Abstract
We consider a spatial bargaining model where players collectively choose a facility location on a two-dimensional rectilinear distance space through bargaining using the unanimity rule. We show that as players become infinitely patient, their stationary subgame perfect equilibrium utilities converge to the utilities that satisfy the lexicographic maximin utility criterion introduced by Sen (Collective choice and social welfare, 1970).
Similar content being viewed by others
Notes
Note that \(v^{\mathrm{T}}=\left( {\frac{b_1-b_4}{2},\frac{b_1+b_4}{2}}\right)\), \(v^{\mathrm{B}}=\left( {\frac{b_2-b_3}{2},\frac{b_2+b_3}{2}}\right)\), \(v^{\mathrm{R}}=\left( {\frac{b_1-b_3}{2},\frac{b_1+b_3}{2}}\right)\), and \(v^{\mathrm{L}}=\left( {\frac{b_2-b_4}{2},\frac{b_2+b_4}{2}}\right)\).
Note that \(d^+=b_1-b_2\) and \(d^-=b_4-b_3\). Furthermore, note that \(v_1^{\mathrm{T}} < v_1^{\mathrm{B}}\) and \(v_2^{\mathrm{R}} < v_2^{\mathrm{L}}\) if \(d^+ < d^-\), \(v_1^{\mathrm{T}} > v_1^{\mathrm{B}}\) and \(v_2^{\mathrm{R}} > v_2^{\mathrm{L}}\) if \(d^+ > d^-\), and \(v_1^{\mathrm{T}} = v_1^{\mathrm{B}}\) and \(v_2^{\mathrm{R}} = v_2^{\mathrm{L}}\) if \(d^+ = d^-\).
Note that \(m^{\mathrm{L}}=\left( {\min \left\{ {v_1^{\mathrm{T}},v_1^{\mathrm{B}}}\right\} ,v_2^{\mathrm{R}}}\right)\) and \(m^{\mathrm{R}}=\left( {\max \left\{ {v_1^{\mathrm{T}},v_1^{\mathrm{B}}}\right\} ,v_2^{\mathrm{L}}}\right)\). Furthermore, note that \(m_1^{\mathrm{L}}<m_1^{\mathrm{R}}\) and \(m_2^{\mathrm{L}}<m_2^{\mathrm{R}}\) if \(d^+ < d^-\), \(m_1^{\mathrm{L}}<m_1^{\mathrm{R}}\) and \(m_2^{\mathrm{L}}>m_2^{\mathrm{R}}\) if \(d^+ > d^-\), and \(m_1^{\mathrm{L}}=m_1^{\mathrm{R}}\) and \(m_2^{\mathrm{L}}=m_2^{\mathrm{R}}\) if \(d^+ = d^-\).
Note that for each \(x,x',x'' \in X\), \(x \in R\left( {x',x''}\right)\) if and only if \(d\left( {x',x''}\right) =d\left( {x',x}\right) +d\left( {x,x''}\right)\); that is, location x is an element of \(R\left( {x',x''}\right)\) if and only if location x is on a shortest rectilinear path between locations \(x'\) and \(x''\). Furthermore, note that for each \(x',x'',x''' \in X\),
$$\begin{aligned}&R\left( {x',x''}\right) \cup R\left( {x',x'''}\right) \cup R\left( {x'',x'''}\right) \\&\quad =\left\{ {x \in X}\mid {\left( {\min \left\{ {x_1',x_1'',x_1'''}\right\} , \min \left\{ {x_2',x_2'',x_2'''}\right\} }\right) \le x \le \left( {\max \left\{ {x_1',x_1'',x_1'''}\right\} ,\max \left\{ {x_2',x_2'',x_2'''}\right\} }\right) }\right\} ; \end{aligned}$$that is, \(R\left( {x',x''}\right) \cup R\left( {x',x'''}\right) \cup R\left( {x'',x'''}\right)\) denotes a rectangle such that locations \(x'\), \(x''\), and \(x'''\) are elements of the rectangle.
Note that as long as the set of lexicographic minimax distance locations is not a singleton, the players’ SSPE proposals do not necessarily converge to a lexicographic minimax distance location.
It is known that in the case of a tree network, a similar result holds for any number of players (see Schummer and Vohra (2002)). In contrast, it is known that in the case of a two-or-more-dimensional rectilinear space, although there exists an anonymous, Pareto optimal, and strategy-proof mechanism for three or less players, there exists no such mechanism for four or more players (see Walsh (2020)).
Note that if \(m_1^{\mathrm{L}}< y_1^{\gamma } < m_1^{\mathrm{R}}\), \(d^+ \not =d^-\).
Note that if \(\min \left\{ {m_2^{\mathrm{L}},m_2^{\mathrm{R}}}\right\}<y_2^{\gamma }<\max \left\{ {m_2^{\mathrm{L}}, m_2^{\mathrm{R}}}\right\}\), \(d^+\not =d^-\).
Note that the horizontal line through player \(\gamma\)’s location and the set of minimax distance locations except for the two endpoints intersect if and only if \(\min \left\{ {m_2^{\mathrm{L}},m_2^{\mathrm{R}}}\right\}< y_2^{\gamma } < \max \left\{ {m_2^{\mathrm{L}},m_2^{\mathrm{R}}}\right\}\). See Figs. 2 and 3.
References
Asheim, G. B. (1992). A unique solution to \(n\)-person sequential bargaining. Games and Economic Behavior, 4, 169–181.
Banks, J. S., & Duggan, J. (2000). A bargaining model of collective choice. American Political Science Review, 94, 73–88.
Baron, D. P. (1991). A spatial bargaining theory of government formation in parliamentary systems. American Political Science Review, 85, 137–164.
Baron, D. P. (1996). A dynamic theory of collective goods programs. American Political Science Review, 90, 316–330.
Baron, D. P., & Ferejohn, J. A. (1989). Bargaining in legislatures. American Political Science Review, 83, 1181–1206.
Binmore, K. (1987). Perfect equilibria in bargaining models. In K. Binmore & P. Dasgupta (Eds.), The economics of bargaining (pp. 77–105). Basil Blackwell.
Cardona, D., & Polanski, A. (2013). Voting rules and efficiency in one-dimensional bargaining games with endogenous protocol. Social Choice and Welfare, 41, 217–240.
Cardona, D., & Ponsati, C. (2007). Bargaining one-dimensional social choices. Journal of Economic Theory, 137, 627–651.
Cardona, D., & Ponsati, C. (2011). Uniqueness of stationary equilibria in bargaining one-dimensional policies under (super) majority rules. Games and Economic Behavior, 73, 65–75.
Chepoi, V., & Dragan, F. F. (1996). Condorcet and median points of simple rectilinear polygons. Location Science, 4, 21–35.
Cho, S., & Duggan, J. (2003). Uniqueness of stationary equilibria in a one-dimensional model of bargaining. Journal of Economic Theory, 113, 118–130.
Cho, S., & Duggan, J. (2009). Bargaining foundations of the median voter theorem. Journal of Economic Theory, 144, 851–868.
Elzinga, J., & Hearn, D. W. (1972). Geometrical solutions for some minimax location problems. Transportation Science, 6, 379–394.
Eraslan, H., & Evdokimov, K. S. (2019). Legislative and multilateral bargaining. Annual Review of Economics, 11, 443–472.
Hansen, P., & Thisse, J. F. (1981). Outcomes of voting and planning: Condorcet, weber and rawls locations. Journal of Public Economics, 16, 1–15.
Herings, P. J. J., & Predtetchinski, A. (2010). One-dimensional bargaining with Markov recognition probabilities. Journal of Economic Theory, 145, 189–215.
Herrero, M. (1985). A strategic bargaining approach to market institutions. Unpublished Ph.D. thesis, London University.
