A metric for quantifying adaptation to successive disruptions

Complex systems face a myriad of stresses and shocks. The ability of complex systems to deal with such shocks is often expressed in terms of the system's resilience. Although resilience is conventionally conceptualised as recovery from a single shock or disruption, in reality, systems often face a sequence of successive shocks, as is the case for earthquake aftershocks or epidemic waves. A system's ability to handle such a sequence of disruptions depends on its capacity to learn from preceding events. To capture adaptation and learning as transformative aspects of resilience, new approaches to account for performance changes across successive disruptions are needed. This paper addresses this gap by developing and empirically testing a novel, performance-based method to quantify adaptation to successive shocks, thereby contributing to the broader discussion on adaptation and resilience. We use the COVID-19 pandemic as a case study which was characterised by multiple waves across different countries and contexts, making it an ideal case to test our theoretical model. We use open-source data from Our World in Data, ranging from 2020 to 2023, and consider every COVID-19 wave as a disruption. Individual waves are identified by fitting mixtures of skew-normal curves to COVID-19 fatality time series, and a performance-based resilience metric, called the Total Performance Loss (TPL), is computed for each wave. Adaptation across successive waves is modelled using an exponential, learning-based metric that captures whether system performance improves or deteriorates over time. The results show that many countries initially showed maladaptive behaviour but adapted in later pandemic waves. The proposed methodology makes adaptation to successive disruptions measurable, and enables systematic comparison of how systems learn, and unlearn over time, which is an essential step towards understanding and improving responses to future crises.