I am a PhD economist with research interests in health, environmental, and behavioral economics. I will be joining RAND Corporation as an Associate Economist in June 2022. My research relates to the behavioral and environmental determinants of health outcomes. I am especially interested in how individuals form beliefs based on incomplete or noisy information in health and environmental contexts with uncertain individual prospects.


Job Market Paper

  • Learning About Subjective Uncertainty: Overinference from Observable Characteristics in Disaggregated Data

    Individuals facing uncertainty frequently use information at varying levels of (dis)aggregation about others' realized outcomes in similar environments to form subjective beliefs about their own prospects. The use of disaggregated information introduces another level of subjectivity: the individual's beliefs about the informativeness of signals based on the category to which they pertain. In this paper, I experimentally study individual belief updating in contexts where outcomes do not meaningfully differ across categories. I find consistent evidence of a bias: individuals incorrectly assume that information about someone in a particular category is more informative about the prospects of others in the same category than about others in different categories, even when they directly observe the underlying process that assigns individual prospects. As a result, when they receive noisy disaggregated information, their posterior frequently features exaggerated differences across categories, especially when the information reinforces a preexisting misperception in their prior. When these incorrect beliefs pertain to their own category, individuals subsequently acting on these beliefs take on a different level of risk than their risk preferences imply they would if their beliefs had been correct. These results suggest that providers of information that may influence individuals' risky behaviors should carefully select a level of (dis)aggregation which balances personal relevance with statistical precision.

Open all Close all


  • Lifestyle and Mental Health Disruptions During COVID-19 (with Osea Giuntella, Silvia Saccardo, and Sally Sadoff), Proceedings of the National Academy of Sciences, 2021


    Using a longitudinal dataset linking biometric and survey data from several cohorts of young adults before and during the pandemic (N=682), we document large disruptions to physical activity, sleep, time use, and mental health. At the onset of the pandemic, average steps decline from 10,000 to 4,600 steps per day, sleep increases by 25-30 minutes per night, time spent socializing declines by over half to less than 30 minutes, and screen time more than doubles to over 5 hours per day. Over the course of the pandemic from March to July 2020, the proportion of participants at risk of clinical depression ranges from 46% to 61%, up to a 90 percent increase in depression rates compared to the same population just prior to the pandemic. Our analyses suggest that disruption to physical activity is a leading risk factor for depression during the pandemic. However, restoration of those habits through a short-term intervention does not meaningfully improve mental well-being.

  • Pandemics, Economic Freedom, and Institutional Tradeoffs (with Vincent Geloso and Ilia Murtazashvili), European Journal of Law and Economics, 2021


    We argue that institutions are bundles that involve trade-offs in the government’s ability to provide public goods that affect public health. We hypothesize that the institutions underlying economic freedom affect the mix of diseases by reducing diseases of poverty relative to diseases of commerce (those associated with free movement of people, such as smallpox or COVID-19). We focus on smallpox and typhoid fever in the late nineteenth century and early twentieth century in order to build on recent work that make arguments similar to ours, especially the framework Werner Troesken sets forth in The Pox of Liberty. Our evidence shows that economic freedom, in multiple periods of time and settings prior to the eradication of smallpox in the second half of the twentieth century, reduced typhoid mortality but had no effect on smallpox deaths. The implication for COVID-19 is that the trade-off between fighting the pandemic and preserving economic freedom may not be too severe in the short run. However, in the long run, the wealth benefits from economic freedom are likely to be crucial in reducing vulnerability to diseases of commerce primarily from their impact on comorbidities (such as diabetes and heart disease). Thus, economic freedom is on balance good for public health, which suggests that it, while requiring trade-offs, might be the best institutional bundle for dealing with pandemics.

Open all Close all

Working Papers

  • The Regressive Costs of Drinking Water Contaminant Avoidance

    Up to 45 million Americans in a given year are potentially exposed to contaminated drinking water, increasing their risk of adverse health outcomes. Existing literature has demonstrated that individuals respond to drinking water quality violations by increasing their purchases of bottled water and filtration avoidance, thereby avoiding exposure to contaminants. This paper demonstrates that poorer households, for whom the costs of avoidance comprise a greater share of disposable income, bear disproportionate costs of water quality violations in the United States. Following a health-based water quality violation, poor households’ expenditure on nutritious grocery products in a nationally representative panel differentially decreases by approximately $7 per month. This is associated with a decrease of about 1,500 calories per household member per month, placing these individuals at a higher risk of food insecurity. This finding suggests that the indirect costs of drinking water contamination through economic channels exacerbate health disparities associated with poverty.

