UC Berkeley

Throughout life, consumers face constraints that impact all aspects of economic decision-making. Government and agency policies can impose resource constraints, limiting the quantity of a natural resource that a consumer can use or imposing fees that shift individuals away from their preferred consumption levels. Households often face financial constraints, restricting their ability to purchase an optimal consumption basket and preventing families from eating optimal diets. Additionally, social constraints can limit behavior through adherence to social norms, with those who violate these norms suffering disutility due to social pressure or public shame from their peers. In this dissertation, I explore individual and household responses to a range of constraints affecting natural resource use, diet, and mobility patterns.

In the first chapter, I explore the responses of urban water customers to price and nonprice water conservation measures. Growing urban water demand combined with a shifting global climate present urban water districts with an acute need for policy approaches that can induce both long- and short-run conservation behavior. Using quasi-experimental long-run variation in prices and exposure to short-run price and non-price policies during severe drought, I estimate the effectiveness of these policy approaches for urban water conservation. This paper is the first to not only identify the water conservation impact of public shame, but also isolate its effect from those due to moral suasion and price-based measures. By following the universe of single family water customers served by a major water district over time, I compare the impacts of different policies on the same households and avoid common sample selection issues. First, I utilize rich administrative panel data to characterize the evolution of causal price elasticities across multiple stages of drought. Next, I leverage quasi-experimental variation in exposure to fees, moral suasion, and public shame to separately identify consumer responses to each during the drought emergency. Demand models yield causal price elasticity estimates of −1 prior to adaptation that attenuate to −0.4 to −0.7 during and −0.2 to −0.5 post-drought, with high-usage households displaying the greatest responsiveness. While subject to behavioral interventions, top water users display no conservation response to excessive use fees. Moral suasion and public shaming show substantial short-run conservation impacts but display immense sensitivity to emergency messaging and consumer beliefs in crisis. Households called out by name in news stories display sizable reductions in water use even after prior exposure to moral suasion and fees, demonstrating the resilience of public shaming’s conservation effect to crowding out by other concurrent policies. These findings yield important implications for the design of future urban water policies that can balance short and long-run conservation goals.

In the second chapter, I investigate the extent to which Supplemental Nutrition Assistance Program (SNAP) participants utilize food bank resources to smooth expenditures throughout the benefit month. Many recent works have documented the existence of the “SNAP benefit cycle” in which a non-negligible portion of enrolled households exhaust their benefits shortly after receipt. Lack of benefits late in the month carries large consequences, with negative impacts ranging from reduced energy intake and nutritional content, to increased likelihood of hypoglycemia and pregnancy-related emergency room admissions, along with lowered test performance and increased disciplinary action for students. These effects diminish or are otherwise nonexistent for SNAP households that exhibit consumption smoothing. The ability for low-income households to complement resources from federal poverty programs with a local public good has major implications for household welfare and the value of food bank networks. Using the USDA Food Acquisition and Purchasing Survey (FoodAPS) dataset, I provide evidence suggesting differential behavior patterns immediately upon receipt of benefits for SNAP households that utilize food banks. After summarizing prior studies documenting the presence of the SNAP benefit cycle and the expenditure patterns within the FoodAPS sample, I discuss the observed differences in household characteristics and acquisition behavior for those using and forgoing food banks. Next, I present visual evidence of variation in food bank utilization over time among surveyed households and of stark differences in day-zero SNAP benefit utilization for households that visited food banks during the survey period. Following the presentation of visual evidence, I leverage variation in timing since SNAP benefit receipt and food bank use among surveyed households to estimate empirical models of daily food-at-home expenditures. I find that, while typical SNAP households spend $91-92 more on the day of benefit receipt than on a typical day in the final week of the benefit month, households that use food banks spend $50-54 less that day. This decrease in day-zero expenditure by food bank clients is offset by higher average spending on the next six days, indicating that the expenditure smoothing exhibited by these households primarily occurs in the first week of the benefit month. Subsequent analyses exploring alternate timing specifications and payment methods confirm that the observed expenditure patterns are driven by payments out of SNAP benefits, and that the expenditure smoothing observed by food bank households is largely isolated to the first week following benefit issuance.

In the the third chapter, co-authored with Sofia Villas-Boas, J. Miguel Villas-Boas, and Vasco Villas-Boas, we estimate the mobility responses prior and in response to COVID -19 stay-at-home mandates. The recent spread of COVID-19 across the U.S. led to concerted efforts by states to “flatten the curve” through the adoption of stay-at-home mandates that encouraged individuals to reduce travel and maintain social distance. Combining data on changes in travel activity and human encounter rates with state policy adoption timing, we first characterize the overall changes in mobility patterns that accompanied COVID-19’s spread. We find evidence of dramatic nationwide declines in mobility and human encounters prior to adoption of any statewide mandates. Then, using difference-in-differences along with weighted and unweighted event study methods, we isolate the portion of those reductions directly attributable to statewide mandates. Once states adopt a mandate, we estimate further mandate-induced declines between 2.1 and 7.0 percentage points relative to pre-COVID-19 baseline levels. While residents of mandate states soon returned to prior business visitation patterns, the impacts on distances traveled and human encounter rates persisted throughout the observed mandate periods. Our estimates of early mobility reductions and the responses to statewide stay-at-home policies convey important policy implications for the persistence of mobility behavior changes and states’ future re-openings.