Skip to main content
Log in

Binary strategy-proof social choice functions with indifference

  • Research Article
  • Published:
Economic Theory Aims and scope Submit manuscript

Abstract

We prove a representation formula that gives a new characterization of the coalitionally strategy-proof binary social choice functions. We give this characterization in the case of social choice functions selecting one of two alternatives (i.e., binary social choice). The domain of the functions we consider consists of profiles of preferences over a society of arbitrary cardinality and indifference is admitted. Strategy proofness is meant to be coalitional: No group of agents has incentives to form a coalition that can manipulate the social choice for their own advantage with false reporting.

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. Say the range is \(\{a, b\}\subseteq A\). In cases like this, we talk of two-valued SCF.

  2. However, notice that in Lahiri and Pramanik (2019, Section 5), where Lahiri and Pramanik compare their own result with those of Ju (2003), one can also read: “the description of the rules in this paper is easier than Ju (2003).”

  3. Also observe that, when we consider the collection \(\varvec{{{\mathcal {F}}}}\) consisting of a single family \({{\mathcal {F}}}\) and restrict to strict profiles, \((++)\) generalizes the representation in Theorem 1.1 in a straightforward manner.

References

  • Barberà, S., Sonnenschein, H., Zhou, L.: Voting by committees. Econometrica 59, 595–609 (1991)

    Article  Google Scholar 

  • Barberà, S., Berga, D., Moreno, B.: Individual versus group strategy-proofness: when do they coincide? J. Econ.Theory 145, 1648–1674 (2010)

    Article  Google Scholar 

  • Barberà, S., Berga, D., Moreno, B.: Group strategy-proof social choice functions with binary ranges and arbitrary domains: characterization results. Int. J. Game Theory 41, 791–808 (2012)

    Article  Google Scholar 

  • Basile, A., Rao, S., Bhaskara Rao, K.P.S.: The structure of two-valued strategy-proof social choice functions with indifference (2020). arXiv:2002.06341v1 [econ.TH]

  • Campbell, D.E., Kelly, J.S.: Relaxing Pareto optimality in economic environments. Econ. Theory 10, 115–130 (1997). https://doi.org/10.1007/s001990050149

    Article  Google Scholar 

  • Ju, B.: A characterization of strategy-proof voting rules for separable weak orderings. Soc. Choice Welf. 21, 469–499 (2003)

    Article  Google Scholar 

  • Kirman, A.P., Sondermann, D.: Arrow’s theorem, many agents and invisible dictators. J. Econ. Theory 5, 267–277 (1972)

    Article  Google Scholar 

  • Lahiri, A., Pramanik, A.: On strategy-proof social choice between two alternatives. Soc. Choice Welf. (2019). https://doi.org/10.1007/s00355-019-01220-7

    Article  Google Scholar 

  • Larsson, B., Svensson, L.G.: Strategy-proof voting on the full preference domain. Math. Soc. Sci. 52, 272–287 (2006)

    Article  Google Scholar 

  • Manjunath, V.: Group strategy-proofness and voting between two alternatives. Math. Soc. Sci. 63, 239–242 (2012)

    Article  Google Scholar 

  • Peleg, B.: Game Theoretic Analysis of Voting in Committees. Cambridge Univesity Press, Cambridge (1984)

    Book  Google Scholar 

  • Rao, S., Basile, A., Bhaskara Rao, K.P.S.: On the ultrafilter representation of coalitionally strategy-proof social choice functions. Econ. Theory Bull. 6, 1–13 (2018)

    Article  Google Scholar 

  • Svensson, L.G.: Coalitional strategy-proofness and fairness. Econ. Theory 40, 227–245 (2009). https://doi.org/10.1007/s00199-008-0366-4

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Achille Basile.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix

Appendix

1.1 Proof of Theorem 4.2 for an infinite V

In V is not finite, the process described for proving Theorem 4.2 may not stop in finitely many steps.

Suppose that for an ordinal \(\beta\) we have produced a double collection \(\langle \varvec{{{\mathcal {I}}}},\varvec{{{\mathcal {F}}}}\rangle\), with \(\varvec{{{\mathcal {I}}}}=(I_\alpha )_{\alpha <\beta }\), \(\varvec{{{\mathcal {F}}}}=({{\mathcal {F}}}_\alpha )_{\alpha <\beta }\), and the corresponding sets \(({{\mathcal {P}}}_\alpha )_{\alpha <\beta }\) such that

$$\begin{aligned} \{\sim _V\}\cup \bigcup _{\alpha <\beta } {{\mathcal {P}}}_\alpha \subseteq \{\phi =\psi _{\langle {{\mathcal {I}}},{{\mathcal {F}}}\rangle ,\, x}\}. \end{aligned}$$

One of the following is true.

  1. 1.

    For every profile \(P\notin \{\sim _V\}\cup \bigcup _{\alpha <\beta } {{\mathcal {P}}}_\alpha\), the value \(\phi (P)\) is x.

  2. 2.

