Student Research
A True Emergency? Estimating the Effect of Nuisance Property Ordinances on Eviction and Homelessness
Ethan Berner ('22)
Municipalities across the United States use nuisance property ordinances to impose fines on property owners if their tenants cause an excessive degree of police attention. While proponents argue that the policy allows police to cut costs and focus on “true emergencies,” critics argue that it incentivizes landlords to evict their tenants, contributing to a range of social issues. I build a novel dataset by collecting nuisance ordinance information for the 75 most populous U.S. cities and linking it with eviction and homelessness data. Exploiting differences in policy enactment, I estimate that the policy increases the eviction rate by 20 percent and the number of evicted individuals per 100,000 by 29 percent. The effect on homelessness is ambiguous. My results suggest that nuisance property ordinances have a negative effect on housing security.
Patrick Stevenson ('22)
This study investigates the Black Lives Matter protests that erupted after the death of George Floyd. Focusing on the incidence of police violence deaths and support for Democrats as representations of the aims of the movement, I assess the efficacy of these protest events. Utilizing rainfall during the protest window as a source of exogenous variation in protest size, I find that larger BLM protests in a city increase support for the 2020 Democratic presidential candidate and decrease in the number and likelihood of Black deaths caused by police violence. I discover several threats to identification and address these concerns in the preferred specifications, which imply that BLM protests were effective in decreasing police violence deaths and increasing support for the Democratic presidential candidate.
Kavita Patel, ‘22
In the United States, black-white racial disparities in incarceration rates are believed to have
endured largely because of targeted drug arrests. Drug reform, specifically recent marijuana
legalization at the state level, has the potential to reduce these disparities. To assess this hypothesis, I combine county-level jail population data from the Vera Institute of Justice with state marijuana legalization data to estimate the causal impact of marijuana legalization on incarceration trends. I employ both differences-in-differences and event study models to estimate the impact of legalization on black and white jail rates and the racial disparity in jail rates. Results show that the most extreme disparities lie in counties with low black populations and high black jail rates. The event studies using state linear time trends show a reducing effect of marijuana legalization on both the black jail rate and the jail disparity; however, pre-trends reveal that common trends may be a strong assumption and estimates are biased downwards. The event studies using a restricted sample of counties with black population sizes in the top quartile reveal evidence in support of a null effect of legalization on all incarceration indicators, suggesting that marijuana legalization has had a limited effect on incarceration.
Is AI Deleting Your Major? Associative Patterns of AI Exposure and Choice of College Major
Maxwell Auborn, ‘22
How can we explore the relationship between AI in the workplace and college major
decisions? I create a list of the five top-employing occupations for each college major and link it to Webb’s (2020) occupational AI exposure indexes. This enables me to measure the amount of AI exposure that a recent graduate is facing in the labor market for a given college degree. Using this measurement, I investigate the yearly pattern of bachelor’s degrees conferred and median wages for recent graduates for each major. I find that there is a trend that suggests college majors with higher AI exposure have seen larger increases in the share of graduates from 2010 to 2019. Whereas, from 2010 to 2014, AI exposure is not indicative of a college major’s change in the share of total bachelor’s degrees conferred. I find that majors like computer and information science, health professions, and architecture have high AI exposure, and other fields like psychology, theology, and communications, journalism, and other related programs have low AI exposure.
Xiaoyu Wu, ‘22
The COVID-19 vaccination rate in the US lags behind other countries, attributed to vaccine hesitancy amongst Republican voters. Republican leaders such as Trump and governors began encouraging people to get vaccinated, but their message contradicts anti-vaxxers’ beliefs. I examine the role of Republican leaders’ endorsement on COVID-19 vaccine uptake when the message is counter-attitudinal to its audience. Using variations of the timing and location of endorsements, I use difference-in-differences and event study analysis. Results show that Trump’s endorsement had a positive effect on vaccine uptake in some rallies, though smaller than studies examining vaccine willingness. This study also finds no effect from Republican governors’ tweets. These findings suggest that Republican leaders had limited effects on vaccine uptake.
Diversity in Start-up Fundraising: Gender, Race & Education
Christian Poulsen, ‘22
This study represents the most recent and comprehensive data work analyzing the cross section of gender, race, and education within venture capital allocation over the 2009-2021 period, and documents the initiatives by venture capital firms to mitigate inequity. The paper measures the magnitude of inequity in venture capital start-up funding, and how it has evolved over time by quantifying and mapping the dynamics of gender and racial inequity in the entrepreneurial sector over the 2009–2021 period. In addition to exploring the time-series evolution, the study uncovers how education influences founders’ fundraising success across gender and race. The paper shows that for entrepreneurs who are women or persons of color, accessing funding for their start-ups is markedly more difficult than for white, male founders.
