When Complexity Pollutes and When It Cleans: A Bayesian Spatial Panel Analysis of Trade, Technology and Research Complexity Effects on Carbon Emissions

Economic complexity is often assumed to support environmental sustainability, yet existing evidence remains mixed because complexity is typically treated as a single aggregate construct. This paper revisits the complexity–emissions nexus by decomposing the Economic Complexity Index into trade, technology and research dimensions and examining their effects on carbon emissions within a Bayesian spatial panel framework covering 57 countries from 1999 to 2019. The results reveal that aggregate measures conceal fundamentally different environmental consequences. Trade complexity is consistently associated with higher per capita carbon emissions, and technology complexity is generally associated with higher emissions, reflecting the energy demands of increasingly sophisticated production systems. In contrast, research complexity is associated with lower emissions, identifying scientific knowledge creation and innovation capacity as key mechanisms for decoupling economic development from environmental degradation. The analysis also uncovers negative spatial dependence across alternative network structures in the static specifications, suggesting that carbon dynamics are characterised by competitive displacement rather than positive spillovers. These findings indicate that not all forms of economic sophistication are environmentally equivalent and that research-based complexity offers the most sustainable pathway to low-carbon development.