Thursday, 2 February 2017

Sugar taxes in the real world

Dr Jayson Lusk has a review paper in the current issue of the International Journal of Obesity which is well worth reading if you have access. It is the most succinct explanation of why it is folly to use economic levers in an attempt to control obesity. A few quotes below highlight some of the main points.

On price elasticity and hurting the poor...

A number of public health professionals have advocated for various forms of taxes on fat, sugar, or other ―unhealthy ingredients (e.g., 18, 19), and a variety of locales have implemented such taxes. Presuming a downward sloping demand curve, an increase in the price of sugar or fat will likely lower consumption of these items, but the key question is: by how much? It is inappropriate to focus only on the own-price elasticity of demand. There are substitutes for, say, taxed sodas, such as milk, juice, or beer. The availability of caloric substitutes will lower the weight impacts of a tax beyond that implied by the own-price elasticity (20). That is one reason why some analyses show that only across-the-board food taxes will significantly affect weight (21).

A problem with food taxes is that they are regressive because the poor spend a higher proportion of their income on food than the rich. Because of concerns over regressivity of food taxes, others have proposed subsidies on health foods like fruits and vegetables. While such subsidies can improve uptake of such foods (22), questions about cost-effectiveness remain as do concerns about equity. Muller et al. (23) show that both unhealthy food taxes and healthy food subsidies favor higher income consumers. They find that unhealthy food taxes increase prices paid more for low than higher income women and that healthy food subsidies reduce the prices paid more for higher than lower income women.

On why computer models fail to predict the impact of sin taxes on obesity correctly...

... An issue often overlooked in economic analyses of ―fat taxes relates to metabolism. Metabolically, one must translate changes in consumption caused by a tax to changes in weight to fully appreciate the impacts of taxes or subsidies. The most common approach taken in the economic literature is to apply some version of the simply linear rule that ―3500kcal = 1 lb. However, as shown by Thomas et al. (26), the rule overstates actual weight losses resulting from a reduction in caloric consumption. These and other reasons help explain the seeming paradox that simulation models show sugared soda taxes to have pronounced effects on obesity prevalence, when actual variation in soda taxes across locations seems to have had no impact on body weight in the ―real world (27-30).

On why taxes on food and drink reduce consumer welfare...

... More fundamentally, it is useful to return to the conceptual basis being used to assert that sugar or fat taxes increase consumers‘ welfare. Taxing food or soda is analogous to reducing consumers‘ real income, which harms the consumer under the standard economic modeling framework (i.e., consumers do not like to pay higher prices). Even if taxes induce a change in body weight, presumably many consumers already consider health impacts when they choose what to eat and drink. If consumers suffer from an aforementioned behavioral bias or lack of information, it is possible to construct a model in which a soda or fat tax benefits the consumer. For example Lusk and Schroeter (34) showed that if consumers ignore future health impacts at the time of purchase, a tax can increase consumers‘ long-term welfare but only in the limited case when prices are very high and where the only consumption taking place is because ultimate health impacts are ignored.

It is true that taxes are required if we are to have public roads, public schools, and national defense. However, the typical economic approach is to identify taxes that create the least deadweight loss, with additional considerations given to tax progressivity. Even if revenue from ―fat taxes could be directed toward health education programs, it is far from clear that these sorts of taxes are the most efficient way to fund such education. Moreover, one would need to show how the deadweight loss of the tax is offset by the benefits of extra information; however, evidence on the monetary benefits of such proposed education programs is typically assumed rather than carefully estimated.

He concludes as follows:

This article presented a somewhat pessimistic view on the ability of government policy to substantively influence obesity prevalence. Obesity is a complicated and multifaceted issue. So too are the effects of anti-obesity policies. One response is to argue for an all-out ―war on obesity. It probably true that government policy mandating what farms grow, restricting the supply and type of food to consumers, and controlling prices, offerings, and advertisements by food manufacturers could reduce obesity prevalence. But, is this the type of coercive society in which we‘d like to live? Society faces very real tradeoffs between economic freedom, technological progress, and obesity prevalence. These sorts of tradeoffs are the unfortunate, but they reflect very real constraints to effective economic policy making.

The same issue of the journal carries an article by Martin Binks and Shao-Hua Chin which is also well worth reading. In particular, I was pleased to see that they have noticed the tendency of sugar tax campaigners to focus on changes in consumption when their predictions about obesity fail to materialise:

...clarity is needed in defining a priori what is or is not a reasonable and or desirable impact of obesity policy interventions. A most recent example can be found in the various regional efforts to reduce sugar sweetened beverage (SSB) consumption as a way of impacting obesity in France, Brazil, Mexico and more recently in Berkley California. In each case the stated primary goal was to impact obesity.

However, as the results were presented in the literature and summarized in a recent meta-analysis it is evident that there was no significant impact on BMI in terms of the actual population-based studies. Furthermore, simulations based on available data identify only a minimal potential impact (fractions of BMI points).

As a result of this failure to impact BMI, the dialogue has shifted towards highlighting reductions in consumption as an indicator of success. While the broader implications of this assumption are beyond the scope of this paper, it is important to note that

1) the goal was to impact obesity, this did not happen even over reasonable time periods
2) By focusing towards a secondary endpoint as a success, we may miss a real opportunity to uncover why consumption dropped yet BMI remained unchanged.

In short, it is necessary to consider why this primary endpoint was not met, and develop new approaches that will impact obesity, as opposed to defending those that have proven ineffective. The challenge faced is one involving the willingness to objectively evaluate the science (even when it opposes strongly held ideological beliefs), and to pursue comprehensive evidence-based solutions to obesity and the policies that may support them.

No comments: