Effects of restaurant menu labels

Brenna Ellison, David Davis, and I have a paper forthcoming in the journal Economic Inquiry and that is finally available online.

Here's are the study objectives:

The overall purpose of this research is to perform an in-depth examination of menu labeling and pricing policies in a full-service, sit-down restaurant. Specifically, this research determines: (1) whether caloric labels in a fullservice restaurant influence food choice, (2) whether symbolic calorie labels are more/less influential than numeric calorie labels, (3) how effective menu labels are relative to “fat taxes” and “thin subsidies” at reducing calories ordered, and (4) the economic value of menu labels.

Our projections of the short-run effects of different policies (numeric label, symbolic label, 10% tax on high calorie items, or 10% subsidy on low calorie items) on the number of calories ordered at the restaurant we studied are as follows:

The only policy option which resulted in a statistically significant change in calories (where the 95% confidence interval on the change didn't include zero) was a "symbolic" calorie label - essentially a traffic light label with a red dot next t the highest calorie items, a yellow dot next to medium calorie items, and a green dot next to lowest calorie items.  We put the point estimate on the value of the symbolic label at about $0.13/person/meal.

It is important to note that the symbolic label policy option was also the one that had the most detrimental effect on restaurant revenues (these results are in Brenna's dissertation).  Also, curiously enough, Brenna's surveys suggest most people say they don't want the symbolic label.  Here's what we wrote in a different paper discussing a post-meal survey conducted with some of the diners:

Interestingly, despite the calorie+traffic light label’s effectiveness at reducing calories ordered, it was not the labeling format of choice. When asked which labeling format was preferred, only 27.5% of respondents wanted to see the calorie+traffic light label on their menus. Surprisingly, 42% preferred the calorie-only label which had virtually no influence on ordering behavior.

Is portion size to blame for obesity?

I've often seen presentations where the authors show the size of an average hamburger or soda in the 1960s, 70s, 80s, and 90s as a way of making the point that portion sizes have increased, and thus contributes to the rise in obesity.  Changes in portion size probably have played some role, but at least according to this experiment published in the journal Obesity, it may play less of a role than first meets the eye. The researchers recruited over 233 people working in a large medical complex and randomly assigned them to treatments that differed according to the size of the free lunch they were given (one control group was given no free lunch at all).  Here's what they found:

Adults (n = 233) were randomly assigned to one of three lunch size groups (400 kcal, 800 kcal, and 1,600 kcal) or to a no-free lunch control group for 6 months. . . .

Body weight change at 6 months did not significantly differ at the 5% level by experimental group (1,600 kcal group: +1.1 kg (SD = 0.44); 800 kcal group: −0.1 kg (SD = 0.42); 400 kcal group: −0.1 kg (SD = 0.43); control group: 1.1 (SD = 0.42); P = 0.07). Weight gain over time was significant in the 1,600 kcal box lunch group (P < 0.05).

A remarkable increase in portion size from 400 kcal to 1600 kcal for lunch over a 6 month time period resulted in no statistically significant differences in weight across groups at the end of the period.  If you compare the pre- and post-weight of the people in the 1600 kcal group, there was a slight increase (0.19kcal/month) in weight for people in that group but not for people in the 400 kcal group or the 800 kcal group.  Curiously, those in the control treatment, which included people who were not given a free lunch, gained a statistically significant 0.24 kcal/month - more than those in the 1600kcal free lunch group! 

The trouble with many "interventions" such as this (similar to those that happen at school lunches) is that people can substitute across time.  If I eat a big lunch, I'm likely to eat less for dinner. And vice versa.  If I eat a 400 kcal lunch, I'm more likely to grab a snack in the afternoon than if I eat a 1,600 kcal lunch.  

Apply this thinking to related policy ideas.  Ban sodas or transfats. What will people drink and eat instead?  Tax restaurants.  What will people eat at home?  Add more veggies to the plate at school.  What will happen to veggie consumption at home?  I'm not saying that such policies might not have some effect on weight, only that because of substitutes and compensating behavior, they will often have less effect than is expected.

What causes obesity?

According to Gary Taubes, writing in the New York Times Sunday Review, we don't really know.

I agree.

