scieee AI-readable full text Open interactive document viewer

Product Differentiation and the Intensity of Price Competition

Baake, Pio,Oechssler, Jörg

Abstract

EconStor is a publication server for scholarly economic literature, provided as a non-commercial public service by the ZBW.

Full text

Baake, Pio; Oechssler, Jörg Article Product Differentiation and the Intensity of Price Competition Zeitschrift für Wirtschafts- und Sozialwissenschaften (ZWS) - Vierteljahresschrift der Gesellschaft für Wirtschafts- und Sozialwissenschaften, Verein für Socialpolitik Provided in Cooperation with: Duncker & Humblot, Berlin Suggested Citation: Baake, Pio; Oechssler, Jörg (1997) : Product Differentiation and the Intensity of Price Competition, Zeitschrift für Wirtschafts- und Sozialwissenschaften (ZWS) - Vierteljahresschrift der Gesellschaft für Wirtschafts- und Sozialwissenschaften, Verein für Socialpolitik, ISSN 0342-1783, Duncker & Humblot, Berlin, Vol. 117, Iss. 2, pp. 247-256, https://doi.org/10.3790/schm.117.2.247 This Version is available at: https://hdl.handle.net/10419/291886 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. https://creativecommons.org/licenses/by/4.0/ Zeitschrift für Wirtschafts- u. Sozialwissenschaften (ZWS) 117 (1997), S. 247 - 256 Duncker & Humblot, Berlin Product Differentiation and the Intensity of Price Competition* By Pio Baake and Jorg Oechssler* * 1. Introduction The Hotelling model of spatial competition is a convenient tool for modelling product differentiation. It can literally describe the choice of locations by firms when customers are distributed along a one-dimensional geographic entity, like a road or a river. But more frequently, it is used to model the choice of product characteristics by firms under the simplifying hypothesis that tastes of consumers differ only in one dimension.1 In Hotelling's original article (1929) and in much of the subsequent work consumers are supposed to be distributed uniformly along the line. However, as Neven (1986) points out, increasing densities toward the center of the distribution have more appeal in many situations. For example, most cities exhibit areas of greater agglomeration at the center. Likewise, if the distribution describes consumers' tastes for certain product characteristics, it seems reasonable to suppose that more extreme tastes are less frequent than 'average' tastes. In this note we show that with such reasonable consumer distributions one encounters problems with the existence of equilibrium, which are different from the ones described earlier in the literature. D'Aspremont et al. (1979) were first to point out that there was a flaw in Hotelling's original argument. Namely, the existence of a pure strategy equilibrium in the price subgame cannot be assured for all location choices of the two firms if consumers are distributed uniformly and transportation costs are linear. They also demonstrate that the problem disappears if transportation costs are quadratic. More generally, Caplin and Nalebuff (1991) show that a pure strategy equilibrium in the price game exists and is unique for a broad * Verantwortlicher Herausgeber / editor in charge: U. S. * * We thank an anonymous referee for very helpful comments. 1 Generalizations to higher dimensional spaces are possible to some extent (see e.g. Tabuchi, 1994). ZWS 117 (1997) 2 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.117.2.247 | Generated on 2023-04-04 12:23:17 248 Pio Baake and Jörg Oechssler range of consumer utility functions (including the case of quadratic transportation costs) if the consumer density function f(x) is log-concave. We assume a log-concave density function and quadratic transportation costs to make sure that there will be no problems with the existence of equilibrium in the price subgame. Nevertheless, a pure strategy equilibrium in the two-stage game may still fail to exist. This observation was already made by Tabuchi and Thisse (1995) and Goeree and Ramer (1994), in particular for the case of a triangular density of consumers. Tabuchi and Thisse (1995) conjecture that the non-existence also holds for densities which are differentiate but very steep at the midpoint. This note is an extension of their work and we prove that their conjecture was correct. We concentrate on completely symmetric games, that is, games with symmetric consumer densities and symmetric cost structures for both firms. We explain why a symmetric equilibrium need not exist by decomposing the consequences of a small change in location by one firm into three effects. We identify a price effect, a quantity effect and a competition intensity effect. What turns out to be essential for the necessary condition for an equilibrium to hold is that the competition intensity effect must not be too large. The latter effect measures how strongly the intensity of price competition decreases when a firm moves away from its competitor. Usually, when one considers a symmetric game, the conjecture is that if an equilibrium exists at all, then a symmetric equilibrium can be found, too. However, in Section III we give an example of a Hotelling game with a symmetric density in which no symmetric equilibrium exists, despite the fact that there are two asymmetric equilibria. In the asymmetric equilibria one firm locates outside the support of the customer distribution, which provides just another example in which Hotelling's 'principle of minimum differentiation' fails to hold. 2. The Model We consider the standard Hotelling location-price game with quadratic transportation costs and unrestricted