Abstract
The circular city model and the linear city model are extended to allow for asymmetric directional transportation costs. A two-stage location-then-quantity model is proposed. We show that in the circular city model, maximal dispersion arises in equilibrium, while in the linear city model, the unique equilibrium is represented by both firms agglomerating in a non-central point of the segment.
Similar content being viewed by others
Notes
Here and after, we shall refer to asymmetric transportation costs to describe transportation costs which differ across directions, and not across firms.
We are grateful to an anonymous reviewer for suggesting this example.
This example has been provided by another reviewer.
Other examples of directional constraints regard TV broadcasting times, bus/air schedules, one-way streets, highways, flows of natural resources as water and oil. For other examples, see Lai (2001).
Sun (2010) also allows for the possibility that the two firms have opposite directional constraints, in the sense that one firm can only move in the clockwise direction, while the rival can only move in the anticlockwise direction. In this case, it is shown that the unique equilibrium is characterized by firms agglomerating in one point of the circle. In the present article, the case of opposite non-extreme directional constraints is not considered.
When \({\partial \bar{{\Pi }}_A }/{\partial a}=0\) is solved with respect to \(a\), there are two roots: \(a*(b)\) and: \(\hat{{a}}(b)=\frac{2r^{2}-r-t-brt-bt^{2}-\sqrt{(r-2r^{2}+t+brt+bt^{2})^{2}+r(2r^{2}-rt-3t^{2})[2-r(1+2b-b^{2})+tb^{2}]}}{2r^{2}-rt-3t^{2}}\). However, \(\hat{{a}}(b)\) can be excluded as \(\hat{{a}}(b)\notin [0,b]\).
Similarly, when \({\partial \bar{{\Pi }}_B }/{\partial b}=0\) is solved with respect to \(b, \) there are two roots: \(b*(a)\) and \(\hat{{b}}(a)=\frac{2r^{2}-r-t+art+ar^{2}-\sqrt{[r^{2}(2+a)-t+r(at-1)]^{2}-(3r^{2}+rt-2t^{2})[r^{2}(1+2a)+a^{2}t^{2}+r(a^{2}t-2)]}}{3r^{2}+rt-2t^{2}}\). However, \(\hat{{b}}(a)\) can be excluded as \(\hat{{b}}(a)\notin [a,1]\).
It can be easily verified that the second-order conditions are satisfied in equilibrium.
Note that when \(b=1/2\), only Configuration 2 can arise.
References
Andree K (2011) Product differentiation and spatial agglomeration in a unidirectional Cournot model. Lett Sp Resour Sci 4:167–172
Cancian M, Bills A, Bergstrom T (1995) Hotelling location problems with directional constraints: an application to television news scheduling. J Ind Econ 43:121–124
Colombo S (2009) The unidirectional Hotelling model with spatial price discrimination. Econ Bull 29:3031–3040
Colombo S (2011) Spatial price discrimination in the unidirectional Hotelling model with elastic demand. J Econ 102:157–169
Hamilton JH, Thisse JF, Weskamp A (1989) Spatial discrimination: Bertrand vs. Cournot in a model of location choice. Reg Sci Urban Econ 19:87–102
Kharbach M (2009) A unidirectional Hotelling model. Econ Bull 29:1814–1819
Lai FC (2001) Sequential location in directional markets. Reg Sci Urban Econ 31:535–546
Nilssen T (1997) Sequential location when transportation costs are asymmetric. Econ Lett 54:191–201
Pal D (1998) Does Cournot competition yields spatial agglomeration? Econ Lett 60:49–53
Sun CH (2010) Spatial Cournot competition in a circular city with directional delivery constraint. Ann Reg Sci 45:273–289
Sun CH (2012) Sequential location in a discrete directional market with three or more players. Ann Reg Sci 82:569–580
Acknowledgments
I thank the Editor Borje Johansson and two anonymous referees for their comments.
Author information
Authors and Affiliations
Corresponding author
Appendix
Appendix
1.1 Equilibrium quantities and prices (circular city)
-
Configuration 1
-
\(x\in [0,b]\): Firm \(A\)’s and Firm \(B\)’s profits are: \(\pi _A (x)=(p_x -tx)q_{A,x} \) and \(\pi _B (x)=[p_x -r(b-x)]q_{B,x} \), respectively. Maximization of the profits functions yields the following equilibrium quantities: \(\dot{q}_{A,1} ={[1+rb-x(r+2t)]}/3\) and \(\dot{q}_{B,1} ={[1-2rb+x(2r+t)]}/3\). Therefore, the equilibrium price is: \(\dot{p}_1 ={(1+br-rx+tx)}/3\).
-
\(x\in [b,x_A ]\): Firm \(A\)’s and Firm \(B\)’s profits are: \(\pi _A (x)=(p_x -tx)q_{A,x} \) and \(\pi _B (x)=[p_x -t(x-b)]q_{B,x} \), respectively. Maximization of the profits functions yields the following equilibrium quantities: \(\dot{q}_{A,2} ={[1-t(b+x)]}/3\) and \(\dot{q}_{B,2} ={[1+t(2b-x)]}/3\). The equilibrium price is: \(\dot{p}_2 ={[1-t(b-2x)]}/3\).
