Off-campus UMass Amherst users: To download campus access dissertations, please use the following link to log into our proxy server with your UMass Amherst user name and password.

Non-UMass Amherst users: Please talk to your librarian about requesting this dissertation through interlibrary loan.

Dissertations that have an embargo placed on them will not be available to anyone until the embargo expires.

Author ORCID Identifier



Open Access Dissertation

Document Type


Degree Name

Doctor of Philosophy (PhD)

Degree Program

Resource Economics

Year Degree Awarded


Month Degree Awarded


First Advisor

Christian Rojas

Second Advisor

Emily Yucai Wang

Third Advisor

Arindrajit Dube

Subject Categories

Industrial Organization


This dissertation empirically explores the economics of advertising, sodium intake, and sweetened beverage taxes in food and beverages industries in the United States. I use recent advances in econometrics and statistics and provide causally interpretable results to deepen our understanding of how individuals respond to changes in advertising and public health policies in consumer packaged good industries.

In the first chapter, titled “Is traditional advertising effective? New evidence from mass-produced lager beer in the United States,” I present new evidence regarding the effectiveness of traditional advertising on product demand for mass-produced lager beer in the United States using data on weekly sales and advertising for 122 brands in 7,906 stores from 2011 to 2015. While the existing literature has examined between-industry heterogeneity or the effect of TV advertisements alone, I report within-industry heterogeneity in own advertising effectiveness and explore heterogeneity in all traditional advertising channels and among demographic groups (as opposed to pooled adults). The empirical exercises are based on a comprehensive sample of products and a credible research design. The type of within-industry heterogeneity I focus on is whether advertising effectiveness differs across light and non-light beer segments. In the baseline specification, the long- run elasticity of demand with respect to own traditional advertising is 0.030 for light and 0.015 for non-light mass-produced lager beer. This finding implies that doubling advertising expenditures would result in a 3.42 percent average revenue in the light segment but only a 1.66 percent increase in the non-light segment. I also find that print and outdoor advertisements are important contributors to light beer sales and to the differential in advertising effectiveness across the two segments. Surprisingly, TV advertising accounts for only 33 percent of the effectiveness differential between light and non-light beer. Finally, I find that light beer advertisements are more effective among a specific set of adults: non-ethnic, white, millennial, and urban dwellers.

In the second chapter, titled “Sodium intake and reformulation in the United States: Evidence from scanner data,” co-authored with Christian Rojas, we construct detailed, barcode-level information on the near-universe of packaged food products to assess the impacts of the 2009 National Salt Reduction Initiative (NSRI), a voluntary pledge to reduce sodium content in food products by 2014. First, we isolate and quantify the role that product reformulation, vis-a`-vis consumer purchasing behavior, played in recently documented declines in sodium intake in the United States. Second, we evaluate whether the NSRI played a causal role in reformulation efforts to reduce sodium content. We find that reformulation efforts alone would have contributed to a 53 percent reduction in sodium intake between 2007 and 2015; however, changes in consumer shopping behavior negated almost the entire improvement, permitting only a 4.73 percent net decrease in sodium intake. Using a kink regression discontinuity design, we show that the NSRI played a causal role in the observed reformulation efforts: On average, products with sodium content above NSRI targets decreased excess sodium by approximately 40 percent. We also document nutritional inequality (as it pertains to sodium intake) across socioeconomic and demographic groups and discuss the implications of our findings for effective diet improvement policies.

In the third chapter, titled “How much do people consume after a soda tax? Evidence from the Philadelphia, PA, tax event,” I compile a comprehensive household-level sample and show empirical evidence of changes in price and barcoded beverage purchases as a result of the Philadelphia, PA, sweetened beverage tax. The longitudinal panel includes 24,793 households and their purchases of 3,729 barcoded beverages across 11,994 stores in the City of Philadelphia and neighboring localities from 2006 to 2017. To construct a counterfactual Philadelphia, I use both traditional difference-in-differences models and frontier causal machine learning methods, as in Athey et al. (2021). Results show that 64 percent of the tax is passed on to consumers (as opposed to the 97 percent pass-through rate reported in Seiler, Tuchman and Yao, 2021) and that households reduced their consumption by 21 percent–39 percent, depending on the estimator. I also find evidence that consumers avoid the tax via cross-border shopping, although the magnitude is one-third of that reported by Seiler, Tuchman and Yao (2021): 8 vs. 24 percentage points.


Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.