Economie et Sociologie Rurales

LORIA
Laboratoire d'Organisation Industrielle Agro-alimentaire

On the Existence and uniqueness of Price Equilibrium with Multi-Store Firms

Eric Giraud-Hraud Hakim Hammoudi Shyama V. Ramani

Septembre 2005

Cahier du LORIA n 2005-03

On the Existence and uniqueness of Price Equilibrium with Multi-Store Firms

Eric Giraud-Hraud1 Hakim Hammoudi2 Shyama V. Ramani3

Septembre 2005

Cahier n 2005-03
Rsum:

Nous donnons une preuve d'existence et d'unicit d'un quilibre en prix dans un contexte o il existe un nombre quelconque de firmes, dtenant chacune un nombre quelconque de magasins. Nous montrons comment les prix sont spatialement diffrencis en fonction des localisations de ces magasins.

Abstract:
This paper gives a proof of the existence and uniqueness of price equilibrium in a multi-store, multi-firm competition framework. It illustrates how the price levels depend on the store localization within a firm's portfolio and between different firm store portfolios.

Mots cls : Key Words :

Equilibre en prix, diffrenciation horizontale, firmes multi-produits Price equilibrium, Horizontal differentiation; Multi-store firms.

Classification JEL: D21, L11, L13

1 2

INRA-LORIA. E.mail. giraud@ivry.inra.fr et Laboratoire dEconomtrie de lEcole Polytechnique. ERMES, Universit Panthon-Assas et INRA-LORIA. E.mail hammoudi@ivry.inra.fr 3 INRA-LORIA. E.mail. ramani@ivry.inra.fr et Laboratoire dEconomtrie de lEcole Polytechnique. La version lectronique du cahier est disponible ladresse : http://www.ivry.inra.fr/loria/cahierspdf/2005.03pdf

1.

Introduction

The impact of store location of a multi-store firm, or that of its competitors, on the price setting strategies of firms, is a subject that has been little explored in the industrial organization literature1. The standard models of horizontal differentiation (e.g. Hotelling, 1929, Salop, 1979) constitute an appropriate framework to treat this question. However, not many models have retained the original structure to study pricing strategies in the context of multi-store competition. Levy and Reitz (1992) is an exception. They analyse the pricing strategies of firms in a multiproduct framework, in which one firm controlling two stores is in competition with two monostore competitors, using the circular model of Salop (1979). More recently, Giraud-Hraud et al. (2003) have studied price settings in the context of a single multi-product firm interacting with several mono-product competitors. The originality of the present paper lies in its generalization of these two contributions. This is the first formal proof of the existence and characterization of a unique price equilibrium, in the context of many multi-product firms, with each firm offering a number of brands, without any constraint on the localization of stores or the pricing strategy of firms2. The paper is organized as follows. Section 2 presents our model. Section 3 contains our main existence and uniqueness results, and equilibrium properties. Section 4 concludes.

The terms store and brand are used equivalently in this paper. A number of articles have explored this question modifying certain assumptions of these models, either with respect to the preferences of consumers (Klemperer, 1992 , Janssen et al., 2005) or with respect to strategic conjectures on the nature of competition (e.g. Cournot competition in Debashis and al., 2002). In addition, Klemperer (1992) and Janssen et al. (2005) identify the equilibrium assuming that firms have to charge the same price for all of their brands.
2

2.

Bertrand-Nash equilibrium with multi-store firms

Consider a standard circular model of spatial differentiation generalized for multi-store (or multiproduct) firms. There are N differentiated products
x1,,xN,

offered at prices p1,,pN in a circle

of circumference d. There are Q firms, Fq (q = 1,,Q), and each firm Fq owns nq stores distributed along the circle. The total number of stores is N, where N =nq . Each store sells one
q =1 Q

differentiated product such that xi represents the location of store i selling product xi[0,d] with

xi x j if i j ; and zero (marginal) production cost. For any market price vector,
p = (p1,,pN), let Di(p) and Bi(p) = pi Di(p) denote the demand and the profit of a store i

respectively. There is a continuum of consumer types distributed with unit density over [0,d]. The transportation cost incurred by consumer of type , when he consumes product xi rather than is given by C(| xi|) and r is the reservation price common to all consumers. Consumers are allowed to buy one unit of a differentiated good, so that a consumer of type purchasing a product xi at price pi obtains a utility of:

U(r,xi,pi )=r pi C(| xi|)

(1) (2)

with C(| xi|)=a ( xi )2 , a>0 .

