Abstract
We study the implications for monopoly pricing strategies and product diversity of consumers’ temptation when the differentiation of the product is horizontal. Consumers have an ex-ante ideal product (“commitment preferences”), but they and the monopolist are aware that consumers may fall prey to “temptation preferences” ex-post with some probability. Our results indicate that when consumers are aware of their dynamic change in preferences, the firm cannot take advantage of consumers’ temptation but instead, in order to attract them into the store, the firm must compensate ex-ante consumers for the possibility of yielding to temptation once inside the store. As a result, this paper shows that the firm narrows the variety of products, not offering those products close to temptation preferences. Moreover, it is shown that product prices and firm’s profits decrease with the probability of temptation and with the consumers’ awareness of their dynamic inconsistency.
Similar content being viewed by others
Notes
The Strotz (1955) representation is a particular case of these preferences where \(\pi =1\).
Some papers have studied other temptation representations. Gul and Pessendorfer (2001), for instance, consider that consumer’s utility from a choice set equals the realized commitment utility minus the linear “self-control costs,” i.e., the realized temptation utility minus the maximum value of the temptation utility over the choice set. Fudenberg and Levine (2006) allow for non-linear self-control costs, as they argue that we can consider self-control as a limited resource such that the “cognitive load” leads to agents falling into temptation more easily. Dekel et al. (2008) representation covers situations in which agents face uncertainty about the “strength of temptation.” Moreover, under the Chatterjee and Krishna (2009) representation, the agent has two different selves, and temptation is thought of as the choice of a “virtual alternate self” or alter ego, who appears with a positive probability which depends on the choice set. If the temptation probability is not menu-dependent, then the latter representation coincides with the one that we use in our model.
Our interpretation of the Hotelling line is different from the standard “linear city” model where consumers are uniformly distributed along the line and several firms or stores, which sell the same physical good, are located at different points in the line. Different locations in the line can also be interpreted as many firms selling imperfect substitutes (horizontal differentiation). In this paper, we use this last interpretation but instead of considering several firms offering only one product, we consider one firm (for instance the only retail store in a small town) offering several horizontally differentiated products.
Note that \(U((\phi ,0); \theta )=V((\phi , 0))=0\) is a standard normalization. If consumers do not derive any intrinsic satisfaction from entering the store, for instance by just looking around, then \(U((\phi , 0); \theta )=V((\phi ,0))=0\) implies \(W(NE,\theta )=0\).
Thus, it is assumed that if a consumer \(\theta \) is indifferent between choosing from \(\bar{M}\) or NE, i.e., if \({W}(\bar{M};\theta )=0\), then he ends up choosing from \(\bar{M}\).
Note that if instead of different consumers located in a Hotelling line, we interpret \(\theta \) as possible realizations of the commitment preferences of one consumer; this simple example could be interpreted as the perfect information case.
Note that since \(\pi =1\), the consumer will always take the tempting choice once in the store so, in this case, we can ignore all other products with \(q>1/2\).
In an online appendix, we provide three additional extensions: General temptation ideal product \(0\le \theta ^{v}\le 1\), heterogeneity in temptation preferences, and uncertainty about the consumers’ temptation ideal product.
Benabou and Pycia (2002) show that Gul and Pessendorfer (2001) representation can be expressed in terms of a dual-self temptation representation. In particular, let us consider \({x}^{{v}}\in {\hbox {argmax}}_{{x}\in \bar{M}} {V}\left( {x} \right) \) and \(\hat{x} \in {\hbox {argmax}}_{{x}\in \bar{M} } \left\{ {{U}\left( {x} \right) +{V}\left( {x} \right) } \right\} \). Therefore, it is immediate that
$$\begin{aligned} \mathop {{\max }}\limits _{{x}\in \bar{M} } \left[ {{U}\left( {x} \right) +{V}({x})} \right] -\mathop {{\max }}\limits _{{x}\in \bar{M} } {V}({x})=\left( {1-{\pi }^{{GP}}} \right) {U}\left( {\hat{x} } \right) +{\pi }^{{GP}}{U}\left( {{x}^{{v}}} \right) , \end{aligned}$$where \({\pi }^{{GP}}=\left( {\frac{{V}\left( {{x}^{{v}}} \right) -{V}\left( {\hat{x} } \right) }{{U}\left( {\hat{x} } \right) -{U}\left( {{x}^{{v}}} \right) }} \right) \). Thus, while in the random Strotz representation temptation probability is constant, under the Benabou-Pycia interpretation of the Gul-Pessendorfer representation, this probability depends on the relative intensity of consumer’s preferences
Dekel and Lipman (2012) showed that every self-control representation has a random Strotz representation with a continuous of states. However, with discrete states (as we assume in our model), the converse is not true. Lemma 7 shows that the monopoly pricing is different under self-control and random Strotz preferences with only two states.
