COVID-19 isn’t the first and won’t be the last pandemic to threaten the well-being of the global population, but it will likely turn out to be the worst in the last 100 years.
Executive SummaryBest practices that have evolved around the use of natural catastrophe models have value for managing future pandemics, writes risk modeling expert Karen Clark. Among them is the use of probability distributions around important model outputs—fatalities and hospitalizations in the case of pandemics—reflecting the uncertainty around model assumptions such as transmission and fatality rates. Continuous model revisions to incorporate new data and the use of random data samples are other takeaways from hurricane and earthquake modeling that would be useful for pandemics.
It would be hard to argue that the handling of this crisis has been ideal, and there is much to be learned and improved upon before the next virus strikes. Policy responses should be based on the science and the data, but because there is so much uncertainty around the data in the early phases of a pandemic, and in particular the data used in the pandemic models, a more robust framework for leveraging the science is called for.
As the insurance industry knows very well, despite wide inherent uncertainty, robust models used correctly can be valuable tools for decision-makers. Insurers also know that models can produce widely divergent results, so it’s imperative to understand the model assumptions, the variability around those assumptions and what’s ultimately driving the model output. Best practices that have evolved around the use of catastrophe models have value for managing future pandemics.