Lights out, decisions on: How households adapt to power outages across regions and events

Abstract

The capacity to adapt to disturbances is a distinguishing feature of a resilient system. A number of recent power outages have forced households to adapt to service disruptions. Households adapt to power outages in various ways, yet most existing research focuses on a single event in a specific location. This study expands the scope by analyzing four datasets to examine household adaptations across different events and locations: general outages in Los Angeles, California (CA); the 2022 North American winter storm in North Carolina (NC) and New York (NY); the 2021 Texas (TX) winter storm; and hypothetical future events in the first three locations. Using mixed logit models that integrate revealed preference (RP) and stated preference (SP) data, the study investigates household adaptation behavior across regions and events. The analysis addresses three key questions: (1) How common are different adaptations? (2) Which household adaptations tend to occur together, and which do not? (3) How do adaptations vary with household characteristics, outage duration, and geographic location? Results show that, as outage duration increases, people are more likely to consider multiple relocation adaptations and/or use a generator. Some household characteristics affect adaptation differently depending on location. For example, more prepared individuals are more likely to go to hotels in NC but less likely in CA. This study leverages mixed logit models in a novel way to estimate adaptation behavior during power outages. The models can estimate the percentage of people implementing adaptations with sufficient accuracy for practical purposes.