Here is Taubes:

Here’s another possibility: The 600,000 articles — along with several tens of thousands of diet books — are the noise generated by a dysfunctional research establishment. Because the nutrition research community has failed to establish reliable, unambiguous knowledge about the environmental triggers of obesity and diabetes, it has opened the door to a diversity of opinions on the subject, of hypotheses about cause, cure and prevention, many of which cannot be refuted by the existing evidence. Everyone has a theory. The evidence doesn’t exist to say unequivocally who’s wrong.

and

One lesson of science, though, is that if the best you can do isn’t good enough to establish reliable knowledge, first acknowledge it — relentless honesty about what can and cannot be extrapolated from data is another core principle of science — and then do more, or do something else. As it is, we have a field of sort-of-science in which hypotheses are treated as facts because they’re too hard or expensive to test, and there are so many hypotheses that what journalists like to call “leading authorities” disagree with one another daily.

If I'm not mistaken, Taubes is leading an effort to raise funds to experimentally test his conjecture  (related to effects of sugars and other carbs) about a cause of obesity.  He has written persuasively about his views, but rather than leaving it at that, he's putting his money where his mouth is, and using scientific experiments to test his theory.  Good for him.

Are Sugar-Sweetened Beverage Taxes a Cost-Effective Means of Reducing Weight?

That was the title of a short paper I just published (ungated version) in the Canadian Journal of Diabetes.  The piece was written in response to a prior article by Buhler et al. arguing that a consensus had been reached on the need for soda taxes.  I pointed out that their consensus didn't include any economists.

 A few snippets:

More fundamentally, one must ask what conceptual basis is being used to assert that SSB taxes will increase consumers' welfare? Presumably, some consumers already consider health impacts when they choose what to eat and drink. More generally, taxing food or SSBs is analogous to reducing consumers' real income, which almost certainly harms the consumers (9 ). . .If the argument is that people do not understand the risks of SSBs, then the appropriate policy response is information provision, not a tax.

and

One of the most common assertions is that SSB taxes are required because one individual's choices impose costs on others because of the existence of public healthcare programs. However, forgotten in such claims is the fact that many of the obesity-related costs are private, not public (12). Moreover, the costs to the public health programs are actually transfers among people in an insurance pool, not an economic deadweight loss to society that reduces Pareto efficiency (12). 

In conclusion:

In sum, Buhler et al (1) are correct that obesity is a complicated and multifaceted issue. So too are the consumption, weight and economic-welfare effects of SSB taxes. SSB taxes often appear to be a simple (if partial) solution for a big problem but, as witnessed by Denmark's recent decision to rescind its versions of the “fat tax,” the consequences and impacts of such taxes are anything but simple.

Longer School Lunch = Less Obesity?

A while back, I wrote:

. . . many school children have to eat lunch as early as 10am! In many schools (including my own kids’ school), children have to be run through the cafeteria so quickly they hardly have time to eat. Couple that with the new federal guidelines limiting the number of calories that can be served, and it is no wonder many kids are starving by the time school gets out and beg to go to McDonalds!

In addition all the above, I'd also add that because of increased curricula requirements, PE has been cut to the bone in most schools.

Alas, it seems most of the discussions I hear about improving childhood health in schools revolve around "sexier" headline-grabbing issues like serving more fruits and veggies, serving more local foods, zoning rules, banning sodas, teaching gardening, and so on. It may just be that the less "sexy" (and potentially less costly) issues like encouraging exercise, increasing cafeteria time or size, or giving a small afternoon snack, may be more promising.

Of course, we'd want empirical evidence that length of lunch had a substantive impact on dietary choice and weight.  I see one piece of evidence was just published in the Southern Economic Journal this month.  The piece is by Rachana Bhatt entitled "Timing is Everything: The Impact of School Lunch Length on Children's Body Weight."  The abstract

The large number of overweight children in the United States has prompted school administrators and policy makers to identify practices in schools that contribute to unhealthy weight outcomes for children and develop strategies to prevent further increases. Advocates for school nutrition reform have suggested that it is important for children to have an adequate amount of time to eat meals in school in order to maintain a healthy weight. This article examines whether the length of time children are given to eat lunch in school has an impact on their weight. I find evidence that an increase in lunch length reduces the probability a child is overweight, and this finding is robust across various econometric specifications, including a two-sample instrumental variable model and difference-in-differences model that account for the potential endogeneity of lunch length.

The paper indicates:

extending lunch length by 10 minutes is associated with a 1.2% reduction in BMI, and it reduces the probability a child is considered overweight for his/her age and gender by 2.4 percentage points.  

I'll be curious to see if these results hold up in randomized controlled trials.