The Causal Impact of Colonial Legacies on Environmental Performance Dataset

Description

This dataset, accompanying “The Causal Impact of Colonial Legacies on Environmental Performance” by Hiroaki Iwama (Kyoto University), tests the hypothesis that countries with lower settler mortality during colonization developed stronger institutions, which in turn led to higher contemporary environmental performance. It combines cross-country data for 63 nations, including the 2024 Environmental Performance Index (EPI), institutional quality indicators from the International Country Risk Guide (1984–2024), historical settler mortality rates (Acemoglu et al., 2001), and controls such as GDP, population, and trade openness. Using a two-stage least squares estimation, the study finds that lower settler mortality predicts stronger institutions, which causally improve environmental outcomes, particularly in environmental health and climate change mitigation. The dataset enables replication and further research on how historical institutions continue to shape modern environmental governance.

Steps to reproduce

The dataset was constructed by integrating publicly available cross-country data from multiple reputable sources. Environmental performance was measured using the 2024 Environmental Performance Index (EPI), published by Yale University and Columbia University. Institutional quality was derived from the International Country Risk Guide (ICRG) dataset (1984–2024), combining the indicators Corruption, Law and Order, and Democratic Accountability into a normalized 0–100 Political Risk Rating index. Historical settler mortality rates were obtained from Acemoglu, Johnson, and Robinson (2001) as an exogenous instrument for institutional quality. Contemporary control variables—GDP per capita (IMF, 2025), trade openness (WTO, 2025), and population (World Bank, 2025)—were merged using ISO country codes. Data cleaning, transformation (e.g., log conversions), and analysis were conducted in R (version 4.3.2) using standard econometric packages for 2SLS estimation. The entire workflow is replicable by following the variable definitions, sources, and model specifications described in the accompanying paper.

Institutions

• Kyoto Daigaku

  Kyoto
• University of Sussex

  Brighton and Hove, Brighton

Categories

Funders

• Kyoto University

  Kyoto, Japan

Licence