How Unanticipated COVID Cases Impact Stock Returns

One of the most interesting academic papers on pandemics is a new NBER paper on COVID (and SARS) case predictions and stock market returns. The essence of the paper is that “unanticipated” case increases lead to large stock market declines. For example, in the cases of the SARS outbreak, if the latest estimated cases was double the day earlier estimate, then the Hang Seng would fall between 8%-11%. And in the case of the current COVID outbreak, a similar doubling sees a 4%-10% decline in the Wilshire 5000 stock index for the US. The study is still ongoing and so updates are coming soon.

Picking The Right Compounding Approach

The starting point of their work to come up with a model that predicts cumulative cases. For this, they use both exponential and logistic specifications – both are frequently used in biology and epidemiology. These specifications allow for the explosive dynamics often seen with disease outbreaks. An example of using both functions is shown in the first chart. The actual cases (black), exponential model (green) and logistic model (red) initially move together but by around day 100, they start to diverge. In this case, the logistic function follows the actual cases better. For SARS, the logistic function worked best, while for COVID, as of the time of publication, the exponential function works better.

Calculating “Unanticapated” Cases

The crux of the study hinges around the uncertainty of disease outbreak. Therefore to capture this, they try to predict the cumulative cases on day t, by first using the logistic/exponential function with data up to two days earlier (t-2). They then come up with another prediction for day t but this time using data up to the day before (t-1). If the disease is behaving “well” then the predictions should be the same as it is following a known logistic or exponential path. However, if there is a large gap between the two predictions, then it means the path of the disease is still not well understood. This is the “unanticipated” cases or change in predicted cases.

They run regressions of these unanticipated cases against stock returns to establish whether they have any explanatory power.

How Did It Work Over SARS?

For the SARS outbreak they use HK rather the China due to quality of data issues. They find a statistically significant relationship between changes in predicted cases and Hang Seng returns (see chart). A doubling in the change in predicted cases would lead to 8% to 11% declines in the Hang Seng depending on the regression specification.

How Does It Work For COVID-19?

Given its current widespread interest, the authors took a more elaborate approach in this case compared to the SARS analysis. Not only did they analyse the equity returns with both market opening and market closing prices, but they also introduced policy elements in the regression. These policy elements include:

a) SIndex – Government Response Stringency Index developed by Oxford University, which tracks travel restrictions, trade patterns, school openings, social distancing and other such measures, by country and day.

b) Fiscal stimulus variable – in the form of a dummy variable which takes the value of 1 every time a major fiscal package was voted in Congress

As it turns out, neither of these variables are found to be significant, though this may change in the future according to the authors. The regressions are run from the 22^nd of January to the 27^th of March 2020, on a daily basis. Like with the SARS example, they find a similar downwards relationship between changes in predicted cases and stocks (see chart).

In terms of numbers, a doubling in predicted cases leads to declines of 4% to 10% depending on how the regression is specified. Using closing stock market prices show larger moves, but this is likely due to more information being available.

Bottom Line

This work suggests that the critical driver of stock market returns during an epidemic is the certainty around the path of the disease. The more stable and predictable, the less the sensitivity of stock markets. Equally, the more unstable the path of the disease, the larger the stock market sensitivity.

There are some puzzling features of the work – it appears that neither health guidance, such as social distancing, nor fiscal policy announcements have much impact. This could be due to such announcements reflecting concern around the epidemic and so any potential positive impact is offset, or it could be investors simply looking at the number of cases as the true test of success. On the policy side, the authors didn’t account for monetary policy, which would likely be more significant than fiscal policy.

Stefan Posea is a Research Analyst at Macro Hive. His research interests lie in macro-financial interactions and monetary policy analysis. Stefan graduated with an MSc in Economics at Birkbeck, University of London and previously held roles in M&A and the Public Sector.