Replication data and code for: Omitted natural capital and the measured productivity of energy exporters

Published: 16 November 2026| Version 3 | DOI: 10.17632/2rnhhfxc2v.3
Contributor:

Description

This dataset contains the data and replication code for the article "Carbon Intensity, Energy Use, and the Producer–Consumer Divide: Cross-Country Evidence on Total Factor Productivity". The article tests three hypotheses on 128 countries over 1990–2023: whether the relationship between the carbon intensity of energy and total factor productivity (TFP) depends on energy import dependence (H1), whether the production technology of import-dependent economies is stable over time (H2), and whether the TFP of net energy exporters depends on the extent of their net energy exports (H3). Files 1. analysis_data.xlsx - readme: author information and overview. - codebook: definition, source and role of every variable. - analysis_panel: country-year panel of 217 countries, 1980–2024 (9,765 rows); the baseline estimation sample is 2,591 observations from 128 countries, 1990–2023. - prices: annual real energy prices, 1980–2025 (World Bank Pink Sheet). - fuel_position: initial (1990–94) net import position by fuel (Our World in Data). - wdi_rents: total natural resources rents, % of GDP, 1970–2021 (World Bank WDI, NY.GDP.TOTL.RT.ZS). - oecd_eamfp: environmentally adjusted multifactor productivity growth, 52 countries, 1996–2018 (OECD). Sources: World Bank World Development Indicators and Pink Sheet, IMF World Economic Outlook, Penn World Table 11.0, Our World in Data, Global Carbon Project, ILOSTAT, IEA, BIS and OECD. 2. analysis_code.py Python code that reproduces every table and figure in the article and its Online Resource 1: step-1 estimation of TFP under four constant-returns production functions (Cobb–Douglas, CES in capital and labor, nested CES (KL)E and (KE)L); step-2 two-way fixed-effects regressions with continuous and binary energy-trade positions; wild cluster bootstrap, two-step bootstrap and randomization inference; threshold, influence and specification diagnostics; tests of technology stability by group and period; multiple-testing corrections by family (Holm, Benjamini–Hochberg, Romano–Wolf); event studies around energy-price shocks; coefficient-stability bounds; natural-capital checks; dynamic-panel GMM; and policy scenarios. Results are associations, not causal effects. Third-party source data remain subject to the terms of their original providers.

Files

Categories

Environmental Economics, Energy Economics, Productivity, Panel Data Analysis

Licence