Long-term temperature shifts and their variability are anticipated to significantly affect heating and cooling demands in buildings, subsequently impacting the electric sector's capacity and capital investments. Prior research has provided insights into these impacts but suffers from several limitations. Firstly, many multi-sector models focus on annual electricity demands, overlooking the influence of temperature changes at sub-annual scales (daily and seasonal peak demands). Secondly, econometric and empirical studies often neglect changes in socioeconomics (population, income), technological advancements, fuel prices, and infrastructure development. Lastly, power-sector-focused tools frequently disregard multi-sectoral interactions, including the effects of broader socioeconomic trends on energy service demands and economic competition among fuels. This study addresses these limitations by using the United States as an example, given its projected significant increase in electricity demand (up to 60% from 2015 to 2050) and its wide climate and fuel mix variations. The research aims to provide a comprehensive understanding of temperature-induced capacity increases at the state level, considering broader socioeconomic development and multi-sectoral interactions—a gap in existing literature.

The paper reviews existing literature on the impact of climate change on electricity systems, highlighting three key methodological limitations: a focus on annual rather than sub-annual electricity demand; a failure to account for socioeconomic changes, technological advances, and fuel price fluctuations; and a lack of consideration for multi-sectoral interactions. The authors cite numerous studies that have investigated these impacts, both globally and for specific countries, but argue that these studies fall short of a comprehensive analysis due to these limitations. They note that while some studies use multisector models, they often lack the detail necessary to capture the subannual dynamics of electricity use. Other studies, using econometric and empirical methods, fail to fully account for the effects of changing socioeconomic conditions. Finally, power sector-focused studies often neglect the interaction between the electricity sector and the rest of the economy.

The study utilizes a modified version of the Global Change Assessment Model (GCAM)-USA, a global multi-sector human-Earth systems model with state-level detail for the US. This model endogenizes the impacts of temperature change on building cooling and heating service demands, sub-annual (monthly day and night) electricity load profiles, and associated electric capacity and investment requirements. The analysis is based on two variants of the RCP 8.5 scenario: one without temperature impacts and one with temperature impacts, both broadly consistent with the SSP2 scenario. The model incorporates spatially gridded (1/8th of a degree), hourly temperature projections to estimate heating degree hours (HDH) and cooling degree hours (CDH) by state. These are used as inputs into GCAM-USA to calculate heating and cooling service demands. The year is divided into 25 time segments (one for each month's day and night, plus a super-peak segment representing the top 10 hours of the year) to capture subannual dynamics. The model uses a linearly optimal least-cost approach to dispatch installed capacity across these segments and incorporates a 15% reserve margin. A logit-based calculation estimates the mix of capacity investments, considering existing stock and retirements. Sensitivity analyses are performed using SSP3 and SSP5 assumptions for population and economic growth to assess the impact of alternative socioeconomic trajectories. The model also includes a representation of electricity trade across the U.S., allowing for the analysis of how trade influences temperature-induced investments.

The study finds that while temperature-induced increases in national annual electricity generation in 2100 may be relatively small (0.5–8.3% across states), the increases in installed capacity (3–20%) and capital investments (3–22%) could be substantial. This discrepancy highlights the importance of considering peak demands, rather than just annual averages, when planning for future electricity systems. The spatial distribution of these increases is heterogeneous across the U.S., driven by factors including socioeconomic development, electricity trade, regional fuel prices, resource endowments, and the characteristics of existing capital stock. The largest increases in installed capacity and investments are concentrated in California, Illinois, Pennsylvania, and Texas. Analysis shows that while temperature-induced increases in capacity and investments are largely driven by increases in cooling demand, their spatial distribution does not precisely mirror that of peak CDH due to the flexibility offered by electricity trade. Sensitivity analysis reveals that temperature-induced capital investments are highly sensitive to socioeconomic assumptions, with significant variations under different population and economic growth scenarios (SSP3 and SSP5). Under SSP5, temperature-induced investments are 34% (California) to 65% (Montana) higher, while under SSP3, they are 27% (Rhode Island) to 36% (Florida, Montana, Nevada) lower. The largest increases in cumulative capital investments (2015–2100) by fuel are in gas, followed by coal and solar.

The findings emphasize the need for electric sector capacity expansion planning to consider the impacts of temperature changes on both mean and peak electricity loads, along with sub-annual demand profiles. The study underscores the importance of incorporating broader socioeconomic drivers and multi-sectoral interactions in planning, as these significantly influence electricity demand and investment requirements. The research highlights the role of flexibility mechanisms, such as electricity trade, in mitigating the impact of temperature changes on individual states' investment needs. The significant sensitivity to socioeconomic assumptions underscores the necessity of considering a range of future scenarios when planning for long-term electricity system reliability.

This study demonstrates the significant impact of long-term temperature changes on electricity investments, exceeding the impact on annual generation. The necessity of considering sub-annual load profiles and peak demand, along with socioeconomic factors and multi-sectoral interactions, is highlighted. Future research should explore alternative climate scenarios, incorporate state-level policies, analyze the role of electricity storage and demand-side management, and investigate the effects of transportation sector electrification.

The study focuses on a single temperature projection (from the CCSM4 model within the CMIP5 ensemble), limiting the full consideration of temperature uncertainty. It also does not explore all possible combinations of SSP and RCP scenarios, and does not incorporate all state-level policies or detailed analysis of electricity storage and demand-side response mechanisms. The model is myopic, using partial equilibrium concepts, limiting the consideration of long-term investor responses to uncertainties.