References
Benabou, R., & Pycia, M. (2002). Dynamic inconsistency and self-control: A planner-doer interpretation. Economic Letters, 77, 419–424.
Chatterjee, K., & Krishna, R. V. (2009). A dual-self representation for stochastic temptation. American Economic Journal: Microeconomics, 1, 148–167.
Dekel, E., & Lipman, B. L., (2007). Self-control and random Strotz representation. New York: Mimeo.
Dekel, E., & Lipman, B. L. (2012). Costly self-control and random self-indulgence. Econometrica, 80, 1271–1302.
Dekel, E., Lipman, B. L., & Rustichini, A. (2008). Temptation-driven preference. Review of Economic Studies, 76, 937–971.
DellaVigna, S., & Malmendier, U. (2004). Contract design and self-control: theory and evidence. Quarterly Journal of Economics, 119, 353–402.
Eliaz, K., & Spiegler, R. (2006). Contracting with diversely naive agents. Review of Economic Studies, 73, 689–714.
Esteban, S., Miyagawa, E., & Shum, M. (2006). Nonlinear pricing with self-control preferences. Journal of Economic Theory, 135, 306–338.
Esteban, S., & Miyagawa, E., (2006a). Optimal menu of menus with self-control preferences. New York: Mimeo.
Esteban, S., & Miyagawa, E. (2006b). Temptation, self-control, and competitive nonlinear pricing. Economic Letters, 90, 348–355.
Fudenberg, D., & Levine, D. K. (2006). A dual-self model of impulsive control. American Economic Review, 96, 1449–1476.
Gul, F., & Pessendorfer, W. (2001). Temptation and self-control. Econometrica, 69, 1403–1436.
Kreps, D. M. (1979). A representation theorem for preference for flexibility. Econometrica, 47, 565–577.
O’Donoghue, T., & Rabin, M. (1999). Incentives for procastinators. Quarterly Journal of Economics, 114, 769–816.
Sarafidis, Y., (2005). Inter-temporal price discrimination with time-inconsistent consumers. New York: Mimeo.
Strotz, R. H. (1955). Myopia and inconsistency in dynamic utility maximization. Review of Economic Studies, 23, 165–180.
Acknowledgments
This paper is a modified version of Chapter 2 of my doctoral dissertation submitted at Universidad Carlos III de Madrid in July, 2011. I am exceedingly grateful to my advisors \(\hbox {M}^{\mathrm{a}}\) Ángeles de Frutos and Susanna Esteban for their valuable advice and suggestions. All remaining errors are attributable to me..
Author information
Authors and Affiliations
Corresponding author
Electronic supplementary material
Below is the link to the electronic supplementary material.
Appendix
Appendix
Proof of Lemma 2
We show that a gap with no products in \(\left( {\underline{\theta }, \bar{\theta }} \right) \subseteq \left[ {{\theta }_{k},1} \right] \) can be improved upon by designing a product with \(\hat{q} \in \left( {\underline{\theta },\bar{\theta }} \right) \). Let \(\underline{p}\) be the price of the product \(\underline{\theta } \) and \(\bar{p}\) be the price of product \(\bar{\theta }\). Assume with no loss of generality that consumer \(\hat{\theta } \in \left( {\underline{\theta } ,\bar{\theta }} \right) \) buys product \(\bar{\theta }\). Then, product \(\hat{x} =\left( {\hat{\theta },\hat{p} } \right) \) where \(\hat{p} ={\max }\left\{ {\underline{p} +{t}\left( {\hat{\theta } ,\underline{\theta } } \right) ;\bar{p}+t\left( \hat{\theta },\bar{\theta }\right) } \right\} \) is feasible and yields more profits. Thus, repeating this argument we get that, in equilibrium, \(q(\theta )=\theta \) for all \({\theta }\in (\underline{\theta },\bar{\theta })\).
Since \(q(\theta )=\theta \) for all \(\theta \ge \theta _{k}\) by U-IC constraints, we have that \(p(\theta )=p^{*}\) for all \(\theta \ge \theta _{k}\).