The consumers product choice problem can then be written as :

( xi ) i=1,., N

Min { pi + a ( xi )2 }
distance between two stores
i

(3) and
j

Let

Zi,j

be

the

shortest

(for

j>i

Zi, j =Z j,i =inf{(x j xi ),(d x j + xi )} ). For two adjacent stores i and i-1, let i = 1 , for i = 1,,N Zi 1,i

with Z0,1 = Z1,N. Then by determining the location of the consumers who are indifferent between buying from store i-1 or i, and between store i and i+1, the demand faced by store i can be expressed as follows:

Di(p)= 1 [-(i + i +1 )pi+(i pi 1 + i +1 pi+1 )+a( 1 + 1 )] 2ad i i +1

i =1,,N .

(4)

Since the demand Di(p) depends only on pi, pi-1 and pi+1 among all the components of price vector p, it is useful to distinguish all the adjacent stores owned by a same firm Fq in order to investigate the Bertrand-Nash equilibrium between the Q firms. In order to do this we introduce some definitions. A k-partition of stores: This is a partition of the N stores into K ranges N =nk .
k =1 K

Range k of firm i, k : k is the kth set of neighboring brands belonging to the same firm i such that only two of these brands (the peripheral brand of the range) are directly exposed to brands owned by other firms. The central store(s) of the range k , m( k ): m(k ) = ( j 1) +

nk for even values of nk (in 2

this case there are two central stores) and m(k ) = ( j 1) + case there is only one central store).
Degree of exposure to competition of store ik , d(i):
i j d (i ) = n + j i k

nk + 1 for uneven values of nk (in this 2

if i m(k ) (5) if i m(k )

Peripheral stores of the range k : The stores i , i=1,,N such that d(i) = 0.

Finally,

let

(p)= Bi(p)
k

denote

the

profit

of

range

ik

and

F (p)= Bi(p)=
q

i Fq

k Fq

(p) denote the profit of a firm Fq.

From the above, it is clear that the Bertrand-Nash equilibrium between the Q firms can be analyzed through the Bertrand-Nash equilibrium between the K ranges. Each firm is indifferent about the number of ranges possessed by the other firms. With the assumption that all of products have positive market shares, the best response for each range k ={j,, j+nk 1} only depends on the two prices of its neighboring stores, namely pj-1 and p j +nk . Moreover, as will be shown, the equilibrium prices of each store will depend only on its exposure to competition. Recall that a Bertrand-Nash equilibrium is defined by a N component vector of prices
* p* =(p1 ,,p* ) such that: N

Fq p{p=(p1,,pN ) / pi = pi* , iFq } Fq(p* ) Fq(p)

(6)

Then we have the following proposition.

Proposition When ( x1 ,..., xn ) are fixed, there exists a unique price equilibrium with a non-zero
market share for each store. Proof : Given that store i's profit only depends on its own pricing and on the prices of its neighbors, the best response for the ranges can be isolated. For k = {j,j+1,,j + nk -1} we can show that k is concave in (p j ,,p j +nk 1 ) , since the Hessian matrix of k is a constant equal to

2( k ) :

0 ( j + j+1 ) j +1 j +1 2 1 ( k )= i (i +i+1 ) i +1 ad j +nk 1 0 j +nk 1( j+nk 1+ j+nk )

(7)

Let m (m = 1,,nk) denote the principal minor of 2( k ) of order m. We have 1=( j + j +1 ) and m =(m+ j 1+m+ j )m12 + j 1 m2 , for m = 2,,nk where 0 =1 . Therefore, we can easily m demonstrate by induction that m = j + m m1 +(1)m j +i 1 for m = 1,,nk.
i =1 m

For odd values of m, we have m < 0 (since i > 0 and 0 > 0 ) and for even values of m, we

have m > 0.Thus, the matrix 2( k ) is negative definite. Then we can write the first order conditions setting