Kalandrakis, T. (2016). Pareto efficiency in the dynamic one-dimensional bargaining model. Journal of Theoretical Politics, 28, 525–536.
Kawamori, T., & Yamaguchi, K. (2010). Outcomes of bargaining and planning in single facility location problems. Mathematical Social Sciences, 59, 38–45.
Moulin, H. (1980). On strategy-proofness and single peakedness. Public Choice, 35, 437–455.
Predtetchinski, A. (2011). One-dimensional bargaining. Games and Economic Behavior, 72, 526–543.
Rubinstein, A. (1982). Perfect equilibrium in a bargaining model. Econometrica, 50, 97–110.
Schummer, J., & Vohra, R. V. (2002). Strategy-proof location on a network. Journal of Economic Theory, 104, 405–428.
Sen, A. K. (1970). Collective choice and social welfare. Holden-Day.
Sutton, J. (1986). Non-cooperative bargaining theory: An introduction. Review of Economic Studies, 53, 709–724.
Walsh, T. (2020). Strategy proof mechanisms for facility location in Euclidean and Manhattan space. arXiv:2009.07983v1 [cs.AI].
Zapal, J. (2016). Markovian equilibria in dynamic spatial legislative bargaining: Existence with three players. Games and Economic Behavior, 98, 235–242.
Zapal, J. (2020). Simple Markovian equilibria in dynamic spatial legislative bargaining. European Journal of Political Economy, 63, 101816.
Acknowledgements
The author is grateful to Tomohiko Kawamori, Yoshihiro Ohashi, an anonymous editor, and three anonymous referees for their useful comments.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
In this appendix, we prove Theorems 1 and 2. Hereafter, since we fix the value of \(\delta\), we abbreviate \(G\left( {\delta }\right)\), \(r^M\left( {\delta }\right)\), \(r^L\left( {\delta }\right)\), and \(A_*\left( {\delta }\right)\) as G, \(r^M\), \(r^L\), and \(A_*\), respectively.
1.1 Appendix 1: Preliminary lemmas for Theorems 1 and 2
If \(m_1^{\mathrm{L}}< y_1^{\gamma } < m_1^{\mathrm{R}}\),Footnote 8 define
Then, \(n^{\mathrm{V}}\) denotes the intersection of the vertical line through player \(\gamma\)’s location (that is, \(\left\{ {x \in X}\mid {x_1=y_1^{\gamma }}\right\}\)) and the set of minimax distance locations except for the two endpoints (that is, \(M {\setminus } \left\{ {m^{\mathrm{L}},m^{\mathrm{R}}}\right\}\)).Footnote 9 Furthermore, if \(\min \left\{ {m_2^{\mathrm{L}},m_2^{\mathrm{R}}}\right\}<y_2^{\gamma }<\max \left\{ {m_2^{\mathrm{L}},m_2^{\mathrm{R}}}\right\}\),Footnote 10 define
Then, \(n^{\mathrm{H}}\) denotes the intersection of the horizontal line through player \(\gamma\)’s location (that is, \(\left\{ {x \in X}\mid {x_2=y_2^{\gamma }}\right\}\)) and the set of minimax distance locations except for the two endpoints (that is, \(M {\setminus } \left\{ {m^{\mathrm{L}},m^{\mathrm{R}}}\right\}\)).Footnote 11 Remember that \(\ell ^{\mathrm{L}}\) and \(\ell ^{\mathrm{R}}\) denote the locations that satisfy \(\ell ^{\mathrm{L}}, \ell ^{\mathrm{R}} \in \arg \min _{x \in M} d^{\gamma }\left( {x}\right)\) and \(\ell _1^{\mathrm{L}} \le x_1 \le \ell _1^{\mathrm{R}}\) for each \(x \in \arg \min _{x \in M} d^{\gamma }\left( {x}\right)\). Furthermore, remember that \(\underline{d}=\min _{x \in M} d^{\gamma }\left( {x}\right)\). Then, we obtain the following lemma from Figures 12 and 13 based on Elzinga and Hearn’s (1972) construction of minimax distance locations (or tedious calculations).
Lemma 1
(1) \(\ell ^{\mathrm{L}}\), \(\ell ^{\mathrm{R}}\), and \(\underline{d}\) are given in Table 1if \(d^+<d^-\). (2) \(\ell ^{\mathrm{L}}\), \(\ell ^{\mathrm{R}}\), and \(\underline{d}\) are given in Table 2if \(d^+>d^-\). (3) \(\ell ^{\mathrm{L}}=m^{\mathrm{L}}\), \(\ell ^{\mathrm{R}}=m^{\mathrm{R}}\), and \(\underline{d}=d^{\gamma }\left( {m^{\mathrm{L}}}\right) =d^{\gamma }\left( {m^{\mathrm{R}}}\right)\) if \(d^+=d^-\).
For each \(Z \in 2^X\) and \(r \in {\mathbb {R}}_+\), define \(O\left( {Z;r}\right) :=\cup _{x' \in Z}\left\{ {x \in X}\mid {d\left( {x,x'}\right) < r}\right\}\). Then, \(O\left( {Z;r}\right)\) denotes the r-open neighborhood of Z.
Lemma 2
(1) Suppose that \(x \in C\left( {M;r^M}\right)\). Then, for each \(i \in I\), \(d^{i}\left( {x}\right) \le \overline{d}+r^M\). (2) Suppose that \(x \in C\left( {L;r^L}\right)\). Then, \(d^{\gamma }\left( {x}\right) \le \underline{d}+r^L\). (3) Suppose that \(x \in O\left( {M;r^M}\right)\). Then, for each \(i \in I\), \(d^{i}\left( {x}\right) <\overline{d}+r^M\). (4) Suppose that \(x \in O\left( {L;r^L}\right)\). Then, \(d^{\gamma }\left( {x}\right) <\underline{d}+r^L\). (5) Suppose that \(x \in X {\setminus } C\left( {M;r^M}\right)\). Then, for some \(i \in I\), \(d^{i}\left( {x}\right) >\overline{d}+r^M\). (6) Suppose that \(x \in C\left( {M;r^M}\right) {\setminus } C\left( {L;r^L}\right)\). Then, \(d^{\gamma }\left( {x}\right) >\underline{d}+r^L\).
Proof
Statements (1) and (3) immediately follow from Proposition 1, and Statements (2) and (4) immediately follow from Corollary 1.
As shown in Fig. 14, note that for each \(z^{\mathrm{L}}, z^{\mathrm{R}} \in X\) such that \(z_1^{\mathrm{L}} \le z_1^{\mathrm{R}}\) and \(|z_1^{\mathrm{L}}-z_1^{\mathrm{R}}|=|z_2^{\mathrm{L}}-z_2^{\mathrm{R}}|\) and \(r \in {\mathbb {R}}_+\),
Then, by Proposition 1, \(C\left( {M;r^M}\right) =C\left( {\left[ {m^{\mathrm{L}},m^{\mathrm{R}}}\right] ;r^M}\right) =\cap _{j=1}^4 M_j\) where
Furthermore, by Corollary 1, \(C\left( {L;r^L}\right) =C\left( {\left[ {\ell ^{\mathrm{L}},\ell ^{\mathrm{R}}}\right] ;r^L}\right) =\cap _{j=1}^4 L_j\) where
Remember that \(C\left( {M;r^M}\right) =\cap _{j=1}^4 M_j\). Thus, \(X {\setminus } C\left( {M;r^M}\right) =\cup _{j=1}^4 M_j^c\).
