Experimental duopolies under price guarantees

Abstract

In a symmetric differentiated experimental duopoly we test the ability of Price Guarantees (PGs) to raise prices above the competitive levels. Different types of PGs (‘aggressive’ and ‘soft’ price-beating and price-matching) are implemented either as an exogenously imposed market rule or as a business strategy. Our results show that PGs may lead close to the collusive outcome, depending on whether the interaction between duopolists is repeated and provided that the guarantee is not of the ‘aggressive’ price-beating type.

Keywords:

• price guarantees
• experimental duopolies
• product differentiation

JEL Classification:

• C91
• L11

Acknowledgements

This research was funded by the Spanish Ministry of Education (ECO 2008-04636/ECON), Bancaixa (P1 1B2007/14; P1 1A2007-06) and the Junta de Andalucia (P07-SEJ-03155). Part of this research was carried out when the second author was visiting the University of Cyprus and the LEM (University Paris II) whose hospitality is gratefully acknowledged. The authors thank Dan Kovenock, Jordi Brandts, Tibor Neugebauer, Antonio Fatás, Bruno Broseta, Henrik Orzen, Silviano Esteve and seminar attendants at Naples, Barcelona, Valencia, Reus, Strasbourg, Torino, Helsinki, Pamplona and Malaga for helpful comments.

Notes

¹ Deck and Wilson (2003) have studied a general family of algorithms based on the comparison of prices across firms, without explicitly linking their paper to price matching and price beating. More recently, a growing literature deals with the related topic of pricing under specific strategic commitments, some of which aim at signalling as a complement or substitute of advertising. See, for example, Huang et al. (2010). Although this suggests that a more general approach to this type of commitments is necessary, here we restrict our attention to two specific competitor-based pricing commitments, leaving a more general approach for future research.

² As a result, such guarantees are not treated as illegal per se by either European or US legislation.

³ Using this evidence Arbatskaya et al. (2006) find that the majority of paired observations involving firms that have PMG (PBG) are (not) consistent with what one would expect if firms were using them to discourage price cutting. This suggests that PMG and PBG may be serving different purposes.

⁴ See Srivastava (2001), Srivastava and Lurie (2004) and Lurie and Srivastava (2005) for parallel analyses.

⁵ See Gwin and VanHoose (2011) exploring the relationship among consumer search costs, the relative degree of price stickiness and firm mark-ups.

⁶ The puzzling result that most subjects kept on offering the nonprofitable undercutting guarantee and abandoning the profitable price matching one seems to be context dependant: subjects simultaneously choose the extent of the different guarantees.

⁷ Dugar (2005) confirms the predicted collusive effects of PMGs in the absence of hassle costs.

⁸ See Hviid and Shaffer (1999), proposition 1, for a proof of equilibrium boundaries.

⁹ In symmetric markets all firms must adopt PMG for prices to rise above the Bertrand price.

¹⁰ Both firms cannot increase profits by setting a different price (or adopting PMG). See Proposition 1.B in Doyle (1988) and Fatás and Mañez (2007) for a complete theoretical analysis.

¹¹ In their Experiment C, full coordination is reached more than half of the times when subjects repeatedly interact with the same opponent while it is never observed when interacting with a different one each round.

¹² It exactly includes 4710 different choices; while van Huyck et al. (1990) has only seven choices and Brandts and Cooper (2006) and Brandts et al. (2006) had five.

¹³ See Fatás et al. (2005) for a review of recent experimental results on coordination games.

¹⁴ From the 900 decisions (50 pricing decisions times 18 subjects) of each experimental session with a ‘S’ protocol, we obtained a single independent observation. Thus, under this protocol we ran twice as many sessions as under a ‘P’ protocol in which 18 independent observations were obtained per session.

¹⁵ The translated instructions are included in Appendix I.

¹⁶ Experimental Economics Laboratory at the University of Valencia.

¹⁷ Z-Tree was originally developed by Fischbacher (2007) at the Institute for Empirical Research of Economics (University of Zürich).

¹⁸ Similar results are obtained (detailed results are available upon request) by regressing for each experimental session per-period SD on a dummy variable that takes the value of 1 for the last 25 rounds and 0 otherwise. All coefficients estimated are negative and significant (depending on the treatments, they range between −21.9 and −94.2) indicating a lower dispersion of prices over time. As García-Gallego and Georgantzís (2001) point out, this is partly a consequence of subjects’ lack of information about the true demand model, forcing them to randomly choose their initial price strategies which, following some try-and-error learning, get closer to the theoretical values as the session proceeds.

¹⁹ An alternative descriptive analysis is provided in Appendix II, with estimations of density functions for every treatment.

²⁰ The observed patterns of adoption of PBGs contrasts with excessively high and certainly counterintuitive percentages of adoption (over 90%) obtained in Deck and Wilson (2003). Our control for the effects of endogenous guarantees indicates that the negative effects of PBG's on profits are perceived by the subjects and this is reflected on a low frequency of PBG adoption.

²¹ Table A1 in Appendix II shows how to obtain estimated average prices from the estimation of Equation 5. We obtain estimated price differentials using these estimated average prices.

²² In this case there is no difference between the estimates of the fixed and random effects models. The reason is that given the way we have constructed all the independent variables they do not show any between-groups variation (in this case each experimental subject is a group). Note that fixed effects model estimates are obtained from the within-group estimator whereas random effects model estimates are a weighted average of the within and between-group estimators. Thus, if there is no between-group variation, the estimates of both models are identical.

²³ At a conventional 5% significance level.

²⁴  are positive and significant. However, is negative and significant.

²⁵ This result is due to the fact that BSL-P average profit (116 439) is in an intermediate situation with respect to the lower BSL-S average profit (112 308) and the higher BNE profit (117 600).