Statewide Load Impact Evaluation of California Capacity Bidding Programs (CBP) Case Study

The CBP is a statewide price-responsive and aggregator-managed program launched in 2007. It is available at the three CA Investor Owned Utilities (IOUs) – Pacific Gas & Electric, Southern California Edison, and San Diego Gas & Electric. However, each utility’s program differs in program features and operations. The program is locationally dispatched upon CAISO market awards to target geographic areas of high demand. The 2022 program included 722 participants (residential and non-residential) and provided approximately 32 MW of load reduction.

AEG has been conducting the impact evaluation of the CBP program since 2015. Over that time, we have seen the program transform from a peak shaving peak capacity program into the locational non-wires alternative program that it is today. This case study presents some of the key findings from the PY2022 evaluation.

Study goals included the following analysis in adherence to the CA load impact protocols.

• Estimate hourly ex-post load impacts for each utility program and dispatched event in the relevant program year.
• Develop a monthly 11-year normalized ex-ante forecast of load impacts for utility each program.

We used a consistent methodology across all programs. Each program is modeled independently, modifying assumptions to account for program design and implementation specific to each IOU’s CBP tariff.

• The Residential ex-post analysis used a matched control group and aggregate hourly regression models. This approach is the best practice for participant populations with less variable loads, which can leverage the higher statistical power with more customers included in each model. A matched control group also more effectively estimates the counterfactual load without a randomized control trial.
• The Non-residential ex-post analyses used a within-subject design and customer-specific hourly regression models. Non-residential customers often vary significantly from one another in load shape, weather response, and overall size. Customer-specific regressions allow us to control for variation in load due to weather conditions, geography, time-related variables, and other unobservable customer-specific effects.
• The ex-ante forecast leverages ex-post hourly regression models to predict impacts under the Utility and CAISO 1-in-2 and 1-in-10 weather scenarios.