUGANBAATAR NINJBAT DEPARTMENT OF MATHEMATICS THE NATIONAL UNIVERSITY OF MONGOLIA SAAM 2015 Outline 1. Introduction to voting theory 2. Probability and voting 2.1. Aggregating

More information

Constitutional Rights and Pareto Efficiency

Constitutional Rights and Pareto Efficiency Journal of Economic and Social Research, 1 (1) 1999, 109-117 Constitutional Rights and Pareto Efficiency Ahmet Kara 1 Abstract. This paper presents a sufficient condition under which constitutional rights

More information

A DISTANCE-BASED EXTENSION OF THE MAJORITY JUDGEMENT VOTING SYSTEM

A DISTANCE-BASED EXTENSION OF THE MAJORITY JUDGEMENT VOTING SYSTEM A DISTANCE-BASED EXTENSION OF THE MAJORITY JUDGEMENT VOTING SYSTEM EDURNE FALCÓ AND JOSÉ LUIS GARCÍA-LAPRESTA Abstract. It is common knowledge that the political voting systems suffer inconsistencies and

More information

The Beach Party Problem: An informal introduction to continuity problems in group choice *

The Beach Party Problem: An informal introduction to continuity problems in group choice * The Beach Party Problem: An informal introduction to continuity problems in group choice * by Nick Baigent Institute of Public Economics, Graz University nicholas.baigent@kfunigraz.ac.at Fax: +44 316 3809530

More information

THREE BRIEF PROOFS OF ARROW S IMPOSSIBILITY THEOREM JOHN GEANAKOPLOS COWLES FOUNDATION PAPER NO. 1116

THREE BRIEF PROOFS OF ARROW S IMPOSSIBILITY THEOREM JOHN GEANAKOPLOS COWLES FOUNDATION PAPER NO. 1116 THREE BRIEF PROOFS OF ARROW S IMPOSSIBILITY THEOREM BY JOHN GEANAKOPLOS COWLES FOUNDATION PAPER NO. 1116 COWLES FOUNDATION FOR RESEARCH IN ECONOMICS YALE UNIVERSITY Box 208281 New Haven, Connecticut 06520-8281

More information

Economics Bulletin, 2012, Vol. 32 No. 1 pp Introduction. 2. The preliminaries

Economics Bulletin, 2012, Vol. 32 No. 1 pp Introduction. 2. The preliminaries 1. Introduction In this paper we reconsider the problem of axiomatizing scoring rules. Early results on this problem are due to Smith (1973) and Young (1975). They characterized social welfare and social

More information

Game Theory: Spring 2017

Game Theory: Spring 2017 Game Theory: Spring 2017 Ulle Endriss Institute for Logic, Language and Computation University of Amsterdam Ulle Endriss 1 Plan for Today In this second lecture on mechanism design we are going to generalise

More information

Math 541 Fall 2008 Connectivity Transition from Math 453/503 to Math 541 Ross E. Staffeldt-August 2008

Math 541 Fall 2008 Connectivity Transition from Math 453/503 to Math 541 Ross E. Staffeldt-August 2008 Math 541 Fall 2008 Connectivity Transition from Math 453/503 to Math 541 Ross E. Staffeldt-August 2008 Closed sets We have been operating at a fundamental level at which a topological space is a set together

More information

Positive Political Theory II David Austen-Smith & Je rey S. Banks

Positive Political Theory II David Austen-Smith & Je rey S. Banks Positive Political Theory II David Austen-Smith & Je rey S. Banks Egregious Errata Positive Political Theory II (University of Michigan Press, 2005) regrettably contains a variety of obscurities and errors,

More information

Sufficient Conditions for Weak Group-Strategy-Proofness

Sufficient Conditions for Weak Group-Strategy-Proofness Sufficient Conditions for Weak Group-Strategy-Proofness T.C.A. Madhav Raghavan 31 July, 2014 Abstract In this note we study group-strategy-proofness, which is the extension of strategy-proofness to groups

More information

Social choice theory, Arrow s impossibility theorem and majority judgment

Social choice theory, Arrow s impossibility theorem and majority judgment Université Paris-Dauphine - PSL Cycle Pluridisciplinaire d Etudes Supérieures Social choice theory, Arrow s impossibility theorem and majority judgment Victor Elie supervised by Miquel Oliu Barton June