    There is a profile \(Q\notin \{\sim _V\}\cup \bigcup _{\alpha <\beta } {{\mathcal {P}}}_\alpha\) such that \(\phi (Q)\ne x.\)

In case 1, we stop since we have obtained \(\phi =\psi _{\langle \varvec{{{\mathcal {I}}}},\varvec{{{\mathcal {F}}}}\rangle ,\, x}\).

In case 2, let Q be a profile such that \(Q\ne \sim _V\)\(Q\notin {{\mathcal {P}}}_\alpha \,\, \forall \alpha <\beta ,\) and \(\phi (Q)\ne x.\)

Let \(I = I(Q)\) be the set of voters that are indifferent under Q. The set I is a proper coalition distinct from all previous \(I_\alpha\). Indeed, since \(Q\ne \sim _V\), the set I is a proper subset of V. The implication    \(I=\varnothing (\Leftrightarrow Q\) is strict) \(\Rightarrow Q\in \cup _{\alpha <\beta }\, {{\mathcal {P}}}_\alpha\)    tells us that I is nonempty and if \(I=I_\alpha\) for some \(\alpha\), then Q is a strict profile over \(I^c_\alpha\) hence \(Q\in {{\mathcal {P}}}_\alpha\).

Consider the restriction of \(\phi\)

$$\begin{aligned} \phi _{I^c}(P_{I^c}):=\phi (\sim _I, P_{I^c}). \end{aligned}$$

It is CSP and takes two values (x for \(P_{I^c} =\sim _{I^c} ,\,\, \phi (Q)\) for \(P_{I^c}=Q_{I^c}\)). Apply Proposition 4.1 to \(\phi _{I^c}\) and let \({{\mathcal {F}}}\) be a SSC family on \(I^c\) such that

$$\begin{aligned}&D(a, P_{I^c})\in {{\mathcal {F}}}\Rightarrow \phi _{I^c}(P_{I^c})= a, \\&D(b, P_{I^c})\in {{\mathcal {F}}}^{\,\circ} \Rightarrow \phi _{I^c}(P_{I^c})= b. \end{aligned}$$

The new double collection \(\langle \varvec{{{\mathcal {I}}}}^*,\varvec{{{\mathcal {F}}}}^*\rangle\), consisting of \(I_\alpha\) for \(\alpha <\beta\), \(I_\beta :=I\), \({{\mathcal {F}}}_\alpha\) for \(\alpha <\beta\),    \({{\mathcal {F}}}_\beta :={{\mathcal {F}}}\), has the property

$$\begin{aligned} \{\sim _V\}\cup \bigcup _{\alpha \le \beta } {{\mathcal {P}}}_\alpha \subseteq \{\phi =\psi _{\langle {{\mathcal {I}}}^*,{{\mathcal {F}}}^*\rangle ,\, x}\}; \end{aligned}$$

therefore, by transfinite induction we can conclude that a well-ordered set \(\Lambda\) exists and a double collection \(\langle \varvec{{{\mathcal {I}}}},\varvec{{{\mathcal {F}}}}\rangle\) indexed on \(\Lambda\) such that \(\phi =\psi _{\langle \varvec{{{\mathcal {I}}}},\varvec{{{\mathcal {F}}}}\rangle ,\, x}\). \(\square\)

1.2 On the equality \(M_x=\phi _{{ \varvec{{{\mathcal {F}}}}},\, x}\)

Let V be a set of cardinality n. For \(k=1, \dots , n\), let us denote by \({{\mathcal {G}}}_k\) the SSC family consisting of the subsets of V with cardinality at least k. We know that \({{\mathcal {G}}}_k^{\,\circ} ={{\mathcal {G}}}_{n-k+1}\).

We show the existence of a collection \(\varvec{{{\mathcal {F}}}}\) such that the SCF \(M_x\) coincide with the SCF \(\phi _{{ \varvec{{{\mathcal {F}}}}},\, x}\). The collection \(\varvec{{{\mathcal {F}}}}\) consists of exactly n SSC families \({{\mathcal {F}}}_0, {{\mathcal {F}}}_1, \dots , {{\mathcal {F}}}_{n-1}\) on V. If we compare this result with the structure of the SCF \(\phi _2\) of Example 2.4 (this function can be defined for a finite V as well), we see that the structure we obtain for \(M_x\) precisely gives \({{\mathcal {F}}}_i\)’s the role physical agents have in \(\phi _2\) in dictating the social choice corresponding to a given profile.

