Cleaned analysis dataset for "The State-Dependent Energy–Growth Nexus: Evidence from a Conditional Production Technology"
收藏资源简介:
File includedpanel_df_clean.rds — R serialized object containing the cleaned country–year panel used in the empirical analysis reported in the manuscript. DescriptionThis file contains the final cleaned and merged analysis dataset used to estimate the production-function models with emissions moderation and the associated marginal effects and panel causality tests. The dataset is a country–year panel assembled from publicly available sources and filtered to the sample used in the manuscript. Underlying public sourcesThe dataset is constructed from the following openly available databases: Penn World Table (PWT) version 11.0 (macroeconomic variables including GDP, capital stock, employment, total factor productivity). World Development Indicators (WDI), World Bank (energy use per capita, renewable energy share, greenhouse-gas emissions). Unit of observation and coverage Unit of observation: country–year. Coverage: global panel; unbalanced. Time span and number of countries/observations correspond to the estimation sample reported in the manuscript (e.g., 2006–2023; 110 countries; 1,764 observations), subject to availability after merging. VariablesThe dataset includes the following variables (as used in the manuscript). Variable names reflect the final analysis file and may include transformed versions used in estimation. Identifiers country: country name (character). isocode: ISO country code used for merging (character). year: year (numeric/integer). Macroeconomic outcomes and inputs (PWT) Y: log real GDP (expenditure-side), used as the main dependent variable. TFP: total factor productivity (manuscript uses a log transform). L: log employment. K: log capital stock. Energy and environmental variables (WDI) E_pc: energy use per capita (original scale, as drawn from WDI). E: log-transformed energy use per capita used in the models (constructed as log(E_pc + 1) in the manuscript). CO2: log-transformed greenhouse-gas emissions (constructed as log(GHG + 1) in the manuscript). Renew: renewable energy share in total final energy consumption, expressed as a fraction in [0, 1]. NonRenew: 1 − Renew. Additional fields (if included in the file) Pop: population (if retained in the analysis file). E_agg: aggregated energy measure (if retained; not central to the baseline specification). How to load the dataIn R, load the dataset as follows: Download panel_df_clean.rds from Zenodo. In R:panel_df_clean <- readRDS("panel_df_clean.rds") Notes on transformations used in the manuscript Real-valued variables are used in natural logarithms for elasticity interpretation. Energy is entered as log(E_pc + 1) to handle zeros/missingness in the original series. Greenhouse-gas emissions are entered in logs as log(GHG + 1). Renewables are expressed as a share (fraction) and NonRenew is constructed as its complement. Missing values and panel structureThe panel is unbalanced due to source coverage differences across indicators. Missing values reflect underlying database availability after merging and sample filtering. Intended useThis dataset is provided for transparency and to enable replication of the empirical results. Users can reconstruct the panel structure as a pdata.frame (plm) using index = c("country", "year").