Suppose that \(x \in X {\setminus } C\left( {M;r^M}\right)\). Then, \(x \in \cup _{j=1}^4 M_j^c\). Note that since
it follows that \(d\left( {x,x'}\right) >\overline{d}+r^M\) for each \(x' \in \left\{ {x \in X}\mid {x_2=-x_1+b_1}\right\}\) if \(x \in M_1^c\), for each \(x' \in \left\{ {x \in X}\mid {x_2=-x_1+b_2}\right\}\) if \(x \in M_2^c\), for each \(x' \in \left\{ {x \in X}\mid {x_2=x_1+b_3}\right\}\) if \(x \in M_3^c\), and for each \(x' \in \left\{ {x \in X}\mid {x_2=x_1+b_4}\right\}\) if \(x \in M_4^c\). Furthermore, note that \(\left\{ {y^i}\right\} _{i \in I} \cap \left\{ {x \in X}\mid {x_2=-x_1+b_1}\right\} \not =\emptyset\), \(\left\{ {y^i}\right\} _{i \in I} \cap \left\{ {x \in X}\mid {x_2=-x_1+b_2}\right\} \not =\emptyset\), \(\left\{ {y^i}\right\} _{i \in I} \cap \left\{ {x \in X}\mid {x_2=x_1+b_3}\right\} \not =\emptyset\), and \(\left\{ {y^i}\right\} _{i \in I} \cap \left\{ {x \in X}\mid {x_2=x_1+b_4}\right\} \not =\emptyset\). Thus, for some \(i \in I\), \(d^{i}\left( {x}\right) >\overline{d}+r^M\). Hence, Statement (5) is true.
Remember that \(C\left( {L;r^L}\right) =\cap _{j=1}^4 L_j\). Let
Then, since \(r^M \le r^L\), for each \(j \in \left\{ {1,2,3,4}\right\}\), \(M_j \subset G_j\), and thus, \(M_j \cap G_j^c=\emptyset\). Let
Then, if \(d^+ < d^-\), it follows from Lemma 1 that (i) if \(y_1^{\gamma } \le m_1^{\mathrm{L}}\) or \(y_2^{\gamma } \le m_2^{\mathrm{L}}\), since \(\ell _1^{\mathrm{L}}+\ell _2^{\mathrm{L}}=m_1^{\mathrm{L}}+m_2^{\mathrm{L}}\), \(L_1=G_1\), and if \(m_1^{\mathrm{L}}<y_1^{\gamma }\) and \(m_2^{\mathrm{L}}<y_2^{\gamma }\), since \(\ell _1^{\mathrm{L}}+\ell _2^{\mathrm{L}}=y_1^{\gamma }+y_2^{\gamma }-\underline{d}\), \(L_1=H_1\); (ii) if \(m_1^{\mathrm{R}} \le y_1^{\gamma }\) or \(m_2^{\mathrm{R}} \le y_2^{\gamma }\), since \(\ell _1^{\mathrm{R}}+\ell _2^{\mathrm{R}}=m_1^{\mathrm{R}}+m_2^{\mathrm{R}}\), \(L_2=G_2\), and if \(y_1^{\gamma }<m_1^{\mathrm{R}}\) and \(y_2^{\gamma }<m_2^{\mathrm{R}}\), since \(\ell _1^{\mathrm{R}}+\ell _2^{\mathrm{R}}=y_1^{\gamma }+y_2^{\gamma }+\underline{d}\), \(L_2=H_2\); (iii) since \(-\ell _1^{\mathrm{L}}+\ell _2^{\mathrm{L}}=\frac{b_3+b_4}{2}=-m_1^{\mathrm{L}}+m_2^{\mathrm{L}}\), \(L_3=G_3\); and (iv) since \(-\ell _1^{\mathrm{R}}+\ell _2^{\mathrm{R}}=\frac{b_3+b_4}{2}=-m_1^{\mathrm{R}}+m_2^{\mathrm{R}}\), \(L_4=G_4\). If \(d^+ > d^-\), it follows from Lemma 1 that (i) since \(\ell _1^{\mathrm{L}}+\ell _2^{\mathrm{L}}=\frac{b_1+b_2}{2}=m_1^{\mathrm{L}}+m_2^{\mathrm{L}}\), \(L_1=G_1\); (ii) since \(\ell _1^{\mathrm{R}}+\ell _2^{\mathrm{R}}=\frac{b_1+b_2}{2}=m_1^{\mathrm{R}}+m_2^{\mathrm{R}}\), \(L_2=G_2\); (iii) if \(y_1^{\gamma } \le m_1^{\mathrm{L}}\) or \(m_2^{\mathrm{L}} \le y_2^{\gamma }\), since \(-\ell _1^{\mathrm{L}}+\ell _2^{\mathrm{L}}=-m_1^{\mathrm{L}}+m_2^{\mathrm{L}}\), \(L_3=G_3\), and if \(m_1^{\mathrm{L}} < y_1^{\gamma }\) and \(y_2^{\gamma } <m_2^{\mathrm{L}}\), since \(-\ell _1^{\mathrm{L}}+\ell _2^{\mathrm{L}}=-y_1^{\gamma }+y_2^{\gamma }+\underline{d}\), \(L_3=H_3\); and (iv) if \(m_1^{\mathrm{R}} \le y_1^{\gamma }\) or \(y_2^{\gamma } \le m_2^{\mathrm{R}}\), since \(-\ell _1^{\mathrm{R}}+\ell _2^{\mathrm{R}}=-m_1^{\mathrm{R}}+m_2^{\mathrm{R}}\), \(L_4=G_4\), and if \(y_1^{\gamma } <m_1^{\mathrm{R}}\) and \(m_2^{\mathrm{R}} < y_2^{\gamma }\), since \(-\ell _1^{\mathrm{R}}+\ell _2^{\mathrm{R}}=-y_1^{\gamma }+y_2^{\gamma }-\underline{d}\), \(L_4=H_4\). If \(d^+=d^-\), it follows from Lemma 1 that (i) since \(\ell _1^{\mathrm{L}}+\ell _2^{\mathrm{L}}=m_1^{\mathrm{L}}+m_2^{\mathrm{L}}\), \(L_1=G_1\); (ii) since \(\ell _1^{\mathrm{R}}+\ell _2^{\mathrm{R}}=m_1^{\mathrm{R}}+m_2^{\mathrm{R}}\), \(L_2=G_2\); (iii) since \(-\ell _1^{\mathrm{L}}+\ell _2^{\mathrm{L}}=-m_1^{\mathrm{L}}+m_2^{\mathrm{L}}\), \(L_3=G_3\); and (iv) since \(-\ell _1^{\mathrm{R}}+\ell _2^{\mathrm{R}}=-m_1^{\mathrm{R}}+m_2^{\mathrm{R}}\), \(L_4=G_4\). Thus, for each \(j \in \left\{ {1,2,3,4}\right\}\), since \(M_j \cap G_j^c=\emptyset\), \(\left( {\cap _{j'=1}^4 M_{j'}}\right) \cap L_j^c \subset H_j^c\). Hence, since \(C\left( {M;r^M}\right) =\cap _{j'=1}^4 M_{j'}\) and \(C\left( {L;r^L}\right) =\cap _{j=1}^4 L_j\), \(C\left( {M;r^M}\right) {\setminus } C\left( {L;r^L}\right) =\left( {\cap _{j'=1}^4 M_{j'}}\right) {\setminus } \left( {\cap _{j=1}^4 L_j}\right) =\left( {\cap _{j'=1}^4 M_{j'}}\right) \cap \left( {\cap _{j=1}^4 L_j}\right) ^c =\left( {\cap _{j'=1}^4 M_{j'}}\right) \cap \left( {\cup _{j=1}^4 L_j^c}\right) =\cup _{j=1}^4\left( {\left( {\cap _{j'=1}^4 M_{j'}}\right) \cap L_j^c}\right) \subset \cup _{j=1}^4 H_j^c\).
