Files in this item

FilesDescriptionFormat

application/pdf

application/pdfSTEVENS-DISSERTATION-2019.pdf (12MB)
(no description provided)PDF

Description

Title:The socioeconomic impacts of the Superfund program
Author(s):Stevens, Alexander Nishi
Director of Research:Christensen, Peter
Doctoral Committee Chair(s):Christensen, Peter
Doctoral Committee Member(s):Baylis, Kathy; Dell’Erba, Sandy; McMillen, Daniel
Department / Program:Agr & Consumer Economics
Discipline:Agricultural & Applied Econ
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:Ph.D.
Genre:Dissertation
Subject(s):Superfund
Housing Market
Difference in Difference
Machine Learning
Conditional Average Treatment Effects
Average Treatment Effects
Environmental Justice
Duration Models
Abstract:Who Benefits from the Cleanup of Superfund Landfill Sites? Evidence from New York State uses a difference-in-difference approach to estimate the average treatment effects of different groups that benefit from Superfund landfill cleanup. The benefits of Superfund landfill cleanup go to those with higher value homes and those with well water. These projects are partially financed with state and local property taxes, which are flat taxes within New York municipalities (New York State Department of Taxation and Finance, 2018). This chapter provides evidence that the cleanup of these sites is regressive. Also, there is an opportunity for land value capture, as states and local governments can tax houses with these identifiable characteristics more than other households to recuperate the costs of cleanup. Estimating Heterogeneous Treatment Effects in the Housing Market with Machine Learning Techniques uses an empirical Monte Carlo experiment to compare the performance of machine learning and standard econometric methods in estimating both average treatment effects and geographically heterogeneous treatment effects. This chapter finds that Double Machine Learning (DML) performs similarly to standard parametric methods in full randomization but shows large performance gains when there is unobserved selection into treatment. For geographically heterogeneous treatment effects, this chapter finds that Conditionally Parametric regressions (CPAR) have the best performance when treatment is randomized. However, when there is unobserved selection into treatment machine learning methods outperform CPAR. This chapter also finds that ensembles of various methods can outperform individual methods. This chapter finds that machine learning methods perform better than standard methods when there is unobserved selection into treatment. Determinants of Superfund Cleanup Duration characterizes the relationship between demographics and funding of Superfund sites and the duration of cleanup, and how these relationships change over time. This chapter finds evidence that demographics are not orthogonal to cleanup duration, suggesting that demographics of the community influence cleanup duration. The pattern is consistent with the hypotheses that white communities lobby for more complete cleanup and project managers are more careful because of liabilities in white communities during the construction phase. White communities also get faster deletion times. For funding, this chapter finds that responsible parties cause substantial delays in construction duration. After 2000, sites with state funding have construction completed and are deleted from the NPL faster than sites without state funding.
Issue Date:2019-07-08
Type:Text
URI:http://hdl.handle.net/2142/105655
Rights Information:Copyright 2019 Alexander Stevens
Date Available in IDEALS:2019-11-26
Date Deposited:2019-08


This item appears in the following Collection(s)

Item Statistics