### Artificial sweeteners: behavior matters

9/20/2013 05:55:00 PM
Tweetable

NBC reports that artificial sweeteners are bad for you. The report is based on a review published in the September issue of Trends in Endocrinology and Metabolism.

The basic puzzle here, which doctors and researchers have been aware of for some time now, is that people who switch to artificial sweeteners instead of sugar-sweetened beverages (ie, diet instead of regular) don't appear to loose weight any faster, and don't appear to have lower incidences of health problems like Type 2 diabetes. Given that they don't loose weight, the latter part isn't surprising--it's well established that Type 2 diabetes is an obesity-related disorder, and we wouldn't expect much improvement without weight loss. But the big puzzler is why replacing beverages with non-caloric sweeteners doesn't reduce weight loss.

The review cites two kinds of studies: "prospective studies" where they just compare drinkers of sugary sodas to drinkers of non-sugary sodas over time, and "intervention studies" where they actually assign people to a sugar-sweetened beverage group and an artificially sweetened beverage group. Neither finds increased weight loss in the artificial sweetener groups. The intervention studies are better than the prospective studies because we know that the group assignment was exogenous, which means we can estimate a causal effect. That is, in the prospective studies we might have hypothesized that it's the less health individuals who switch to artificially sweetened beverages, in a bid to become healthier, thereby biasing the estimates. This hypothesis is proven false in the intervention studies.

However, NBC erroneously reports that the intervention studies controlled for the number of calories consumed. This appears to be false. The intervention studies were not randomized-controlled trials, and the participants in the two groups were allowed to adjust their diets in other ways, including by increasing the number of calories consumed. Such a study design cannot rule out the multicolinearity problem, so the results are consistent with the hypothesis that artificial sweeteners trigger a behavioral response of eating more, higher-calorie foods (though, it would still be fair to argue that the estimated effect is "causal" because artificial sweeteners caused the behavioral response). This is the conclusion that the journal review argues.

Why am I blogging this result? Because it has Health Economics consequences (in general, whenever a result shows that behavioral responses matter, economics is involved). The literature shows that consumers treat sweet calories as complementary to artificial sweeteners, and do not treat the latter as a substitute for the former. What is important, though, is that while sugar is complementary to artificial sweeteners, sweeteners are clearly not complementary to sugar. That implies that the slutsky matrix is not symmetric. If you do the math, it turns out that these preferences cannot be economically rational; though, all is not lost: they could still satisfy the weak axiom of revealed preference, homogeneity of degree zero, and Walras' law [1].

So there you have it. Nutritionists may have just proved that the standard economic assumption of rational behavior is wrong, at least in this instance.

[1] Just to clarify my thinking, if $l$ is non-caloric sweetener and $k$ is sweet-tasting calories in other foods, then the review article argues the substitution effect is negative so that $s_{k,l}<0$ in the slutsky matrix--more non-caloric sweetener causes people to consume more sweet calories elsewhere. However, I think we can clearly assume that $s_{l,k}\geq 0$, because it is not the case that consuming more sweet-tasting calories causes people to want more artificial sweetener. At the very least, the biological mechanism presented in Trends in Endocrinology and Metabolism clearly suggests that sweeteners have an effect on appetite for caloric foods and not the otherway around, so that it is almost certainly true that $s_{l,k}\neq s_{k,l}$. This violates the rationality assumption and leads to intransitive preferences--see Mas-Colell, Whinston, and Green section 3.H.

Pro-tip to nefarious food producers: if this is true, you should be able to goose your profits up by adding artificial sweeteners to foods. Intransitive preferences are exploitable. Just sayin'.
richard paterson 10/03/2013 06:03:00 AM
I was extremely pleased to find this website. I wanted to thank you for this good knowledge and I definitely enjoying every single small bit of it and I am looking forward to check out new stuff you post.
Dr. Jay Potter 11/20/2013 11:20:00 AM
With all the controversy around artifical sweetners, I've been looking for a product with the fewest "cons". It seems that all the substitutes have a significant aftertaste which turns most people off. There are a couple things that seem important. (1) reduce the amount of chemicals used and (2) minimized the aftertaste. I've finally found a product that does both these things and its pretty amazing. FireAngel sweetner is a relatively new product that cut way back on the amounts used yet was able to keep the sweetness. The taste is very close to sugur without the metalic or bitter aftertaste. You mix a pre-measured amount with water and 1-2 drops for a cup of coffee.