locations. In a first step, both firms simultaneously choose their locations on the real line, which we denote by a and b, respectively, (with a < b). In a second step, given these locations and the known distribution of consumers F(x), firms simultaneously choose their prices. Consumers buy one unit of the good from the supplier with the lower price including transportation cost. Thus, the total price a consumer located at x e R has to pay is pi + (x - a)2 when buying from firm 1 (located at a) or p2 + (x - b)2 when buying from firm 2 (located at b). ZWS 117 (1997)2 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.117.2.247 | Generated on 2023-04-04 12:23:17 Product Differentiation and Intensity of Price Competition 249 For given locations and prices the position £ of the consumer who is indifferent between buying from firm 1 or firm 2 is given by (1) «a,b,Pi,P2)=i(^f + b + a). Obviously, all consumers to the left of £ will buy from firm 1 and all consumers to the right will buy from firm 2. Assuming, for simplicity, that production costs are zero, profits are then given by IIi(a, b,pi,p2) = PiF(£) and n2(a,b,p1,p2)=p2(l-F(0). Since we want to consider symmetric game situations, we assume that the consumer density /Or) is symmetric and uni-modal. In order to guarantee the existence of a unique solution in the price-subgame we will additionally assume that/Or) is log-concave. Assumption 1: The consumer density f(x) = F'(x), with convex support B C R, is log-concave, twice differentiate, and symmetric with a mode at x = 0. Applying the Brekopa-Borell Theorem, Caplin and Nalebuff (1991) have shown that log-concavity of the consumer density is sufficient to guarantee existence and uniqueness of an equilibrium in the price game. Lemma 1: Given Assumption 1 there exists a unique Nash equilibrium of the price game given by the solutions p* (a, b) and p* (a, b) to •n(ah\ 9 (h = ~~ /(£(a> &»Pi»P2)) and l-F(£(a,5,Pi,P2)) p2{a,b) = 2(b-a)- f(((a,b,pi,p2)) Proof: The formulas for pi(a, b) and p2(a, b) follow directly from the first order conditions of profit maximization. The existence of a Nash equilibrium follows from Theorem 2 in Caplin and Nalebuff (1991) and the uniqueness from Proposition 7 in Caplin and Nalebuff (1991). Using the result of Lemma 1 together with (1) we can implicitly derive the location of the indifferent consumer at the equilibrium prices «-»-¿'••»•HW1- Note that in a symmetric equilibrium (in pure strategies) b = a and therefore c = o. ZWS 117 (1997)2 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.117.2.247 | Generated on 2023-04-04 12:23:17 250 Pio Baake and Jorg Oechssler Finally, the reduced form profit functions are given by (3) U1(aib) = 2{b-a)^- and (4) II2(a,b) = 2(b-a)(l The first order condition for profit maximization is ei2 €« = 0, where £a denotes the partial derivative of C(a>b) with respect to a. Straightforward differentiation of (2) yields £a = 6/2-2(2F-l)/' ' which implies, using symmetry, that £a\-a=b = 1/6 because F(0) = 1/2. The same holds for Hence, -a* = b* = 3 4/(0) is the only candidate for a symmetric equilibrium in pure strategies. What matters for the existence of a symmetric equilibrium are the consequences a change in location by one firm has on prices and quantities. We can decompose those effects by defining the following elasticities. Let dp*(a,b) a r)p*va =—— ^ ^ denote the elasticity of the equilibrium price with 1 ' respect to the location of firm 1, a dF(C) = v — I p*=p*=const, denote the elasticity of quantity sold with respect to F OCL 1 2 location for given prices, d2p* a (vA = v denote the elasticity of the reaction in price to location as location is varied. The first elasticity indicates the direct price effect of a change in location, the second the direct quantity effect and the last one the indirect competition intensity effect. ZWS 117 (1997)2 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.117.2.247 | Generated on 2023-04-04 12:23:17 Product Differentiation and Intensity of Price Competition 251 Proposition: A symmetric equilibrium in pure strategies exists only if Proof: We prove the Proposition by demonstrating that the symmetric candidate solution can be a maximum only when it satisfies the above condition. Simple calculations show that for all densities satisfying Assumption 1 Vp\,a = 1/4 and TJF,a = -3/4 at the symmetric candidate solution. The necessary condition for the existence of a symmetric equilibrium depends, therefore, entirely on the size of the competition intensity effect. It is satisfied if and only if To check whether the candidate solution represents in fact a (local) maximum, we calculate the second order conditions. Using the fact that /'(0) = 0 and F(0) = 1/2, further differentiation of (2) yields £aa\-a=b = 0- When we substitute the above calculated values for a*, b*, F(0), f (0) and Ul-a=b> we find that 1 (2F - j/') + 2(b - a) (2F - Fp)u + «2 (2/ - 2Ff , f2 - Ff F2 P f2 ^|a*<0^24/(0)3>-/"(0). Consider now 77 . Since and one finds that at the symmetric solution ZWS 117 (1997)2 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.117.2.247 | Generated on 2023-04-04 12:23:17 252 Pio Baake and Jörg Oechssler Generally speaking there is always a trade-off in the choice of location. One the one hand, a firm has an incentive to locate as close as possible to the center since the bulk of customers is located there. On the other hand, as firms get closer to each other, competition in the price subgame becomes fiercer. The steeper the consumer density is at the center (which is the location of the indifferent consumer in the