Skip to main content
Log in

Weak independence and the Pareto principle

  • Original Paper
  • Published:
Social Choice and Welfare Aims and scope Submit manuscript

Abstract

In this paper, the independence of irrelevant alternatives and the Pareto principle are simultaneously weakened in the Arrovian framework of social choice. Moreover, we also relax transitivity of social preferences. We show that impossibility remains under weaker versions of Arrow’s original conditions. Our results complement the recent work by Coban and Sanver (Soc Choice Welf 43(4):953–961, 2014).

This is a preview of subscription content, log in via an institution to check access.

Access this article

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Similar content being viewed by others

Notes

  1. The assumption that there exist at least four alternatives is needed to obtain the existence of a vetoer (Campbell and Kelly 2000a). See also Cato (2015).

  2. Pareto decisiveness requires that if there exists a unanimous agreement about judgement on two alternatives, the rule is “decisive” over the choice between the two in the sense that it should clarify which alternative is socially better. That is, a clear social judgement should be made under a “Paretian consensus.” This can be regarded as a necessary function of the decision process because a judgement under a conflict is hardly probable when the axiom is violated.

  3. Our two-step approach is originally due to Murakami (1968) and is developed by Cato (2012).

  4. Positive responsiveness requires that social preference must monotonically respond to the change of individual preferences. A monotonic response of social preferences is desirable for “democratic” decision processes, and indeed this axiom is satisfied by most voting methods, such as the simple majority rule and the absolute majority rule.

  5. Coban and Sanver (2014) refer to weak independence as quasi IIA.

  6. For a comprehensive treatment of these rationality concepts, see Bossert and Suzumura (2010).

  7. Murakami (1968) employs SNI and Wilson (1972) employs NI, while Fountain and Suzumura (1982) employ PD and Cato (2012) introduces PSD.

  8. The transitive closure of R is defined as follows: \((x,y) \in tc(R)\) if and only if there exist \(K \ge 1\) and \(x^0,x^1,\dots , x^K\) such that \((x^{k-1},x^k) \in R\) for all \(k \in \{ 1,\dots , K \}\).

  9. Bordes (1976) employs the choice-functional framework. A systematic treatment of the framework can be found in Suzumura (1983).

  10. For an argument on triple consistency, see Cato and Hirata (2010), which provide a characterization of a triple consistent CCR satisfying anonymity and neutrality.

  11. Blair et al. (1976) show an impossibility result under triple acyclicity, and Cato and Hirata (2010) provides a characterization of a triple acyclic CCR.

  12. For classical works of social choice with semi-transitive properties, see Blair and Pollak (1979) and Blau (1979).

  13. Fountain and Suzumura (1982, Theorem 2) extend the analysis of Mas-Colell and Sonnenschein (1972, Theorem 3) to the case without the Pareto principle. They examine a CCR that satisfies PD, IIA, and PR. Proposition 10 can be regarded as a counterpart of Theorem 2 of Fountain and Suzumura (1982).

  14. A set \(A \subseteq N\) is decisive for (resp. inversely decisive for) f if, for all \(x,y \in X\) and for all \({\mathbf {R}} \in {\mathcal {R}}^N\), \((x,y) \in \bigcap _{i \in N} P(R_i) \Rightarrow \) \((x,y) \in P(f({\mathbf {R}}))\) (resp. \((y,x) \in P(f({\mathbf {R}}))\)).

  15. Although PSD is not explicitly employed by Murakami (1968, Theorem 6-1) and Wilson (1972, Theorem 3), the axiom is implied by the completeness of social preference, which they assume.

  16. Campbell and Kelly (1993) and Cato and Sekiguchi (2012) employ Theorem 5 of Wilson (1972).

  17. Domain richness conditions are implicitly imposed.

  18. Gaertner (2001) contains a review on these works.

  19. For an argument on the free-triple assumption, see Kelly (1994). Malawski and Zhou (1994) assume the assumption to establish their results.

