Annual time‑series data for CO₂ emissions, energy‑human development composite index, GDP, trade openness, population density, and structural breaks in Saudi Arabia (1970–2023)

This dataset supports the manuscript "A PCA‑Based Energy–Human Development Index and CO₂ Emissions in Saudi Arabia: Evidence from Symmetric, Asymmetric, and Quantile ARDL Models". It contains annual observations for Saudi Arabia from 1970 to 2023 (n = 54). All variables except the PCA‑based composite index (LNEHD) are natural‑log‑transformed. Variables Variable Definition Unit / Notes Year Calendar year 1970–2023 lnc Log of CO₂ emissions per capita Metric tons per capita (dependent variable) lng Log of real GDP Constant 2015 international USD lnehd PCA‑based Energy–Human Development Composite Index Constructed from total energy supply per capita, urban population share, and Human Capital Index (standardised, mean = 0, SD = 1). Positive values indicate higher energy‑human development. lnto Log of trade openness (Imports + Exports) as % of GDP lnp Log of population density Persons per km² d1 Structural break dummy = 1 for 1991 onwards (Gulf War / post‑Cold War regime shift) d2 Structural break dummy = 1 for 2010 onwards (post‑global‑financial‑crisis oil‑price regime shift, identified by Zivot–Andrews test) Data sources World Bank World Development Indicators (WDI) BP stats United Nations Development Programme (UNDP) Human Development Reports Purpose of the study The study examines the long‑run and asymmetric effects of the Energy–Human Development Composite Index (LNEHD) on CO₂ emissions in Saudi Arabia, while accounting for GDP, trade openness, population density, and structural breaks. The dataset is used to estimate symmetric (ARDL), asymmetric (NARDL), quantile (QARDL) models, and Toda–Yamamoto Granger causality tests. The composite index resolves severe multicollinearity among energy supply, urbanisation, and human capital. Reuse notes LNEHD is a z‑scored PCA score; its coefficient in log‑linear models is a semi‑elasticity (percentage change in CO₂ per one‑standard‑deviation change in LNEHD). Due to near‑perfect collinearity between LNEHD and lnp (ρ = 0.98), the individual coefficients of these two variables are not separately identified in the long run. Structural break dummies d1 and d2 are included as exogenous controls; they should not be re‑estimated without verifying break dates. Missing values: none. Full annual coverage from 1970 to 2023.