symmetric candidate solution), the higher will be the intensity of competition for customers located there. Our condition states that for a symmetric equilibrium to exist the increase in the price that can be charged is not too steep as a firm moves away from the center. In other words, the intensity of competition must not decrease too much when moving away from the center. The equivalent condition 24/(0)3 > -/"(0) demands that the consumer density must not be too steep at the center in the sense that /"(0) must not be too large in absolute value relative to/(0). If the condition is violated, we get a profit minimum instead of a maximum. This implies that when firm 2 locates at the symmetric candidate solution b*, firm 1 has an incentive to deviate from a* in both directions. A deviation to the center is profitable because so many new customers can be gained. And a deviation away from the center is profitable because the competition intensity is reduced so drastically that the loss of customers is more than offset by the possibility to charge a higher price. Remark 1: In order to check which symmetric densities f(x) satisfy the condition stated in the Proposition, it is easier to examine the equivalent condition 24/(0)3 > -/"(0). As pointed out above, the distribution must not be too steep at the center. In particular, it must be differentiate at the center (see Tabuchi and Thisse, 1995; and Goeree and Ramer, 1994, for this result) as non-differentiability can be seen as the limit case of a differentiate but very steep density. For example, with a triangular distribution a symmetric equilibrium does not exist. On the other hand, it may easily be checked that many common distributions, like the Normal distribution, do satisfy the condition. Remark 2: Note that the condition given in the Proposition is only necessary but not sufficient for a symmetric equilibrium. As the condition is derived from the second order conditions for profit maximization, it guarantees only a local maximum. To guarantee that a candidate solution represents in fact a global maximum, one has to impose stronger conditions (see Goeree and Ramer (1994) for such a (rather complicated) sufficient condition). ZWS 117 (1997)2 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.117.2.247 | Generated on 2023-04-04 12:23:17 Product Differentiation and Intensity of Price Competition 253 3. An Example As mentioned above a simple, but surprising, example for the non-exis- tence of a symmetric equilibrium in a completely symmetric Hotelling location game is the case of a triangular consumer density, e.g. f(x) = 1 - \x\.2 Since the location of the indifferent consumer is of decisive importance in the analysis and since the density is non-differentiable at exactly this point in the case of symmetric locations, we thought it at first plausible to suspect the non-differentiability to cause the non-existence.3 However, as the analysis of the last section showed there is a non-trivial necessary condition for the existence of a symmetric equilibrium. Intuitively, the condition requires that the density be not too steep at the peak. We will now give an example of a density that violates this condition. The following density satisfies all aspects of Assumption 1: it is log-con- cave, symmetric and twice differentiate everywhere. It was constructed by smoothing the triangular density at the peak by inserting a polynomial for some range at the center of the density. where c > 1/2 is a parameter and n(c) = 2oc?-i a normalization factor such that f^f(cc, c)dx = 1, Vc > 1/2. It is straightforward to check that the condition of 24/(0)3 > - /"(0) is violated for all c > 7.42, which implies that the symmetric candidate solution is a local profit minimum instead of a maximum. Equivalently, it can be checked that the competition intensity effect represented by 77 becomes larger than 1/2 for all c > 7.42. Figure 1 displays the profit function of firm 1 for the case of c = 20 if firm 2 is located at the symmetric candidate solution bs = 0.7570. It is apparent that ITi (a, bs) fails to be quasi-concave. Due to the lack of quasi-concavity of the profit functions the reaction functions4 (i.e., the optimal location choice of firm i for a given location of firm j assuming that prices are chosen optimally) are discontinuous (see Figure 2) and therefore no symmetric equilibrium exists.5 Furthermore, it can 2 See Goeree and Ramer (1994) or Tabuchi and Thisse (1995) for a proof. 3 However, Tabuchi and Thisse (1995) conjectured, but did not prove, that the nonexistence also occurs with differentiate densities. 4 Strictly speaking, they are reaction correspondences since at the jump both endpoints are best replies. 5 The reaction functions are not piece-wise linear although it might appear so in the graph. ZWS 117 (1997)2 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.117.2.247 | Generated on 2023-04-04 12:23:17 254 Pio Baake and Jörg Oechssler be seen that despite the non-existence of a symmetric equilibrium two asymmetric equilibria exist (where the reaction functions intersect). For the case of c = 20 the asymmetric equilibria are located at aj = -1.5443, b\ = 0.3607 and a\ = 0.3607, b*2 -1.5443, respectively. Note that one firm locates outside of the support of the customer distribution. n,(a,b*) Figure 1: Profit function of firm 1 with 6s = 0.7570 and c = 20 Figure 2: Reaction functions for c = 20 ZWS 117 (1997)2 OPEN ACCESS | Licensed under CC BY 4.0 | https://creativecommons.org/about/cclicenses/ DOI https://doi.org/10.3790/schm.117.2.247 | Generated on 2023-04-04 12:23:17