Optimal price setting during a currency changeover: theory and evidence from french restaurants

Abstract

This article studies firms’ price-setting decision during a currency changeover. Buyers’ difficulties with the new nominal price level may create incentives to raise prices temporarily but doing so comes at the risk of damaging a seller’s standing as a fair retailer. We model firms’ trade-off and study conditions under which increasing or decreasing prices is optimal. A difference-in-differences analysis based on micro-data of French restaurants strongly supports the model’s predictions. Prices during the 2002 changeover in the European Monetary Union were less likely to rise in larger restaurants, nontourist restaurants and when prices were advertised.

Keywords:

• price setting
• changeover
• euro
• inflation

JEL Classification:

• E31
• F33
• M39

Acknowledgements

We would like to thank an anonymous referee for helpful comments. We are also grateful to Philippe Andrade, Isa Kirchmaier, Jürgen Eichberger, Zeno Enders, Jose Jorge, Hervé Le Bihan, Michel Simioni, Daniel Rittler, Patrick Sevestre, Lenno Uusküla, as well as participants in the 29th Journées de Microéconomie Appliqué (Brest, 2012), the 61st AFSE Congress (Paris, 2012) and seminars at University of Heidelberg and at Foundation of the Banque de France for insightful comments and suggestions.

The views expressed in this article are those of the authors and do not necessarily represent those of Banque de France.

Notes

¹ Using micro consumer price data, several empirical studies for other countries find that the frequency of both price increases and decreases rose at the euro changeover (see Glatzer and Rumler (2007) and Rumler et al. (2011) for Austria, Hoffmann and Kurz-Kim (2006) for Germany, or Lunnemann et al. (2010) for Luxemburg).

² If consumers perceive a price increase as an attempt of the firm to take advantage of the changeover, it is not necessary that the firm actually had the intention to ‘cheat’. Any rounding up, thus, bears this risk to some extent.

³ In this set-up, the price needs to be restricted from above to avoid firms choosing an infinite price.

⁴ Two spikes in the fast food series (6 months before and 6 months after the changeover) are visible in Fig. 1. A possible explanation is that the reluctance of the average fast food restaurant to raise prices during the months around the changeover drove these restaurants to anticipate and postpone price changes.

⁵ Another possibility would be to assume that consumers use the price of the period immediately before the changeover as reference. Assuming such a behaviour has the interesting consequence that it creates incentives to raise the pre-changeover price in order to be able to reduce the price at the changeover. Since we do not find evidence for this kind of behaviour in the data we do not pursue this possibility here.

⁶ An example of such a pricing strategy in the retail sector are loss-leader products.

⁷ In the New Keynesian literature, the G_i entering the household’s consumption basket is sometimes interpreted as an exogenous, firm-specific taste shock.

⁸ We abuse notation slightly by using the same symbol to denote the set and its cardinality.

⁹ The fixed costs may be interpreted as the costs of setting up branches. A firm that serves more customers needs to open more branches.

¹⁰ This way of introducing firm heterogeneity in the Dixit-Stiglitz framework is simpler than in, say, Melitz (2003) or Lancaster (1966) and serves a different purpose. Unlike in Melitz (2003), a firm’s size in our model is exogenous (chosen by ʽnatureʼ). For the agents, nature’s choice of the indicator I is random and does therefore not affect behaviour in equilibrium. However, from a general equilibrium perspective, nature’s choice is constrained by our assumptions that markets clear. Equation 7 assures that the produced goods equal the demanded goods. Lancaster (1966) model has some formal similarities with our model but the weights (the indicator function I) in his model are not simply 0 or 1 as in our case and, more importantly, they are determined by the consumer rather than ʽnatureʼ.

¹¹ Baudry et al. (2007) provides a detailed description of the data set.

¹² Contrary to countries like Germany or Italy where the exchange rate was close to respectively 2 and 2000, the exchange rate in France makes quite complex for consumers the conversion of prices between euro and francs quite complex for consumers and there was no predominant rule of thumb used. Ehrmann (2011) finds evidence that the complexity of exchange rate is positively related to a rise in food inflation after the euro cash changeover, but does not find effects in other sectors.

¹³ The original database begins in July 1994, but in August 1995 a VAT increase occurred concerning only traditional restaurants, so we restrict our sample to January 1996–February 2003.

¹⁴ As robustness check, we will consider several definitions of the period around the changeover (see below for details).

¹⁵ In Austria, Greece, Finland or Portugal this dual display of prices was mandatory with different practices (see Eife, 2006).

¹⁶ To our knowledge, only Austria implemented a law which explicitly forbade price increases; the ʽEuro Waehrungsangabengesetzʼ allowed the Ministry of Economics to set prices for a period of 6 months in the event of unjustified price increases.

¹⁷ This has implications for the plausibility of the menu-cost explanation. See the discussion in Eife (2011).

¹⁸ In order to avoid overlapping in the 18 month window, we exclude from the sample price changes calculated between March 2000 and September 2001, between March 1998 and September 1999 and between March 1996 and September 1997.

¹⁹ In principle, with a sufficiently long sample such a comparison could shed light on the hypothesis of Adriani et al. (2009) and Eife (2011) of a persistent and structural impact of the changeover.

²⁰ Fast food restaurants can be directly owned by the chain or be franchises. In the latter case, they are free to set their prices independently of the chain, so that price setting decisions are not fully coordinated at the chain level. Indeed, in our data the distribution of price changes for meals in fast food restaurants is quite dispersed.

²¹ Based on sectoral national accounts and a report by Parniere and Pollet (2003), sales of chains represented 85% of total sales in the fast-food restaurant sector (including cafeterias) in the 2001–2002 period. In the traditional restaurant sector this share is only 16%. In addition, the fast food market is relatively concentrated, with the first five largest chains having a market share of 75%. The name of the chain is not available in our data set but since the sample of prices collected should be representative of the whole sector, big chains should be well represented in our price sample.

²² We also consider cafeterias as part of the fast food category. Cafeterias often belong to larger chains and, like fast food restaurants, typically have no waiting staff table service.

²³ As robustness checks, we also consider March year t − 1 and September year t − 1 as initial dates for the calculations of price changes and we find that estimation results remain very similar.

²⁴ See Puhani (2012) and Ai and Norton (2003) for a discussion on difference-in-differences estimates in nonlinear models.

²⁵ See Table A4 in the Appendix for estimates for the 18-month window.

²⁶ In the marketing literature, these fixed-price meals are an example of product bundling. The bundle is cheaper than the sum of its individual components.

²⁷ The coefficient is negative and nonsignificant, implying that the euro effect tends to disappear and prices go back to their pre-changeover path.

²⁸ We introduce dummy variables for the position of prices in the price distribution (in terms of quartiles).

²⁹ See Table A5 in the Appendix for estimates for the 18-month window.

³⁰ A département is an administrative zone of which there are 96 in France. Each has approximately the same geographical size (6000 km²). Around a quarter of the French départements are on the seacoast.

³¹ First empirical investigations however do not show clear and significant differences across products in pricing point strategy before and after the changeover.