[Updates and corrections below original post.]

In The 2020s and Beyond I mentioned that “black swans” (unpredictable, extreme events) can seriously mess up any prediction. In that article I sketched in broad strokes what I expect to be the main trends for the coming two decades. Rising fascism, economic crisis, (civil) war, and climate disaster or environmental collapse could serve as keywords summarizing that sketch. I also mentioned three possible black swans – a solar storm, nuclear war, and technological breakthroughs – but a kind of black swan that I didn’t mention is a pandemic. COVID-19 may very well develop into a pandemic, which made me wonder what kind of effects it could have, or in other words, how it could mess up my predictions. (On a side note: I briefly considered giving this article the title “Why you should be worried about the new corona virus”, but I don’t like clickbaity titles, so I decided against it.)

Obviously, before considering any kinds of effects, we should know how likely it actually is that COVID-19 becomes a pandemic. There are a number of reasons to believe this is very likely. Firstly, the disease appears to be extremely infectious. A single woman visiting a church meeting in Korea infected hundreds of others at that same meeting. Assuming that she didn’t go around coughing in all of their faces, mere presence in the same room for a few hours is apparently sufficient for the infection to spread. Similarly, in Japan a single taxi driver infected many others at a party. Secondly, symptoms of the disease start very late after infection. An infected person might not know that they are infected for more than a week or even two weeks because they don’t have any obvious symptoms yet, and all that time they can apparently infect (many) others. And thirdly, some infected people never develop any symptoms at all, but still infect other people.1

For these reasons, I don’t believe that COVID-19 can be contained. It will continue to spread. I don’t think that any of the experts involved really believe that it can be contained either. The purpose of trying to contain it is to slow down its spread to create sufficient time to develop vaccines, antivirals, and treatment protocols, and to make other preparations. (Unfortunately, it seems that politicians and bureaucrats still believe that COVID-19 can be contained, and thus continue to waste time rather than make such preparations.)

Obviously, if COVID-19 becomes a pandemic, that would result in a significant peak in mortality. How tall that peak is going to be is hard to say because there isn’t much reliable data (publicly) available yet. The most useful source of information is a report published by the Chinese Center for Disease Control and Prevention (CCDC) on February 11th.2 That report breaks down all known cases in China (mostly in Hubei) into a number of categories, most importantly by age. It also gives the case fatality rate (CFR) per age group. This varies from 0% for children under 10 (because there had been no deaths in that age group) to almost 15% for people over 80. (There may be a time lag effect in these numbers. If that is the case, real mortality will be higher.)

It is important to realize that CFR is not the total mortality rate per age group, but only the percentage of diagnosed cases that died. The CCDC report also gives the number of diagnosed cases per age group, but does not give infection rates – that is, the percentage of people in an age group that got infected. Given available data, the closest would something like a diagnosed case rate (DCR), the number of diagnosed cases divided by the population size in each age group. Calculating this would require data about the total number of people in each age group in Hubei. Unfortunately, I could not find that data, so the next best thing would be to assume that Hubei’s age pyramid looks similar to that of China as a whole, with a slight adjustment for Wuhan’s enormous student population (because the city is a major higher education hub). If DCR is a good proxy for the infection rate, then the data suggests that the chance for someone over 60 to get infected is more than 20 times higher than the chance for someone under 20 (which I find a bit hard to believe). If this is right, then old people have a much higher chance to get infected, and if they get infected, they have a much higher chance to die from COVID-19.

However, this isn’t exactly right, as DCR is not really the same thing as the infection rate. People who are infected, but show no typical symptoms (or no symptoms at all) and have no other reason to believe they might be infected may never be diagnosed. Perhaps, there are many more young people infected, but because of their complete lack of symptoms they were never tested. Hence, we need to know the ratio of diagnostic cases to total infections, and that number doesn’t exist, of course, because the first number aims to approximate the second.

As mentioned, DCRs differ significantly between age groups, but the larger the percentage of people that gets infected, the smaller the differences between age groups. Obviously, in a hypothetical case of 100% infection, the infection rates for all age groups would be 100% as well, and thus there would be no difference at all. Taking this into account, age-relative DCRs and age-relative CFRs can be combined to give expected COVID-19-caused mortality rates at different total (population) infection rates (again, assuming that DCR is a decent proxy for the infection rate). These are shown for 35%, 50%, and 65% total infections in a population with a similar age structure as that of Hubei in the following figure:

mortality rates per age group for different infection/case rates

mortality rates per age group for different infection/case rates

If in such a population 65% of people get infected, then COVID-19 would kill 3.5% of people between 60 and 70 and 13.4% of people over 80. If only 35% of people get infected, then these percentages would be 2.3% and 7.8%, respectively. But I must emphasize again that if the number of diagnosed cases is much lower than the total number of infections, then these percentages will be lower as well. The blue bars in the figure take this into account – they are based on a different approach of calculating age-dependent mortality. In the UD/NAI model, it is assumed that only one third of infections are diagnosed,3 and that there are no differences between age groups in the chance to get infected. Hence, in this model, age-differences in DCRs are entirely explained by differences in the numbers of diagnoses rather than the numbers of infections.4 By implication, total infection rates do not affect the mortality in case of UD/NAI, and consequently, one version of this model is sufficient. An important difference between the blue bars and the other bars in the figure is that the blue bars show mortality per infections, while the others show mortality relative to the population. Hence, percentages shown by blue bars must be multiplied with infection rates to get total mortality for that age group.

