Color Me Surprised

Update: The paper isn't exactly new. It's entitled "Are Restaurants Really Supersizing America?" and the working paper is dated December, 2007. But new to me (and apparently to Cowen).

Tyler points to a new economic study that finds fast food doesn't make you fat. The authors of the study are Michael Anderson (UC-Berkeley) and David Matsa (Northwestern). They use what appears to be a "natural experiment" methodology to identify the relationship between fast food availability and body weight. The author of the article writes:

To investigate the link, Matsa and Anderson turned to a natural experiment unknowingly created by the government decades ago. The Federal Highway Act of 1956 paved the way for over 42,000 miles of turnpike to be constructed in the United States. One of the unintended consequences of the new Interstate system was that, in rural areas, shopping centers that wanted to attract traveling drivers began to cluster around highways. For the residents of those areas, this development meant easier access to restaurants.

So, in other words, the federal government built interstates, which in turn, led to the development of restaurants along those interstates. But it also put new restaurants (presumably fast food ones, with lower health benefits) in closer proximity to residents located near the interstates, and therefore lowered the transportational costs of consuming fast food. Ideally, the location of these new fast food restaurants around the interstate highways is not correlated with the eating habits of residents, otherwise it is not a valid falsification of the hypothesis. It must be the case that the people living in areas located near the highway system are statistically the same as people located further from the highway system, because the goal is to simply introduce fast food to one group and compare their body mass index outcomes to some control group who was left "untreated." Apparently, that's exactly what they did too. The author again notes:

"To see if this new supply changed obesity rates, Matsa and Anderson compared the Body Mass Index levels of those living close to highways (0 to 5 miles) with those living farther away (5 to 10 miles) and had less access to restaurants. Using survey data from 11 states, Matsa and Anderson found no difference in the number of overweight, normal, and underweight people in the two areas, suggesting that access to restaurants was not making people fatter. "

This seems on the surface impossible, because as the writer notes, if they are consuming more calories, then why isn't that translating into higher BMI. Two possibilities I can think of. One, they are increasing expenditure of calories at the same rate, and thus the net increase calories is zero. Two, they are substituting other kinds of foods so as to cancel out the increase in calorie intake from the fast foods. Looking ahead in the article, that sounds like what the authors suggest was happening, too:

Matsa and Anderson next looked at data on individual eating habits from a survey conducted between 1994 and 1996. When eating out, people reported consuming about 35 percent more calories on average than when they ate at home. But importantly, respondents reduced their caloric intake at home on days they ate out (that's not to say that people were watching their weight, since respondents who reported consuming more at home also tended to eat more when going out). Overall, eating out increased daily caloric intake by only 24 calories. The results for urban and suburban consumers were similar.

From my casual reading of the economics of obesity literature, the main culprits for rising obesity is a change in the market from manufacturing to services, which caused indirectly a decrease in calorie expenditure through more sedentary jobs. In other words, when the workforce was primarily in manufacturing, they were being paid to exercise, but now that they work at desks, they are not. Some of this transition is coincident to the rise in fast food industry, and thus suggests the correlation is mainly spurious and not causal. The other culprit I read regularly has to do with technology. In the Cutler, Glaeser and Shapiro paper, they write:

The increase in food consumption is itself the result of technological innovations which made it possible for food to be mass prepared far from the point of consumption, and consumed with lower time costs of preparation and cleaning. Price changes are normally beneficial, but may not be if people have self-control problems.

I am not entirely sure what the technology shocks are precisely, but I always thought it meant things like the microwave, improved storage with the fridge, tupperware, and things like that. Things that lowered the cost on eating bad food, in other words. This is not exactly identical with fast food restaurants, but to me always seemed to include fast food. So, this paper means I need to read this more closely to better understand. But the patterns we're seeing in the economics of obesity is emphasizing more general labor market changes (including an increase in female labor force participation rates, which affects child and family eating habits), and technology which made eating bad food cheaper.

I also think, as I told J and H one day, that giving the rise in incomes, there's two effects. If healthy food is a normal good, then higher incomes means we want to eat better, and I think that explains some of the renaissance in healthy eating we're seeing. But secondly, higher incomes means the cost of food preparation has also risen, because the time spent preparing foods is a scarce resource. Time is not something we normally consider as an important input in the preparation of food - we normally focus just on the actual ingredients - but watching my wife cook, I know first hand how much time and skill it takes to eat well. For many Americans, the technology that made it cheaper to make bad food meant it was now more expensive to eat healthy food. This, I think, is more or less what is behind the Cutler, Glaeser and Shapiro paper. There's also the human capital (ie, skill) component to eating healthy. You have to have studied this, not only to know the benefits, but to collect the recipes and the know-how involved in making excellent food. If this is true, and if we're paternalistic about eating, then the solution is I think to consider a pigouvian tax on bad food so as to make eating healthy less expensive. In other words, to cause a substitution towards healthier eating, you need to make eating bad food more expensive.

Interesting paper. I'm printing it out now.