The Coronavirus Emergency: A Quantitative View

This analysis was written on March 20, 2020, click here to access the full Northfield research.

The coronavirus pandemic represents an unprecedented challenge to all investors. The global situation is deeply uncertain and subject to change in conditions by the hour. However, Northfield owes it to our clients to say something we believe substantive and sensible about the crisis, rather than stay silent. In the current atmosphere of crisis, it is critical that we help institutional investors have a logical rather than emotional basis for their actions.

The worst-case scenarios for global coronavirus mortality are certainly unimaginable as a human tragedy. Our estimate of the worst case is that the entire world ignores the pandemic (obviously already untrue) in which everyone in the world becomes infected with the coronavirus, leading to mass mortality of around 60 million. While terrifying, the 60 million casualty estimate is about .8% of the current world population. While casualties of war are hard to tally with precision, that figure of .8% is around the higher end of the estimated range for each of the eight plus years of the World Wars. When we add in the massive number of deaths (median estimate 35 million) from the Spanish Flu epidemic of 1918, the maximum likelihood estimate for abnormal mortality for that year was about 1.9% of the then global population.

Implications for Long Horizon Investment Institutions

The impact of the variation in mortality rates from conflict casualties on financial markets is summarized in this essay from 2015. Even if the implausible worst case of 60 million deaths were to come about, this would represent about a 9% increase in global mortality over a ten-year time horizon (much smaller than multiple years of sustained large-scale war). When we did the war study cited above, the correlation of decade-long returns with the variations in mortality rates was on the order of -.35 (mortality rates up, equity market returns down).

As a concept, risk is always in the future. Volatility of asset values has occurred in the past, and volatility of asset values will occur in the future, but there is no risk in the past. What is done is done. The problem is that most organizations equate past volatility and future risk in wholly unsound ways. The potential for rare extreme events of any kind (think Pompeii in 79 AD) is routinely ignored by financial institutions.

Let’s walk through the financial algebra for a moment for a typical institutional investor. Initial assumptions are:

1. Future equity returns would be 6% in an average year with a volatility of 15% (this is a geometric mean return of 4.875%) or a cumulative return of 59% over a decade
2. Future fixed interest returns would be 2% with a volatility of 7% (this a geometric mean return of 1.755% annually)
3. The correlation of equities and fixed interest returns is .3
4. The investor is 60% equities and 40% fixed interest by asset value
5. The total portfolio expected arithmetic return works out to 4.4% with a volatility of 10.67.
   1. This equates to a geometric mean return of 3.83% annually or a cumulative return of 45.6% over a decade

So, if an investor had a 10-year time horizon and the pandemic effects are similar to war, the expectation of the cumulative return of their portfolio would decline by -1.44%. This arises from the expected return of 45.6% over 10 years times the 9% total increase in mortality over the decade times the correlation coefficient of -.35 from the 2015 study. So, the expectation of cumulative return over a decade declines from 45.6% to 44.16% which is a net geometric mean return of 3.72%. As you can see, the expectation of the geometric mean annual return on the investor’s portfolio has declined only by a very modest one-tenth of 1% per annum, conditional on an extremely grim scenario for mortality.

There are also other considerations at play which suggest this is a pessimistic assessment in other ways as well. War destroys physical capital (factories, roads, trucks) and war is expensive to wage, heavily impacting bond markets. In one year of World War II, the U.S. military budget reached 35% of GDP. Nothing like that is plausible in this situation. Even the massive financial steps just proposed by the U.S. government represent a one-time event of about 9% of GDP. War casualties are also skewed toward younger persons who would otherwise be the most productive members of an industrial society. The coronavirus mortality is skewed toward the elderly.

As physicist Niels Bohr said, “it is always hard to forecast, especially about the future,” so I usually decline to prognosticate. However, it seems that the current crashing of markets around the world cannot be explained by rational actions of long-term investors conditioned on the only available data on large variations of mortality rates. We must therefore fall back onto a couple possible explanations.

1. Investors are very short sighted so that nobody is thinking logically about 10-year horizons right now. Obviously, if you think about a worst-case scenario of 60 million dead over a span of a few months it is very, very, scary, as are the likely economic impacts of virus mitigation efforts.
2. Investors switching from risky assets to riskless assets (e.g. cash, sovereign debt) have a relatively simple decision as they don’t have to decide what the risk-free asset is. However, investors choosing to move from riskless assets to risky assets must make decisions about what risky assets they believe are appropriate. This means that selling is always faster than buying which leads to crashes which are eventually made up by long growth periods (e.g. the GFC was followed up by an 11-year bull market in global equities). A similar pattern of a long expansion was experienced in the “roaring 20’s” after the Spanish flu pandemic in 1918 was roughly coincident with the end of World War I.
3. Investors are relatively indifferent to small changes in their wealth level but extremely sensitive to larger changes in wealth so the current a large amount of selling by investors may be partly logical but mostly not. You can also make a “behavioral” argument in the form of “Cumulative Prospect Theory.”

Investor Risks in the Short Run

As described in the section above, one explanation for the large recent decline in equity markets is that investors are not thinking long term, but rather are thinking about their portfolio assets “one day at a time.” While Northfield takes no position on the likely success or failure of measures taken by various governments to stimulate economic activity and calm investors, we can certainly shed light on daily variations in investor confidence.

To assess very short-term risk, Northfield has maintained our U.S. Short Term Risk Model since 1997. In this approach, a statistical factor model is adjusted daily for changes in the implied volatility of options traded on equities in the U.S. A mathematical process then maps the day to day changes in security level volatility across the factors of covariance so that changes in market conditions are applied to all securities, not just equities in which options are traded. Security coverage includes all non-U.S. equities traded on U.S. exchanges in ADR form, so the model does cover most large publicly traded firms globally. Mathematical details of the model are presented in this February 2002 research, which was subsequently published in 2005. Although volatility levels are presented in the usual annualized units, the intended time horizon for the risk forecast is the next trading day.

The chart below shows the expected volatility of the S&P 500 on an equal weighted basis, the S&P 500 on the conventional capitalization weighted basis, and the tracking error of the two portfolios. The time period is from the end of October 2019 through to March 16, 2020, at a daily frequency.

The chart shows that the volatility values for both series increased roughly five-fold over the sample period. Peak annualized volatility values of over 60% were observed on March 13 but decreased to 52% for the capitalization weighted index and 55% on the equal weighted index by March 16. As is common in crisis periods, correlations have increased with the expected correlation of the two portfolios going from .968 at the start to .984 at the end.

{chart}

Conclusions

It should be clear that the coronavirus pandemic is a very serious matter. Allowed to spread entirely without limit the death toll would be massive in number and comparable to the peak year of World War I or World War II as percentage of the population. However, it is also obvious that steps have already been taken to slow the spread of the infection, even if those efforts have seemed laggard in some countries.

While we do not take a position on the matters of morality, ethics, or public policy associated with the pandemic, it is clear that even worst case scenarios for related mortality and health care expenditures, the economic impact for financial market investors should be minimal when observed over long horizons of 10 years or more.

Much greater risk to investors and the world economy at large comes from clumsy handling of mitigation measures than from the virus itself. Of greatest concern is that many organizations, companies, and governments have been engaged in an undesirable game of managing the “optics” of the situation. Whether out of concern for liability in litigation, the ability to enforce business interruption insurance, political expediency, or reputational issues of being thought less prudent than their cohorts, the actions of many entities are being driven by a desire to be perceived as managing risk as opposed to actually managing risk in response to factual information and analysis.

This blog post has been written by a third-party contributor and does not necessarily reflect the opinion of FactSet Research Systems Inc.