= 0 for i = j , , j + nk 1 pi
k

and obtain the best response to the prices (p j 1,p j +nk ) for each range k as below:

j
2

p j1+ ( j + j +1 ) p j j +1 p j +1= a ( 1 + 1 ) 2 j j +1
i = j +1,, j +nk 2

i pi 1 + (i +i +1 ) pi i +1 pi +1= a ( 1 + 1 ) 2 i i +1 j +nk 1 p j +nk 2 +( j +nk 1+ j+nk ) p j +nk 1

(8)

j +n
2

p j +nk = a ( 1 + 1 ) 2 j +nk 1 j +nk

Then from (8), all the first-order conditions for the set of ranges, k ; k = 1,,K can be summarized in the standard form Ap = B where A and B are matrices and p is a vector. We have
B 0. since i > 0 for all i. Applying McKenzie's theorem (1959 theorem 4, p. 50) a necessary

and sufficient condition for Ap =B to have a unique solution p 0 is that A = (ai,h)i=1,.,N;h=1,.,N

5

has a quasi-dominant diagonal or the following properties are verified :

(P1) There exist dh > 0 such that dh |ah,h| di |ai,h| (h=1,,N)

ih

(P2) When ai,h = 0, given hH and iH for some set of indices H, the strict inequality holds

for some hH .

We can verify the property (P1) taking dh = 1 for all h, h = 1,,n and furthermore confirm that there is no set of indices H such that ai,h = 0 given hH and iH . This completes the proof.

3.

Properties

We describe now the main properties of the Bertrand-Nash equilibrium. Without loss of generality, we suppose the products are symmetrically differentiated inside each range :

k ={j, j +1,, j +nk 1} Zk ik {j +nk 1} Zi,i +1 Zk

(9)

There is the same product differentiation between the members of each range, but the differentiation between the ranges (that is between the peripheral stores of two ranges possessed by two different firms) is not necessarily the same. Let Z k indicate the differentiation between the range k = {j,j+1,,j + nk -1} and the range k +1

(Zk =Z j +nk 1, j +nk ) . Therefore, the solution of system (8) takes the form

pi =(i j +1)2 + (i j+1)+ with constant coefficients. Using (8) and after identification of the
constant coefficients, we have the following formulation for the equilibrium prices:

p* = i +

a Zk2 (i j+1)2 2

p*j +nk p*j 1 Zk { +a[(nk +1)Zk + Zk Zk 1 ] } (i j +1) 2 (nk 1)Zk +Zk 1+Zk (Zk 1Zk )p*j +nk +(Zk +nk Zk )p*j 1 a(Zk +nk Zk )(Zk Zk 1 )} (nk 1)Zk +Zk 1+ Zk if nk >1

+ 1{ 2

(10)
p*j =

1 (Z p* + Zk 1 p*j +1 ) + a Zk 1 Zk 2 2 (Zk 1+ Zk ) k j 1

if nk =1

Furthermore, using (4) and (8), it can be shown that the equilibrium market shares
Di* =Di(p* ) verify :

k , nk >2 ik , d(i)0
K

Di* =

Zk 2d

0Di* = 1 + 21 [ (Zk +Zk )] 2 d k =1 d(i)=

(11)

In figures 1 and 2 we focus on a duopoly configuration with only one multi-product firm F1 =1 ={1,,n} and a mono-product F2 =2 ={N} with N = n +1, in the market. We consider the case where any two stores of a firm are equidistant : Z1 = Z. The distance between store 1 and N is Z 2 and the distance between n and N is Z1 and we have (n1)Z + Z1 + Z 2 =d . Figures 1 and 2 show the kind of price and profit differentiation that could emerge in two polar cases: first, the symmetrical situation when Z1 = Z 2 (figures 1a and 1b) and second, the asymmetrical situation, when Z1 is significantly greater than Z 2 (figures 2a and 2b).

