This dissertation consists of three essays in public economics. The three chapters focus on interactions between public and private economic decisions. The first two chapters focus on unemployment insurance (UI) policy in the United States. The third discusses public-private interactions in the education market. The first chapter, a joint work with Arindrajit Dube, Calvin Jahnke, Raymond Kluender, Suresh Naidu, and Michael Stepner estimates the labor supply and spending responses to a large change in UI benefits during the pandemic. We examine the effects of the sudden withdrawal of expanded pandemic unemployment benefits in June 2021 using anonymized bank transaction data for 16,548 individuals receiving UI in April 2021. Comparing the difference in differences between states withdrawing and retaining expanded UI, we find that UI receipt falls 35 p.p. while employment rises by only 4.4 p.p. by early August. Average cumulative UI benefits fall by $1,385 while average cumulative earnings increase by only $93. Heterogeneity by unemployment duration implies that these effects are primarily driven by extensive margin expiration of benefits, rather than intensive margin reductions in the benefit level. The second chapter examines the role of gifts and loans from friends and family during unemployment. These transfers play a largely unstudied informal insurance role in high-income countries, making it difficult to assess their implications for social insurance policy. I present new results on informal insurance paid via person-to-person (P2P) payment platforms using a survey-linked administrative bank transaction dataset covering 130,502 low-income users from the US who were unemployed at least once between July 2019 and September 2020. Event study estimates show average monthly inflows from all P2P platforms increase by $30, or 2% of lost earnings, one month after job loss before returning to baseline over 10 months. Single mothers and the long-term unemployed receive the largest increases, as do those living in high-income areas. I exploit three plausibly exogenous changes to federal pandemic unemployment insurance (UI) policy to estimate that UI benefits crowd out at most $0.04 of informal P2P transfers. Using the social insurance framework introduced in Chetty & Saz (2010), my crowd-out estimates indicate negligible welfare consequences for an additional dollar of benefits. Altogether these results imply that public UI benefits can raise welfare by pooling risk across networks without reducing within-network targeting of informal insurance. The third chapter asks whether public school services fill in gaps left by private school failures. Specifically, it explores what type of schools enter the market and experience an increase in enrollment after reports of abuse by Catholic priests lead to Catholic Schools closures. I use a two-way fixed effects event study method to estimate a change in enrollments and number of different types of schools after a report of priest abuse within the same zip code, school district, or county. I find there are 0.2 fewer Catholic schools and Catholic school enrollment falls by 75 students after six years, which are offset by a 0.2 and 50-student increase in charter school counts and enrollments on average. These increases are unique to charter schools and is not observed in other public or non-Catholic private schools. Altogether, these results suggest that former Catholic schooled families show a preference for charter schools over other public schools, which may be due to the low-cost and similar emphasis on discipline and academic achievement.

In the first chapter of this dissertation, I exploit the Covid-19 pandemic and associated government restrictions as a natural experiment in order to study the resilience of businesses in the United States. I use a border-county identification strategy with data on government restrictions, employment and open small businesses, in order to assess the resilience of small businesses in the United States. In my main results, I find negative impacts of stay-at-home orders on the number of open small merchants. In particular, shutdowns of businesses accelerated 8 weeks after imposition of a stay-at-home order, suggesting that many businesses were only resilient enough to handle adverse conditions for 8 weeks. On average, a county with a stay-at-home order experienced an additional 1.51 percentage points loss in the number of open small businesses, relative to January 2020, 8 weeks later compared to a neighboring county that did not have a stay-at-home order. Firms were quicker to resort to layoffs. On average a county with an active stay at home order in a month experienced an additional 1.19 percentage point loss in employment, relative to January 2020, the following month compared to a neighbor that did not have a stay-at-home order the previous month. My results suggest that in future scenarios where governments consider enacting similar restrictions further aid is needed for businesses in order to help them stay afloat. In particular, more assistance should be delivered to businesses within two months from the enacting of the order. In the second chapter of this dissertation, I study economic spillovers in the context of theCovid-19 associated government restrictions. I use a detailed geolocation dataset to construct data on the number of visitors per-capita between neighboring counties in the early stages of the pandemic, which I use as a proxy for economic