The “Starbucks Effect”: Correlation vs. Causation
Earlier this year, several news organizations reported on a study that found that homes near a Starbucks increased in value at a rate higher than others during a fifteen year period. Did Starbucks cause this larger rise in home values?
The headlines seemed to suggest it had: “What Starbucks Has Done to American Home Values,” “Living Near a Starbucks Might Double Your Home’s Value” and “Starbucks Increases Neighborhood, Home Values”—all imply that the presence of Starbucks led to the increase in home value.
This story caught my eye for a number of reasons. While I’m not a coffee drinker, a new Starbucks just opened down the street (okay, about two miles down the street, so not that close) but if its presence further increased our home values that would be a plus. But more to the point, it drew my attention as a sociologist. The headlines seem to confuse causation with correlation, assuming that one variable had a direct impact on the other, rather than coinciding with a number of other factors that come along with the decision to open a new Starbucks.
To know for certain if Starbucks were the cause, we would have to conduct a randomized experiment: find communities where Starbucks would normally choose to place stores, and then randomly select which location gets a store, and which ones would not.
We would have to control for a host of other factors as well, including things that make a neighborhood more desirable such as the quality of local schools, proximity to other amenities like shops, parks, waterways, public transportation, and neighborhood safety.
We would then measure the value of the homes after the stores were placed in the randomly selected locations. But during its period of rapid expansion, Starbucks opened stores wherever they thought they would be most profitable, and would likely not withhold a new store for the purpose of research. After all, they are in business to make money, not to raise real estate values.
We would also have to consider historical shifts and cohort effects. The same time period measured in the housing study has been a time of growth for Starbucks, which grew from about 2,000 stores in 1998 to about 17,000 in 2011. As the company expanded, it sought to place new stores in areas where they would be most profitable. Apparently they did this rather well. According to a story in the business news site qz.com:
The Starbucks team explained that while they have 20 or so analytics experts around the world poring over maps and geographic information systems data—assessing factors like an area’s traffic patterns and businesses—the company also empowers dozens of regional teams to come to their own conclusions about location, store design, and a host of other issues.
Would the home value disparities continue once Starbucks was no longer as new?
As for cohort effect, since the so-called “millennial generation” has come of age in the past decade—and the time of Starbuck’s massive growth— they have gravitated towards housing in urban centers or walkable communities that are also places where Starbucks are likely to open stores. Starbucks stores also encourage and enable people to spend time just hanging out, either visiting or working, well beyond the time that it takes to consume a cup of coffee or a snack, which may appeal to people who have the leisure time to do these things. These spaces may appeal to younger affluent consumers who are part of the new tech economy and might be accustomed to working in non-traditional spaces. They may also be drawn to the store’s WiFi and comfortable seating.
The cohort effect might also explain some of the rise in housing prices as well. The increase in young adults means an increase in demand for housing. If you take a look at the population pyramid of the United States below, you can see that the bars for ages 20-34 are wider than those 35-49. What this means is that there are more young adults now than there were in the mid-1990s, when there was a smaller youth cohort—and when the study of housing and Starbucks began.
It’s not that there isn’t a relationship between housing prices and Starbucks. But it’s too simple to suggest that a Starbucks alone has raised property rates; while its experts might do a terrific job finding places where their stores will thrive, and thriving neighborhoods often see a rise in property values, it is important to look at the broader neighborhood, historical, and demographic context to understand the so-called “Starbucks Effect.”