Why a Soda Tax Is Unlikely to Work: Yours Truly in the Washington Post

Several European countries also tax sugary drinks, but, as with Mexico, it’s tough to tease out whether, or how much, the taxes affect consumption. Marc Bellemare, assistant professor in the department of applied economics at the University of Minnesota, took a close look at soda sales data (from Euromonitor International, which tracks sales of an astonishing array of food items around the world). He concluded that, depending on how you parse the data, you could claim anything from no impact to about a 2.6 percent decrease.

“In academic parlance, the results are not ‘robust,'” Bellemare says. …

The lack of a clear correlation doesn’t mean sugary drinks aren’t implicated in obesity and disease. … We cannot be sure, not by a long shot, that a tax on soda will result in improved public health.

From a longer column by Tamar Haspel in the Washington Post last week. A few weeks ago, Tamar got in touch with me telling me she had gotten her hands on Euromonitor data for soda sales (supplemented with information about soda taxes in Europe), and she asked me whether I could figure out whether taxes appeared to cause any decrease in soda sales.

The data lent themselves to a nice difference-in-differences analysis, given the variation over time and across countries in the adoption (and, in one instance, disadoption) of sugar taxes. If you buy the parallel trends assumption, at best, I found a 2.6 percent decrease in the logarithm of liters of soda sold per capita; at worst, I found no statistically significant relationship between taxes and soda. When looking at the level (i.e., liters per capita, instead of the log of liters per capita), I also found no relationship. This is what I wrote to Tamar:

In the best-case scenario, I find that taxes decrease sales of soda per capita very, very slightly (on the order of 2.6 percent per year, or about four and half 12-oz cans of soda per capita, which is less than many Americans drink per week). In the worst-case scenario, I find no statistical significance, meaning that for all intents and purposes, the effect is zero.

In academic parlance, the effects are not “robust,” and so I would not stake any money on such policies having an actual effect in practice. This is especially so given that even when significant, the effect is still very small. The reason why I don’t have a unique answer for you is that there are several ways of looking at the problem, and I have accounted for all possible specifications of the equation of interest, given the data you sent me.

The fact that those taxes have no (economically and, often, statistically) significant effect is unsurprising. One, even in the US–which consumes way, way more soda than Europe in per capita terms according to the Euromonitor data–soda represents a minuscule share of the average consumer’s budget. Two, from casual empiricism, the demand for soda strikes me as relatively inelastic; there are few substitutes for sweet, fizzy drinks: club soda does not contain any sugar, fruit juices aren’t fizzy, and many people cannot stand the taste of diet sodas.

(Tamar spoke to a number of other economists for her article. In his post on the topic, Jayson generously referred to those of us quoted in the article as “a slew of top food and agricultural economists,” which I imagine is what it feels like if George Clooney tells you that you’re handsome.)

So, notwithstanding what some people in the public health community seem to to take as an article of faith, taxing soda is unlikely to help with this country’s (or any other country’s, for that matter, given my European estimates) love affair with obesity, though it is certainly likely to contribute to the revenues of governments that levy a tax on soda.

In that sense, it is absolutely no surprise that some politicians seem to love soda taxes:

  1. Cater to your base by looking like you’re doing something for public health (“Think of the children!“)
  2. Make money
  3. ???
  4. Profit!

What’s there not to love?

Generally speaking, much like the obsession with local and organic foods as instruments of public policy, the use of soda taxes often feels to me as a tactic used by some among the wealthy and educated, who are much more likely to abstain from drinking soda,* to wage a proxy culture war on the poor and uneducated, who are more likely to consume soda–but that’s a topic for another, future blog post.

* This describes me, too. The difference is that, based on the evidence at hand, I just don’t believe there is an economic case for soda taxes, though there certainly appears to be fiscal and political cases for them. And with all of that said, I am a firm believer in the claim that sugar is the root of all evil when it comes to obesity!

No related content found.

20 comments

  2. Marc F. Bellemare

    I wish I knew what went on in his mind as well… Beyond that, no comment. That said, thanks for passing this along. I had no idea he’d said that. For me, it’s more evidence of the culture war I mentioned in my post.

  3. Pingback: Assorted links
  4. Anand Hattiangadi

    People get addicted to soda, so it seems like there is a possible difference between short term elasticity and a longer term elasticity. It’s also unclear to me that cost is the relevant margin in all societies. A society where coke is a luxury is quite different from one in which its a water substitute.

  5. Pingback: Just Financial News / Assorted links
  6. BruceB

    Does this differ from the effect of tobacco taxes on tobacco consumption? Alcohol taxes on alcohol consumption? Income taxes (or higher marginal rates) on income generation? Is there a Laffer curve for soda taxes?

  7. Val DeLa

    The analysis only looks at the miniscule taxes already in place on sugar soda. Surely there would be significant decreases in soda drinking if taxes were at 25-50% markup on the cost of soda, no?

  9. Pingback: Assorted links | Homines Economici
  10. Arjun

    Even if there are no effects of soda taxes on actual soda consumption, what about the effects of how the tax revenue is used–especially if the taxes actually go toward public health programs and obesity prevention measures? Given good local governance, soda taxes could arguably still be beneficial.

  11. Tom W

    Even if it doesn’t reduce consumption, I have no problem with using such a tax as a revenue raising mechanism. These fats cost us all in everything from healthcare to comfort on planes, so it’s sensible that they pay to cover their own costs.

  12. Foster Boondoggle

    Have you looked at the data from Mexico? From an article last fall:

    Taxation, just like it did for tobacco, is the most effective way to get people to change their behavior,” said Barry Popkin, an economist who teaches global nutrition at the University of North Carolina.

    Popkin is working with researchers from Mexico’s National Institute of Public Health to document the effects of that country’s tax on sugar-sweetened beverages.

    Their preliminary results show that during the first three months of 2014, purchases of sodas and other taxed beverages declined by 10 percent compared to the same time period last year.

    Meanwhile purchases of untaxed drinks, like 100 percent fruit juice and milk, went up 7 percent, and purchases of bottled water went up 13 percent.

    “Essentially people were substituting plain water, the healthiest beverage we could drink other than milk, for the sugary beverages,” Popkin said.

    http://www.kpbs.org/news/2014/nov/28/mexicos-sugary-drink-tax-shows-initial-results/

  16. Aram

    Ok, I know you mocked thinking of the children, but it seems plausible that kids would be affected more by this tax since they have less money. I remember being annoyed in high school when the 75-cent soda vending machine got replaced by one where drinks cost $1. Is there any data on kids’ soda elasticity?

  17. Cody

    You dismiss the comment above raising the possibility of a more substantial tax having a more substantial effect. But you claim in this post is that “A Soda Tax is Unlikely to Work”. And as you admit, you have no data on whether or not a bigger soda tax would work. So how can you justify the central claim of this post?

    To be more accurate, your post should say “Small taxes on soda have a small or zero effect on consumption.” And this claim is hardly surprising. One might take this claim to suggest the exact opposite of the conclusion of your post, that much larger soda taxes are needed.

  18. Tom West

    Well, if we tax soda at $1,000,000 a bottle, then 0 bottles are sold, so we know at *some* point, the tax must work. So the point isn’t whether such taxes ‘work’, it’s whether they’re an appropriate nudge to proper behavior.

  19. Doug

    So, soda taxes do work? I assume it all from which angle you are looking. If I want revenue generation, then Wow, the taxes work great. If I am interested in modifying behavior, then … not so much.

    Great article!