by Elena Salomone*
When consumption of a good entails transportation costs for end consumers, competition among suppliers of that good is likely to be at a local level. This is the case, for instance, in the markets for the supply of retail grocery: consumers balance the cost they bear or perceive for transportation against the utility they expect to obtain from their grocery purchases and end up restricting their attention to certain local suppliers. In other words, transportation costs limit the geographic extension of the markets.
The definition of the relevant geographic market can be tricky in these cases that involve local competition, and where the exact geographic scope of the relevant market depends on consumers’ willingness to move.
Competition authorities need to address this question before carrying out their assessment of the competitive effects of certain market conducts, including mergers. Some competition authorities have relied upon survey evidence on the time or distance consumers are willing to travel and concluded that a relevant geographic market could be defined for each store as the area surrounding it with a radius corresponding to such time or distance. In other words, they considered that if consumers are willing to travel a certain average time, say 15 minutes, to reach a grocery store, then each store is competitively constrained by other stores within a 15-minute travel time. This approach, for instance, was undertaken by the Italian Competition Authority (ICA) in its review of the acquisition of Auchan by BDC Italia-Conad (case C12247B).
At a first glance, this approach might seem a simple and efficient way to conclude on the extent of the relevant geographic market. But is it reasonable? That is, does it provide a reasonable estimate of the area in which a hypothetical monopolist could successfully raise prices above the competitive level?
Well, no! This solution gives rise to a biased definition of the geographic market.
From the standpoint of a competition authority which is carrying out an investigation, accuracy of results must be balanced against efficiency, as more sophisticated analyses may entail the use of resources which are limited. This trade-off might justify, in some cases, the use of shortcuts (“heuristics”) that lead to some (random) errors, thus sacrificing some accuracy, but saving resources. The approach outlined above for geographic market definition, however, cannot be justified on these grounds. Not only it is simplistic, it will lead to systematic (not random) mistakes.
Remember that the definition of the relevant geographic market to which a given store belongs aims at identifying all the competitors that impose a significant constraint on the store, i.e. that limit the extent to which it can exercise market power.
The approach outlined above would be reasonable, and actually accurate, if all customers of the given store were located exactly where the store is located. This assumption, however, is clearly unreasonable. Under any alternative assumption on the distribution of customers, the extension of the relevant geographic market will be larger. Just to clarify, suppose that all customers are located at a 15-minute travel time from the store: as customers can travel in all directions, in this case the appropriate relevant market for the store would comprise all stores within a 30-minute travel time, namely it would be roughly twice the size envisaged by the competition authority.
These examples help illustrate the point that the extension of the relevant market depends, crucially, on population distribution. Completely disregarding population distribution, competition authorities are victims of a cognitive bias and make a systematic mistake. Different approaches can be adopted to correct for this mistake. If you want to find out more, stay tuned for our Lear Competition Note.
* With contributions from Mohamed Dahmani and Jakob Rustige