Essays in Individual Adaptation to Policy Change

• Main Author: Pennington, Kate
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01-01-2021
• Subjects: Abortion; Affordable housing; Construction; Demographics; Economics; Election results; Employment; Fires; Housing prices; Lead content; Public policy; Race; Rents; Supply & demand; Trends; Urban planning; Voter behavior
• ISBN: 9798535551463

Government policy has a profound impact on many aspects of individual people's lives, limiting, expanding, or changing the choices available to them. My dissertation examines how individuals respond to policy changes through behavioral adaptation and political participation. Policy change is often shaped by political interests and incomplete information that can lead to unintended or unexpected impacts; people must adapt to a changing world that may or not be better for them. In Chapter 1, I study the impact of new market rate and affordable housing construction on nearby housing affordability, displacement, and gentrification. In Chapter 2, I investigate the relationship between government failure and voter turnout after the Flint Water Crisis. In Chapter 3, I examine the role of risk aversion in women's contraceptive choice, demonstrating that women make defensive investments in methods that can shield them from adverse policy change. In each paper, I combine natural experiments with microdata to identify the causal impact of policy change on individual choices. In Chapter 1, I answer a question at the heart of intense debate over how to address housing affordability in San Francisco. How does new construction affect housing prices, displacement, and gentrifcation in the surrounding neighborhood? To separate the causal impact from the well-documented correlation between new development and increasing housing prices, I exploit random variation in the location of new construction in San Francisco caused by serious building fires. These fires are rare and unrelated to trends in rent, displacement, or gentrification, and they increase the probability of construction by a factor of 32. I combine this identification strategy with an original dataset of Craigslist rents, which I create by scraping old rental listings that have been collected by an internet archive nonprofit. Next, I leverage the individual address histories of 1.24 million San Francisco residents to create separate measures of displacement and gentrification. I find that rents fall for housing near the new construction, reducing displacement risk by 17% for nearby renters. At the same time, neighborhood quality improves through increased residential renovations, and gentrification increases as wealthier people move in. In contrast, I show that affordable housing and endogenously located construction do not affect rents, displacement, or gentrification. These findings suggest that building new market rate housing has beneficial spillover effects for incumbent renters, reducing rents and displacement pressures while improving neighborhood quality. In Chapter 2, I study the impact of the Flint water crisis on political participation with coauthor Eleanor Wiseman. Unpacking the motivation to vote is a central goal in political science, political economy, and the pursuit of representative democracy, but it is difficult to study empirically. We examine the impacts of a massive municipal and state government failure that exposed a quasi-random sample of households in Flint, MI to elevated lead in their water. Does personal experience of government quality motivate voting? We combine spatial data on lead exposure with the Michigan Qualified Voter File to explore how the results and timing of household lead tests affect voter turnout. To overcome selection issues, we use a triple differences approach that compares individual turnout for voters with 1) positive versus negative lead test results, 2) who receive their test results before versus after the 2015 mayoral election, 3) for the 2015 mayoral election versus the two mayoral elections before the water crisis. We find that voters who tested positive for lead before the 2015 mayoral election were 3.1 percentage points less likely to vote. However, this average effect masks significant and opposite reactions to lead test results among Black and White voters. We find that Black voters who tested positive before the election became 14.4 percentage points less likely to vote, while White voters who tested positive before the election became 4.5 percentage points more likely to vote. Moreover, we find that the crisis increased turnout, new voter registration, and the preference for a mayoral challenger for Black residents compared to White residents at several times the magnitude of the impacts of individual lead test results. We interpret these findings as evidence that the crisis increased the salience of race, causing divergence in the voting rates and preferences of Black and White voters. Chapter 3 examines the role of the reproductive healthcare policy environment in women's choice of a contraceptive method. Reproductive healthcare policy has become a knife-edge issue in the United States, with narrow legislative majorities passing laws that dramatically change women's access to affordable family planning services and abortion on a state-by-state basis. This paper investigates the role of the policy environment in women's choice of a contraceptive method. I model contraceptive method choice as a dynamic discrete choice under uncertainty about future pregnancy, abortion access, and contraceptive method costs. The model predicts that women are forward-looking and risk averse in their choice of contraception, switching to lower failure-rate methods when they expect abortion access may fall and to longer-lasting methods when they expect costs may rise. Using de-identified patient level data from Planned Parenthood of Wisconsin and Planned Parenthood of Northern New England, I evaluate the impacts of three shocks to the policy environment: Wisconsin's 2015 abortion ban, Vermont's 2016 reproductive healthcare expansion, and the 2016 presidential election. I find strong evidence in support of the model predictions, with the probability of switching increasing by 2-17 percentage points after the policy shocks. Women choose a contraceptive method in response not only to method attributes, but also to the policy environment, making defensive investments in methods that can shield them from adverse policy changes in the future.