spillovers. I employ a similar border-county identification strategy as in the first chapter to identify the causal effect of stay-at-home orders on inter-county movement. Additionally, I provide evidence for an assumption used in chapter one by examining if there are reduced spillovers in county-pairs that lie in the different commute zones. I find that stay-at-home orders caused reductions in inter-county visits in both directions in a county-pair. That is, I find a decrease in travel from the county without a stay-at-home order to the county with one, as well as a decrease in the opposite direction. On average, a county that does not have stay-at-home order will receive 408 fewer weekly visitors from their neighboring county that has a stay-at-home order. I also examine the effect of stay-at-home orders on the ratio of travel between the two directions in order to find evidence of a net spillover effect between the two counties and fail to find evidence of a net spillover effect. I also find that spillover effects are indeed reduced in neighbor county-pairs where the two counties are in different commute zones. The results of this paper imply that residents in counties with stay-at-home orders decreased travel to their neighboring counties even when those counties did not issue their own orders. In future situations where policy makers need to consider similar restrictions, they should focus on acting more quickly and not be concerned if neighboring counties are not cooperative. In the third chapter of this dissertation, I test the predictions of career concerns models by studying Major League Baseball umpires. Major League Baseball games can be dramatically shaped by minor lapses in judgement from the umpires officiating the game. Due to the indefinite length a game may have, this can include having the game shaped in a way that ends it faster. I study whether evidence for the career concerns model can be found in baseball umpires. A career concerns model would suggest that older umpires, whose careers and reputations are much more established than younger ones, would be more inclined to improperly make judgements that favor the end of the game in extra innings. I use data on MLB umpires and extra-innings games from the 2010-2018 seasons to conduct my empirical analysis and use a linear probability model to isolate the impact of the umpires’ tenure on the probability they make a “bad call.” I find evidence supporting the career concerns hypothesis and that the probability that an umpire makes a bad call that shortens the length of the game and allows them to go home increases with their tenure. I show that these results are likely driven by career concerns, rather than carelessness, by showing their error rate does not increase with tenure in situations where it would not reduce their workload.

This dissertation consists of three essays in applied microeconomics.The first chapter investigates the effect of coroner partisanship on COVID-19 death reporting. The politicization of the COVID-19 pandemic in the United States has raised questions about the integrity and accuracy of death reporting, particularly in jurisdictions with elected, partisan coroners. Using mortality data from the CDC and manually collected data on county-level death certification systems and coroner party affiliation where applicable, I examine the parallel systems of appointed medical examiners and elected coroners and analyze the effect of partisanship on reported COVID-19 deaths. Cross-sectional comparisons do not seem to suggest counties with coroners report fewer deaths than those with medical examiners, and difference-in-differences specifications reveal limited evidence of a statistically significant but not economically meaningful effect of partisanship on reported COVID death counts. The second chapter examines the effect of new information on lead water pipes on housing prices. In 2016, the Water and Sewer Authority of Washington, DC released an online map that contains information on lead service lines (LSLs) for all properties in the district. Using the release as a natural experiment, I estimate the effect of the new information on prices of properties with and without LSLs. Recent literature has found that housing lead reduction policies such as remediation mandates have significant price effects. In DC, while the map’s release was followed by a marked increase in requests for water lead tests, neither a difference-in-differences model nor a repeat sales model captures a significant divergence between housing prices of the two types of properties after the release, implying the housing market response to the information was limited. The second chapter considers the effect of the marriage tax subsidy on the marriage decision of same-sex couples. The U.S. Supreme Court’s ruling on United States v. Windsor in June 2013 compelled the federal government to recognize state-sanctioned same-sex marriages, including for tax purposes. The switch in the income tax filing status for same-sex couples meant that the marriage penalty or subsidy as a result of joint filing became a relevant factor that may enter couples’ marriage decisions. I construct a sample of married and cohabiting same-sex couples in 2012 and 2014 from public-use data of the American Community Survey. Using a difference-in-differences methodology, I do not find evidence that same-sex couples who would earn a higher marriage subsidy became more likely to marry after the Supreme Court ruling.