Suppose that \(x \in C\left( {M;r^M}\right) {\setminus } C\left( {L;r^L}\right)\). Then, \(x \in \cup _{j=1}^4 H_j^c\). Note that
Thus, \(d^{\gamma }\left( {x}\right) >\underline{d}+r^L\). Hence, Statement (6) is true. \(\square\)
Lemma 3
(1) For each \(i \in \left\{ {\alpha ,\beta }\right\}\), \(\min _{x \in C\left( {M;r^M}\right) }d^i\left( {x}\right) =\min _{x \in C\left( {L;r^M}\right) }d^i\left( {x}\right) =\min _{x \in A_*}d^{i}\left( {x}\right) =\max \left\{ {\overline{d}-r^M,0}\right\}\). (2) \(\min _{x \in C\left( {M;r^M}\right) }d^{\gamma }\left( {x}\right) =\min _{x \in C\left( {L;r^M}\right) }d^{\gamma }\left( {x}\right) =\min _{x \in A_*}d^{\gamma }\left( {x}\right) =\max \left\{ {\underline{d}-r^M,0}\right\}\).
Proof
(1) First, we show that \(\min _{x \in C\left( {M;r^M}\right) }d^i\left( {x}\right) =\max \left\{ {\overline{d}-r^M,0}\right\}\). Consider the case where \(\overline{d} \le r^M\). Then, since \(y^i \in C\left( {M;r^M}\right)\) by Proposition 1, \(\min _{x \in C\left( {M;r^M}\right) }d^i\left( {x}\right) =d^i\left( {y^i}\right) =0=\max \left\{ {\overline{d}-r^M,0}\right\}\). Consider the case where \(\overline{d}>r^M\). Suppose that for some \(x \in C\left( {M;r^M}\right)\), \(d^i\left( {x}\right) <\overline{d}-r^M\). Then, for some \(x' \in M\), \(d\left( {x,x'}\right) \le r^M\). Note that by Proposition 1, \(d^i\left( {x'}\right) \ge \overline{d}\). Furthermore, note that by the triangle inequality, \(d^i\left( {x'}\right) \le d^i\left( {x}\right) +d\left( {x,x'}\right)\). Thus, \(\overline{d} \le d^i\left( {x'}\right) \le d^i\left( {x}\right) +d\left( {x,x'}\right) <\overline{d}\), which is a contradiction. Hence, for each \(x \in C\left( {M;r^M}\right)\), \(d^i\left( {x}\right) \ge \overline{d}-r^M\). Let \(x \in M\). Then, since \(d^i\left( {x}\right) =\overline{d}\) by Proposition 1, \(d^i\left( {x}\right) >r^M\). Thus, for some \(x' \in X\), \(x' \in R\left( {x,y^i}\right) \cap \left\{ {x' \in X}\mid {d\left( {x,x'}\right) =r^M}\right\}\). Note that since \(R\left( {x,y^i}\right) \cap \left\{ {x' \in X}\mid {d\left( {x,x'}\right) =r^M}\right\} \subset C\left( {M;r^M}\right)\), \(x' \in C\left( {M;r^M}\right)\). Furthermore, note that since \(d^i\left( {x}\right) =d^i\left( {x'}\right) +d\left( {x,x'}\right)\) by \(x' \in R\left( {x,y^i}\right)\), \(d^i\left( {x}\right) =\overline{d}\) by Proposition 1, and \(d\left( {x,x'}\right) =r^M\) by \(x' \in \left\{ {x' \in X}\mid {d\left( {x,x'}\right) =r^M}\right\}\), \(d^i\left( {x'}\right) =d^i\left( {x}\right) -d\left( {x,x'}\right) =\overline{d}-r^M\). Thus, for some \(x' \in C\left( {M;r^M}\right)\), \(d^i\left( {x'}\right) =\overline{d}-r^M\). Hence, \(\min _{x \in C\left( {M;r^M}\right) }d^i\left( {x}\right) =\overline{d}-r^M=\max \left\{ {\overline{d}-r^M,0}\right\}\). Next, we can show that \(\min _{x \in C\left( {L;r^M}\right) }d^i\left( {x}\right) =\max \left\{ {\overline{d}-r^M,0}\right\}\) by replacing “Proposition 1” and “M” in the above proof with “Corollary 1” and “L.” Finally, we show that \(\min _{x \in A_*}d^i\left( {x}\right) =\max \left\{ {\overline{d}-r^M,0}\right\}\). Note that \(L \subset M\), \(r^M \le r^L\), and \(A_*=C\left( {M;r^M}\right) \cap C\left( {L;r^L}\right)\). Then, \(C\left( {L;r^M}\right) \subset A_* \subset C\left( {M;r^M}\right)\). Thus, \(\min _{x \in C\left( {M;r^M}\right) }d^i\left( {x}\right) \le \min _{x \in A_*}d^i\left( {x}\right) \le \min _{x \in C\left( {L;r^M}\right) }d^i\left( {x}\right)\). Hence, since \(\min _{x \in C\left( {M;r^M}\right) }d^i\left( {x}\right) =\max \left\{ {\overline{d}-r^M,0}\right\}\) and \(\min _{x \in C\left( {L;r^M}\right) }d^i\left( {x}\right) =\max \left\{ {\overline{d}-r^M,0}\right\}\), \(\min _{x \in A_*}d^i\left( {x}\right) =\max \left\{ {\overline{d}-r^M,0}\right\}\).
(2) First, we show that \(\min _{x \in C\left( {M;r^M}\right) }d^{\gamma }\left( {x}\right) =\max \left\{ {\underline{d}-r^M,0}\right\}\). Consider the case where \(\underline{d} \le r^M\). Then, since \(y^{\gamma } \in C\left( {M;r^M}\right)\) by Corollary 1, \(\min _{x \in C\left( {M;r^M}\right) }d^{\gamma }\left( {x}\right) =d^{\gamma }\left( {y^{\gamma }}\right) =0 =\max \left\{ {\underline{d}-r^M,0}\right\}\). Consider the case where \(\underline{d}>r^M\). Suppose that for some \(x \in C\left( {M;r^M}\right)\), \(d^{\gamma }\left( {x}\right) <\underline{d}-r^M\). Then, for some \(x' \in M\), \(d\left( {x,x'}\right) \le r^M\). Note that by Proposition 1, \(d^{\gamma }\left( {x'}\right) \ge \underline{d}\). Furthermore, note that by the triangle inequality, \(d^{\gamma }\left( {x'}\right) \le d^{\gamma }\left( {x}\right) +d\left( {x,x'}\right)\). Thus, \(\underline{d} \le d^{\gamma }\left( {x'}\right) \le d^{\gamma }\left( {x}\right) +d\left( {x,x'}\right) <\underline{d}\), which is a contradiction. Hence, for each \(x \in C\left( {M;r^M}\right)\), \(d^{\gamma }\left( {x}\right) \ge \underline{d}-r^M\). Let \(x \in L\). Then, since \(d^{\gamma }\left( {x}\right) =\underline{d}\) by Corollary 1, \(d^{\gamma }\left( {x}\right) >r^M\). Thus, for some \(x' \in X\), \(x' \in R\left( {x,y^{\gamma }}\right) \cap \left\{ {x' \in X}\mid {d\left( {x,x'}\right) =r^M}\right\}\). Note that since \(R\left( {x,y^{\gamma }}\right) \cap \left\{ {x' \in X}\mid {d\left( {x,x'}\right) =r^M}\right\} \subset C\left( {M;r^M}\right)\), \(x' \in C\left( {M;r^M}\right)\). Furthermore, note that since \(d^{\gamma }\left( {x}\right) =d^{\gamma }\left( {x'}\right) +d\left( {x,x'}\right)\) by \(x' \in R\left( {x,y^{\gamma }}\right)\), \(d^{\gamma }\left( {x}\right) =\underline{d}\) by Corollary 1, and \(d\left( {x,x'}\right) =r^M\) by \(x' \in \left\{ {x' \in X}\mid {d\left( {x,x'}\right) =r^M}\right\}\), \(d^{\gamma }\left( {x'}\right) =d^{\gamma }\left( {x}\right) -d\left( {x,x'}\right) =\underline{d}-r^M\). Thus, for some \(x' \in C\left( {M;r^M}\right)\), \(d^{\gamma }\left( {x'}\right) =\underline{d}-r^M\). Hence, \(\min _{x \in C\left( {M;r^M}\right) }d^{\gamma }\left( {x}\right) =\underline{d}-r^M=\max \left\{ {\underline{d}-r^M,0}\right\}\). Next, we can show that \(\min _{x \in C\left( {L;r^M}\right) }d^{\gamma }\left( {x}\right) =\max \left\{ {\underline{d}-r^M,0}\right\}\) by replacing “Proposition 1” and “M” in the above proof with “Corollary 1” and “L.” Finally, we show that \(\min _{x \in A_*}d^{\gamma }\left( {x}\right) =\max \left\{ {\underline{d}-r^M,0}\right\}\). Note that \(L \subset M\), \(r^M \le r^L\), and \(A_*=C\left( {M;r^M}\right) \cap C\left( {L;r^L}\right)\). Then, \(C\left( {L;r^M}\right) \subset A_* \subset C\left( {M;r^M}\right)\). Thus, \(\min _{x \in C\left( {M;r^M}\right) }d^{\gamma }\left( {x}\right) \le \min _{x \in A_*}d^{\gamma }\left( {x}\right) \le \min _{x \in C\left( {L;r^M}\right) }d^{\gamma }\left( {x}\right)\). Hence, since \(\min _{x \in C\left( {M;r^M}\right) }d^{\gamma }\left( {x}\right) =\max \left\{ {\underline{d}-r^M,0}\right\}\) and \(\min _{x \in C\left( {L;r^M}\right) }d^{\gamma }\left( {x}\right) =\max \left\{ {\underline{d}-r^M,0}\right\}\), \(\min _{x \in A_*}d^{\gamma } \left( {x}\right) =\max \left\{ {\underline{d}-r^M,0}\right\}\). \(\square\)
1.2 Appendix 2: Proof of Theorem 1
Let \(i \in I\). Then, for each \(i' \in \left\{ {\alpha ,\beta }\right\}\),
Furthermore
Thus, for each \(i' \in I\), \(x_*^i \in A_*^{i'}\). Hence, in \(\sigma\), player i’s proposal \(x_*^i\) is accepted by all players. Therefore, in \(\sigma\), player i’s utility in the subgame beginning with her proposal is \(u^i\left( {x_*^i}\right)\).
Consider each proposing node of each player i. Let \(x \in X\) such that \(u^i\left( {x}\right) > u^i\left( {x_*^i}\right)\). Then, since \(d^i\left( {x}\right) <d^i\left( {x_*^i}\right)\) and \(x_*^i \in \arg \min _{x \in A_*}d^i\left( {x}\right)\), \(x \not \in A_*\). Thus, since \(A_*=C\left( {M;r^M}\right) \cap C\left( {L;r^L}\right)\), either \(x \in X {\setminus } C\left( {M;r^M}\right)\) or \(x \in C\left( {M;r^M}\right) {\setminus } C\left( {L;r^L}\right)\). Hence, by Lemma 2, either (1) for some \(i' \in I\), \(d^{i'}\left( {x}\right) >\overline{d}+r^M\) or (2) \(d^{\gamma }\left( {x}\right) >\underline{d}+r^L\). Suppose that (1) for some \(i' \in I\), \(d^{i'}\left( {x}\right) >\overline{d}+r^M\). Note that since \(d^i\left( {x_*^i}\right) >0\) by \(d^i\left( {x_*^i}\right) >d^i\left( {x}\right) \ge 0\) and \(d^{i}\left( {x_*^{i}}\right) \le \max \left\{ {\overline{d}-r^M,0}\right\}\) by Lemma 3 and \(\underline{d} \le \overline{d}\), \(\overline{d}>r^M\). Then, since \(\overline{d}>r^M\) and \(d^{i'}\left( {x_*^{i'}}\right) \le \max \left\{ {\overline{d}-r^M,0}\right\}\) by Lemma 3 and \(\underline{d} \le \overline{d}\), \(d^{i'}\left( {x_*^{i'}}\right) \le \overline{d}-r^M\). Thus,
Hence, \(x \not \in A_*^{i'}\). Suppose that (2) \(d^{\gamma }\left( {x}\right) >\underline{d}+r^L\). Then,
Thus, \(x \not \in A_*^{\gamma }\). Hence, in \(\sigma\), x is rejected by some player. Therefore, by a one-stage deviation of proposing x such that \(u^i\left( {x}\right) > u^i\left( {x_*^i}\right)\), player i obtains \(\delta u^i\left( {x_*^{i''}}\right) \le \max _{x \in A_*}u^i\left( {x}\right) = u^i\left( {x_*^i}\right)\) for some \(i'' \in I\). By a one-stage deviation of proposing x such that \(u^i\left( {x}\right) \le u^i\left( {x_*^i}\right)\), player i obtains \(u^i\left( {x}\right) \le u^i\left( {x_*^i}\right)\) if x is accepted by all players, and \(\delta u^i\left( {x_*^{i''}}\right) \le \max _{x \in A_*}u^i\left( {x}\right) = u^i\left( {x_*^i}\right)\) for some \(i'' \in I\) if x is rejected by some player. Thus, in \(\sigma\), player i’s proposal \(x_*^i\) is optimal.
Consider each responding node of each player i following each proposal x. By rejecting x, player i obtains \(\delta u^i\left( {x_*^i}\right)\). By accepting x, player i obtains \(u^i\left( {x}\right) \ge \delta u^i\left( {x_*^i}\right)\) if \(x \in A_*^i\), \(u^i\left( {x}\right) \le \delta u^i\left( {x_*^i}\right)\) if \(x \in \left( {\cap _{\iota \succ i} A_\dag ^\iota }\right) {\setminus } A_\dag ^i\), and \(\delta u^i\left( {x_*^{i'}}\right) \le \delta \max _{x \in A_*}u^i\left( {x}\right) = \delta u^i\left( {x_*^i}\right)\) for some \(i' \in I\) if \(x \notin \cap _{\iota \succ i} A_\dag ^\iota\). Thus, in \(\sigma\), player i’s response is optimal.
Hence, \(\sigma\) is an SPE in G. Therefore, since \(\sigma\) is stationary, \(\sigma\) is an SSPE in G. \(\square\)
1.3 Appendix 3: Proof of Theorem 2
Let \(\sigma\) be an SSPE in G. For each \(i \in I\), let \(x^i\) be player i’s proposal in \(\sigma\). For each \(i \in I\), let \(\pi ^i\) be player i’s utility in the subgame beginning with her proposal in \(\sigma\). For each \(i \in I\), let \(A^i\) be the set of locations that player i accepts in \(\sigma\). For each \(i \in I\), let \(A_O^i := \left\{ {x \in X}\mid {u^i\left( {x}\right) > \delta \pi ^i}\right\}\) and \(A_C^i := \left\{ {x \in X}\mid {u^i\left( {x}\right) \ge \delta \pi ^i}\right\}\). Let \(A := \cap _{i \in I} A^i\), \(A_O := \cap _{i \in I} A_O^i\), and \(A_C := \cap _{i \in I} A_C^i\).
Lemma 4
\(A_O \subset A \subset A_C\).
