I am going to keep harping on Ezra Klein’s analysis of obesity, food intake and public policy because its a prime example of how very smart people crafting very well meaning policy can nonetheless go wrong.

Here Ezra discusses the issue of calorie labeling:

All quite delicious. When I got back to the office, though, I decided to see what it added up to. First, I looked up the cookie. A solid 450 calories, with 19 grams of fat. Yikes. But what might have actually changed my purchase was knowing the content of my sandwich: According to the nutrition calculator, 525 calories.

The calories in the cookie weren’t startling. But their calories relative to my sandwich proved a bit off-putting. I could pretty much have ordered a second sandwich for the caloric cost. Buying them without the information, it was easy enough to just consider them a side dish. As it happened, the cookie was more like a second lunch. I wouldn’t have ordered a second lunch. Good to know.

You can imagine a lot of marginal changes in decision-making after a menu labeling law goes into effect. But would small changes in matter? They might. The following table comes from a Health Impact Assessmentprepared by the County of Los Angeles on calorie labeling laws. It shows how much of the whole county’sprojected weight gain would be averted if calorie labeling got X percent of restaurant patrons to make average decisions that were Y calories smaller:

weightgainlabeling.jpg

There are few good things here. First, is Ezra’s recognition that the cookie has more calories relative to sandwich than he would have thought. This is good because it starts to bring up the important issue of calories necessary for satiation.

However, the entire rest of the piece is written from the calorie consumption hypothesis prospective. That is, that people are obese basically because they consciously or unconsciously choose to eat too many calories.

I should first note that there is serious debate about this within the nutritional community. That alone should give us pause. We are talking about a national policy here, that takes as a given a hotly debated point.

Second, and I am going to give some strong editorial here, but the calorie consumption hypothesis increasingly seems to be the promulgated by those whose who either don’t take the scientific issue of obesity seriously or whose analytical skills are somewhat lacking.

Why?

Well, the core assumption is that body mass is a function of calories-in vs. calories-out and that both of these are independent choice variables. The problem is that a basic system of differential equations built on that model would estimate that people would get much fatter, much faster than they do.

The exact results depend on your activity response function but it is not hard to get the conclusion that people should be 300lbs because they eat a slice of cake for dessert every night. That is, they eat everything else the same as other people do, but they also have this extra slice. As a result they become morbidly obese. That doesn’t seem quite right?

Even more unusual are the weight loss forecasts which tells us that by walking 20 mins a day someone could loose 200lbs or, and this is the clincher for me, a normal sized person should be able to exercise themselves into starvation simply by jogging an extra 10 minutes a day.

To make matters worse these conclusions assume a positive relationship between size and calorie expenditure. The data actually show a negative relationship. That is, bigger people actually burn fewer calories because they are so much less active.

If you pop that little gem into the model you get the conclusion that people should be able to reach infinite weight in finite time. Clearly this is not possible.

Those familiar with model making will instantly recognize that we need a stronger feedback or dampener.  Calories-in must somehow influence calories-out or vice versa. Simply factoring in that a larger person burns more calories in a given activity is not a strong enough feedback to produce realistic results. When you include the fact that larger people actually engage in fewer activities, you need an even bigger dampener.

Ah, you say – maybe the obese are obese because they are engaging in fewer activities. Decreased calorie expenditure is the cause not a consequence of weight gain. Sorry, but no cigar.

Exercise in and of itself shows almost no potential to lower BMI. Exercise as an uncontrolled intervention will show ever so modest decreases in weight but the evidence is that the exercisers are secretly dieting. See, it turns out that you can’t do placebo controlled experiments for exercise because its hard to trick people who are not exercising into thinking that they are.

However, if you compare dieters who exercise to dieters who don’t you get no difference in results. Same weight loss. One possible conclusion is that after being forced to exercise study participants are more careful to watch what they eat.

“Watch what they eat,” what does that mean exactly?

Most of us have a sense that it means, “eat healthy foods.” But, what does that mean? Ezra noted in his piece that he was having a bad day so he added the cookie. Why doesn’t he always have the cookie?

Is it because the cookie is high in calories? Well, its actually lower in calories than the sandwich. Could Ezra watch his weight by skipping the sandwich and only eating the cookie? That seems comically silly, but why?

The cookie doesn’t have nutrients some say. He needs nutrients.

Okay, cookie plus multivitamin. Is that just as good as sandwich?

No, no you say, the cookie lacks fiber and protein. Well, actually I’m not sure its as bad you might guess on that, but fine, cookie plus multivitamin plus fiber pill plus egg albumen, are we good now?

Or, is there something about the cookie that makes it especially bad? Is it not how much food is in the cookie, but what kind of food is in the cookie? Could it be that the feedback mechanisms we need to make our obesity model work take some of their clues from the content of the food we eat? A system feed by sandwiches dampens to a greater extent than one feed by cookies.

The debate rages about what exactly it is in the cookie that effects the feedback mechanism. Is it the fat? Is it the sugar? Is it the aroma, yes this hypothesis exists. Is it the physical density of the cookie, the weight or mass per calorie?

The problem with enacting a public policy is, what happens if the aroma guys are right? Cookies makes us fat because they smell so good. How is calorie labeling going to help that? If calorie labeling fails is our conclusion that there is no hope?

We can update our label requirements as new information comes in but each update undermines public confidence in the label. First, they told us it was calories, then they told us it was fat, then they told us it was carbs, now they tell us its smell – these guys have no idea what they are talking about the public will say. And, they’d be right.