1,4 1,2 1,0 0,8 0,6 0,4 0,2 0,0 1 2 3 4 5 6 7 8 9 10

Figure 1a: Prices'differentiation setting a = d = 2 and Z1 = Z2 = d / 10 , Z = d / 10 , n = 9 (N = 10)

0,3

0,3

0,2

0,2

0,1

0,1

0,0 1 2 3 4 5 6 7 8 9 10

Figure 1b:Profits'differentiation setting a = d = 2 and Z1 = Z2 = d / 10 , Z = d / 10 , n = 9 (N = 10)

1,2

1,0

0,8

0,6

0,4

0,2

0,0 1 2 3 4 5 6 7 8 9 10

Figure 2a: Prices' differentiation setting a = d = 2 and Z1

= d / 2,

Z 2 = 0, Z

= d / 16 ,

=9

(N = 10)

0,1 0,1 0,1 0,1 0,1 0,1 0,0 0,0 0,0 0,0 0,0 1 2 3 4 5 6 7 8 9 10

Figure 2b: Profits'differentiation setting a = d = 2 and Z1

= d / 2,

Z 2 = 0, Z

= d / 16 ,

=9

(N = 10)

The variation of prices along the range depends on the degree of differentiation between the competitors and on the number of stores within the range. A multi-store firm chooses its brands price vector as a function of its degree of exposure to outside competition.

The closer a brand is to outside competition of another firm, the lower its price and vice versa. In the symmetrical situation, the prices inside the range are increasing from the periphery to the center of the range. Peripheral stores offer the lowest prices while central stores protected from the external competition offer the highest prices. We show in figure 1b that for high values of Z 2 and n, the highest profits in the range 1 are those of the peripheral stores because of the greater market share that they capture. The price at store N is always below the prices at any store of the multi-store firm, but its profit is the highest in the industry, if its product is sufficiently differentiated. For the asymmetrical situation, we suppose there is no differentiation between store 1 and store N, taking Z 2 = 0 , while store n is the most differentiated store from the rival N, taking
Z1 = d . Figure 2a shows the variation of prices, which are increasing from store 1 to store n. 2

Store n not only charges the highest price, but also enjoys the largest market share, thereby being the most profitable in the industry.

4.

Concluding Remarks

In this paper, we extended the standard duopoly model of spatial differentiation to a multi-store, multi-firm competition. We resolved the complex problem of identifying and confirming the existence and uniqueness of a price equilibrium for symmetric and asymmetric locations of brand portfolios of multi-product firms. The results obtained with multi-store firms are different from those obtained with single-store firms. Prices vary over stores as a function of the degree of exposure of the firm portfolios to outside competition. Such results are in the line with some observations and empirical works (see for example Barron et al (2000) and Netz and Taylor (2002) for the gasoline retail sector). In many final product markets the multi-store 10

phenomenon is gaining ground. Therefore, the proof of existence of an equilibrium in prices and its characterization contributes to a better understanding of the determinants of pricing strategies of firms in such markets, which in turn could be useful for the formulation of competition policy in the context of multi-store firm competition.

11

References

Barron J.M., B.A. Taylor, and J.R. Umbeck, 2000, A Theory of Quality-Related differences in Retail Margins: Why There is a Premium on a Premium Gasoline, Economic Inquiry, 38: 4.

Debashis P. and J. Sarkar, 2002, Spatial competition among multi-store firms, International Journal of Industrial Organization 20, 163-190

Giraud-Hraud E., H. Hammoudi and M. Mokrane, 2003, Multiproduct Firm Behavior in a Differentiated Market, Canadian Journal of Economics, 36, 41-61.

Hotelling, H. 1929, Stability in competition, Economic Journal 39, 4157. Janssen, M C.W, Karamychev V.A., and P.V. Reeven, 2005, Multi-store competition: Market segmentation or interlacing?, Regional Science and urban Economics, 15, 700-714.

Klemperer, P. 1992, Equilibrium Product Lines: Competing Head-to-Head may be less competitive, American Economic Review, 82, 741-755.

Levy and Reitzes, 1992, Mergers in Markets with Localized Competition, Journal of Law, Economics and Organization, 8, 427-440.

12

Mc Kenzie, L.W., 1960, Matrices with dominant diagonals and economic theory, in: Arrow, K.J, Karlin, S. and P. Suppes eds, Mathematical Methods in the Social Sciences, Stanford University Press, Stanford California 47-62.

Netz J.S., and B.A. Taylor, 2002, Maximum or Minimum Differntiation? Location Patterns of Retail Outlets, The review of Economics and Statistics, 84, 162-175.

Salop, S. C, 1979., Monopolistic competition with outside goods, Bell Journal of Economics, 10, 141-156.