-
\(x\in [x_A ,x_B^1 ]\): Firm \(A\)’s and Firm \(B\)’s profits are: \(\pi _A (x)=[p_x -r(1-x)]q_{A,x} \) and \(\pi _B (x)=[p_x -t(x-b)]q_{B,x} \), respectively. Maximization of the profits functions yields the following equilibrium quantities: \(\dot{q}_{A,3} ={[1+t(x-b)-2r(1-x)]}/3\) and \(\dot{q}_{B,3} ={[1+2t(b-x)+r(1-x)]}/3\). The equilibrium price is: \(\dot{p}_3 ={(1+r-bt-rx+tx)}/3\).
-
\(x\in [x_B^1 ,1]\): Firm \(A\)’s and Firm \(B\)’s profits are: \(\pi _A (x)=[p_x -r(1-x)]q_{A,x} \) and \(\pi _B (x)=[p_x -r(1+b-x)]q_{B,x} \), respectively. Maximization of the profits functions yields the following equilibrium quantities: \(\dot{q}_{A,4} ={[1-r(1-b-x)]}/3\) and \(\dot{q}_{B,4} ={[1-r(1+2b-x)]}/3\). Therefore, the equilibrium price is: \(\dot{p}_4 ={[1+r(2+b-2x)]}/3\).
Therefore, the equilibrium profits of Firm \(B\) are:
-
Configuration 2
-
\(x\in [0,x_B^2 ]\): Firm \(A\)’s and Firm \(B\)’s profits are: \(\pi _A (x)=(p_x -tx)q_{A,x} \) and \(\pi _B (x)=[p_x -t(1-b+x)]q_{B,x} \), respectively. Maximization of the profits functions yields the following equilibrium quantities: \(\ddot{q}_{A,1} ={[1+t(1-b-x)]}/3\) and \(\ddot{q}_{B,1} ={[1-t(2-2b+x)]}/3\). Therefore, the equilibrium price is: \(\ddot{p}_1 ={[1+t(1-b+2x)]}/3\).
-
\(x\in [x_B^2 ,b]\): Firm \(A\)’s and Firm \(B\)’s profits are: \(\pi _A (x)=(p_x -tx)q_{A,x} \) and \(\pi _B (x)=[p_x -r(b-x)]q_{B,x} \), respectively. Maximization of the profits functions yields the following equilibrium quantities: \(\ddot{q}_{A,2} ={[1+r(b-x)-2tx]}/3\) and \(\ddot{q}_{B,2} ={[1-2r(b-x)+tx]}/3\). Therefore, the equilibrium price is: \(\ddot{p}_2 ={[1+tx+r(b-x)]}/3\).
-
\(x\in [b,x_A ]\): the same as Configuration 1. Therefore, \(\dot{q}_{A,2} =\ddot{q}_{A,3} \), \(\dot{q}_{B,2} =\ddot{q}_{B,3} \) and \(\dot{p}_2 =\ddot{p}_3 \).
-
\(x\in [x_A ,1]\): Firm \(A\)’s and Firm \(B\)’s profits are: \(\pi _A (x)=[p_x -r(1-x)]q_{A,x} \) and \(\pi _B (x)=[p_x -t(x-b)]q_{B,x} \), respectively. Maximization of the profits functions yields the following equilibrium quantities: \(\ddot{q}_{A,4} ={[1-t(b-x)-2r(1-x)]}/3\) and \(\ddot{q}_{B,4} ={[1+r(1-x)+2t(b-x)]}/3\). Therefore, the price in equilibrium is given by: \(\ddot{p}_4 ={[1+r(1-x)-t(b-x)]}/3\).
Therefore, the equilibrium profits of Firm \(B\) are:
-
Configuration 3
-
\(x\in [0,x_B^2 ]\): the same as Configuration 2. Therefore, \(\ddot{q}_{A,1} =\dddot{q}_{A,1} \), \(\ddot{q}_{B,1} =\dddot{q}_{B,1} \) and \(\ddot{p}_1 =\dddot{p}_1 \).
-
\(x\in [x_B^2 ,x_A ]\): Firm \(A\)’s and Firm \(B\)’s profits are: \(\pi _A (x)=(p_x -tx)q_{A,x} \) and \(\pi _B (x)=[p_x -r(b-x)]q_{B,x} \), respectively. Maximization of the profits functions yields the following equilibrium quantities: \(\dddot{q}_{A,2} ={[1+r(b-x)-2tx]}/3\) and \(\dddot{q}_{B,2} ={[1-2r(b-x)+tx]}/3\). Therefore, the equilibrium price is: \(\dddot{p}_2 ={[1+tx+r(b-x)]}/3\).
