A recent (seemingly unsuccessful) stint hunting for non-academic articles on spatial aggregation of stores produced curious results. Naturally, many people observe at some point that a.) supposed fierce competitors like Exxon/Mobile and Shell or CVS and Walgreen’s surprisingly frequently appear in tandem, that is to say right next to each other and b.) often even bigger groups of competing stores like car dealerships seem to cluster in certain areas.

Eager for knowledge as human beings (ideally) are, the internet provides massive evidence for users on the quest for the roots of the above described phenomena. Now, the curious part are the answers floating around on the world wide web. Surprisingly, there seems to be no popular trusted media outlet which answers the question (I am thankful if proved wrong and pointed towards the desired direction). Among the attempts showing the most hits is a short SAP article published via Forbes’ BrandVoice feature which can be found here. A few seconds suffice to realize that this article is merely an executive summary of a blog post of author Presh Talwalkar which itself can be found here. These two  pieces are (near) perfect representatives of the majority of (non-academic) justifications of the two above scenarios on the internet.

They, correctly, link scenario a.) above to the basic Hotelling model. Hotelling explains in a simple model why strategic reasoning leads two competing stores to choose an identical location in a local duopoly. In order to stress the old example once more, consider a (linear) beach served by two ice cream stands. Assume the beach to be busy, such that potential customers are distributed uniformly over its entire length. Further assume, for simplicity, that the stands offer the same flavors and quality of ice cream at identical prices and competition is purely executed via location. Since walking over hot sand is uncomfortable, every customer opts to buy ice cream as close as possible to her towel spot. Game theory predicts that both ice cream vendors end up in the middle back to back. Why is that? Suppose one of the vendors to position his stand anywhere north (south) of the middle. Its competitor could swiftly just move an inch south (north) of it to ensure himself more than half of the customers. As a consequence, both stands right next to each other in the middle constitutes the unique Nash equilibrium of the outlined game since in any other scenario at least one of the vendors can make himself better off by unilaterally moving his stand. This is a “strict” Nash equilibrium in the sense that every unilateral deviation causes a vendor to be worse off.

While this simplistic model does not capture the entirety of location choice in duopolies in reality, it is a perfectly reasonable approach to explain why CVS and Walgreen’s can be frequently found across the street of each other, or, as in this special case in Edina, Minnesota, in such close proximity to each other that drive through clerks can wave to each other. In fact, numerous academic articles have claimed that the simple Hotelling approach is a powerful predicition tool in local duopolies. It is, however, crucial to understand that this behavior is nothing specific to representatives of national or global chains as some articles on the internet want us to believe. The simple Hotelling duopoly model applies to strategic competition of national giants CVS and Walgreens in downtown Chicago or San Francisco as much as it does to family operated pharmacies in a rural village in Maine (as long as the village is sufficiently populous to have two pharmacies operate profitably).

The more striking claim to be found in the above referenced articles and the set of pieces they represent is, however, that this logic generalizes to clusters of competing stores like car dealerships. It is straightforward to falsify this claim. Reconsider the above discussed ice cream vendors and bring a third one into play. If this entrant positions himself right in the middle of the beach next to the two incumbents he can expect to serve approximately a third of the beach population drooling over ice cream. Moving slightly north or south of the middle, however, ensures him nearly half the customers. As a result, the basic Hotelling model does not explain why you typically drive by car dealerships for minutes once you passed the first. The logic of the Hotelling model does simply not extend to a scenario featuring a number of competitors in excess of two. Analogously, the same reasoning explains why in a two party system winning the median voter is key, whereas in a multi party system we witness parties positioning significantly to the left or right of each other.

Clustering of car dealerships, furniture stores, etc. is much more likely to be caused by a common effort to reduce customers’ search costs combined with zoning restrictions set by municipalities. Simplified, as car dealerships (to stick to one story) are typically located on the outer fringes of urban areas, it would be mighty costly for a potential customer to drive all over town to compare and test drive vehicles of multiple brands. Thus, she is likely to pre-select a smaller number of manufacturers to visit in the first place. This, in turn, would result in a lower average number of potential customers to enter a car dealership every given day. This reasoning is also consistent with the fact that it has been observed that brands which are closer substitutes to each other are more likely to be found in clusters. Chicago’s northern suburbians for instance may find Subaru, Volkswagen, Mazda, Nissan and Fiat in the Evanston/Skokie area whereas Mercedes, BMW, Infiniti, Volvo and Land Rover are located on the same road in Glencoe.

In the case of dealerships selling bulky and expensive items which require huge parking lots, loading and storing capacity or direct access to roads in the case of cars, regulations passed and incentives provided by local governments seem to further foster clusters. In order to raise their attractiveness to potential customers and residents alike, municipalities might restrict potential storefront locations via zoning or grant tax reliefs to induce the birth of shopping parks. While such factors might also come into play regarding clustering of pharmacies or gas stations, they seem to be less likely the driving forces behind the location coupling of competitors. On the other hand, central strategic positioning predicted in the Hotelling duopoly does not account for suburban clusters of car dealerships as exemplified with three ice cream vendors above.