Date of Graduation

Spring 5-23-2026

Document Type

Thesis

Degree Name

Master of Science in International and Development Economics (MSIDEC)

College/School

College of Arts and Sciences

Department/Program

Economics

First Advisor

Andrew Hobbs

Abstract

This thesis estimates the causal effect of urban parks on land surface temperature in the

American Southeast using a geographic regression discontinuity at park boundaries. The

result is null. Within a CCT-optimal 257-meter bandwidth across five cities, park interiors

are not detectably cooler than the land immediately outside them:−0.023 °C, p = 0.926.

A two-way fixed effects panel across 20 cities and 39 years confirms the well-documented

NDVI–LST relationship—a 0.1-unit increase in NDVI is associated with 3.4 °C lower surface

temperature—while a park size heterogeneity model runs directly into the siting endogeneity

problem, showing why naive distance gradients cannot be interpreted causally. The null RDD

result is most likely a power problem: at 150-meter pixel resolution and a narrow bandwidth,

the boundary signal is too attenuated to surface. Around 57 million pixel-year observations,

drawn from Landsat Collection 2 and Google’s satellite embedding dataset.

Download

Share

COinS