School Feeding Effects on Health & Education: Evidence from India’s ‘Mid-Day Meal’ Program
Shridhar Hari Singhania, ‘22
This study analyzes long-term health and education impacts of the world’s largest school feeding program, with attention to gender inequality. Using DHS data spanning five decades, difference-in-differences ATT estimates indicate positive educational and ambiguous nutritional effects. Results suggest women and men with maximum exposure to India’s mid-day meal program attend school 160 and 306 days longer than do those unexposed. Yet, fully exposed women lose 0.65 kg/m² in BMI; likewise, exposed men shed 0.53 kg/m² and 0.605 cm in BMI and arm circumference. Gender inequality analysis reveals no substantial verdict, and wide variation in educational gender inequality exists across Indian states. Findings suggest the program enjoys some success in bolstering school enrolment and participation but warrants further attention on state-level gender inequity and nutritional improvement.
Derek Ban, ‘24
This study examines the repeal of anti-miscegenation laws, or laws that prevent
interracial marriage, whether done on a state-by-state basis or compelled by the Supreme Court
ruling in Loving v. Virginia. I seek to evaluate the repeals’ effect on the nonwhite marriage rate
to discuss the larger issue of how the United States’ nonwhite population has been
disenfranchised by the denial of equal marriage rights which were enacted by racially
discriminatory legal institutions. Using various difference-in-difference models, I find no
evidence that repealing anti-miscegenation laws has an effect on the nonwhite marriage rate. I
hypothesize that the lack of an effect is due to lingering negative public opinion regarding
interracial marriage and the low prevalence of interracial marriages as a proportion of total
nonwhite marriages.
Huthefa Maalim, ‘24.5
This paper investigates the impact of female bargaining power in deforestation rates
across Sub-Saharan Africa (SSA), where women actively engage in both the exploitation and
conservation of forest resources but often have limited control over these resources. Utilizing
the legal variations and colonial origins of national borders, the study employs spatial regression
discontinuity (RD) analysis to assess these impacts. Contrary to expectations, the findings reveal
that common law countries, characterized by lower female bargaining power, exhibit lower
deforestation rates compared to civil law countries. These results suggest the existence of
compensatory behaviors in decision-making processes related to deforestation in SSA.
Crypto Volatility: Bitcoin Trading Responses to Regulatory Sentiment
Matt Goodman, ‘23.5
This research explores the intricate relationship between central bank speeches and the cryptocurrency market, focusing on the impact of Bitcoin regulatory sentiment on trading volume. I employ a difference-in-differences method to assess if market participants respond to Bitcoin regulatory sentiment. This paper contributes the largest known dataset of Bitcoin references in central bank speeches and classifies the sentiments as positive, negative, or neutral. I find 164 mentions of Bitcoin from 19,130 speeches across 118 institutions spanning 2012 to 2023, and I create a new panel data set with trading volume data from 39 countries. Results show imprecise variation in trading volume in the days following a speech.
Akari Tsurumaki , ‘23.5
This study examines impacts of elderly care on labor market outcomes at both extensive and intensive margins for caregivers in Japan, treating deaths of the cared elderly as exogenous shocks that would reduce elderly care demand. I implement static and dynamic difference-in-difference analyses on individual caregivers using the Japanese Household Panel Survey from 2017 to 2021. Overall, the effects of deaths seem to become positive over time. However, the results are imprecise and statistically insignificant, suggesting that no evidence of effects and the relationship between deaths of the cared elderly and labor supply of caregivers may be weak.
Urban Afforestation and Infant Mortality Outcomes: Evidence from Million Trees New York City
Jack Waitz, ‘24
In light of urbanization trends and an uptick in urban afforestation projects, it is
important to estimate the causal effects of tree plantings on health outcomes. I spatially
aggregate tree planting data from MillionTreesNYC at the community district level to
estimate tree planting effects on NYC infant mortality outcomes. Across a variety of
specifications, including a novel event study specification, I find no relationship between
plantings and infant mortality. However, missing data on park plantings presents a
likely confounder, biasing estimates downward and making it difficult to estimate the
true effect size.
