The case for taxing sugar-sweetened drinks to curb obesity, which is being enthusiastically advanced by some of the country's most prominent public health experts, wildly overestimates weight loss, according to two new studies published in leading academic journals.
And the reason explains why there are no quick fixes when it comes to losing weight.
"Measuring weight outcomes for obesity intervention strategies: the case of a sugar-sweetened beverage tax," (Lin et al.) which is published in the latest issue of the journal Economics and Human Biology, expands on a crucial problem raised by another study, "Quantification of the effect of energy imbalance on bodyweight," (Hall et al.), which was published in the Lancet's special issue on obesity last August.
The problem is this: Obesity research has tended to count projected weight loss from any intervention as cumulative. If you lose 100 calories on Monday and 100 on Tuesday, you've lost 200 by Wednesday.
More to the point, given that there are 3,500 calories per pound of body weight, you can plot how many calories you'd need to lose each day in order to achieve fixed weight loss goals. This is called a static model for weight loss; its virtue is simplicity; and it is the model typically used to make predictions about the impact of soda or fat taxes on obesity.
It means you can conceptualize a pound of bodyweight in terms of numbers of cans of soda or packets of chips and plot the way changes in price will affect consumption.
Unfortunately, the static model appears to be an exercise in wishful thinking, as it does not take into account things we now know about how the body's metabolism changes when a person is trying to lose weight.
The simplest way to understand this is that as you lose weight, your body compensates by expending less energy, thus slowing your rate of weight loss. Another way of looking at this is that weight loss becomes progressively harder to achieve the more you lose weight.
In the Lancet, Hall et al. go so far as to call the static model a "myth" and warn that its widespread use "has led to drastically overestimated expectations for weight loss." Instead of a static model, we need to model weight loss in a dynamic way that captures the body's processes of physiological adaptation.
To illustrate their point, they take one of the most recent and optimistic soda tax studies from the US Department of Agriculture (Smith et al. 2010).
Hall and his colleagues note that the USDA's static model predicted that a 20 per cent tax on soda would translate into a reduction of 40 calories per day, which would add up to almost four pounds per year. Maintain this for five years and the obesity epidemic in the U.S. could be reversed. Average adult bodyweight would return to the 1970s.
It's not hard to see, given that kind of prediction, why soda taxes seem like a miracle pill for obesity. Think of all the health care costs you could eliminate by raising the price of soda by 20 per cent – while creating billions of dollars in tax revenue. As a policy solution to three problems, it sounds too good to be true.
Unfortunately, it is too good to be true.
When you put Smith et al.'s data through Hall et al.'s dynamic simulation, instead of losing four pounds in the first year, the average person loses just 2.2 pounds. Over five years, the rate of loss slows to only four pounds, instead of 20 or so in the static model.
This could still confer health benefits, if you were on the cusp of becoming diabetic, but again, this is still based on one of the most optimistic models of the impact of a soda tax. Other tax models based on static weight loss have estimates of daily weight loss as low as four calories per day, compared to Smith's 40 calories. (There are all sorts of other contestable modeling factors at play in calculating the impact sin taxes, which for the sake of simplicity, we'll ignore.)
To be fair to the USDA's Smith, he is a co-author of the Lin et al. USDA paper, which recalculates a soda tax in light of dynamic modeling.
First, they look at the results of using a static model for a 20 per cent beverage tax and find a linear weight loss of 3.5 pounds at the end of year one and 35 pounds after a decade. But when they switch to dynamic modeling this changes to a two-pound loss at the end of the first year and about four pounds by year five. Their conclusions are almost identical to Hall et al.
Lin et al. also performed a fun calculation, whereby all sugar-sweetened drinks (soda, energy, sport and fruit drinks) were completely replaced by calorie free drinks. The weight loss is substantial: 9.5 pounds for the first year, 15.8 pounds by year five.
There are two notable things about this conclusion. The first is that soda tax models based on static modeling predict greater weight loss by year five than completely eliminating sugared drinks from one's diet.
The second is a related and, as they note, counter-intuitive point: there is a complementary relationship between regular and diet drinks at the household purchasing level where, if the price on sugared drinks is raised, there is also a negative impact on purchases of diet drinks.
There is much more to the new USDA study than just dynamic versus static modeling. Lin et al. combine two datasets – the National Health and Nutrition Examination Survey (NHANES) with Nielsen Homescan purchasing data to aggregate beverage consumption into eight categories, and then average consumption out for low and high-income adults and children.
There are some interesting findings here; for instance, high-income kids drink more sugary drinks per gram than low-income children, while low-income children drink more juices and much more whole milk than high-income kids. As the price of sugar-sweetened drinks rises according to Lin et al.'s modeling, low-income households increasingly substitute juice and whole milk, thus negating the calorie loss.
The USDA study wrestles with the complexities of modeling a tax in a much more thorough and rigorous fashion than the proposals that have gotten the lion's share of the media coverage.
Nevertheless, Lin et al.'s conclusions are still built on the premise that there is a strong relationship between soda consumption and obesity in aggregate terms while, at the same time failing to address the ability of a tax to change the behavior of specific over-consumers, the teens, for instance, who chug the equivalent of 7 to 16 cans of soda a day (and who may also be super consumers of other high calorie foods and super-sedentary). The whole point of sin taxes is that they somehow reach those kinds of people and "encourage" them to change their behavior while minimally inconveniencing everyone else.
Still, these two studies are a wake-up call on the effortless promises of public health experts to solve the obesity crisis through taxation. Unfortunately, they also demonstrate why it's just so darned hard to lose weight and keep it off.
