Experiments concerning the effect of publishing calorie counts on restaurant menus tend to show little effect on choices. In the experiments that I know of, choices before and after publishing calorie counts are compared. But this form of test cannot be considered conclusive. Some people were overestimating the calories and they might cut back, some were underestimating and they might eat more. There is no reason to expect that the aggregate change should be positive or negative.
A better experiment would be to use a restaurant where calorie counts are already published and manipulate them. Will people change their choices when you add 5% to the reported calories? 10%? What is the elasticity? It’s a safe guess that there would be little response for small changes and a large response for very large changes. Any response at all would prove that their is value is publishing calorie counts because it would prove that this information is useful for choices.
The only question that would remain is how those welfare gains measure up against the cost of collecting and publicizing the information.
Cheap Talk 
6 comments
Comments feed for this article
July 21, 2011 at 7:31 am
Heski
Of course another way to go would be if you could track individuals and their purchases … say if you had loyalty cards that they used at purchase.
Unfortunately, I’m not that smart but Bryan Bollinger, Phil Leslie and Alan Sorensen are http://www.stanford.edu/~pleslie/calories.pdf
“The second component is a large sample of anonymous Starbucks cardholders (inside and outside of NYC) that we track over the same period of time, allowing us to examine the impact of calorie posting at the individual level.”
When they presented it here one issue (that they’re up front about) is that they have data only on one store – it may be that individuals feel that they’ve been virtuous at one store and so are free-er in their eating habits at other stores (or have an extra cookie when a million of them are sitting outside the office after the lunchtime seminar…) or maybe give up on riding bikes, or maybe are thinking about health and exercise more, and who cares about calories per se (see http://www.nytimes.com/2011/07/19/health/19brody.html?_r=1&src=me&ref=general). As you suggest in the last sentence what we really care about is health outcomes and to my knowledge – no one has the foggiest idea of that or how to go about assessing it …
July 21, 2011 at 8:15 am
Mild Speculation
A couple extensions of the research I’d be interested in…suppose people view the counts and do consume less at the restaurant. What is the effect on total daily calorie consumption (including eating outside the restaurant)? Could there be a non-behavioral explanation for a strict increase in total calories that doesn’t violate revealed preference?
July 21, 2011 at 12:16 pm
k
Umm…manipulating calorie counts would be…incorrect information…?
Maybe better ways of presenting the information contained in the number would be useful to look into – Chipotle’s calorie counts for instance are not entirely clear; this in addition to the problem of interpreting the number.
In order to counter the problem you pose, perhaps we should think of ways of making the interpretation standard, to a reasonable level at least.
July 21, 2011 at 1:49 pm
Frank
“Umm…manipulating calorie counts would be…incorrect information…?” So? Why is that a problem?
I wonder if there is an “effect of publishing calorie counts on [the content of] restaurant menus”…
July 21, 2011 at 5:21 pm
dan s
the aggregate change is expected to be positive (or neg) under hypothesis that consumers systematically over (or under) estimated cals
July 21, 2011 at 10:12 pm
Scott
I know I look for the highest calorie count when choosing between options, at least when I am hungry. This is a big problem for estimation since almost surely some people choose the low calorie option and some the high calorie option. There could thus be large welfare gains for both both types of eaters even though the aggregate data shows little or no effect. In light of this it is also important to find individual-level data.