This study develops a data-driven forecasting framework based on grammatical evolution (GE) to model carbon dioxide (CO2) emissions across U.S. states within a multilevel governance context characterized by regulatory heterogeneity. The United States exhibits significant variation in state-level climate policy implementation, generating differentiated regulatory environments that directly affect corporate strategic decisions. In this setting, climate-related risks are increasingly intertwined with regulatory uncertainty, financial conditions, and firm-level adaptation strategies. Our approach integrates federal and state-level dynamics in a single multiscale predictive structure, improving forecasting accuracy while maintaining interpretability. The study makes two main contributions: first, it advances the emission forecasting literature by introducing the first multiscale GE application that explicitly captures cross-level policy interactions within a federal system; second, it provides managerial and policy-relevant insights by offering a quantitative framework to assess subnational regulatory risk, institutional coherence, and strategic predictability across jurisdictions. Findings indicate that stronger alignment between state and federal emission trajectories is associated with greater regulatory predictability, which in turn supports corporate location decisions, investment timing, and transition risk management. By complementing existing research focused on causal and structural drivers of emissions, this framework adopts a predictive perspective that captures the combined effects of macroeconomic conditions, institutional settings, and financial dynamics. Overall, the results highlight the role of emission forecasting as a strategic information mechanism embedded in multilevel governance systems, with implications for both policy evaluation and corporate sustainability planning.
Verdugo et al. (Tue,) studied this question.