  • Water Availability and Heat-Related Mortality: Evidence from South Africa

    Rising global surface temperatures threaten to reduce precipitation and evaporate surface freshwater in areas already experiencing water stress. In this paper, I demonstrate that higher upstream water availability significantly reduces the slope of the temperature-mortality relationship during the hot season. This suggests investment in water infrastructure is an effective community-level adaptation to climate change, especially where the status quo of water access is relatively poor. As an example of such investment, I show a transnational water transfer project both increased water availability and reduced hot-season mortality in receiving districts.

  • Drinking Water Contamination and COVID-19 Mortality in the United States

    Over 185,000 deaths have been attributed to the COVID-19 pandemic in the United States as of September 2020. There is growing evidence that the composition of these deaths reflects multiple long-standing health disparities, including environmental quality. In this paper, a county-day level panel of confirmed COVID-19 case and death counts, water quality violations, and demographic variables is constructed to estimate the association between risk of exposure to drinking water contaminants and the COVID-19 case fatality rate (CFR). Counties with more recent violations among major community water systems than average (“treated”) are matched to “control” counties on key demographic and environmental variables using coarsened exact matching (CEM). Three categories of water quality violations are considered: acute health-based violations, which pose an immediate health threat to exposed individuals; health-based violations involving contaminants shown in prior literature to increase the risk of cardiovascular disease (lead, arsenic, cadmium, and copper); and all health-based violations regardless of type. The county-level COVID-19 CFR is significantly associated with acute and cardiovascular-associated health-based violations. On average, the CFR is about 18% higher (0.48 percentage points; p < 0.01) in counties more affected by acute violations than average and about 15% higher (0.42 percentage points; p = 0.037) in counties more affected by cardiovascular-associated violations. There is suggestive evidence of a linear association between the “dose” of violation exposure (the sum of the estimated percentages of the population affected by each respective violation) and the CFR.

Open all Close all

Works In Progress

  • Heat Exposure and Dietary Choices

    Using a panel of American household grocery purchases and daily weather variation, I find that households' revealed food and drink preferences respond to the degree of summer heat they experience. However, the nature of this response is significantly different for households above and below 150 percent of the federal poverty line respectively. During months in which a county's average maximum temperature exceeds its historical average by 2 standard deviations or more, higher-income households purchase more dairy and ready-to-eat deli products, resulting in a higher percentage of observed calories from fat and lowering overall nutritional quality. Lower-income households in the same county during the same month spend more on bottled water, carbonated soft drinks, and alcoholic beverages, crowding out other categories of expenditure as budget constraints bind and thus increasing the risk of nutritional deficiencies. These results suggest that as heat waves become more frequent and more intense as a result of climate change, nutritional outcomes will worsen, and the underlying mechanisms and policy implications will differ by household income. To explore one potential mechanism, I use a national census tract-level measure of lead exposure risk to show that beverage expenditures increase and non-beverage expenditures decrease more sharply in high-risk areas. Thus lower-income households in areas with lower-quality infrastructure are especially susceptible to adverse nutritional consequences of heat and could benefit from targeted mitigation policies.

  • Gender-Specific Shocks & Household Bargaining Power: A Machine Learning Approach to Scanner Data (with Rania Gihleb and Osea Giuntella)

    We examine the effect of several gender-specific labor market shocks documented in the literature on the consumption of heterosexual married couples in the United States. Using machine learning and text analysis techniques, we construct a score of relative gender preference at the product UPC level, culminating in an overall score of each household’s consumption. We find that within households, negative shocks to male labor demand increase the relative share of female-preferred goods, and vice versa, suggesting women gain intra-household bargaining power following these shocks. These effects are mirrored in consumption of children’s goods by gender, suggesting an improved bargaining position for the mother proportionally benefits daughters.

Open all Close all