Proposition 4.3

In order to have the equality \(M_b=\phi _{{ \varvec{{{\mathcal {F}}}}},\, b}\) one can take the collection \(\varvec{{{\mathcal {F}}}}=({{\mathcal {F}}}_i)_{0\le i\le n-1}\) as follows.

$$\begin{aligned} \varvec{{{\mathcal {F}}}}= \left\{ \begin{array} {ll} {{\mathcal {F}}}_{2i}={{\mathcal {G}}}_{p+1+i},{{\mathcal {F}}}_{2i+1}={{\mathcal {G}}}_{p-i}, \text{ with } {0\le i\le p-1}, &{}\quad \text{ if } \; n=2p \\ {{\mathcal {F}}}_0={{\mathcal {G}}}_{p+1}, {{\mathcal {F}}}_{2i+1}={{\mathcal {G}}}_{p+2+i}, {{\mathcal {F}}}_{2i+2}={{\mathcal {G}}}_{p-i}, \text{ with } {0\le i\le p-1}, &{}\quad \text{ if } \; n=2p+1.\\ \end{array} \right. \end{aligned}$$

On the other hand, for having the equality \(M_a=\phi _{{ \varvec{{{\mathcal {F}}}}},\, a}\) one can take the collection \(\varvec{{{\mathcal {F}}}}=({{\mathcal {F}}}_i)_{0\le i\le n-1}\) as follows.

$$\begin{aligned} \varvec{{{\mathcal {F}}}}= \left\{ \begin{array} {ll} {{\mathcal {F}}}_{2i}={{\mathcal {G}}}_{p-i},{{\mathcal {F}}}_{2i+1}={{\mathcal {G}}}_{p+1+i}, \text{ with } {0\le i\le p-1}, &{}\quad \text{ if } \; n=2p \\ {{\mathcal {F}}}_0={{\mathcal {G}}}_{p+1}, {{\mathcal {F}}}_{2i+1}={{\mathcal {G}}}_{p-i}, {{\mathcal {F}}}_{2i+2}={{\mathcal {G}}}_{p+2+i}, \text{ with } {0\le i\le p-1}, &{}\quad \text{ if } \; n=2p+1.\\ \end{array} \right. \end{aligned}$$

Proof

We show that \(M_b=\phi _{{ \varvec{{{\mathcal {F}}}}},\, b}\). For this, let us suppose that n is odd: \(n=2p+1\) (being the case \(n=2p\) similar). Hence, the sequence of SSC families forming \(\varvec{{{\mathcal {F}}}}\) is:

\({{\mathcal {F}}}_0={{\mathcal {G}}}_{p+1}, {{\mathcal {F}}}_1={{\mathcal {G}}}_{p+2}, {{\mathcal {F}}}_2={{\mathcal {G}}}_p, {{\mathcal {F}}}_3={{\mathcal {G}}}_{p+3}, {{\mathcal {F}}}_4={{\mathcal {G}}}_{p-1}, \ldots , {{\mathcal {F}}}_{2i+1}={{\mathcal {G}}}_{p+2+i}, {{\mathcal {F}}}_{2i+2}={{\mathcal {G}}}_{p-i}, \dots , {{\mathcal {F}}}_{n-2}={{\mathcal {G}}}_{2p+1}={{\mathcal {G}}}_n, {{\mathcal {F}}}_{n-1}={{\mathcal {G}}}_1.\)

Notice that \({{\mathcal {F}}}_0\) is self-dual and, pairwise, the families \({{\mathcal {F}}}_{2i+1}, {{\mathcal {F}}}_{2i+2}\) are dual to each other.

Hence, given a profile P, its index \(\lambda (P)\) is the first element of \(\{0,1,2, \dots , n-1\}\) for which the corresponding statement in the list below (where \(i=0,\dots , p-1\)) is true:

  • 0. either \(|D(a, P)|\ge p+1\) or \(|D(b, P)|\ge p+1\)

  • 1. either \(|D(a, P)|\ge p+2\) or \(|D(b, P)|\ge p\)

  • 2. either \(|D(a, P)|\ge p\) or \(|D(b, P)|\ge p+2\)

  • ...

  • \(2i+1\). either \(|D(a, P)|\ge p+2+i\) or \(|D(b, P)|\ge p-i\)

  • \(2i+2\). either \(|D(a, P)|\ge p-i\) or \(|D(b, P)|\ge p+2+i\)

  • ...

  • \(n-2\). either \(|D(a, P)|\ge n\) or \(|D(b, P)|\ge 1\)

  • \(n-1\). either \(|D(a, P)|\ge 1\) or \(|D(b, P)|\ge n\)

if one such element exists, otherwise \(\lambda (P)=\infty\).

We remind that, by definition,

$$\begin{aligned} \phi _{\varvec{{{\mathcal {F}}}}, \, b}(P) = \left\{ \begin{array} {ll} a, &{}\quad \text{ if } \; D(a,P)\in {{{\mathcal {F}}}}_{\lambda (P)}\\ b, &{}\quad \text{ if } \; D(b,P)\in {{{\mathcal {F}}}}_{\lambda (P)}^{\,\circ} \\ b, &{}\quad \text{ if } \; \lambda (P)=\infty \,\,.\\ \end{array} \right. \end{aligned}$$

Now, checking that \(\phi _{\varvec{{{\mathcal {F}}}}, \, b}(P) =M_b(P)\) for every P is straightforward. \(\square\)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Basile, A., Rao, S. & Rao, K.P.S.B. Binary strategy-proof social choice functions with indifference. Econ Theory 73, 807–826 (2022). https://doi.org/10.1007/s00199-020-01273-1

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00199-020-01273-1

Keywords

JEL Classification

Mathematics Subject Classification

Navigation