Berkeley Soda Tax

There have been a number of news stories about this new paper in the American Journal of Public Health that studied the impact of Berkeley's new sugar-sweetened beverage tax (aka "soda tax").  The authors surveyed residents of Berkeley before and after the implementation of the tax and asked about beverage consumption.  They also surveyed people in Oakland and San Francisco (who were presumably not affected by the tax) before and after the tax.  By comparing these two groups before and after, the authors can calculate something like a difference-in-difference estimate of the impact of the tax.

What did they find?

Consumption of SSBs decreased 21% in Berkeley and increased 4% in comparison cities (P = .046). Water consumption increased more in Berkeley (+63%) than in comparison cities (+19%; P < .01).

The results have been largely heralded as indicating that the taxes "work".  Here is a bit from the organization Healthy Food America:

this study provides another key piece of evidence that sugary drink taxes work, not just to raise revenue for important community priorities, but also to reduce consumption and shift sales from an unhealthy product to healthier drinks, such as bottled water,” said Jim Krieger, MD, MPH, executive director of Healthy Food America. “As four other communities consider their own taxes this fall, the Berkeley findings join those from Mexico and elsewhere to show that sugary drink taxes have great benefits – especially in low income communities.”

We shouldn't be too surprised that a tax reduces consumption - more confirmation that the demand curve slopes downward.  Yay Econ 101!  The real question about soda taxes hasn't been "whether" but "how much" consumption falls when prices rise.  

First, I'll touch on some conceptual issue related to the interpretation of the study then offer a few thoughts on the study methods.  

As indicated, I've seen numerous studies showing that this paper "proves" that soda taxes "work."  I'm not sure what "work" means.  There have been scores of studies projecting impacts of soda taxes, and virtually all suggest the taxes will lower soda consumption by some amount (though curiously the evidence is much less clear if you look at studies that have looked at actual sales data before and after taxes).  But the goal isn't to reduce soda consumption for the sake of reducing soda consumption.  The goal, presumably, is to make people's lives better - to reduce obesity or other dietary related diseases.  On this front, the evidence is much less clear that soda taxes will have a substantive effect on body weight.  Moreover, as other studies have shown, we need to be cognizant of what people eat and drink instead when they substitute away from higher-priced soda, and many of these options could have adverse health impacts as well.

The largest conceptual issue in all this is whether a tax can actually make people better off.  Yes, a tax can reduce consumption.  But, does that mean people are happier?  Even if people switch to a lower-priced alternative after a tax, it doesn't follow that they're necessarily better off.  After all, consumers could have chosen the lower-priced alternative before the tax if that's what they really preferred.  Thus, the tax causes people to choose a lesser-preferred option.  This is a standard economic result: consumer welfare falls when prices rise.  Here's what I wrote in a piece for the Canadian Journal of Diabetes

More fundamentally, one must ask what conceptual basis is being used to assert that SSB taxes will increase consumers’ welfare? Presumably, some consumers already consider health impacts when they choose what to eat and drink. More generally, taxing food or SSBs is analogous to reducing consumers’ real income, which almost certainly harms the consumers (9). Perhaps consumers suffer from lack of information or other cognitive biases, but even so, Lusk and Schroeter (10) show that only in very limited cases would a tax increase consumers’ long-term welfare. Sugden (11) further points out the philosophical (not to mention political) problems encountered when attempting to base public policy on the presumption of consumers’ behavioural biases. In particular, asserting that someone else consumes “too much” SSBs presumes that the nutrition expert or politician knows better which factors most impact an individual’s ultimate well being than the individual herself or himself. Such paternalism may be justifiable in the case of children or the mentally impaired, but it is less compelling when considering the general population. It is likely the case that excess consumption of SSBs will lead to health problems; however, people care about tasty, satisfying foods and beverages in addition to health. Life is full of difficult tradeoffs, and it is conceptually problematic for a third party to deem another person’s choices wrong or incorrect, given that different people have different preferences, incomes and constraints (assuming that people are making decisions with accurate information about the risks they face). If the argument is that people do not understand the risks of SSBs, then the appropriate policy response is information provision, not a tax.

I went on to tackle the argument that rising public health care costs justify the tax, but I won't belabor the point: showing a tax "work" involves much more than showing that it reduces consumption.  

Now on to the Berkeley study's methods.  Overall, this is a nicely designed study that uses a difference-in-difference approach to try to tease out a causal effect.  My biggest beef with the study is that it relies on consumers' stated consumption behaviors.   The Berkeley tax has been a high profile event, and no doubt many Berkeley residents were aware of the debate and policy change.  It is possible that what we're picking up is some form of social desirability bias: Berkeley residents know they and their neighbors passed a tax on soda, and now here's this researcher asking about soda consumption.  The social pressure is clear: I don't want to be the kind of person who consumes the product everyone else wants to tax.  I don't want to look like some kind of social outcast, so I'm going to hedge and give a lower level of consumption than I actually consume.  

Why didn't the authors do this study using scanner data based on actual grocery sales?  These data are easily attainable from companies like Nielsen and IRI and can't be much more expensive than the surveys the authors conducted.  Even still, there would be other questions about the longevity of the effect, the substitution toward other food and beverages after the tax, the degree of substitution across city boundaries, and so on.

All that said, I'm more than willing to accept the finding that the Berkeley city soda tax caused soda consumption to fall.  The much more difficult question is: are Berkeley residents better off?