More information

Optimal Mechanism Design without Money

Optimal Mechanism Design without Money Optimal Mechanism Design without Money Alex Gershkov, Benny Moldovanu and Xianwen Shi January, 013 Abstract We consider the standard mechanism design environment with linear utility but without monetary

More information

arxiv: v1 [cs.gt] 9 Apr 2015

arxiv: v1 [cs.gt] 9 Apr 2015 Stronger Impossibility Results for Strategy-Proof Voting with i.i.d. Beliefs arxiv:1504.02514v1 [cs.gt] 9 Apr 2015 Samantha Leung Cornell University samlyy@cs.cornell.edu Edward Lui Cornell University

More information

Generalized Pigeonhole Properties of Graphs and Oriented Graphs

Generalized Pigeonhole Properties of Graphs and Oriented Graphs Europ. J. Combinatorics (2002) 23, 257 274 doi:10.1006/eujc.2002.0574 Available online at http://www.idealibrary.com on Generalized Pigeonhole Properties of Graphs and Oriented Graphs ANTHONY BONATO, PETER

More information

Lexicographic Choice under Variable Capacity Constraints

Lexicographic Choice under Variable Capacity Constraints Lexicographic Choice under Variable Capacity Constraints Battal Doğan Serhat Doğan Kemal Yıldız May 14, 2017 Abstract In several matching markets, in order to achieve diversity, agents priorities are allowed

More information

Non-deteriorating Choice Without Full Transitivity

Non-deteriorating Choice Without Full Transitivity Analyse & Kritik 29/2007 ( c Lucius & Lucius, Stuttgart) p. 163 187 Walter Bossert/Kotaro Suzumura Non-deteriorating Choice Without Full Transitivity Abstract: Although the theory of greatest-element rationalizability

More information

Isomorphisms between pattern classes

Isomorphisms between pattern classes Journal of Combinatorics olume 0, Number 0, 1 8, 0000 Isomorphisms between pattern classes M. H. Albert, M. D. Atkinson and Anders Claesson Isomorphisms φ : A B between pattern classes are considered.

More information

13 Social choice B = 2 X X. is the collection of all binary relations on X. R = { X X : is complete and transitive}

13 Social choice B = 2 X X. is the collection of all binary relations on X. R = { X X : is complete and transitive} 13 Social choice So far, all of our models involved a single decision maker. An important, perhaps the important, question for economics is whether the desires and wants of various agents can be rationally

More information

Implementation in undominated strategies by bounded mechanisms: The Pareto Correspondence

Implementation in undominated strategies by bounded mechanisms: The Pareto Correspondence Implementation in undominated strategies by bounded mechanisms: The Pareto Correspondence Saptarshi Mukherjee Eve Ramaekers Arunava Sen September 5, 2016 Preliminary and Incomplete Abstract We show that

More information

arxiv: v2 [cs.gt] 6 Jan 2015

arxiv: v2 [cs.gt] 6 Jan 2015 Voting with Coarse Beliefs Samantha Leung, Edward Lui, and Rafael Pass Department of Computer Science, Cornell University {samlyy,luied,rafael}@cs.cornell.edu arxiv:1405.5827v2 [cs.gt] 6 Jan 2015 April

More information

Rationality and solutions to nonconvex bargaining problems: rationalizability and Nash solutions 1

Rationality and solutions to nonconvex bargaining problems: rationalizability and Nash solutions 1 Rationality and solutions to nonconvex bargaining problems: rationalizability and Nash solutions 1 Yongsheng Xu Department of Economics Andrew Young School of Policy Studies Georgia State University, Atlanta,

More information

Allocating Public Goods via the Midpoint Rule

Allocating Public Goods via the Midpoint Rule Allocating Public Goods via the Midpoint Rule Tobias Lindner Klaus Nehring Clemens Puppe Preliminary draft, February 2008 Abstract We study the properties of the following midpoint rule for determining

More information

Stability of Jurisdiction Structures under the Equal Share and Median Rules

Stability of Jurisdiction Structures under the Equal Share and Median Rules Stability of Jurisdiction Structures under the Equal Share and Median Rules Anna Bogomolnaia Michel Le Breton Alexei Savvateev Shlomo Weber April 2005 Abstract In this paper we consider a model with multiple