Proof
It suffices to show that for each \(i \in I\), \(\cap _{\iota \succsim i} A_O^\iota \subset \cap _{\iota \succsim i} A^\iota \subset \cap _{\iota \succsim i} A_C^\iota\). We show this by induction. (1) Let \(i \in I\). Suppose that player i is the last responder. Let \(x \in X\). Consider a response of player i to proposal x. If she accepts it, she obtains \(u^i\left( {x}\right)\). If she rejects it, she obtains \(\delta \pi ^i\). Since \(\sigma\) is an SPE in G, \(x \in A^i\) if \(u^i\left( {x}\right) > \delta \pi ^i\), and \(x \notin A^i\) if \(u^i\left( {x}\right) < \delta \pi ^i\). Then, \(A_O^i \subset A^i \subset A_C^i\). Thus, \(\cap _{\iota \succsim i} A_O^\iota \subset \cap _{\iota \succsim i} A^\iota \subset \cap _{\iota \succsim i} A_C^\iota\). (2) Let \(i,i' \in I\) and suppose that player i is the next responder of player \(i'\). Suppose that \(\cap _{\iota \succsim i} A_O^\iota \subset \cap _{\iota \succsim i} A^\iota \subset \cap _{\iota \succsim i} A_C^\iota\). Let \(x \in \cap _{\iota \succsim i'} A_O^\iota\). Consider a response of player \(i'\) to proposal x. If she rejects it, she obtains \(\delta \pi ^{i'}\). Since \(x \in \cap _{\iota \succsim i'} A_O^\iota \subset \cap _{\iota \succsim i} A_O^\iota \subset \cap _{\iota \succsim i} A^\iota\), if she accepts it, she obtains \(u^{i'}\left( {x}\right)\). Note that since \(x \in \cap _{\iota \succsim i'} A_O^\iota \subset A_O^{i'}\), \(u^{i'}\left( {x}\right) > \delta \pi ^{i'}\). Then, since \(\sigma\) is an SPE in G, \(x \in A^{i'}\). Note that \(x \in \cap _{\iota \succsim i'} A_O^\iota \subset \cap _{\iota \succsim i} A_O^\iota \subset \cap _{\iota \succsim i} A^\iota\). Then, \(x \in \cap _{\iota \succsim i'} A^\iota\). Thus, \(\cap _{\iota \succsim i'} A_O^\iota \subset \cap _{\iota \succsim i'} A^\iota\). Let \(x \in \cap _{\iota \succsim i'} A^\iota\). Consider a response of player \(i'\) to proposal x. If she rejects it, she obtains \(\delta \pi ^{i'}\). Since \(x \in \cap _{\iota \succsim i'} A^\iota \subset \cap _{\iota \succsim i} A^\iota\), if she accepts it, she obtains \(u^{i'}\left( {x}\right)\). Note that \(x \in \cap _{\iota \succsim i'} A^\iota \subset A^{i'}\). Then, since \(\sigma\) is an SPE in G, \(u^{i'}\left( {x}\right) \ge \delta \pi ^{i'}\), and thus, \(x \in A_C^{i'}\). Note that \(x \in \cap _{\iota \succsim i'} A^\iota \subset \cap _{\iota \succsim i} A^\iota \subset \cap _{\iota \succsim i} A_C^\iota\). Then, \(x \in \cap _{\iota \succsim i'} A_C^\iota\). Thus, \(\cap _{\iota \succsim i'} A^\iota \subset \cap _{\iota \succsim i'} A_C^\iota\). Hence, \(\cap _{\iota \succsim i'} A_O^\iota \subset \cap _{\iota \succsim i'} A^\iota \subset \cap _{\iota \succsim i'} A_C^\iota\). \(\square\)
Lemma 5
For each \(i \in I\), \(x^i \in A\), \(\pi ^i = u^i\left( {x^i}\right)\), and \(x^i \in \arg \max _{x \in A} u^i\left( {x}\right)\).
Proof
Suppose that for each \(i \in I\), \(x^i \notin A\). Let \(i' \in I\). Then, for each \(i \in I\), since \(\pi ^i = 0\), \(u^{i}\left( {x^{i'}}\right) > 0 = \delta \pi ^{i}\). Thus, \(x^{i'} \in A_O\). Hence, by Lemma 4, \(x^{i'} \in A\), which is a contradiction. Suppose that for some \(i \in I\), \(x^i \notin A\). If \(\pi ^i = 0\), since \(x^{i'} \in A\) for some \(i' \in I\), player i can improve her utility from 0 to \(u^i\left( {x^{i'}}\right) > 0\) by proposing \(x^{i'}\), which is a contradiction. If \(\pi ^i > 0\), since \(\pi ^i = \delta ^t u^i\left( {x}\right)\) for some \(x \in A\) and \(t \in {\mathbb {N}}\), player i can improve her utility from \(\delta ^t u^i\left( {x}\right)\) to \(u^i\left( {x}\right) >\delta ^t u^i\left( {x}\right)\) by proposing x, which is a contradiction. Thus, for each \(i \in I\), \(x^i \in A\). Hence, for each \(i \in I\), \(\pi ^i = u^i\left( {x^i}\right)\). Furthermore, since \(\sigma\) is an SPE in G, for each \(i \in I\), \(x^i \in \arg \max _{x \in A} u^i\left( {x}\right)\). \(\square\)
Lemma 6
Suppose that \(\max _{i \in I} d^i\left( {x^i}\right) =0\). Then, for each \(i \in I\), \(x^i \in \arg \min _{x \in A_*}d^i\left( {x}\right)\).
Proof
Let \(i \in I\). Suppose that \(y^i \not \in A_*\). Let \(x \in L\). Note that since \(y^i \not \in A_*\), \(r^M<d^i\left( {x}\right)\). Furthermore, note that by Corollary 1, \(d^i\left( {x}\right) \le \overline{d}\). Then, since \(r^M=\frac{1-\delta }{1+\delta }\left( {\bar{u}-\overline{d}}\right)\) by the definition of \(r^M\), \(\left( {1 - \delta }\right) \bar{u} =\left( {1 +\delta }\right) r^M+\left( {1 - \delta }\right) \overline{d}<\left( {1 +\delta }\right) d^i\left( {x}\right) +\left( {1 - \delta }\right) \overline{d} \le 2 \overline{d}\). Note that since \(\max _{i \in I} d^i\left( {x^i}\right) =0\), \(x^{\alpha } = y^{\alpha }\) and \(x^{\beta } = y^{\beta }\). Furthermore, note that by Proposition 1 and the definition, \(d^{\beta }\left( {y^{\alpha }}\right) = 2 \overline{d}\) and \(d^{\beta }\left( {y^{\beta }}\right) =0\). Then, since \(u^{\beta }\left( {x^{\alpha }}\right) \ge \delta u^{\beta }\left( {x^{\beta }}\right)\) by Lemmas 4 and 5, \(\left( {1 - \delta }\right) \bar{u} \ge d^{\beta }\left( {x^{\alpha }}\right) -\delta d^{\beta }\left( {x^{\beta }}\right) =d^{\beta }\left( {y^{\alpha }}\right) -\delta d^{\beta }\left( {y^{\beta }}\right) =2 \overline{d}\), which is a contradiction. Thus, \(y^i \in A_*\). Hence, since \(d^i\left( {y^i}\right) =0\) by the definition, \(y^i \in \arg \min _{x \in A_*}d^i\left( {x}\right)\). Therefore, since \(x^i = y^i\) by \(\max _{i \in I} d^i\left( {x^i}\right) =0\), \(x^i \in \arg \min _{x \in A_*}d^i\left( {x}\right)\).
Lemma 7
Suppose that \(\max _{i \in I} d^i\left( {x^i}\right) >0\). Then, for each \(i \in I\), \(x^i \in \arg \min _{x \in A_*}d^i\left( {x}\right)\).
