Crunching the Numbers

Where’s the optimal place to park a food truck? In the Fall 2015 issue of Chicago Booth Magazine, we asked three experts who think about this professionally (not just at lunchtime): doctoral student Eliot Abrams, BA ’12, professor Chad Syverson, and alumnus Drew Davis, ’14.

Eliot Abrams, BA ’12, is working on a PhD in economics at Booth. For his summer project, he studied how food trucks in Chicago choose their locations.

Pursuing a love of food and cooking, I completed the basic pastry certificate at Le Cordon Bleu in Paris the summer before starting the PhD program. So I was thrilled to see a crêpe truck, Paris Ouh La La, serving lunch during the school year. After several good meals at the food trucks on Ellis Avenue, and observing the variation in the trucks parked each day, I started thinking about how the trucks decide where to park.

Where you choose to locate a business is a fundamental economic question—one that food trucks must re-answer every day. The classic location choice model was offered by the mathematician and economist Harold Hotelling in 1929. Consider two ice cream vendors who parked their carts on a one-mile stretch of beach. Assuming the vendors offer roughly the same treats, beachgoers will naturally choose to walk to the closest cart. The vendor on the left will serve all the beachgoers to its left, and the vendor on the right will serve all the beachgoers to its right. Knowing this, the vendors should both choose to locate at the middle of the beach in order to maximize their respective market shares and profit. Thus, Hotelling found that competition will lead vendors to congregate in the center.

His law explains why food trucks park right next to each other rather than spreading out along a block or across several city blocks. However, the law doesn’t explain why food trucks choose different parking locations on different days. The parking patterns of food trucks suggest an expanded model that incorporates additional influences: how consumers vary their lunch options, and also how they may develop, and then may lose the habit of patronizing a food truck. 

Modeling food truck parking locations is complex, as there are around 70 active  food trucks parking at more than 30 locations in Chicago. Thankfully, there is plenty of data because food trucks need to advertise their locations. Since 2011, Andrew Violette, who runs, has been tracking the city’s food trucks based on their Twitter feeds. I pulled 34,328 parking records from Violette’s website and created a simulation of how food trucks park.

This exercise translates the observed food truck movements into information on the relative number of customers a truck serves at each location on a given day. The number of customers is a function of many variables, such as the day of the week and the location chosen by the truck. I focus on estimating how customer traffic is impacted by the number and diversity of other trucks parking at the location and by the past frequency with which the truck has parked at the particular location. Just like Hotelling’s ice cream vendors, food trucks should (and do) choose their locations in response to these dynamics in order to maximize their profit.