Course paper for ECO481: Health and Economic Inequality. Fall 2023. Taught by Prof. Michael Stepner.
How does time spent commuting affect one's health?
The correlation between commute time and worse health outcomes is well-known, but what about the causal effect of commute duration on one's health? Using pooled cross-sectional data from the American Time Use Survey for 2009-2015, I present evidence to confirm the correlation between commute duration and health outcomes and behaviours. I attempt an instrumental variables analysis using Uber's staggered entry across U.S. Metropolitan Statistical Areas between 2011 and 2015 as an instrument for commute duration. My results show no significant causal effect between commute duration and health outcomes.
Data sources are documented and stored in data/raw. Data that is generated by the code (ex. cleaned or merged data) is not tracked in version control. Instead, the main do-file will create the directory data/derived that stores all data generated by the project.
All code is written in the Stata language. Required Stata packages are documented in stata-requirements.txt. The main do-file that runs all the code is build.do, which calls three do-files stored in code/ that load, clean, and analyze the data.
The cleaned data, log files, figures, and graphs are not tracked in version control. Instead, main do-file code will create the directory results/ that stores all the figures and graphs for the paper. You can recreate them by running the code. All numbers reported in the paper are exported to the results/results.yaml file or to one of the figures and tables.
Files for the paper creation (a Latex file, a Latex class, and a bibliography file) are stored in paper/. You can recreate the paper by setting the \opath command in paper/eco481_paper.cls to the results/ directory on your computer.
You can download the paper and replication code by clicking on "Releases" on the right of the GitHub interface.