Distribution of BMI for people living in US towns adjacent and non adjacent to interstate highways (note there are 2 curves)

It’s commonly taken for granted that a disproportionately large number of fast food stores in an area is a key reason why the local population often has high rates of obesity.

The California Centre for Public Health Advocacy, for example, argues that “there is growing scientific evidence that what people eat—and their likelihood of being obese—is influenced by the food environment in which they live”.

More people are going to fast food restaurants where the food is high in calories and portions are large. Many respectable studies have shown a clear correlation between average body weight and eating out. According to one, growth in the number of restaurants accounts for as much as 65% of the rise in the percentage of Americans who are obese.

Various policies have been proposed to break the evident connection between fast food restaurants and obesity. These include proposals to limit the number of fast-food restaurants, ‘fat taxes’, limits on fast food advertising and mandatory calorie counts on menus.

But do fast food restaurants really cause obesity? As with many things, appearances may be deceiving. Correlation doesn’t mean causation.

This study, Are Restaurants Really Supersizing America?, by UC Berkeley researchers Michael Anderson and David Matsa, poses this question: “do more restaurants cause obesity or do preferences for greater food consumption lead to an increase in restaurant density?”. 

It is not obvious, they say, that the empirical link between eating at restaurants and obesity is causal. “If consumers’ lifestyles”, they argue, “are increasingly conducive to excess energy intake and positive energy balance, the increasing prevalence of restaurants may simply reflect a greater demand for calories”.

They compare indicators of body mass for two rural populations in eleven States of the US. The ‘treatment’ group lives within five miles of an interstate highway where the density of fast food restaurants is high. The ‘control’ group live between five and ten miles from the highway where accessibility to restaurants is lower.

They find that there is no causal link between obesity and restaurant food – the effect on BMI of restaurants is minimal at best (see chart).

The distributions of BMI in highway and non-highway areas are virtually identical, and point estimates of the causal effect of restaurants on obesity are close to zero and precise enough to rule out any meaningful effects.

But they also find that notwithstanding that more calories are consumed in a typical restaurant meal than at home, lowering restaurant prices does not increase obesity. There are two reasons:

First, there is selection bias in who eats at restaurants; people who eat at restaurants also consume more calories when they eat at home. Second, when eating relatively large portions at restaurants, people tend to reduce other calorie consumption at other times during the day. After accounting for these factors, eating a meal at a restaurant is associated with only 24 additional calories.

The authors suggest that the same outcome may apply to other targeted obesity intervention programs. They point to two recent studies of school-based programs where improvements to the nutritional content of cafeteria food had no effect on student’s weight because there was no control over what students ate outside of school.

This may also apply to other targeted interventions where consumers have other ways of compensating. The authors cite a study that contends mandating automobile safety devices does not reduce traffic fatalities because motorists respond by driving less carefully. Another study found that smokers react to cigarette taxes by smoking fewer cigarettes more intensively.

In the case of obesity, consumers can choose from multiple sources of cheap calories. Restricting a single source – restaurants – is therefore unlikely to affect obesity, as confirmed by our findings. This mechanism may also underlie the apparent failure of so many targeted obesity interventions……despite their ineffectiveness, such policies have the potential to generate considerable deadweight loss.

They measure the potential deadweight loss of policies targeted at restaurants in the US at up to $33 billion annually.

There are some possible shortcomings in this research. It might be that behaviour in rural areas does not translate to urban areas as easily as the authors assume. It might be that the people who live more than five miles away from fast food restaurants work or go to school closer to the highway.

Even so, it sounds a cautionary note not to over-sauce the physical environment angle (an issue I’ve discussed specifically in the context of obesity before – herehere and here). And as always, correlation doesn’t mean causation.