“It’s tough to make predictions, especially about the future.” — Yogi Berra
Existence is fragile. Our modern world makes it difficult to remember this until circumstances take a turn for the worst. If history can promise us anything, it’s that this fragility isn’t going away anytime soon, so it’s best to be humble to this principle and embrace it in the furtherance of mitigating risk. Which brings us to the coronavirus.
It seems that like most things in today’s media, the opinions people have on the coronavirus are extremely polarized. Either you’re in the “coronavirus is just like the flu” camp or you’re in the “we’re all gonna die, this is the end of society” camp. Regardless of where you fall on the spectrum, it is interesting to see how the understanding of complex systems, tail risk, and incentives have revealed both professional illiteracy of mathematics and risk, as well as the complete incompetence of many politicians.
We have a perfect storm. Politicians and health authorities acting too late, don’t test don’t tell, and a centralized bureaucratic healthcare system all create the necessary conditions for increased mortality and economic collapse. To me this is not an argument about who’s right and who’s wrong, it’s about mitigating the potential for disaster. Taking the exact same steps at different times can have vastly different outcomes, and the decisions that we as individuals can make to influence these outcomes are critical regardless of whether the health authorities and politicians take heed.
It’s Just Like The Flu
No, it’s not. The numbers above demonstrate a best estimate of important metrics and how the coronavirus is dramatically different from the annual flu numbers. Based on other estimates that I’ve seen, the numbers above may be very optimistic given what we’ve seen from Italy and China, especially for the hospitalization rate and R0. Remember that R0 is influenced by the proactive measures that a country takes. The asymptomatic transmission rate makes the coronavirus particularly concerning relative to the flu. The idea that one could take a new virus with a significant amount of unknowns and nonlinear effects, and somehow map that onto an annual phenomena is indicative of the myopia that many people have.
Summer Will Kill It
Maybe, but if past pandemics are indicative of possible outcomes, then perhaps we could see a reduction in cases over the summer, followed by complacency and an even stronger resurgence of cases in the fall/winter. This is exactly what happened with the Spanish flu from one hundred years ago.
If we look at the graph above we see that both the second and third wave were far more deadly than the first. One could argue that the variables surrounding the Spanish Flu were worse (they had worse medical care, etc), but you could make similar arguments of concern for the coronavirus (we are a more globalized society, etc). It doesn’t matter. What matters is that we are dealing with the uncertainty of an outcome X, and we don’t even know what X is. That’s why this needs to be taken seriously.
Healthcare System Collapse
Let’s take New York City as an example. There are somewhere between 22000–26000 hospital beds depending on the data. We’ll be generous and take the upper bound. Even if we do that we have to factor in the existing utilization of hospital beds and the fact that the beds for coronavirus patients have to be completely isolated. What does that leave us with? De Blasio says he can make 1200 beds available immediately. Let’s give him the benefit of the doubt and say that NYC could muster a total of 5000 beds through extensive restructuring to allow for the new accommodations to still be of the highest isolation (big if). We’ll say NYC has 8 million people and take the lower bounds of the all figures above to be as optimistic as possible: 10 days in the hospital, 5% requiring hospitalization, and a 30% total infection rate. That’s 120,000 hospitalizations. Again, this is in all likelihood far too low (if we take the lower end of the World Health Organization’s infection rate of 40% and 10% hospitalizations, that translates to 320,000 hospitalizations, and this would still assume the lower bound of days in the hospital). “But wait!” — you say, “that’s misrepresenting the actual number of hospitalizations at any given time because not everyone will be hospitalized at the same time.” True, and this is why we are going to have a problem. People, including health professionals and politicians, think about hospitalization in linear terms — limiting their extrapolation to numbers like the uniform distribution throughout the year or whatever time period they have in mind. This mistake is what will cause people to die. Initially the cases will trickle into the hospitals 1 at a time, then 10 at a time, then 100 at time, and so on. Barring effective containment, which looks extremely unlikely at this point, the best way to predict the hospital bed shortage will be to look at how fast the infection is doubling. Based on the above graphic, we’ll say it doubles weekly. Once the hospital system gets 1,000 hospitalizations, by the end of that month that number swells to 16,000. Planning a triage unit to accommodate linear future expectations will not work to fight a virus defined by exponentials.
There have been many comments on mortality and the numbers above make this statistic very disconcerting. The virus mortality may very well be as low as 1% (I’m skeptical), but how much is this mortality rate conditional on adequate medical care, especially when it comes to hospitalization? How many deaths from people with conditions unrelated to the virus die because the health care system cannot function properly due to the sudden increase in coronavirus patients? These are the things we cannot see until they become statistics of the past.
Beyond Health
“Of course the word chaos is used in rather a vague sense by a lot of writers, but in physics it means a particular phenomenon, namely that in a nonlinear system the outcome is often indefinitely, arbitrarily sensitive to tiny changes in the initial condition.” — Murray Gell-Mann
So after all this fear mongering I’ve written above, you somehow manage to survive contracting the virus or never get infected. Well, that’s only one of the risks and arguably the one with the shortest duration probabilistically speaking. Let’s talk about the supply chain — you know — the one that was supposed to factor in risk by having multiple sourcing plans in different countries to combat uncertainty from government, natural disaster, and other acts of God. Oh yeah, very few companies did that because it didn’t pay off in the short term and they were incentivized to distribute this risk into the future rather than address it in the present.