  20. See Le Breton and Weymark (2011) for a survey.

  21. See Border (1983) and Campbell (1990).

  22. In this respect, our problem is closely related to that of Fleurbaey et al. (2005), which introduce various weakenings of IIA in economic environments.

References

  • Arrow KJ (1963) Social choice and individual values, 2nd edn. Wiley, New York

    Google Scholar 

  • Baigent N (1987) Twitching weak dictators. J Econ 47(4):407–411

    Article  Google Scholar 

  • Banks JS (1995) Acyclic social choice from finite sets. Soc Choice Welf 12(3):293–310

    Article  Google Scholar 

  • Blair DH, Pollak RA (1979) Collective rationality and dictatorship: the scope of the Arrow theorem. J Econ Theory 21(1):186–194

    Article  Google Scholar 

  • Blair DH, Bordes G, Kelly JS, Suzumura K (1976) Impossibility theorems without collective rationality. J Econ Theory 13(3):361–379

    Article  Google Scholar 

  • Blau JH (1979) Semiorders and collective choice. J Econ Theory 21(1):195–206

    Article  Google Scholar 

  • Border KC (1983) Social welfare functions for economic environments with and without the Pareto principle. J Econ Theory 29(2):205–216

    Article  Google Scholar 

  • Bordes G (1976) Consistency, rationality and collective choice. Rev Econ Stud 43(3):451–457

    Article  Google Scholar 

  • Bossert W, Suzumura K (2010) Consistency, choice and rationality. Harvard University Press, Cambridge

    Google Scholar 

  • Brown DJ (1975) Aggregation of preferences. Q J Econ, 89(3):456–469

  • Campbell DE (1976) Democratic preference functions. J Econ Theory 12(2):259–272

    Article  Google Scholar 

  • Campbell DE (1990) Intergenerational social choice without the Pareto principle. J Econ Theory 50(2):414–423

    Article  Google Scholar 

  • Campbell DE, Kelly JS (1993) \(t\) or \(1-t\). That is the trade-off. Econometrica: J Econom Soc 61(6):1355–1365

  • Campbell DE, Kelly JS (2000a) Weak independence and veto power. Econ Lett 66(2):183–189

    Article  Google Scholar 

  • Campbell DE, Kelly JS (2000b) Information and preference aggregation. Soc Choice Welf 17(1):3–24

    Article  Google Scholar 

  • Cato S (2012) Social choice without the Pareto principle: a comprehensive analysis. Soc Choice Welf 39(4):869–889

    Article  Google Scholar 

  • Cato S (2013) Remarks on Suzumura consistent collective choice rules. Math Soc Sci 65(1):40–47

    Article  Google Scholar 

  • Cato S (2014) Weak independence and social semi-orders. Jpn Econ Rev 66(3):311–321. doi:10.1111/jere.12051

  • Cato S (2015) Weak independent decisiveness and the existence of a unique vetoer. Econ Lett 131:59–61

    Article  Google Scholar 

  • Cato S, Hirata D (2010) Collective choice rules and collective rationality: a unified method of characterizations. Soc Choice Welf 34(4):611–630

    Article  Google Scholar 

  • Cato S, Sekiguchi Y (2012) A generalization of Campbell and Kelly’s trade-off theorem. Soc Choice Welf 38(2):237–246

    Article  Google Scholar 

  • Coban C, Sanver MR (2014) Social choice without the Pareto principle under weak independence. Soc Choice Welf 43(4):953–961

    Article  Google Scholar 

  • Fleurbaey M, Suzumura K, Tadenuma K (2005) Arrovian aggregation in economic environments: how much should we know about indifference surfaces? J Econ Theory 124(1):22–44

    Article  Google Scholar 

  • Fountain J, Suzumura K (1982) Collective choice rules without the Pareto principle. Int Econ Rev 23(2):299–308

    Article  Google Scholar 

  • Gaertner W (2001) Domain conditions in social choice theory. Cambridge University Press, Cambridge

    Book  Google Scholar 

  • Kalai E, Muller E, Satterthwaite MA (1979) Social welfare functions when preferences are convex, strictly monotonic, and continuous. Public Choice 34(1):87–97

    Article  Google Scholar 

  • Kelly JS (1994) The free triple assumption. Soc Choice Welf 11(2):97–101

    Article  Google Scholar 

  • Le Breton M, Weymark JA (2011) Arrovian social choice theory on economic domains. In: Arrow KJ, Sen AK, Suzumura K (eds) Handbook of social choice and welfare, vol 2. North Holland, Amsterdam, pp 191–299

    Chapter  Google Scholar 

  • Malawski M, Zhou L (1994) A note on social choice theory without the Pareto principle. Soc Choice Welf 11(2):103–107

    Article  Google Scholar 

  • Mas-Colell A, Sonnenschein HF (1972) General possibility theorems for group decisions. Rev Econ Stud 39(2):185–192

    Article  Google Scholar 

  • Miller MK (2009) Social choice theory without Pareto: the pivotal voter approach. Math Soc Sci 58(2):251–255

    Article  Google Scholar 

  • Murakami Y (1968) Logic and social choice. Routledge, London

    Google Scholar 

  • Suzumura K (1983) Rational choice, collective decisions, and social welfare. Cambridge University Press, Cambridge

    Book  Google Scholar 

  • Wilson R (1972) Social choice theory without the Pareto principle. J Econ Theory 5(3):478–486

    Article  Google Scholar 

Download references

Acknowledgments

I thank an anonymous referee of this journal and Marc Fleurbaey for their valuable comments. This paper was financially supported by JSPS KAKENHI Grant Number 26870477. I also supported by Postdoctoral Fellowships for Research Abroad of JSPS.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Susumu Cato.

Additional information

The author is a Postdoctoral Fellow for Research Abroad of the Japan Society for the Promotion of Science.

Appendix: Proof of Proposition 9

Appendix: Proof of Proposition 9

Let us introduce some definitions. A set \(A \subseteq N\) is said to be decisive over (xy) for f if, for all \({\mathbf {R}} \in {\mathcal {R}}^N\),

$$\begin{aligned} (x,y)\in \bigcap _{i \in A}P(R_i) \Rightarrow (x,y) \in P(f({\mathbf {R}})). \end{aligned}$$

A decisive set A over (xy) for f is minimal if there exists no proper subset \(A'\) of A that is decisive over (xy) for f: for every proper subset \(A'\) of A, there exists \({\mathbf {R}}\in {\mathcal {R}}^N\) such that

$$\begin{aligned} (x,y) \in \bigcap _{i \in A'}P(R_i)\quad \hbox {and}\quad (y,x) \in f({\mathbf {R}}). \end{aligned}$$

In the presence of PR, it is the case that for all \(j \in A \), \((y,x) \in f({\mathbf {R}})\) when \((y,x) \in \bigcap _{i \in A {\setminus } \{j\}}P(R_i)\) and \((x,y) \in \bigcap _{i \in \{ j \} \cup (N {\setminus } A) }P(R_i)\).

A coalition \(A \subseteq N\) is said to be almost weakly decisive over (xy) for f if, for all \({\mathbf {R}} \in {\mathcal {R}}^N\),

$$\begin{aligned} \left[ (x,y) \in \bigcap _{i \in A}P(R_i) \hbox { and } (y,x) \in \bigcap _{i \in N {\setminus } A}P(R_i)\right] \Rightarrow (x,y) \in f({\mathbf {R}}). \end{aligned}$$

Let \(\mathcal {W}^*_f(x,y)\) be the family of coalitions almost weakly decisive over (xy) for f.

The proof of Proposition 9 proceeds in a similar manner to that of Theorem 3 of Mas-Colell and Sonnenschein (1972).

Lemma 2

Suppose that \(\# X \ge 3\). If a Paretian QSWF f satisfies WI and PR,  there exists \(v \in N\) such that \(\{ v \} \in {\mathcal {W}}^*(x,y)\) for some pair (xy) of alternatives.

