Causal inference to scope environmental impact assessment in multisector systems: the case of trans-border hydropower exports


Poster (3.29 MB)
Downloads: 46

TR Number



Journal Title

Journal ISSN

Volume Title



Decarbonization of the United States’ electricity sector will require trillions of dollars of investment in generation and transmission infrastructure. The National Environmental Policy Act (NEPA) requires proponents of many major projects to complete environmental impact statements (EIS) that address reasonably foreseeable impacts, regardless of where these impacts occur. There has been controversy over the cause-effect relationships among electrical supply, electrical demand, apparent cost, and other variables given the complex interactions between them. Therefore, the range of environmental impacts attributable to new infrastructure projects is subject to frequent disagreements. In this work, we address increasing U.S. imports of Canadian hydropower in the setting of falling prices and surplus generation. There has been controversy as to whether new transmission capacity stimulates new generation capacity, and thus whether generation-side environmental and health impacts must be assessed in the scope of incremental transmission projects. We have developed a rich longitudinal database of variables related to generation capacity, export volume, retail prices, and climate over the period 1979 to 2021. We have applied a novel multivariable wide neural network machine learning methodology to evaluate alternative causal models for the evolution of the electricity system and the role of new transmission infrastructure. We find no evidence that transmission capacity stimulates generation capacity. Rather, generation capacity growth in Canada is triggered primarily by domestic price signals and climate parameters, with trans-border transmission capacity developed primarily to absorb excess generation potential. This work supports a relatively narrow scope for EIS related to trans-border transmission projects. More generally, this analysis demonstrates how causal inference methods may help build consensus around the appropriate scope of EIS for highly interconnected energy and infrastructure projects.