Methodology & Sources

The EPI ranks 180 countries on environmental health and ecosystem vitality using 58 performance indicators across 11 issue categories. Scores range from 0 to 100, where higher values indicate better environmental performance. Categories include air quality, water & sanitation, heavy metals, biodiversity, ecosystem services, fisheries, climate change, pollution emissions, agriculture, and water resources.

A comprehensive database of 12,470+ climate and energy policies across 217 jurisdictions maintained by the International Energy Agency. Each policy record includes the policy type (sector), instrument mechanism (e.g., tax, subsidy, standard), legal binding status, quantified targets, and current enforcement status. We classify policies into 7 sectors: Renewable Energy (RE), Cross-cutting/Framework (CF), Energy Efficiency (EEF), Fossil Phase-Down (FPD), Green Transport (GRT), Carbon Pricing & Markets (CPM), and Land Use (LU).

National CO2 emissions data sourced from the Global Carbon Project's annual budget. Metrics include total emissions, emissions per capita, and emissions per unit GDP (carbon intensity). Time-series data allows trend analysis from 1990 to the most recent available year.

Installed renewable energy generation capacity by country and technology type. Covers solar PV, onshore and offshore wind, hydropower, geothermal, biomass, and marine energy. Used to compute clean energy share as a percentage of total installed capacity.

Policy Attribute Extraction

We transform the raw IEA/OECD Climate Policy Tracker export into a standardized policy table used by the frontend. Each policy is assigned a sector code using rule-based mapping from the IEA topic fields with a keyword fallback when topic metadata is missing. We then classify key policy attributes—instrument type, legal binding status, and whether the policy includes quantified targets—using an LLM-assisted workflow (GPT-4o-mini) and cache the results for reuse. Policy status (e.g., "In force" vs "Ended" or "Announced") is retained so we can compute "currently in effect" counts and portfolio shares consistently across countries.

Correlation Analysis

Pearson correlation coefficient (r) measures linear association between each policy sector's share of a country's portfolio and its CO2 emissions (per capita or per GDP). Statistical significance tested via two-tailed t-test. A negative r indicates that countries with a larger share of that policy type tend to have lower CO2.

Lift Score

Lift measures how over-represented a policy characteristic is among top-performing countries.

"Top quartile" = countries in the lowest 25% of CO2 per GDP among those with 5+ policies. A lift of 1.2x means a country emphasizing that sector is 20% more likely to be a top performer than the base rate.

Sector Density Normalization

To avoid bias toward large economies that naturally have more policies, we normalize by computing each sector's share as (sector count / total policies) per country. This density measure ensures we capture policy emphasis rather than raw count.

Policy Cocktails

Countries are grouped by their two most-represented policy sectors (by count). Average CO2/GDP is compared across groups containing 3+ countries to identify which sector combinations are associated with the lowest carbon intensity.

Instrument Effectiveness

For each policy instrument type (subsidies, carbon taxes, standards, bans, etc.), we split countries into "users" vs "non-users" and compare average CO2 per capita. Point-biserial correlation tests whether the presence of an instrument is associated with lower emissions.