Fa(s)t Food Analysis

The economic research paper described here pretty cleverly exploits a natural experiment to suggest that fast food does not cause obesity. The main finding is that people in rural areas next to highways (where there are more fast food restaurants) are no more likely to be fat than rural people farther away from highways. Since the building of these restaurants was not driven by local demand (but rather to accommodate through traffic), they take this to imply that the mere existence of convenient fast food does not itself cause people to gain weight. They suspect that the proliferation of fast food and other restaurants, serving large portions, may be a reflection of people's preferences for eating a lot of food and not a causal factor for obesity.

I wondered whether their measure of convenience was poor because it does not account for the fact that people who live in more distant areas may work in areas that are convenient to a higher density of restaurants, but they examined employed and unemployed people separately and did not find any differences.

One thing I found rather hard to believe was that based on survey data from 1994 - 1996, Americans (living in rural areas) increase their daily calorie intake on days they eat in a restaurant by a net of 24 calories. Though "when a given individual eats out, he consumes 233 more calories per meal than when he eats at home," this increased intake is compensated for by eating less the rest of the day. Can this be right? That data also seems pretty out of date, doesn't it? It's 2008 now. Was the Super Size phenomenon as prevalent at that time? Were sit-down restaurant meals as huge and full of freebies (e.g. mammoth and endless free bread) as they are now? I wish they had reported more recent numbers (though I recognize that the economists were only reporting the secondary data they could find).

They also found that "the between-individual coefficient is significantly larger than the fixed effects coefficient (235 versus 24), implying that individuals who frequent restaurants also eat more at home. This difference suggests that selection may explain why a number of observational studies have found a link between caloric intake and food away from home." So basically, the idea is that fat people eat a lot in restaurants and at home and thus any correlation between restaurant eating and weight is showing that fat people like to eat out, not that eating out makes a given person fat.

The authors suggest that legislation intended to limit the availability of fast food or to increase its price through a "fat tax" (this is discussed in the paper itself, but not the description of the research) will not have much if any effect on obesity levels.

In the economic jargon, "Regulating specific inputs into health and safety production functions is unlikely to be effective when optimizing consumers can compensate along other margins." In other words, keeping people from eating a lot in restaurants isn't going to do much when people can meet their desires for porking out by buying a bunch of craptacularly high-cal food at the supermarket. This rings pretty much true to me and suggests that either (1) the regulation of what people eat is going to have to be a lot stricter than what most people would be willing to tolerate or (2) people's preferences (and habits) need to be changed in a more widely reaching way. Of course, once you start talking about "preferences," I start to wonder what preferences we mean? The preference of the Hungry (or Stressed or Bored or Whatever) Person Who Wants a Large Supreme Pizza Right Now or the Overly Stuffed Person Who Regrets the Trip to Pizza Hut and Never Wants to Eat Again?

Because one thing this analysis does not address is the impact of restaurant eating on people who are generally trying to eat more healthfully / "watch what they eat" / eat less / diet, and they acknowledge that fact when they write: "If consumer preferences are time inconsistent, then regulation that decreases obesity may benefit at-risk individuals. The goal of this paper is not to evaluate how time inconsistency affects the optimality of decisions regarding caloric intake."

Fair enough. But I sure would have been interested in seeing the net calorie difference among "people trying to be careful about eating" for days they eat at home and days they eat at a restaurant because it's my experience, and the experience of approximately 3.6 gazillion people who have lost weight and talked about it on the Internet, in Weight Watchers meetings, etc., that Restaurant Eating is generally Diet Death.

Of course, it is also the case, for me at least, that I am more likely to want to eat in a restaurant when I want to "eat the world" than when I am feeling normal, so it may be a selection problem for me also. But when I successfully resist going to a restaurant in that desperate condition, I do not stay home and eat 1,000 calories at dinner. But I may be more successful than the average person in not having a lot in the way of attractive, convenient calories in my own home. (E.g. I do have the ingredients to make snickerdoodles, but I never do, even when I am strongly desiring going out for a high-calorie dessert or would eat a box of snickerdoodles if they appeared in my kitchen cabinet.)