A proposed robust regression model to study carbon dioxide emissions in Egypt

Abstract

Developing countries face environmental challenges as they rely on non-renewable energy for economic growth. This study examined the dynamic relationship between CO₂ emissions and four economic factors (manufacturing, trade, urban population, and cereal production land) in Egypt from 1990 to 2021. It introduced a new approach to estimate the autoregressive distributed lag (ARDL) model using robust methods (M, MM, and S) and compared them with ordinary least squares (OLS) to determine the best estimate. The ARDL methodology was employed to test short-term and long-term relationships. Results showed ARDL (1,0,2,1,3) as the most suitable model. Manufacturing and Trade variables negatively impacted CO₂ emissions, while Urban population and Land variables had positive long-term effects. The period lags of Trade and Land variables have significantly affected CO₂ levels. The error correction model indicated economic adjustments occur within about 25 months. The study found that robust methods (M, MM, and S) outperformed the non-robust OLS method. Among these, the S estimation method proved most effective, showing the lowest Akaike information criterion (AIC) and Bayesian information criterion (BIC) values.