Proof

Suppose that \(\{ i \} \notin \mathcal {W}^*(a,b)\) for all \(a,b \in X\) and for all \(i \in N\). Let \(M \subseteq V\) be a minimal set decisive over (yz) for f. If M is a singleton, the claim is immediately follows. In the rest of the proof, we suppose that \(\# M \ge 2\). Take any \(v,v' \in M\) (\(v \ne v'\)). Let \({\mathbf {R}}^1 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&\{ (x,y),(y,z) \} \subseteq P(R^1_v),\\&\{ (y,z),(z,x) \} \subseteq P(R^1_i)\quad \hbox {for all } i \in M {\setminus } \{ v \}, \\&\{ (z,x),(x,y) \} \subseteq P(R^1_i)\quad \hbox {for all } i \in N {\setminus } M. \end{aligned}$$

Since M is decisive over (yz) for f, we have \((y,z) \in P(f({\mathbf {R}}^1))\). Since \(\{ v \} \notin \mathcal {W}^*(x,z)\), \((z,x) \in P(f({\mathbf {R}}^1))\). Quasi-transitivity implies that \((y,x) \in P(f({\mathbf {R}}^1))\).

Let \({\mathbf {R}}^2 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&(y,z)\in P(R^2_v)\quad \hbox {and}\ (z,y)\in P(R^2_{v'}),\\&(y,z)\in P(R^2_i)\quad \hbox {for all } i \in M {\setminus } \{ v,v' \},\\&(z,y)\in P(R^2_i)\quad \hbox {for all } i \in N {\setminus } M. \end{aligned}$$

Since M is a minimal decisive set over (yz), it follows that \((z,y) \in f({\mathbf {R}}^2)\).

Let \({\mathbf {R}}^3 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&(y,z)\in I(R^3_v)\quad \hbox {and}\quad (z,y)\in P(R^3_{v'}),\\&(y,z)\in P(R^3_i)\quad \hbox {for all } i \in M {\setminus } \{ v,v' \},\\&(z,y)\in P(R^3_i)\quad \hbox {for all } i \in N {\setminus } M. \end{aligned}$$

Comparing \({\mathbf {R}}^3\) with \({\mathbf {R}}^2\), PR implies that \((z,y) \in P(f({\mathbf {R}}^3))\).

Let \({\mathbf {R}}^4 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&\{ (x,y),(y,z) \} \subseteq I(R^4_v),\\&\{ (z,y),(y,x) \} \subseteq P(R^4_{v'}),\\&\{ (y,x),(x,z) \} \subseteq P(R^4_i)\quad \hbox {for all } i \in M {\setminus } \{ v,v' \},\\&\{ (x,z),(z,y) \} \subseteq P(R^4_i)\quad \hbox {for all } i \in N {\setminus } M. \end{aligned}$$

Since \({\mathbf {R}}^4|_{\{ y,z \}}={\mathbf {R}}^3|_{\{ y,z \}}\), WI implies that \((z,y) \in f({\mathbf {R}}^4)\). Comparing \({\mathbf {R}}^4\) with \({\mathbf {R}}^1\), WI and PR implies that \((y,x) \in P(f({\mathbf {R}}^4))\). Quasi-transitivity implies that \((z,x) \in f({\mathbf {R}}^4)\).

Let \({\mathbf {R}}^5 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&(z,x) \in P(R^5_v)\quad \hbox {and}\ (z,x) \in P(R^5_{v'}),\\&(x,z) \in P(R^5_i)\quad \hbox {for all } i\in M {\setminus } \{ v,v' \}, \\&(x,z) \in P(R^5_i)\quad \hbox {for all } i \in N {\setminus } M. \end{aligned}$$

Comparing \({\mathbf {R}}^5\) with \({\mathbf {R}}^4\), PR implies that \((z,x) \in P(f({\mathbf {R}}^5))\). In the light of PR, \(\{v,v' \}\) is decisive over (zx) for f.

