Why are gas stations clustered together?

By John, 28 July, 2011

Presh Talwalkar over at Mind Your Decisions has noticed that gas stations tend to cluster together in urban areas. It seems like every time you find a gas station, you find another one across the street. Anecdotal evidence suggests this is true.

Presh, being a game theory guy, sought to explain this in terms of game theory. The basic idea is that gas stations are mostly interchangeable and that if people are going to randomly choose your product or the other person's, you might as well locate yourself right in the middle of the area so that the most customers come to your gas station and the other person's gas station. If you get to close to an "edge" then your competitor can move into the space you are "ceding" and gain all of those extra customers.

No doubt there is something to that. I am reminded of voronoi diagrams of McDonald's locations.

He does say in the post that it's a simple model. "There are clearly many factors at play. It’s an optimization involving demand factors, real estate, estimates of population growth, and supply considerations—like the ease of refueling a station. The answer is complex, and any explanation I offer will have its problems."

True, but it doesn't mean we can't take a stab at it. Here are a few other hypotheses:

1. Traffic levels, lane configurations and even one way streets may make it impossible for me to pick which gas station to go to, even if I really want to. If the gas stations are "across the street" on the map it may be nearly impossible in practice to substitute one for the other without taking a long detour.

2. Certain types of businesses tend to cluster together because the cost of leasing the space and operating the businesses are similar. Kinkos is more likely to be in the same strip mall as Batteries Plus because they have similar fixed cost structures and size requirements. Perhaps the location of gas stations is similarly affected.

3. What about zoning regulations and local planning commissions?!

4. Gas buddy relies on a system of volunteers to report gas prices. It's possible that this is not a fully systematic, comprehensive data set and that smaller neighborhood gas stations are getting underrepresented.

5. Maybe gas stations really are randomly distributed, but they are randomly distributed on the busiest roads. This causes them to "bunch up" together more than they would otherwise, because they're all seeking a spot on the roads with the most traffic. If traffic levels were not a factor in business success, perhaps they would spread out more in suburban areas instead of clustering together.

#1 is the first thing I thought of, and #3 is probably the most likely.

Tags