Scientific comment on "Tail risk of contagious diseases"

TL;DR

This paper critically reexamines claims that major epidemics have a fat-tailed fatality distribution, showing that the data can also be explained by other distributions like the log-normal, thus questioning the original conclusion.

Contribution

It provides a reanalysis of epidemic fatality data, demonstrating that fat-tailed distributions are not conclusively supported and alternative models fit the data equally well.

Findings

Data may be compatible with power-law tails but are not conclusive.

Log-normal distribution can replicate the empirical data's statistics.

Original claims of fat-tailed epidemic fatalities are not definitively supported.

Abstract

Cirillo and Taleb [Nature Phys. 16, 606-613 (2020)] study the size of major epidemics in human history in terms of the number of fatalities. Using the figures from 72 epidemics, from the plague of Athens (429 BC) to the COVID-19 (2019-2020), they claim that the resulting fatality distribution is ``extremely fat-tailed'', i.e., asymptotically a power law. This has important consequences for risk, as the mean value of the fatality distribution becomes infinite. Reanalyzing the same data, we find that, although the data may be compatible with a power-law tail, these results are not conclusive, and other distributions, not fat-tailed, could explain the data equally well. Simulation of a log-normally distributed random variable provides synthetic data whose statistics are undistinguishable from the statistics of the empirical data.