More information

Strategy-Proofness and the Core in House Allocation Problems

Strategy-Proofness and the Core in House Allocation Problems Strategy-Proofness and the Core in House Allocation Problems Eiichi Miyagawa Department of Economics, Columbia University 420 West 118th Street, New York, NY 10027 Email: em437@columbia.edu July 28, 1999

More information

Dynamics of Stable Sets of Constitutions

Dynamics of Stable Sets of Constitutions Dynamics of Stable Sets of Constitutions HOUY Nicolas February 26, 2007 Abstract A Self-Designating social choice correspondence designates itself when it is implemented as a constitution and when the

More information

Mechanism Design without Money

Mechanism Design without Money Mechanism Design without Money MSc Thesis (Afstudeerscriptie) written by Sylvia Boicheva (born December 27th, 1986 in Sofia, Bulgaria) under the supervision of Prof Dr Krzysztof Apt, and submitted to the

More information

Recognizing single-peaked preferences on aggregated choice data

Recognizing single-peaked preferences on aggregated choice data Recognizing single-peaked preferences on aggregated choice data Smeulders B. KBI_1427 Recognizing Single-Peaked Preferences on Aggregated Choice Data Smeulders, B. Abstract Single-Peaked preferences play

More information

Separability and decomposition in mechanism design with transfers

Separability and decomposition in mechanism design with transfers Separability and decomposition in mechanism design with transfers Debasis Mishra, Swaprava Nath, and Souvik Roy August 9, 2017 Abstract In private values quasi-linear environment, we consider problems

More information

Theories of justice. Harsanyi s approach, Rawls approach, Unification

Theories of justice. Harsanyi s approach, Rawls approach, Unification Theories of justice Section 6.4 in JR. The choice of one swf over another is ultimately a choice between alternative sets of ethical values. Two main approaches: Harsanyi s approach, Rawls approach, Unification

More information

Judgement Aggregation

Judgement Aggregation Judgement Aggregation Stanford University ai.stanford.edu/ epacuit/lmh Fall, 2008 :, 1 The Logic of Group Decisions Fundamental Problem: groups are inconsistent! :, 2 The Logic of Group Decisions: The

More information

Extension of continuous functions in digital spaces with the Khalimsky topology

Extension of continuous functions in digital spaces with the Khalimsky topology Extension of continuous functions in digital spaces with the Khalimsky topology Erik Melin Uppsala University, Department of Mathematics Box 480, SE-751 06 Uppsala, Sweden melin@math.uu.se http://www.math.uu.se/~melin

More information

Separable Discrete Preferences

Separable Discrete Preferences Separable Discrete Preferences W. James Bradley Department of Mathematics and Statistics Calvin College Grand Rapids, MI 49546 USA Jonathan K. Hodge Department of Mathematics Grand Valley State University

More information

Resource-Monotonicity for House Allocation Problems

Resource-Monotonicity for House Allocation Problems Resource-Monotonicity for House Allocation Problems Lars Ehlers Bettina Klaus This Version: March 2004 Abstract We study a simple model of assigning indivisible objects (e.g., houses, jobs, offices, etc.)

More information

Chapter 12: Social Choice Theory

Chapter 12: Social Choice Theory Chapter 12: Social Choice Theory Felix Munoz-Garcia School of Economic Sciences Washington State University 1 1 Introduction In this chapter, we consider a society with I 2 individuals, each of them endowed

More information

The Coordinate-Wise Core for Multiple-Type Housing Markets is Second-Best Incentive Compatible

The Coordinate-Wise Core for Multiple-Type Housing Markets is Second-Best Incentive Compatible The Coordinate-Wise Core for Multiple-Type Housing Markets is Second-Best Incentive Compatible Bettina Klaus October 2005 Abstract We consider the generalization of Shapley and Scarf s (1974) model of

More information

Monotonicity and Nash Implementation in Matching Markets with Contracts

Monotonicity and Nash Implementation in Matching Markets with Contracts Monotonicity and Nash Implementation in Matching Markets with Contracts Claus-Jochen Haake Bettina Klaus March 2006 Abstract We consider general two-sided matching markets, so-called matching with contracts

More information