Reducing Obesity by Taxing Soft Drinks: Tax Salience and Firms’ Strategic Responses

Abstract

In the fight against obesity, governments have pursued various policies to increase soda prices in the hope of discouraging consumption. Previous studies examining the efficacy of taxing sodas have generalized sales taxes to other taxes and assumed that imposing a tax is as salient as imposing a direct price hike. In this research, the authors recognize that most consumers do not pay much attention to how taxes differ in application. Thus, the authors disassociate the effects of sales taxes from the effects of direct price increases on soda demand, estimate pass-through rates of excise taxes in the soda market, and compare the efficacy of the two types of tax at three tax levels. They find that, for all three tax levels, imposing an excise tax on sodas drives much steeper declines in overall demand than a sales tax does. The findings suggest that failing to distinguish between tax types leads to a biased estimation of the effect of taxes on soda consumption. This research provides important implications to help policy makers adopt the most applicable tax instrument to battle obesity.

The impact of the Oakland SSB tax on prices and volume sold: A study of intended and unintended consequences

“…3 As we discuss later, the tax did not apply to SSBs that have 25 calories or fewer per 12 ounces nor chocolate milk. 4 Although not the goal here, several papers specifically examine tax pass-through across different types of taxes (ad valorem vs. specific) and market structures in both theoretical and empirical settings (Allcott et al, 2019;Anderson et al, 2001;Bonnet & Réquillart, 2013;Delipalla & Keen, 1992;Weyl & Fabinger, 2013;Zheng et al, 2019). These papers provide guidance not only on the conditions which yield low versus high tax pass-though (including over-shifting) but also on the design of optimal tax policies.…”

Simulating international tax designs on sugar-sweetened beverages in Mexico

“…Following previous structural models on SSB taxes [ 26 , 32 , 33 ], we implement a random-coefficients logit demand model [ 34 , 35 ]. The indirect utility function for household i ∈(1,…, H ) when consuming beverage j from the choice set j ∈(0,…, J ) at market t ∈(1,…, T ) is: where Sugar j is beverage j ’s sugar content in terms of grams per liter, and p jt is beverage j ’s price per liter at market t .…”

“…For sensitivity analyses of the demand model, we use 2,000 and 5,000 draws of households’ characteristics for Eq ( 3 ) to test for the precision of the estimated coefficients. Moreover, we also re-run the demand model including seasonal dummies following the equivalent approach in the studies by Zheng et al [ 33 ] and Liu et al [ 41 ]. Finally, we extend the set of instruments by including average monthly prices of beverage j in cities that are not part of Nielsen CPS and correspond to urban locations from which INEGI [ 31 ] collects price information to calculate the consumer price index.…”

“…We report the tax effect as the average difference in the outcome of interest across tax designs but are unable to provide uncertainty ranges of the estimated difference (e.g., standard errors). This approach is standard in simulations based on structural models of demand and supply as the one we used in this study (for example, see Zheng, Huang, and Ross [ 33 ], Liu, Lopez, and Zhu [ 26 ], and Bonnet and Requillart [ 25 ]). Obtaining uncertainty ranges in our estimates through methods such as bootstrapping can result in unrepresentative shares by firms given the sampling procedures and, therefore, it can result in non-convergence of the models as well as implausible values.…”

Untitled

“…Following previous structural models on SSB taxes [26,32,33], we implement a random-coefficients logit demand model [34,35]. The indirect utility function for household i2(1,. .…”