Finally, by V-IC constraint, \(p^{*}- p^{v }\ge t(q^{v}, \theta ^{v}) - t(\theta _{k}, \theta ^{v}) \ge 0\), where the last inequality follows from the fact that \(q^{v }\le \theta _{k}\). \(\square \)
Proof of Lemma 3
We show that, if \(q^{v }< \theta _{k}\), the monopolist finds it optimal to decrease \(\theta _{k}\). By definition of \(\theta _{k}, U(x(\theta _{k}); \theta _{k})=U(x^{v}; \theta _{k})\). Thus, using Lemma 2, \(p^{v}=p^{*}- t(q^{v}, \theta _{k})\). Therefore, the monopolist’s profits are
where \(\left( {\frac{{\mathrm{d}\Pi }}{{\mathrm{d}\theta }_{k} }} \right) <0\) for all \(q^{v}<\theta _{k}\). Thus, in equilibrium, \(q^{v}=\theta _{k}\), which implies \(p^{v}=p ^{*}\) by incentive compatibility. \(\square \)
Proof of Corollary 1
Note that social welfare is given by
Using the optimal menu, \(t(q^{v},0)=\pi t(q^{v},1)\) and \(q^{v}=\theta _{k}\), we get
Let T be the antiderivative of t. Then, applying the fundamental theorem of calculus, we can write,
where
where the last equality holds because \(t(q^{v},0)=\pi t(q^{v},1)\) at the optimal menu. \(\square \)
Proof of Lemma 4
It is immediate, from the analysis in Sect. 4, that in equilibrium \(V(x^{v})={W}(\bar{M} ;{\theta }^{{NFC}})=0\), so \(t(q^{v}(\pi , \theta ^{NFC}),0)={\pi }t(q^{v}(\pi , \theta ^{NFC}), \theta ^{NFC})\). Hence, the tempting choice when the market is non-fully covered depends on the temptation probability \(\pi \) and the number of consumers served \(\theta ^{NFC}\). Where, \(\frac{{\mathrm{d}q}^{{v}}({\pi },{\theta }^{{NFC}})}{{\mathrm{d}\pi }}>0\) and \(\frac{{\mathrm{d}q}^{{v}}({\pi },{\theta }^{{NFC}})}{{\mathrm{d}\theta }^{{NFC}}}>0\).
To compute the optimal menu, we just need to calculate \({\theta }^{{NFC}}\) solving the following maximization problem:
st.,
Thus, the implicit equation for \({\theta }^{{NFC}}\) is given by
\(\square \)
Proof of Lemma 5
Suppose by contradiction that the monopolist designs commitment products in the interval \([\underline{\theta },1]\), where \(\underline{\theta } \in [0,1)\). Then, for any commitment product, the price is \(\underline{p} ={s}-{t}\left( {\underline{\theta },0} \right) \), which is unique by the incentive compatibility constraints. Therefore, the ex-ante surplus of consumers with \(\theta \ge \theta _{k}\), is given by
Since \(\left( \frac{\mathrm{d}W(\bar{W};\theta )}{\mathrm{d}\theta }\right) <0\), for all \(\theta >q^{v}\), the “worst consumer type” from the point of view of the monopolist is \(\theta =1\). Note that by setting \(\underline{\theta } =1\), the monopolist maximizes \({W}(\bar{M},1)\) and thus allows for the lowest \(q^{v}\) which yields the highest price for the tempting product and hence the highest profits.
Finally, note that since \(\underline{\theta } =1, p = s - t(1,0)\). Moreover, since \(V(x^{v})=0, p^{v}= s - t(q^{v},0)\). Therefore, from the definition of \(\theta _{k}, U(x(\theta _{k}); \theta _{k})=U(x^{v}; \theta _{k})\), we get that \(\theta _{k}=0\). \(\square \)
Proof of Lemma 6
Since the monopolist wants to charge the highest possible p\(^{v}\), we need to check that \(p^{v}=s\) (i.e., \(q^{v}=\theta ^{v}=0\)) is feasible. In our case, it is sufficient to check that \({W}\left( {\bar{M} ;{\theta }} \right) \ge 0\) for all \(\theta \in [0,1]\), where
Thus, applying Lemma 5 and letting \(p^{v}=s\), we get
Note that
Due to the properties of the transportation cost function, this condition implies that for all \(\pi \in [0,1], \exists \theta ^{*}\in [0,1]\) such that for all \(\theta <\theta ^{*} (\theta >\theta ^{*}), \frac{\mathrm{d}W(\bar{M}; \theta )}{\mathrm{d}\theta }>0 (<0)\). Therefore, since \({W}\left( {\bar{M}}; 0\right) =0\), a necessary and sufficient condition for \(W(M;\theta )\ge 0\) for all \(\theta \in [0,1]\) is that \(W(M;1)\ge 0\). Since
then \({W}\left( {\bar{M}},1 \right) \ge 0\) iff \({\pi }\le \frac{1}{2}\). This implies that, when \({\pi }>\frac{1}{2}\), the monopolist has to locate the tempting choice beyond \(\theta ^{v}\), i.e., \(q^{v}>0\), which implies charging a \(p^{v}<s\) to attract consumer \(\theta =1\) into the store. Since \({W}\left( {\bar{M}}; 1\right) =0\), then
Thus,
\(\square \)
Proof of Lemma 7
To prove this lemma, we use some of the results and intuition of our basic model. Let us start with the candidate of optimal menu with self-control preferences (See Fig. 7). Thus, consumers in the interval \((2q^{v},1]\) purchase a product which is a compromise between commitment and temptation ideal points \(({q}\left( {\theta } \right) =\frac{{\theta }}{2})\), while all consumers in the interval \([0,2q^{v}]\) purchase the tempting choice \(x^{v}=(q^{v},p^{v})\). The proof consists of several steps.
Step 1 Note that IC constraints of consumers \(\theta \in (q^{v},1]\) imply that \(p(\theta )=p\).
Step 2 We need to check that in equilibrium consumers with \(\theta \in (2q^{v},1]\) do not want to deviate and choose the offer \(x^{v}\). By the IC constraint of consumer \(\theta =2q^{v}\), this implies that \(p^{v}=p\).
Step 3 In equilibrium \({W}^{{SC}}\left( {\bar{M}}; 1 \right) =0\), because \(\theta =1\) is the consumer with the highest self control cost. Since \({W}^{{SC}}\left( {\bar{M}}; 1 \right) =s-p-f({q}^{{v}},1)\), we get that in equilibrium \(p=s-f(q^{v},1)\).
Step 4 Let us define \(\hat{\theta } \in \left[ {{q}^{{v}},\frac{1}{2}} \right] \). We need to check that consumer \(\hat{\theta }\) do not want to deviate and consume \({q}_{d} =\hat{\theta }\). Note that \({W}^{{SC}}\left( {\bar{M}}; \hat{\theta } \right) =s-p-t(q^{v},\hat{\theta })\) while under the deviation \({W}_{d}^{{SC}} \left( {{M}^{{d}};\hat{\theta }} \right) ={s}-{p}-\left[ {{t}\left( {\hat{\theta },0} \right) -{t}\left( {{q}^{{v}},0} \right) } \right] \). Where \({W}^{{SC}}\left( {\bar{M}};\theta \right) >W^{SC}_{d}(M^{d};\theta )\) iff \({t}\left( {{q}^{{v}},0} \right) +{t}\left( {{q}^{{v}},\hat{\theta } } \right) <t\left( {\hat{\theta },0} \right) \) which is true because of strict superadditivity.
Step 5: Moreover, we need to assure participation of the low tail of the distribution \(({W}^{{SC}}\left( {\bar{M}};0 \right) \ge 0)\). Note that \({W}^{{SC}}\left( {\bar{M}}, 0 \right) =s-p-t(q^{v},0)\), and hence applying Step 3, we get \({W}^{{SC}}\left( {\bar{M}},0 \right) =f\left( q^{v},1\right) -t(q^{v},0)= t\left( \frac{1}{2},1\right) +t\left( \frac{1}{2},0\right) -2t(q^{v},0)= 2\left[ t\left( \frac{1}{2},0\right) -t(q^{v},0)\right] \), where the last equality comes from the symmetry of the transportation cost function. Therefore, \({W}^{{SC}}\left( {\bar{M}},0\right) \ge 0\) for all \({q}^{{v}}\le \frac{1}{2}\).
Step 6: Finally, since \(p=s-f(q^{v},1)\) and \({q}^{{v}}\le \frac{1}{2}\), the firm maximizes profits offering a single product \({x}^{{SC}}=\left( {\frac{1}{2},{s}-{t}\left( {\frac{1}{2},1} \right) } \right) \), which, as we have shown in the previous steps, is a feasible menu. \(\square \)
Rights and permissions
About this article
Cite this article
Gómez-Miñambres, J. Temptation, horizontal differentiation and monopoly pricing. Theory Decis 78, 549–573 (2015). https://doi.org/10.1007/s11238-014-9437-0
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11238-014-9437-0