Environmental sustainability transitions require robust indicator-based evidence to evaluate how technological, financial, and governance factors shape progress toward carbon neutrality. However, the environmental sustainability indicators literature still offers limited evidence on how these structural drivers jointly influence a core environmental indicator within a single advanced economy context . This study examines Canada’s carbon transition by assessing the long- and short-run effects of artificial intelligence (AI) innovation, stock market capitalization, fiscal decentralization, renewable energy consumption, and economic growth on CO 2 emissions over the period 1990 to 2023. Grounded in the integrated insights of the Environmental Kuznets Curve, Ecological Modernization Theory, and the Technology-Environment Nexus, the study employs autoregressive distributed lag (ARDL) bounds testing, which is well suited to mixed orders of integration and relatively small annual time-series samples , complemented by FMOLS, DOLS, and CCR estimators. The findings show that AI innovation and financial system expansion are associated with higher emissions in the long run, whereas fiscal decentralization and renewable energy consumption contribute to emissions reduction. These results suggest that technological and financial advancement do not automatically improve environmental performance unless supported by effective governance and sustainability-oriented policy coordination. The findings offer policy-relevant insights for designing governance and monitoring frameworks that better align innovation, finance, and decentralized decision-making with long-term environmental sustainability goals. • Uses CO 2 emissions as a core environmental sustainability indicator • Integrates AI innovation, finance, governance, energy, and growth in one model • Provides original Canada-specific evidence on carbon transition dynamics • Shows that innovation and financial expansion may also increase emissions • Applies robust ARDL, FMOLS, DOLS, and CCR estimation techniques
Roshid et al. (2026) studied this question.