-
\(x\in [x_A ,b]\): Firm \(A\)’s and Firm \(B\)’s profits are: \(\pi _A (x)=[p_x -r(1-x)]q_{A,x} \) and \(\pi _B (x)=[p_x -r(b-x)]q_{B,x} \), respectively. Maximization of the profits functions yields the equilibrium quantities: \(\dddot{q}_{A,3} ={[1-r(2-b-x)]}/3\) and \(\dddot{q}_{B,3} ={[1+r(1-2b+x)]}/3\). Therefore, the equilibrium price is: \(\dddot{p}_3 ={[1+r(1+b-2x)]}/3\).
-
\(x\in [b,1]\): Firm \(A\)’s and Firm \(B\)’s profits are: \(\pi _A (x)\!=\![p_x \!-\!r(1-x)]q_{A,x}\) and \(\pi _B (x)=[p_x -t(x-b)]q_{B,x} \), respectively. Maximization of the profits functions yields the following equilibrium quantities: \(\dddot{q\!}_{A,4} =[1-t(b-x)-2r(1- x)]/3\) and \(\dddot{q}_{B,4} ={[1+r(1-x)+2t(b-x)]}/3\). Therefore, the price in equilibrium is given by: \(\dddot{p}_4 ={[1+r(1-x)-t(b-x)]}/3\).
Therefore, the equilibrium profits of Firm \(B\) are:
Proof of Proposition 1
First, we have \(\dot{\Pi }_B (b=\frac{t}{r+t})=\ddot{\Pi }_B (b=\frac{t}{r+t})\) and \(\ddot{\Pi }_B (b=\frac{r}{r+t})={\Pi }_B (b=\frac{r}{r+t})\). As a consequence, the profits function of Firm \(B\) is continuous in \(b\). Next, by differentiating \(\dot{\Pi }_B \) and \(\dddot{\Pi }_B \) with respect to \(b\), we have: \(\frac{\partial \dot{\Pi }_B }{\partial b}=\frac{2b[2rt-b(r+t)^{2}]}{9}\ge 0\), \(\forall b\in [0,\frac{t}{r+t}]\), and \(\frac{\partial {{\dddot{\Pi }}}_B }{\partial b}=-\frac{2(1-b)[2brt-(1-b)(r^{2}+t^{2})]}{9}\le 0\), \(\forall b\in [\frac{r}{r+t},1]\). On the other hand, by differentiating \(\ddot{\Pi }_B \) with respect to \(b\), we get: \(\frac{\partial \ddot{\Pi }_B }{\partial b}=\frac{2t^{2}(1-2b)}{9}\), and, \(\frac{\partial ^{2}\ddot{\Pi }_B }{\partial b^{2}}=-\frac{4t^{2}}{9}<0\). Therefore, \(\dot{\Pi }_B \) increases with \(b\), \({\dddot{\Pi }}_B \) decreases with \(b\), while \(\ddot{\Pi }_B \) is maximized when \(b=\frac{1}{2}\).Footnote 10 It follows that the profits of Firm \(B\) continuously increase when \(b<\frac{1}{2}\), continuously decrease when \(b>\frac{1}{2}\), and are maximized when \(b=\frac{1}{2}\).\(\square \)
Proof of Proposition 2
Consider first the overall transportation costs under Configuration 1. They are:
The overall transportation costs under Configuration 2 are given by:
Finally, the overall transportation costs under Configuration 3 are given by:
Next, note that \(\dot{T}(b=\frac{t}{r+t})=\ddot{T}(b=\frac{t}{r+t})\) and \(\ddot{T}(b=\frac{r}{r+t})={\dddot{T}}(b=\frac{r}{r+t})\). This implies that the transportation costs vary continuously with \(b\). By differentiating \(\dot{T}\) and \({\dddot{T}}\) with respect to \(b\), we get: \(\frac{\partial \dot{T}}{\partial b}=-\frac{b[2rt-b(r+t)^{2}]}{3}\le 0\), \(\forall b\le \frac{t}{r+t}\), and \(\frac{\partial {\dddot{T}}}{\partial b}=\frac{(1-b)[2brt-(1-b)(r^{2}+t^{2})]}{3}\ge 0\), \(\forall b\in [\frac{r}{r+t},1]\). On the other hand, by differentiating \(\ddot{T}\) with respect to \(b\), we have: \(\frac{\partial \ddot{T}}{\partial b}=-\frac{t^{2}(1-2b)}{3}\), and: \(\frac{\partial ^{2}\ddot{T}}{\partial b^{2}}=\frac{2t^{2}}{3}\ge 0\). Therefore, \(\dot{T}\) decreases with \(b\), \({\dddot{T}}\) increases with \(b\), while \(\ddot{T}\) is minimized when \(b=\frac{1}{2}\). As a consequence, the overall transportation costs continuously decrease when \(b<\frac{1}{2}\), continuously increase when \(b>\frac{1}{2}\), and are minimized when \(b=\frac{1}{2}\).\(\square \)