Business-cycle dynamics in the AI–green growth nexus: evidence from artificial intelligence investment in OECD economies

This study examines the regime-dependent association between artificial intelligence investment (AIINV) and green growth (GGR) across business-cycle phases in OECD economies during 2012–2024. It investigates whether the AIINV–GGR relationship differs between boom and recession periods and whether AIINV is positively associated with the business cycle. Business-cycle conditions are identified using OECD and Hodrick–Prescott filter output gaps. The empirical analysis applies Augmented Mean Group and Common Correlated Effects Mean Group estimators within a pooled interaction framework, complemented by Wald tests. Results show that AIINV is positively associated with GGR during booms but negatively associated with GGR during recessions, while AIINV is positively associated with the output gap. Inflation and real interest rates have negative coefficients, whereas labour, physical capital, and human capital have positive coefficients. A fixed-T factor-proxy IV-GMM sensitivity analysis treats regime-specific AIINV interactions as endogenous and instruments them with their second and third lags. The GMM estimates reproduce the same directional pattern under both business-cycle classifications. However, weak recession-side instruments and residual cross-sectional dependence limit causal interpretation.