Let \({\mathbf {R}}^6 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&\{(y,z),(z,x)\} \subseteq P(R^6_v),\\&\{(z,x),(x,y)\} \subseteq P(R^6_{v'}),\\&\{(x,y,(y,z)\} \subseteq P(R^6_i)\quad \hbox {for all } i \in N {\setminus } \{ v,v' \}. \end{aligned}$$

Since \(\{ v,v'\}\) is decisive over (zx) for f, we have \((z,x) \in P(f({\mathbf {R}}^6))\). Since \(\{ v \} \notin \mathcal {W}^*(y,x)\), it follows that \((x,y) \in P(f({\mathbf {R}}^6))\). Quasi-transitivity implies that \((z,y) \in P(f({\mathbf {R}}^6))\). This implies that \(\{ v' \} \in \mathcal {W}^*(z,y)\). The proof is complete. \(\square \)

Lemma 3

Suppose that \(\# X \ge 3,\) \(\# N \ge 4,\) and a Paretian QSWF f satisfies WI and PR. Let \(x,y \in X\). If \(\{ v \} \in {\mathcal {W}}^* (x,y),\) then \(\{ v \} \in {\mathcal {W}}^* (x,z) \hbox { for all } z \in X \).

Proof

Suppose that \(\{ v \} \in \mathcal {W}^* (x,y)\). Since \(\# N \ge 4\), there exists a partition \((A_1,A_2,A_3)\) of \(N {\setminus } \{ v \}\). Let \({\mathbf {R}}^1 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&\{(x,y), (y,z) \} \subseteq P(R^1_v),\\&[ (x,y) \in I(R^1_i) \wedge (y,z) \in P(R^1_i)]\quad \hbox {for all } i \in A_1,\\&\{(y,z),(z,x) \} \subseteq P(R^1_i)\quad \hbox {for all } i \in A_2,\\&\{(y,z),(z,x) \} \subseteq P(R^1_i)\quad \hbox {for all } i \in A_3. \end{aligned}$$

Since \(\{ v \} \in \mathcal {W}^* (x,y)\), PR implies that \((x,y) \in P(f({\mathbf {R}}^1))\). Since f is Paretian, we have \((y,z) \in P(f({\mathbf {R}}^1))\). By quasi-transitivity, we have \((x,z) \in P(f({\mathbf {R}}^1))\).

Let \({\mathbf {R}}^2 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&\{(y,x), (x,z) \} \subseteq P(R^2_v),\\&\{(y,x), (x,z) \} \subseteq P(R^2_i)\quad \hbox {for all } i \in A_1,\\&\{(z,y),(y,x) \} \subseteq P(R^2_i)\quad \hbox {for all } i \in A_2,\\&[(x,z) \in I(R_i) \wedge (y,x) \in P(R^2_i)]\quad \hbox {for all } i \in A_3. \end{aligned}$$

Comparing \({\mathbf {R}}^2\) with \({\mathbf {R}}^1\), PR implies that \((x,z) \in P(f({\mathbf {R}}^2))\). Since f is Paretian, \((y,x) \in P(f({\mathbf {R}}^2))\). Thus, \((y,z) \in P(f({\mathbf {R}}^2))\).

Let \({\mathbf {R}}^3 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&\{(x,y), (y,z) \} \subseteq P(R^3_v),\\&\{(y,z), (z,x) \} \subseteq P(R^3_i)\quad \hbox {for all } i \in A_1,\\&\{ (x,y),(y,z) \} \subseteq I(R^3_i)\quad \hbox {for all } i \in A_2,\\&\{(y,z), (z,x) \} \subseteq P(R^3_i) \quad \hbox {for all } i \in A_3. \end{aligned}$$

Since \(\{ v \} \in {\mathcal {W}}^* (x,y)\), PR implies that \((x,y) \in P(f({\mathbf {R}}^3))\). Comparing \({\mathbf {R}}^3\) with \({\mathbf {R}}^2\), WI and PR imply that \((y,z) \in P(f({\mathbf {R}}^3))\). Quasi-transitivity implies that \((x,z) \in P(f({\mathbf {R}}^3))\).