13

CAHIERS DU LORIA
Anne 2002
2002-01 C. Chambolle et E.Giraud-Hraud E. Giraud-Hraud, L.G. Soler et H. Tanguy Certification de la qualit par une AOC : un modle danalyse. Concurrence internationale dans le secteur viticole : Quel avenir au modle dAppellation dOrigine Contrle ? Le fonctionnement des marchs entre vignoble et ngoce dans le secteur des vins AOC : trois tudes rgionales. Equilibre de Cournot-Nash avec contraintes de capacit : une formule explicite. Influence de la structure financire sur les choix stratgiques : tude de cas dans lindustrie du vin. Harmonisation des taxes la consommation : le cas des boissons alcoolises. Les effets des politiques de promotion sur la fiabilit des outils dvaluation des prix de dtail. Principales tendencias del mercado mundial de vinos. Multiproduct Firm Behavior in a Differentiated Market Les relations entre la grande distribution et ses fournisseurs : bilan et limites de trente ans de rgulation Stratgies de revente perte et rglementation Faut-il interdire la revente perte ? Standards de qualit minimum et marques de distributeurs : un modle d'analyse

2002-02

2002-03

L.G. Soler et H. Tanguy

2002-04

C. Caron et J. Laye

2002-05

O. Saulpic et H. Tanguy

2002-06

C. Arnaud, E. Giraud-Hraud et A.Hammoudi M. Mainsant et F. Porin

2002-07

2002-08 2002-09

R. Green et L. Pierbattisti E. Giraud-Hraud, H. Hammoudi et M. Mokrane M.L. Allain et C. Chambolle

2002-10

2002-11 2002-12 2002-13

C. Chambolle C. Chambolle E. Giraud-Hraud, L. Rouached, L.G. Soler

Anne 2003
2003-01 2003-02 R. Green, M. Hy R. Green, M. Rodriguez Zuniga et Leandro Pierbattisti E. Giraud-Hraud, J. Mathurin, L.G. Soler Scurit alimentaire et traabilit Global Market changes and business behavior in the wine sector Quelle lgitimit des mcanismes de rgulation de loffre dans les Appellations dOrigine Protge ?

2003-03

2003-04

C. Chambolle, E. GiraudHraud S. Poret

Certification of Origin as a Non-Tariff Barrier

2003-05

A Note on the Location of the Anti-Drug Law Enforcement Policy Do Vertical Extension and Horizontal Mergers Interact Lindustrie de transformation de la viande bovine en France : Une approche historique (1950-2003) Bilan conomique du stockage en rserve qualitative de vin de Champagne Concentrations Horizontales et Puissance dAchat

2003-06

C. Chambolle, E. GiraudHraud P. Sans, G. de Fontguyon

2003-07

2003-08

J.A. Laye

2003-09

C. Chambolle, L. Muniesa M.A Ravon

Anne 2004
2004-01 2004-02 M.L. Allain, C. Chambolle M.L. Allain, C. Chambolle Below cost pricing laws as vertical restraints Forbidding Resale at a Loss : a Strategic Inflationary Mechanismv Les politiques publiques des drogues : vers une analyse conomique Lchec de la guerre la drogue : une revue de la littrature thorique Multinationals and the training of workers in the host country : the tariff jumping argument revisited Forces et faiblesses de la filire AOC Maine Anjou Promotions and producer-retailer relationships with perishable products

2004-03

S. Poret

2004-04

S. Poret

2004-05

H. Hammoudi, W. Hatit

2004-06 2004-07

J.M. Noury, G. de Fontguyon R. Lahmandi-Ayed, L. Rouached, L. G. Soler

Anne 2005
2005-01 H. Hammoudi, W. Hatit Menace de dbauchage et incitation des multinationales la formation de la main doeuvre dans les pays htes : un modle thorique Consommation de vin dans le monde : Tour dhorizon des tudes et recherches et Mta-analyse de la demande

2005-02

P. Bazoche, E. Giraud-Hraud, Y. Surry

2005-03

E. Giraud-Heraud, H.Hammoudi, S.V. Ramani

On the existence and uniqueness of Price Equilibrium with multi-store firms

INRA LORIA 65, Bd de Brandebourg 94205 IVRY Cedex (33) 1 49 59 69 87 lampenoi@ivry.inra.fr http://www.ivry.inra.fr/loria/