Sugar Sweetened Beverage Taxes: Demand and Supply Side Effects

Taxation has long provided policy makers a tool to influence consumer behavior. This policy tool has been particularly popular and effective at reducing consumption of so-called “vices” such as alcohol and cigarettes. Several jurisdictions have imposed consumption taxes on sugar sweetened beverages (SSB) in hopes of reducing obesity rates and its effects on health, worker productivity and healthcare costs. Studying the effect of SSB taxes on consumption patterns (and, ultimately, on health status) is of obvious public policy interest. It provides a setting to study if and to what extent consumers undertake tax-avoidance strategies. For examples, consumers may stock pile soon-to-be-taxed items, move to non-taxed items (e.g., 100% juice or artificially sweetened beverages(ASB)), switch to cheaper brands (e.g., store brands), move to low-price-per-ounce or multipack formats, or cross-border shop in lower-taxed/tax-free jurisdictions. The same setting can also be used to study supply-side responses which may help firms mitigate these policies’ intended consequences. More specifically, one can study how sellers set prices across both taxed and untaxed goods, across different formats as well as across taxed and non- taxed jurisdictions. In this paper, we estimate demand and supply-side responses to a sugar-sweetened beverage (SSB) tax that took effect July 1, 2017 and covered the city of Oakland California. The Oakland tax has several interesting features, from both pricing and identification perspectives, including: (i) its per-unit (i.e., per ounce of liquid), rather than ad valorem, nature, and (ii) its invariance to sugar content per ounce. Our made-to-order geo-coded Neilson data provide weekly sales and price information, along with detailed product characteristics, on the near universe of beverages sales – all at the UPC level. The data include identical information on sales and prices within a 2-mile radius around the taxed area, as well as for the city of Sacramento, California. Information on sales and prices within the 2-mile radius around Oakland allow us to consider cross-border shopping effects while Sacramento serves as a control. The model we estimate allows for differential effects of the tax over time. This is particularly important in our setting where the tax imposed was announced prior to being put into effect (allowing consumers to stock-pile soon-to-be-taxed goods) and where short and long-term effects may be different. Our framework also explicitly consider the endogeneity of tax pass-through using an instrumental variable approach.