Let \({\mathbf {R}}^4 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&\{(y,x), (x,z) \} \subseteq P(R^4_v),\\&\{(z,y), (y,x) \} \subseteq P(R^4_i)\quad \hbox {for all } i \in A_1,\\&\{ (y,x),(x,z) \} \subseteq P(R^4_i)\quad \hbox {for all } i \in A_2,\\&\{(z,y), (y,x) \} \subseteq P(R^4_i)\quad \hbox {for all } i \in A_3. \end{aligned}$$

Comparing \({\mathbf {R}}^4\) with \({\mathbf {R}}^3\), WI and PR imply that \((x,z) \in P(f({\mathbf {R}}^4))\). Since f is Paretian, \((y,x) \in P(f({\mathbf {R}}^4))\). Quasi-transitivity implies that \((y,z) \in P(f({\mathbf {R}}^4))\).

Let \({\mathbf {R}}^5 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&\{(x,y), (y,z) \} \subseteq P(R^5_v),\\&[(z,x) \in P(R^5_i) \wedge (x,y) \in I(R^5_i)] \quad \hbox {for all } i \in A_1,\\&\{ (y,z),(z,x) \} \subseteq P(R^5_i) \quad \hbox {for all } i \in A_2,\\&[(y,z) \in I(R^5_i) \wedge (z,x) \in P(R^5_i)]\quad \hbox {for all } i \in A_3. \end{aligned}$$

Since \(\{ v \} \in \mathcal {W}^* (x,y)\), PR implies that \((x,y) \in P(f({\mathbf {R}}^5))\). Comparing \({\mathbf {R}}^5\) with \({\mathbf {R}}^4\), WI and PR implies that \((y,z) \in P(f({\mathbf {R}}^5))\). By quasi-transitivity, we have \((x,z) \in P(f({\mathbf {R}}^5))\). From WI, if \((x,z) \in P(R^5_v)\) and \((z,x) \in P(R_i)\) for all \( i \in N {\setminus } \{ v \}\), we have \((x,z) \in f({\mathbf {R}})\). Thus, it follows that \(\{ v \} \in {\mathcal {W}}^* (x,z)\). \(\square \)

The proof of the following lemma is similar to Lemma 3. We omit it.

Lemma 4

Suppose that \(\# X \ge 3,\) \(\# N \ge 4,\) and a Paretian QSWF f satisfies WI and PR. Let \(x,y \in X\). If \(\{ v \} \in \mathcal {W}^* (x,y),\) then \(\{ v \} \in {\mathcal {W}}^* (z,y) \hbox { for all } z \in X \).

Proof of Proposition 9

By Lemma 2, there exists \(v \in N\) such that \(\{ v \} \in \mathcal {W}^*(x,y)\) for some (xy). Lemmas 3 and 4 implies that \(\{ v \} \in \mathcal {W}^* (w,z) \hbox { for all } w,z \in X\). PR implies that \(\{ v \}\) is a vetoer for f.

Now, we prove the uniqueness. Suppose that there exists two vetoers, v and \(v'\). Let \({\mathbf {R}}^1 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&(x,y) \in P(R^1_v)\quad \hbox {and}\quad (y,x) \in P(R^1_{v'}),\\&(x,y) \in I(R^1_i)\quad \hbox {for all } i \in N {\setminus } \{ v,v' \}. \end{aligned}$$

Since both v and \(v'\) are vetoers, it follows that \((x,y) \in f({\mathbf {R}}^1)\) and \((y,x) \in f({\mathbf {R}}^1)\). Let \({\mathbf {R}}^2 \in {\mathcal {R}}^N\) such that

$$\begin{aligned}&(x,y) \in P(R^2_v)\quad \hbox {and}\quad (y,x) \in P(R^2_{v'}),\\&(x,y) \in P(R^2_i)\quad \hbox {for all } i \in N {\setminus } \{ v,v' \}. \end{aligned}$$

By PR, we have \((x,y) \in P(f({\mathbf {R}}^2))\). This contradicts the fact that \(v'\) is a vetoer. \(\square \)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Cato, S. Weak independence and the Pareto principle. Soc Choice Welf 47, 295–314 (2016). https://doi.org/10.1007/s00355-016-0960-5

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00355-016-0960-5

JEL Classification

Navigation