The numbers in parenthesis in the figure key are total mortality rates due to COVID-19 in a population similar to that of Hubei (in the DCR-based models): 1.2% of people would die if 65% get infected, and 0.7% would die if only 35% would get infected (once more, with the same provision). These numbers would be a bit different in populations with a very different age structure. In older populations (like Japan, for example) they would be higher; in younger populations they would be lower. Applying the UD/NAI model to the global population results in 0.7% overall mortality in case of 100% infection, 0.45% mortality in case 65% of the world population gets infected, and so forth. There are considerable margins of uncertainty for all of these numbers, however.

These numbers seem to imply that if COVID-19 spreads worldwide and infects 65% of people, then that could result in between 35 and 80 million deaths, and if it infects 35% of the global population, it would probably be close to the low end of that estimate. This doesn’t take into account that at some point during the virus’s spread we may find better ways to treat patients or even a working vaccine (that can be produced quickly and abundantly). Such developments would lower the numbers. On the other hand, when the disease starts to spread seriously beyond the few present hot-spots, medical services, supplies, and facilities will be unable to cope, and mortality rates will start to increase fast. Presently, many severe cases are being treated in intensive care facilities, but that will not be possible if there are too many cases. And drugs to treat pneumonia may also not suffice when the number of cases explodes. For these reasons, the total death toll could be easily twice the numbers mentioned.

indirect effects

While this is obviously a reason for concern, what is even more worrying than such a direct effect of COVID-19 are the indirect effects. One indirect effect is economic damage, but not economic damage of the disease itself. COVID-19 mainly kills (very) old people who are no longer productive, and consequently, the direct economic effect of COVID-19 is almost negligible. The economic damage that matters is the damage caused by attempts at containment, treatment, as well as by other indirect effects. Containment attempts are especially disruptive, putting enormous numbers of people out of work, and hurting many businesses more than they can manage. The more areas are effected by containment policies, the greater the damage will be.

Since the world is already heading for a major financial and economic crisis (and China is especially vulnerable), it is possible – perhaps even likely – that COVID-19 pushes the world economy over the edge and causes another major recession or even depression. Depending on how deep that recession is going to be (and it might be deep, as governments don’t have the funds to keep their economies afloat anymore, especially not if they also need to divert funds to COVID-19-related expenses), it may even kill more people (through poverty, stress, civil unrest, and so forth) than the virus itself. But even this is not the main reason for concern.

When governments give themselves new powers in times of crisis, they rarely give those powers back after the crisis is over. Governments will ban mass gatherings in an effort to slow down (or prevent, although that is futile) contagion, for example, but that new power of banning mass gatherings will be oh-so useful to ban politically unwelcome mass gatherings such as demonstrations as well. And as long as there are some cases of COVID-19 (which may linger for many years or even indefinitely!) governments will have the perfect excuse to forbid any kind of demonstration or collective political action.

On the same grounds, governments may try to restrict or ban any kind of event or activity that increases the risk of contagion. This would be economically disastrous, but it could also be politically disastrous. Elections, for example, make huge numbers of people pass through the same small space and are thus possible causes of very many new infections. Given that the world has already seen a rise in the number of politicians that have little respect for democracy, it doesn’t seem at all implausible that some of them will abuse COVID-19 to cancel elections or otherwise further undermine democracy. What if Trump would abuse COVID-19 fears to cancel the presidential elections in the US?5

What might be even more worrisome is that many people won’t even be bothered by an effective abolition of democracy under the guise of combating disease. COVID-19 is already causing an increase in nationalist and racist sentiments, as well as an increase in support for authoritarian policies and politicians. Economic crises tend to produce fascism as well, and therefore, a COVID-19-related recession or depression is also likely to further reinforce the global right-wing extremist trend.

And lastly, on top of all that, COVID-19 is a distraction. People cannot handle too many crises at the same time, and neither can the press and politicians. The more attention for COVID-19, the less attention for climate change. And given that the rich and powerful (who control the press) don’t want us to give much attention to climate change, COVID-19 is a perfect excuse to make sure that most of us won’t be paying attention indeed.

So, this is what worries me: a further undermining of (or even fatal blow to) democracy and civic rights, economic disaster, the rise of right-wing authoritarianism, and a decline of climate change awareness and climate activism. Of course, the direct human suffering is also reason for concern, but on the longer run these indirect effects are infinitely more worrying. Tens of million deaths due to a disease is terrible, but billions of deaths due to a refusal to deal with climate change – and COVID-19 may very well cement that refusal – is even worse.