Supply Chain: What We Know So Far
Before we go into detail, it’s important to know the difference between a tier 1 and tier 2 supplier. A tier 1 supplier is an organization that supplies directly to a company, whereas a tier 2 supplier is an organization that supplies to a tier 1 organization. It suffices to say that many companies are very interconnected in ways that they don’t even know due to the levels of dependencies that have developed in our globalized society. To give you an example, Volkswagen has 5,000 tier 1 suppliers, each with an average of 250 tier 2 suppliers. That’s over 1,000,000 tier 2 suppliers for one company. Based on the best data available right now, we know that 51,000 companies have at least one tier 1 supplier in Wuhan and 5,000,000 companies have a tier 2 supplier in the region. That includes 938 of the fortune 1000 companies. We are only beginning to feel the true magnitude of this disruption now.
Particularly concerning is the supply chain for pharmaceuticals. India has already placed limits on the export of 26 pharmaceutical ingredients, including several antibiotics. They also source about 20% of the world’s generic drug supply and 70% of India’s pharmaceuticals are sourced from China. There is also a transparency issue when it comes to tracking supply chain risk for pharmaceuticals. While drug manufacturers must report to the FDA if they have any potential disruptions to their suppliers, the same requirement does not extend to FDA-registered active pharmaceutical ingredient facilities.
It’s important to note, as it goes with the theme, supply chain effects are also nonlinear. Localized effects with small amounts of tier 1 or tier 2 disruption are better managed and bad effects can be mitigated by potential substitution or other delaying measures. However, past a certain critical point, if too many companies’ supply chains are disrupted, then the problem magnifies exponentially, as there are too many interdependencies that become disrupted. Hidden weaknesses are only exposed when it becomes a serious problem.
If we go into an economic depression or face other severe economic conditions, it’s important to note that the coronavirus is NOT the underlying cause, but rather the straw that broke the camel’s back. In this context, there are a plethora of other events that could’ve achieved the same effect, but it just so happens to be the coronavirus in the current reality. Through bad monetary policy and the financial engineering of companies, society has been playing hot potato with a stick of dynamite, fantasizing about how to extend the fuse forever.
“People take the longest possible paths, digress to numerous dead ends, and make all kinds of mistakes. Then historians come along and write summaries of this messy, nonlinear process and make it appear like a simple, straight line.” — Dean Kamen
Incentives
Of the many problems to the response of the outbreak, the incentives and game theory behind what a politician or country should do are extremely challenging, both in part due to the lack of understanding of the complexity described above, but also due to the lose-lose position in which they find themselves. The politician that panics early and contains the virus may be vilified by his opponent for overreacting and disrupting the daily lives of the people (historian’s fallacy). Panic too late and your opponents accuse you of negligence. So what are you to do? So far we’ve seen lots of paralysis from the politicians in the countries with the earliest outbreaks, as panicking would’ve been the contrarian decision that carried more risk. As outbreaks continue to occur in new countries, it will no longer be contrarian for politicians to implement containment measures stemming from panic. The risks change as the panic response becomes conventional. This is a case for optimism as time passes.
Conclusion
Based on the points above, it’s hard to support the “it’s just the flu” argument. Okay fine, but the defeatist in you says,“we’re all going to get it anyway so why do anything?” (this assumes you’re healthy and young). Well, for one thing, there’s still a lot we don’t understand about the virus. So even if you wanted to roll the dice and hope that you don’t need hospitalization, or that if you do that you will get a bed, there are still a host of other risks that you would incur by throwing caution to the wind (unknown long term complications anyone?). If we’re all to get it at some point, I would prefer to do what is in my control to delay this as long as possible in the hopes that we will know much more about the virus as time progresses.
Panicking early is not the same as panicking late because of the exponential risks the virus presents. Panic is reflexive in that the less people who care and downplay the risk, the greater the chance of worse outcomes and vice versa. Protective measures slow the growth of the virus, so even if it continues to spread, it smooths out the risk to hospital systems that would otherwise have to accommodate larger exponential increases in patients requiring hospitalization. This is literally a matter of life and death, not because of the virus, but because we fail to heed the precautionary principle. Our perception of the coronavirus is not an independent variable, it is part of the system.
Evolution favors those who are paranoid and those who panic. This is not an argument for panicking at every little thing that happens, but it is important to understand the difference between localized risks (car accidents) and global risks (pandemics). For global risks, the costs of panicking in which the outcome is benign is far less than the cost of not panicking in which the outcome is severe. Your ancestors would agree for it’s the reason you exist at all.
Sources:
https://www.statnews.com/2020/03/03/who-is-getting-sick-and-how-sick-a-breakdown-of-coronavirus-risk-by-demographic-factors/
https://thehill.com/changing-america/well-being/prevention-cures/484942-japan-confirms-first-case-of-person-reinfected
https://hbr.org/2020/03/coronavirus-is-proving-that-we-need-more-resilient-supply-chains
https://www.dnb.com/content/dam/english/economic-and-industry-insight/DNB_Business_Impact_of_the_Coronavirus_US.pdf
https://www.theguardian.com/world/2020/mar/04/india-limits-medicine-exports-coronavirus-paracetamol-antibiotics
https://www.modernhealthcare.com/supply-chain/coronavirus-strains-fragile-pharmaceutical-supply-chain
