Lynne Kiesling
One of the most fascinating phenomena to be analyzed using industrial organization is the clustering of competing retailers. Ask yourself: why do Oriental rug stores all cluster in the same block? And what about car dealers? Don’t you want to be farther away from your competitors?
Enter Harold Hotelling, who developed a simple model of the retailer location decision. Take a straight line. Label one end 0 and one end 1. Suppose your possible customers are distributed uniformly along that line. Say that you are a retailer and you have to decide where on that line to locate to maximize your revenue (assume constant per unit costs, so irrelevant to location decision). Where do you choose? Duh-halfway, 0.5. Now suppose there are two of you retailers. Where do you each locate? As close to halfway as possible, 0.5+epsilon and 0.5-epsilon. But that puts you right next to each other. And so on.
I love the Hotelling model; it remains one of my favorite simple IO models that explains a lot of meaningful real-world phenomena.
The latest application of the Hotelling model is in Chicago on Michigan Avenue. I was in Streeterville last week for a doctor’s appointment, and I noticed that Hershey’s is opening a new soda and ice cream shop right next to … Ghiradelli! Chicagoist noticed the same thing:
Funny thing, though. The location of the Hershey’s store is pretty much right across the street from the Ghirardelli shop.
Textbook application of the Hotelling model, with the only hitch being that the corner of Michigan and Pearson is not exactly in the middle of the stretch of Michigan Avenue between the river and Oak Street. It is, however, right smack dab on the plaza where the old water tower is located, so it’s a prime tourist and shopping location, which makes it the non-uniform equivalent of “the middle of the line”.
Chicagoist also noted the other important dimension that adds interesting complexity to the Hotelling model: product differentiation.
We know that Ghirardelli is more upscale than Hersheys, that it tries to be more like a soda fountain with sundaes and shakes and huge ass cookies and Hersheys is more.. well.. Hershey’s… but can that little water tower square on North Michigan Avenue really support TWO chocolate shops?
Answer: yes. Have ya *seen* the throngs of tourists around there at 3PM on summer Saturdays? Between the concentration of customers and the product differentiation, I predict profits for both enterprises.
Although I agree with Rachelle at Chicagoist: go for the hometown favorite, Vosges, a retail outlet of which is located south on Michigan Avenue at the North Bridge shops. *That’s* serious product differentiation, and serious chocolate.
Lynne,
A very interesting post, but I believe you may be missing a trick.
In reality, there is rarely such an obvious line extending from 0 to 1. When the line goes from – infinity to + infinity, and then blithely assuming a smooth demand distribution along the line (or 2D place as the case may be), the context of the median point becomes meaningless.
I propose, instead, the HBG theory of car-dealership clustering, as follows:
Let us assume that, there is a dealership somewhere on our line of demand.
the next person to start (or want to start) another dealership does not now see a uniform distribution of demand. Why? Because the existing dealership draws demand to it. Thus, although the source of demand may originally have been spread evenly, the satisfaction of that demand is now concentrated at the source of supply. The existing dealership is a footfall generator. Thus, the next man in the market will want to place his dealership where he knows (or believes) he will get the highest concentration of consumers who have an active desire to purchase a car.
This has a secondary beneficial effect on the operation of the market as it becomes easier for consumers to make an informed choice (they do not have to traipse all over town to make comparisons between brands of car).
There is clearly a limiting factor: if 8 chocolate shops opened on the same terrace, the suppliers would suffer (unless they engaged in a little underhand bit of price fixing, in which case it is the poor benighted consumer who loses).