Proof
Let \(\tilde{\alpha } \in I\) such that
Suppose that \(x^{\tilde{\alpha }} \in A_O\). Then, since \(x^{\tilde{\alpha }} \not =y^{\tilde{\alpha }}\) and \(u^i\left( {x}\right)\) is continuous in X for each \(i \in I {\setminus } \left\{ {\tilde{\alpha }}\right\}\), there exists \(x \in X\) such that \(u^{\tilde{\alpha }}\left( {x}\right) >u^{\tilde{\alpha }}\left( {x^{\tilde{\alpha }}}\right)\) and \(x \in A_O^i\) for each \(i \in I {\setminus } \left\{ {\tilde{\alpha }}\right\}\). Note that since \(x \in A_O\) and \(A_O \subset A\) by Lemma 4, \(x \in A\). Then, there exists \(x \in A\) such that \(u^{\tilde{\alpha }}\left( {x}\right) >u^{\tilde{\alpha }} \left( {x^{\tilde{\alpha }}}\right)\). This contradicts Lemma 5. Thus, \(x^{\tilde{\alpha }} \not \in A_O\). Note that by Lemmas 4 and 5, \(x^{\tilde{\alpha }} \in A_C\). Furthermore, note that by Lemma 5, \(x^{\tilde{\alpha }} \in A_O^{\tilde{\alpha }}\). Then, for some \(i',i'' \in I{\setminus } \left\{ {\tilde{\alpha }}\right\}\) such that \(i' \not =i''\), either (i) \(x^{\tilde{\alpha }} \in A_O^{\tilde{\alpha }}\), \(x^{\tilde{\alpha }} \in A_C^{i'} {\setminus } A_O^{i'}\), and \(x^{\tilde{\alpha }} \in A_O^{i''}\), or (ii) \(x^{\tilde{\alpha }} \in A_O^{\tilde{\alpha }}\), \(x^{\tilde{\alpha }} \in A_C^{i'} {\setminus } A_O^{i'}\), and \(x^{\tilde{\alpha }} \in A_C^{i''} {\setminus } A_O^{i''}\). Consider the case where (i) \(x^{\tilde{\alpha }} \in A_O^{\tilde{\alpha }}\), \(x^{\tilde{\alpha }} \in A_C^{i'} {\setminus } A_O^{i'}\), and \(x^{\tilde{\alpha }} \in A_O^{i''}\). Suppose that \(x^{\tilde{\alpha }} \not \in R\left( {y^{\tilde{\alpha }},y^{i'}}\right)\). Then, since \(x^{\tilde{\alpha }} \not \in R\left( {y^{\tilde{\alpha }},y^{i'}}\right)\) and \(u^{i''}\left( {x}\right)\) is continuous in X, there exists \(x \in X\) such that \(u^{\tilde{\alpha }}\left( {x}\right) >u^{\tilde{\alpha }} \left( {x^{\tilde{\alpha }}}\right)\), \(u^{i'}\left( {x}\right) >u^{i'} \left( {x^{\tilde{\alpha }}}\right)\), and \(x \in A_O^{i''}\). Note that since \(x \in A_O\) and \(A_O \subset A\) by Lemma 4, \(x \in A\). Then, there exists \(x \in A\) such that \(u^{\tilde{\alpha }}\left( {x}\right) >u^{\tilde{\alpha }}\left( {x^{\tilde{\alpha }}}\right)\). This contradicts Lemma 5. Thus, \(x^{\tilde{\alpha }} \in R\left( {y^{\tilde{\alpha }},y^{i'}}\right)\). Consider the case where (ii) \(x^{\tilde{\alpha }} \in A_O^{\tilde{\alpha }}\), \(x^{\tilde{\alpha }} \in A_C^{i'} {\setminus } A_O^{i'}\), and \(x^{\tilde{\alpha }} \in A_C^{i''} {\setminus } A_O^{i''}\). Suppose that \(x^{\tilde{\alpha }} \not \in R\left( {y^{\tilde{\alpha }},y^{i'}}\right)\), \(x^{\tilde{\alpha }} \not \in R\left( {y^{\tilde{\alpha }},y^{i''}}\right)\), and \(x^{\tilde{\alpha }} \not \in R\left( {y^{i'},y^{i''}}\right)\). Then, since \(x^{\tilde{\alpha }} \not \in R\left( {y^{\tilde{\alpha }},y^{i'}}\right) \cup R\left( {y^{\tilde{\alpha }},y^{i''}}\right) \cup R\left( {y^{i'},y^{i''}}\right)\), there exists \(x \in X\) such that \(u^{\tilde{\alpha }}\left( {x}\right) > u^{\tilde{\alpha }}\left( {x^{\tilde{\alpha }}}\right)\), \(u^{i'}\left( {x}\right) >u^{i'}\left( {x^{\tilde{\alpha }}}\right)\), and \(u^{i''}\left( {x}\right) >u^{i''}\left( {x^{\tilde{\alpha }}}\right)\). Note that since \(x \in A_O\) and \(A_O \subset A\) by Lemma 4, \(x \in A\). Then, there exists \(x \in A\) such that \(u^{\tilde{\alpha }}\left( {x}\right) >u^{\tilde{\alpha }}\left( {x^{\tilde{\alpha }}}\right)\). This contradicts Lemma 5. Suppose that \(x^{\tilde{\alpha }} \not \in R\left( {y^{\tilde{\alpha }},y^{i'}}\right)\), \(x^{\tilde{\alpha }} \not \in R\left( {y^{\tilde{\alpha }},y^{i''}}\right)\), and \(x^{\tilde{\alpha }} \in R\left( {y^{i'},y^{i''}}\right)\). Then,
Thus, \(x^{i''} \not \in A_C\). Hence, since \(A \subset A_C\) by Lemma 4, \(x^{i''} \not \in A\). This contradicts Lemma 5. Thus, either \(x^{\tilde{\alpha }} \in R\left( {y^{\tilde{\alpha }},y^{i'}}\right)\) or \(x^{\tilde{\alpha }} \in R\left( {y^{\tilde{\alpha }},y^{i''}}\right)\). Hence, for some \(i \in I{\setminus } \left\{ {\tilde{\alpha }}\right\}\), \(x^{\tilde{\alpha }} \in R\left( {y^{\tilde{\alpha }},y^i}\right)\) and \(x^{\tilde{\alpha }} \in A_C^i {\setminus } A_O^i\). Note that \(x^{\tilde{\alpha }} \in R\left( {y^{\tilde{\alpha }},y^i}\right)\) if and only if \(d\left( {y^{\tilde{\alpha }},y^i}\right) =d^{\tilde{\alpha }} \left( {x^{\tilde{\alpha }}}\right) +d^i\left( {x^{\tilde{\alpha }}}\right)\). Furthermore, note that since \(\pi ^i=u^i\left( {x^i}\right)\) by Lemma 5, \(x^{\tilde{\alpha }} \in A_C^i {\setminus } A_O^i\) if and only if \(u^i\left( {x^{\tilde{\alpha }}}\right) =\delta u^i\left( {x^i}\right)\). Therefore, for some \(i \in I{\setminus } \left\{ {\tilde{\alpha }}\right\}\), \(d\left( {y^{\tilde{\alpha }},y^i}\right) =d^{\tilde{\alpha }} \left( {x^{\tilde{\alpha }}}\right) +d^i\left( {x^{\tilde{\alpha }}}\right)\) and \(u^{i}\left( {x^{\tilde{\alpha }}}\right) =\delta u^i\left( {x^i}\right)\).
