Faculty Sponsor: Professor François Seyler
Live Poster Session: Zoom link will be added soon.
Steven Li
I am a student researcher at Wesleyan University and an apprentice at the Hazel Quantitative Analysis Center (QAC). My work combines economic history, applied econometrics, GIS, and data engineering to explore how infrastructure shapes migration and political change. In my current project, supervised by Professor François Seyler, I examine the relationship between railway expansion, immigrant settlement, and abolition-era voting in nineteenth-century Brazil. I have validated historical station and municipality crosswalks, constructed spatial measures of railway access, and prepared district-level panel data for regression and instrumental-variable analysis. My research emphasizes reproducibility, transparent quality assurance, and cautious interpretation of evidence. More broadly, I am interested in using quantitative methods to transform fragmented historical records into credible insights about long-term social, economic, and political development.
Abstract: This project introduces a new station-level dataset tracing the expansion of Brazil’s railway network from 1872 to 1927. We harmonize 2,527 historical station records, retaining 2,461 validated observations, and link them to geographic coordinates, historical boundaries, census measures, legislator biographies, and thirteen emancipation-related parliamentary votes. We use the dataset to examine whether railways influenced slavery-related political outcomes by reshaping immigrant settlement. Because immigrant concentration may be endogenous—reflecting unobserved local economic, political, or demographic conditions that also affected legislative behavior—we complement baseline regressions with instrumental-variable strategies designed to isolate plausibly exogenous variation in immigrant exposure. Railway station intensity positively predicts the foreign-born population share in 1890, supporting the proposed demographic channel. Baseline, Bartik IV, and combined Rail-Bartik specifications suggest that immigrant exposure was associated with more abolitionist voting where slavery was locally more salient. However, clustering and spatial-HAC adjustments reduce statistical precision, so the findings remain suggestive rather than definitive causal evidence. The dataset provides reusable infrastructure for studying transportation, migration, labor systems, and political change in historical Brazil.
Animated Railway Map: