Guest Contribution: “Black Swans Like COVID-19 are Forecastable”

Today, we present a guest post written by Jeffrey Frankel, Harpel Professor at Harvard’s Kennedy School of Government, and formerly a member of the White House Council of Economic Advisers. A shorter version appeared in Project Syndicate  March 27th.


Events like the COVID-19 pandemic of 2020, the US housing crash of 2007-09, and the terrorist attack of September 11, 2001, can be called “black swans”: in each case, few people were able to predict them reliably.  But they were known unknowns, not unknown unknowns.  That is, in each case, knowledgeable analysts were fully aware that such a thing could happen, even that it was likely to happen eventually.  They could not predict that the event would happen with high probability in any given year.  But the consequences of each of these events were severe, and predictably so.  Thus, policymakers should have listened to the warnings and should have taken steps in advance. They could have helped avert or mitigate disaster if they had done so.

Listen to the Cassandras

After the danger of the new coronavirus had become apparent to all, US President Donald Trump repeatedly said that such a pandemic was “an unforeseen problem” that “nobody ever thought would be a problem.”  But of course epidemiology experts had warned about the danger for years, and as recently as last October.

After the attacks of September 11, 2001, President George W. Bush said, “Nobody in our government at least, and I don’t think the prior government, could envisage flying air planes into buildings.”  National Security Advisor Condoleezza Rice claimed: “no-one could have predicted that they would try to use an airplane as a missile.”  But in fact experts had warned about the possibility of terrorists flying planes into buildings like the Pentagon and even about the possibility of terrorists bringing down the World Trade Center.  The Cassandras included the Director of the counter-terrorism unit at the President’s National Security Council, Richard Clarke, the man who in August 2001 sent Bush a memo titled “Bin Laden Determined to Strike in the United States”.

It is tempting to attribute such failures entirely to the incompetence of some of our leaders.  But that is too facile an explanation.  Also caught completely by surprise were the public in general and the financial markets in particular.

The stock markets were at historic highs shortly before they crashed in the global financial crisis of late 2008 and again just before the current crash that began in late February 2020.  In both cases, there were plenty of candidates on the list of potential black swans.  Indeed, housing prices had already been in decline for over a year in the first case and the COVID-19 virus had already surfaced in Wuhan, China, by the start of 2020.

The worst misperceptions of investors did not just lie in their optimistic baseline forecasts. The markets did not even see a risk:  the VIX, which reflects investor perceptions of risk or uncertainty, was very low in advance of both 2007-09 and 2020.

Black swans versus narrow data sets

I see several systematic factors that help explain why such extreme events have come as complete surprises to our leaders, to the public, and to the financial markets.   Even the technical experts can be myopic: they sometimes fail to cast the net wide enough when they analyze the data.  They look only at recent data sets – which are by definition short data sets.  They think that because the world is changing rapidly, events from 100 years ago are irrelevant.

Americans often add a second set of blinders to their forgetfulness of history: they focus excessively on what has happened inside the United States.  They discount what happens in other countries, perhaps out of a belief in American exceptionalism (“the US is different”).

In 2006, finance whizzes who priced US mortgage-backed securities relied on the recent history of housing prices for their statistics.  They effectively applied a rule that housing prices never fall in nominal terms.  What they really meant was that they themselves had no experience with housing prices falling in nominal terms.  In reality, it had happened in the US in the 1930s, and it had happened in Japan in the 1990s, but it had not happened in the lifetimes of US analysts in the US.  If the analysis had included a longer or broader data set, the statistical estimates would have assigned a distinctly positive probability that housing prices would fall, and mortgage-backed securities would therefore crash.  (Government policy got it wrong too, of course, from subsidizing housing debt to inadequate financial regulation.)

Financial analysts who limit their data to their own country and their own time period are like 19th century British philosophers who concluded that all swans were white, by induction from their personal observation.  They had never been to Australia, where black swans had been discovered in a previous century, nor had they consulted an ornithologist.

Plagued by unpreparedness

Even when the experts get it right, the leaders often don’t listen to the experts.  One reason may be that the political system does not respond appropriately to estimates like a 5 % chance of disaster per year, even when the consequences of failing to prepare are large in expected value (that is, the probability of the event, multiplied by consequences of the event if it occurs, is large).

In advance of 2020, the experts got the risk of a serious pandemic right. Bill Gates and many other astute observers got it right. Even a multitude of Hollywood movie producers got it right.  And yet the US federal government was not prepared. The current Administration in 2018 eliminated the unit at the National Security Council that had been set up by the Obama Administration to deal with the risk of pandemics. It worked to cut sharply the budget for the Centers for Disease Control and Preparedness and other public health agencies (including the CDC office in China). That obliviousness helps explain why the US finds itself less well prepared than many other advanced countries like Singapore, lacking enough testing kits, respirators, ventilators, hospital beds, or and capacity.

Just as important, the White House didn’t have a plan ready, and did not know that it was important to develop one even during the months after the new coronavirus surfaced in Wuhan and it started to spread worldwide.  (Or perhaps the plan was: if we administer few tests, then we can report few confirmed cases and so keep the stock market up for another week.

Although Trump on March 19 said “Nobody has ever seen anything like this before,” there had been plenty of precedents.  The very deadly Ebola virus killed 11,000 people in 2014-16; but the victims were far away in Africa.  The influenza pandemic of 1918-19 killed 675,000 Americans (among some 50 million worldwide); but that wasn’t in living memory.  If you haven’t seen a black swan with your own eyes, then it must be that nobody ever has.

Apparently, in order to make an impression on our leaders, a disaster has to kill lots of fellow citizens within their own memory spans.  Such a lesson is underway, worldwide, today.  Let us hope that the price of the lesson, in the number of deaths, is not too high.


This post written by Jeffrey Frankel.

17 thoughts on “Guest Contribution: “Black Swans Like COVID-19 are Forecastable”

  1. 2slugbaits

    Isn’t the COVID-19 pandemic an example of a “gray swan” rather than a “black swan”? We had some previous limited experience with smaller scale events, so anticipating the COVID-19 pandemic only required us to scale up previous pandemics.

    1. Barkley Rosser

      Nassim Taleb, the person who coined these terms, I think would agree with you, 2slug. In The Black Swan he specificallly argued that 9/11 was a gray swan, something unusual but not unpredictable, with many people forecasting that something like that could happen. Likewise he later said 2008 crash and Great Recession were also grey swans, foreseen by many.

      For him, the still-largest one day crash ever, 22 percent of Dow on Oct. 19, 1987, was a black swan. Came out of nowhere. He also said that the collapse of his home nation Lebanon into civil war in the 1970s was for him a black swan.

      Regarding this pandemic, it is true none of us alive have seen anything like it, but the Spanish flu was only a century ago, and quite a few medical observers have been warning of something like this. Indeed, apparently the Trump admin itself considered hypothetical cases that looked very much like it, a major flu-like disease coming out of hina. But even as they could see the possibility of it, they did nothing about it. Anyway, much more of a gray swan than a black one.

  2. pgl

    “in order to make an impression on our leaders, a disaster has to kill lots of fellow citizens within their own memory spans. Such a lesson is underway, worldwide, today”.

    Of course not everyone is as narrowly focused as President MAGA. One would hope an operation as large as the Federal government could keep on staff people smart enough to retain the lessons from history around the world as well as good science. But then – such people would not sit around all day reaping praises on the genius of Donald John Trump so he get rid of these expert advisers. And here we are.

  3. Anonymous

    Nobody gets a bailout for foreseeable problems, therefore all major dislocations become “unforeseeable” regardless of how obvious it is that they will occur *sometime*. In the case of epidemics, they happen every 7-10 years.

      1. Anonymous

        Fair enough … we’ll know in the fullness of time just how dangerous this is compared to prior viral epidemics. It took a couple of years before we got a full accounting of the swine flu. The map is not the territory, the policy response is not the virus.

        In any event, what I have seen very often in my lifetime is extremely highly-paid CEOs claim ignorance of problems that were very foreseeable, and were in fact foreseen by others (who weren’t nearly as highly-paid) well in advance. They usually do this when asking for a bailout.

      2. Moses Herzog

        @ Menzie and Anon
        Can I say something while remaining relatively neutral on this?? I think both of you are “right”. It’s not that either one of you doesn’t have in fact a very strong point to make on both sides of it. It’s really the same crash scene viewed from different angles. Here is my question: If some Catholic Cathedral had the foresight for this (and remember I’ve made it clear on this blog before I’m no fan of the Catholic Church), If Goldman Sachs had the foresight for this (ditto on my dislike of some of GS behaviors), and if Mark Zuckerberg had the foresight for this (ditto on my low regard for Zuckerberg’s lack of morals), then why were hospital administrators (who BTW, sometimes command 8 figures or more in salary) across this entire nation caught with their pants down around their ankles??

        Nurses and doctors are even whining about it. My question to nurses and doctors is: “You couldn’t even keep a personal stock of twenty to thirty N-95 masks when scientists have been warning us for the last 30 years it was a ticking time-bomb that pandemics would happen, to keep yourself and YOUR OWN FAMILY safe?? NEVERMIND you knew those hospital administrators on 8+ figure salaries have never given a damn about you from the get-go. If you couldn’t even do that, what sentimental emotion am I supposed to feel towards you here???

          1. Moses Herzog

            I’m saying hospital administrators should have had a deep backlog of N-95 masks, and medical workers should have purchased twenty to thirty masks on their own, as a personal stock. I am saying those who work in the medical/health field should have had masks. And it’s not a “fat-tail” when you’re viewing it over a long time horizon. Which is how hospitals and hospital administrators should have viewed this. This may be super hard for people to “get”, but it’s a life and death event here. You have 3,500+ Americans dead now. What do you think that equates to in value of a backlog of N-95 masks??

            Have you ever gone through a fire drill when you were in grade school?? So how many schools did you go to that burned down?? Why would it be that school districts and school administrators do that would you guess??

          2. 2slugbaits

            Moses Herzog I think you misunderstood my question. I wasn’t asking if you thought we should have insurance stocks of critical medical supplies, I asked if you thought people should self-insure against the possibility of fat-tailed events. Let me give you an analogy in a field that I actually happen to know a lot about. In the Army there has always been a tension between front line “killer” units that want to stock as much as they can against every imaginable contingency. That’s understandable given that those units are the ones whose life literally depends upon having enough materiel. That’s an example of individual battalions and brigades wanting to self-insure to fight any contingency. But from an operations research perspective that approach is crazy. The federal government should procure and store critical medical supplies in the same way that the Army centrally procures and stores critical materiel as part of its war reserve stocks. In fact, you might even remember this, but many years ago 60 Minutes had a supposed scandal story that explored the huge number of medical gowns and masks that DoD had stored at one its depots. The depot commander looked foolish because he couldn’t explain it; however, that’s to be expected because it was never his job to know why DoD felt we needed that many gowns and masks. The commander’s job was to store the stuff, not determine how much was too much. In the end it caused a political scandal and gave DoD a disastrous “war on excess.” So DoD threw out all those gowns and masks in order to avoid further political heat even though it made no sense in terms of managing a crisis.

            One other lesson from my professional experience working with wartime consumption rates. The demand for COVID-19 medical equipment reflects a fat-tailed distribution, but most people (as well as most commercial ERP software) look at the world in terms of a normal distribution. In a normal distribution a 7 standard deviation event is essentially impossible; but in a fat-tailed world 7 standard deviation events are entirely possible and cannot be ignored. In modeling wartime casualties and consumption rates you ignore fat-tailed distributions at your peril. And unlike most commercial and peacetime oriented steady-state models, wartime models require a dynamic approach. And dynamic math is a lot more complicated than some steady-state Lagrangian model. We need to apply that same kind of thinking to future pandemics.

          3. Baffling

            Moses, many health care professionals had an emergency supply. The problem is they already went through their supply. The hope was that reinforcements would eventually arrive. But the cavalry got lost, or given poor directions. At any rate, have to disagree with blaming medical professionals who risk their lives for not storing up ppe’s. I also wont blame a soldier for lack of resupply bullets.

          4. Barkley Rosser


            Oh dear, I just fully read your comment above. You state, “And it’s not a ‘fat tail’ when you are viewing it over a long time horizon.”

            You could not be more wrong. It takes a lot of data, which one gathers over a long time horizon, to clearly identify what is a fat tail and what is not. The longer the time horizon, the better you know what is and is not a fat tail. You could not be more wrong.

            As it is, this is a reminder of your massive ignorance regarding skewed distributions that you have repeatedly reminded us about here. This looks more of the same. You may have gotten a degree in finance, but I find it hard to believe that you managed to pass the statistics course you were almost surely required to take to get that degree.

  4. anon_coward

    The FBI had been following the 9/11 hijackers for months before the event. they were taking flying lessons in cessnas and asking about the difference for an airliner and didn’t care about learning how to land. the laws prior to 9/11 put limits on the FBI and what they could do

    lots of bloggers were following the 2007-2009 crash for years before it happened. I used to read The Housing Bubble Blog

    1. Menzie Chinn Post author

      anon_coward: Sure, the FBI had been following the hijackers, but the chief White House adviser, Richard A. Clarke, advocated escalating the alert status much higher, and was rebuffed during the G.W. Bush administration. (Like Trump and the coronavirus, the previous administration’s warnings were dismissed by the incoming administration.) Knowledge at the mid-level is no substitute for attention at the highest policy levels.

  5. William Jefferies

    This is a very pertinent article – Middle East respiratory syndrome: current status and future prospects for vaccine development 2015 Lanying Du & Shibo Jiang from “Expert Opinion on Biological Therapy” – “”The outbreaks of Middle East respiratory syndrome (MERS) previously in Middle East and recently in South Korea have raised serious concerns worldwide, reinforcing the importance of developing effective and safe vaccines against MERS-coronavirus (MERS-CoV). A number of vaccine candidates have been developed on the basis of viral vectors, recombinant proteins, DNAs, nanoparticles, and recombinant MERS-CoV, and some of them have shown efficacy in laboratory animals. However, the paucity of financial support has made it difficult to transfer effective candidates from the preclinical stage to clinical trials. Here, we summarize currently available MERS vaccine candidates and illustrate strategies for future development, with the aim of provoking government agencies and Big Pharma to invest more funds for developing efficacious and safe MERS vaccines.” They conclude

    “Big Pharma has consistently refused to invest funds for manufacturing or developing in pre-clinical and clinical studies based on the uncertainty of immediate return on such investments.”

    1. baffling

      NIH should be funding this work. uncertain government budgets make that very difficult.

Comments are closed.