Let \(\tilde{\beta } \in I{\setminus }\left\{ {\tilde{\alpha }}\right\}\) such that
Then,
Note that by the triangle inequality, \(d\left( {y^{\tilde{\alpha }},y^{\tilde{\beta }}}\right) \le d^{\tilde{\alpha }}\left( {x^{\tilde{\beta }}}\right) +d^{\tilde{\beta }}\left( {x^{\tilde{\beta }}}\right)\). Thus,
Furthermore
Note that by Lemmas 4 and 5, \(u^{\tilde{\alpha }}\left( {x^{\tilde{\beta }}}\right) \ge \delta u^{\tilde{\alpha }}\left( {x^{\tilde{\alpha }}}\right)\). Thus,
By Statements (2), (3), (4), and (5),
Thus, by Statements (1) and (6),
Hence, for each \(i,i' \in I\),
Therefore
Note that for each \(i \in I\),
where equality holds if \(i \in \left\{ {\tilde{\alpha },\tilde{\beta }}\right\}\). Suppose that \(x \not \in C\left( {M;r^M}\right)\), that is, \(x \in X {\setminus } C\left( {M;r^M}\right)\). Then, for some \(i \in I\), \(d^i\left( {x}\right) >\overline{d}+r^M\) by Lemma 2, and thus, \(u^i\left( {x}\right) =\bar{u}-d^i\left( {x}\right) <\bar{u}-\overline{d}-r^M \le \delta \pi ^i\), which implies that \(x \not \in A_C\). Thus,
Suppose that \(x \in O\left( {M;r^M}\right)\). Then, for each \(i \in \left\{ {\tilde{\alpha },\tilde{\beta }}\right\}\), \(d^i\left( {x}\right) <\overline{d}+r^M\) by Lemma 2, and thus, \(u^i\left( {x}\right) =\bar{u}-d^i\left( {x}\right) >\bar{u}-\overline{d}-r^M=\delta \pi ^i\), which implies that \(x \in A_O^i\). Thus, for each \(i \in \left\{ {\tilde{\alpha },\tilde{\beta }}\right\}\),
Let \(\tilde{\gamma } \in I {\setminus } \left\{ {\tilde{\alpha },\tilde{\beta }}\right\}\). Note that by definition and Statement (8), \(\left\{ {\tilde{\alpha }, \tilde{\beta }, \tilde{\gamma }}\right\} = I\) and \(d\left( {y^{\tilde{\alpha }},y^{\tilde{\beta }}}\right) =\max _{i,i' \in I}d\left( {y^i,y^{i'}}\right)\). Suppose that \(d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right) <\underline{d} -r^M\). Then, for each \(x \in M\), since \(\underline{d} \le d^{\tilde{\gamma }}\left( {x}\right)\) by Proposition 1 and \(d^{\tilde{\gamma }}\left( {x}\right) \le d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right) +d\left( {x^{\tilde{\gamma }},x}\right)\) by the triangle inequality, \(r^M < \underline{d}-d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right) \le d^{\tilde{\gamma }}\left( {x}\right) -d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right) \le d\left( {x^{\tilde{\gamma }},x}\right)\), and thus, \(x^{\tilde{\gamma }} \not \in C\left( {M;r^M}\right)\). However, since \(x^{\tilde{\gamma }} \in A\) by Lemma 5, \(A \subset A_C\) by Lemma 4, and \(A_C \subset C\left( {M;r^M}\right)\) by Statement (9), \(x^{\tilde{\gamma }} \in C\left( {M;r^M}\right)\), which is a contradiction. Thus, \(d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right) \ge \underline{d} -r^M\). Hence, since \(d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right) \ge 0\), \(d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right) \ge \max \left\{ {\underline{d}-r^M,0}\right\}\). Suppose that \(d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right) > \max \left\{ {\underline{d}-r^M,0}\right\}\). Note that by Lemma 3, \(\min _{x \in C\left( {M;r^M}\right) }d^{\tilde{\gamma }}\left( {x}\right) =\max \left\{ {\underline{d}-r^M,0}\right\}\). Then, \(\min _{x \in C\left( {M;r^M}\right) }d^{\tilde{\gamma }}\left( {x}\right) <d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right)\). Thus, since \(d^{\tilde{\gamma }}\left( {x}\right)\) is continuous in X, there exists \(x \in O\left( {M;r^M}\right)\) such that \(d^{\tilde{\gamma }}\left( {x}\right) <d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right)\). Note that for each \(i \in \left\{ {\tilde{\alpha },\tilde{\beta }}\right\}\), since \(O\left( {M;r^M}\right) \subset A_O^i\) by Statement (10), \(x \in A_O^i\). Furthermore, note that since \(u^{\tilde{\gamma }}\left( {x}\right)>u^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right) >\delta u^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right) =\delta \pi ^{\tilde{\gamma }}\) by Lemma 5, \(x \in A_O^{\tilde{\gamma }}\). Then, \(x \in A_O\). Thus, since \(A_O \subset A\) by Lemma 4, \(x \in A\). Hence, there exists \(x \in A\) such that \(d^{\tilde{\gamma }}\left( {x}\right) <d^{\tilde{\gamma }}\left( {x^{\tilde{\gamma }}}\right)\). This contradicts Lemma 5. Therefore
Note that
Suppose that \(x \not \in X {\setminus } \left( {C\left( {M;r^M}\right) {\setminus } C\left( {L;r^L}\right) }\right)\), that is, \(x \in C\left( {M;r^M}\right) {\setminus } C\left( {L;r^L}\right)\). Then, \(d^{\tilde{\gamma }}\left( {x}\right) >\underline{d}+r^L\) by Lemma 2, and thus, \(u^{\tilde{\gamma }}\left( {x}\right) =\bar{u}-d^{\tilde{\gamma }}\left( {x}\right) <\bar{u}-\underline{d}-r^L=\delta \pi ^{\tilde{\gamma }}\), which implies that \(x \not \in A_C\). Thus,
Suppose that \(x \in O\left( {L;r^L}\right)\). Then, \(d^{\tilde{\gamma }}\left( {x}\right) <\underline{d}+r^L\) by Lemma 2, and thus, \(u^{\tilde{\gamma }} \left( {x}\right) =\bar{u}-d^{\tilde{\gamma }}\left( {x}\right) >\bar{u} -\underline{d}-r^L=\delta \pi ^{\tilde{\gamma }}\), which implies that \(x \in A_O^{\tilde{\gamma }}\). Thus,
Let \(i \in I\). Suppose that \(x^i \not \in \arg \min _{x \in A_*} d^i\left( {x}\right)\). Note that since \(A_*=C\left( {M;r^M}\right) \cap C\left( {L;r^L}\right)\) by the definition of \(A_*\), \(x^i \not \in \arg \min _{x \in C\left( {M;r^M}\right) \cap C\left( {L;r^L}\right) } d^i\left( {x}\right)\). Furthermore, note that since \(x^i \in A\) by Lemma 5, \(A \subset A_C\) by Lemma 4, and \(A_C \subset C\left( {M;r^M}\right) \cap C\left( {L;r^L}\right)\) by Statements (9) and (12), \(x^i \in C\left( {M;r^M}\right) \cap C\left( {L;r^L}\right)\). Then, \(\min _{x \in C\left( {M;r^M}\right) \cap C\left( {L;r^L}\right) } d^i\left( {x}\right) <d^i\left( {x^i}\right)\). Thus, since \(d^i\left( {x}\right)\) is continuous in X, there exists \(x \in O\left( {M;r^M}\right) \cap O\left( {L;r^L}\right)\) such that \(d^i\left( {x}\right) <d^i\left( {x^i}\right)\). Note that since \(O\left( {M;r^M}\right) \cap O\left( {L;r^L}\right) \subset A_O\) by Statements (10) and (13), and \(A_O \subset A\) by Lemma 4, \(x \in A\). Then, there exists \(x \in A\) such that \(d^i\left( {x}\right) <d^i\left( {x^i}\right)\). This contradicts Lemma 5. Thus, \(x^i \in \arg \min _{x \in A_*} d^i\left( {x}\right)\). \(\square\)
The statement of Theorem 2 follows from Lemmas 5, 6, and 7. \(\square\)
Rights and permissions
About this article
Cite this article
Yamaguchi, K. Spatial bargaining in rectilinear facility location problem. Theory Decis 93, 69–104 (2022). https://doi.org/10.1007/s11238-021-09838-9
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11238-021-09838-9