Flash back a century, to the Spanish flu
The raging Covid-19 pandemic has whetted researchers’ interest in pandemic modelling.
The interest is partly because modelling multiple factors that link the pandemic to market performance is challenging. Usually, researchers have to also contend with uncertainty over the quantum and timing of policy interventions. Moreover, pandemic effects are linked to health system responses, immunity that people might develop over a period, and the possibility of a vaccine. These reactionary measures are uncertain, with no dependency on the past actions, and hence challenging to model.
In this paper, we address some of the challenges around modelling market risk factors and compare market situations between the two pandemics.
Of all the pandemics in the past century, Covid-19 has striking parallels with the Spanish flu (1918-1920), not just on the health and epidemiological side, but also in how financial markets have reacted to the two pandemics. Therefore, we have tried to identify risk comparison metrics – for use in market risk and stress-testing practices – for evaluating the severity of Covid-19, using the Spanish flu as a frame of reference.
Given the similarities and differences between the two, it is important to compare market behaviour during these pandemics to understand the factors present at the time better.
Only a few indicators were available during the Spanish flu to indicate market performance and reaction. If we look at the foreign exchange asset class, forex rates were fixed for most currencies during the Spanish flu, making any comparison inherently flawed. Prices prior to 1957, when the Commodity Research Bureau index – a representative indicator of global commodity markets was launched – are unreliable.
However, the Dow Jones Industrial Average (DJIA), which was trading at the time of the Spanish flu, does provide a reliable indicator, and we use this as a central measure of comparative performance between the two periods. As such, equity is typically the quickest to react to pandemics compared with other asset classes.
We use statistical parameters such as annualised volatility over 30 days, peak-to-trough fall, peak volatility, and correlation checks to compare the two pandemics. These parameters provide a comprehensive overview of index performance during pandemics, because they capture multiple parameters to gauge the indicator performance, e.g. uncertainty over a period, market correction, uncertainty during the peak of a flu, maximum drawdown of the DJIA index, and correlation between the death rate and index performance.