I suppose that if you’re vengeful in nature and believe in a religious version of the Gaia hypothesis, you might have a different view. Then, COVID-19 may seem like Earth’s revenge on the generation who is killing the planet and refuses to stop. And indeed, that generation will be hardest hit by the pandemic. But I don’t believe in revenge and I don’t believe that Earth is capable of revenge (or of any other intentional action). So all I see is this huge, very black swan that may not kill many children directly, but that may very well end up helping to kill almost all of our children a bit further down the road.

updates and corrections

(March 9) — Contrary to what I suggested above, asymptomatic spread appears to play no significant role. However, something very similar is happening: the first symptoms of COVID-19 are often so minor that an infected person doesn’t realize that he or she is infected, and many people never develop more serious symptoms and thus never become aware that they are infected, unless they are tested. Spread by people who have such minor symptoms and who are not identified as being infected is not asymptomatic infection, but the effect on the spread of the disease is the same.

A preprint on the infection and transmission COVID-18 published online on March 4 (but not yet published in an academic journal) suggests that children are as susceptible to infection as adults, but are far less likely to develop symptoms.6 If this is right (and I see no reason to doubt this result), then the correct interpretation of the Chinese data mentioned in my original post is the second one leading to the blue bars in the graph(i.e. the UD/NAI model).

However, those calculations do not yet take the time gap between identification of a case and death into account. Now the Chinese epidemic has progressed further, we can make a better estimate of the case fatality rate. Of 80,699 cases (at the time of writing) 3,097 have died. Neither number has been increasing fast for a couple of weeks, while recoveries continue to increase. Hence, the current case fatality rate (CFR; number of deaths divided by number of identified cases) is 3.8% in China.

South Korea’s epidemic started much later, but also seems to be almost under control (for now, at least, but things can change rapidly). There the number of deaths is 50 out of 7,314 cases, or 0.7%. That is a huge difference with the Chinese percentage. Several factors may partially explain this difference. Firstly, South Korea made a massive effort to track down and test cases and will have found a much higher percentage of infected cases. If there would be no other differences between the two situations, then this would imply that there are about 400,000 unidentified cases in China, or in other words, that only about one fifth of infections are identified.7 There are other differences, however. Hubei’s health care system was and is completely overwhelmed leading to more patients not receiving needed care. And Chinese cities tend to have a lot of air pollution, which is bad news for people with pneumonia. Probably there are more important differences explaining the high case death rate in China. Nevertheless, the biggest difference must be a significant difference in the percentage of cases that were identified in China and South Korea.8

As mentioned before, the case death rate is not the same as the total mortality from COVID-19. The media (including Wikipedia) systematically confuse identified case with infected cases, but those are two very different numbers. South Korea found a higher percentage of infections, but still lags behind, so the “infections death rate” (number of death divided by number of infections) will probably be a bit higher than the Korean 0.7%. However, it will not be nearly as high as Chinese case fatality rate of 3.8%. Above, I estimated the infections death rate or to be 0.7 in the UD/NAI model (blue bars in the graph), similar to the current case fatality rate in South Korea. Currently available date suggest that that estimate was a bit too low; most likely it will be around 1% (except in countries with seriously deficient health care). This also means that the age-specific fatality rates will probably turn out to be a little higher than the blue bars/columns in the graph, but again, not very much. Perhaps it will be 10% for people over 80; half that for people over 70; again half that for people over 60; and so forth.

Because of the time gaps between infection and identification and between identification and death. Mortality rates start low and then gradually increase. (And to what level they increase depends mostly on the percentage of infections that is identified.) Current data suggests that in the first two weeks or so, the case fatality rate is near 0.5%, which means that if in some new local outbreak there is significant number of deaths, the total number of infected people is likely to be up to 200 times the number of deaths. If this is right, then Italy and Iran have many more infected people than currently identified.

New outbreaks have started in very many countries in the past week (mostly imported from Italy or Iran). All of these are still in the first phase of exponential growth. All of them will grow to many thousands and probably even ten-thousands of infections in less than two weeks. Panic and fear seem to spreading even faster than the virus itself, however, and the economic and social disruptions are getting more severe everywhere. On the longer term, those economic and social disruptions will be far more damaging than the virus itself (and they are even harder to control). This is what should worry you (and surely is what worries me) much more than COVID-19 itself. This pandemic strengthens and speeds up the rise of neo-fascist authoritarianism and undermines climate action. It might kill a few million people directly, but its indirect effects may be fatal to human civilization and lead to the death of billions.

(March 23) — I wrote the original post above on February 23th, one month after Hubei went in lockdown. It is now March 23th, again one month later, and time for some reflection.

When Hubei went in lockdown on January 23th (i.e. two months ago), they had 770 identified cases and 25 deaths. The median time between identification and death in fatal cases is about 14 days. On February 6th, the region had 563 deaths, which suggests that there may have already been 50,000 infections on the day of the lockdown. (Assuming that the South-Korean case fatality rate approaches mortality relative to total number of infections. See above.) Both new cases and deaths continued to increase sharply for another month after the start of the lockdown. The total number of infections probably reached around 300,000 (again based on the South-Korean case fatality rate), of which only about 80,000 were identified. Today, two months have passed since the lockdown went into effect and Hubei has reported almost no new cases for several days.

This teaches us a number of important things. Firstly, total lockdown works. When the epidemic is too widespread for contact tracing etc., then forcing (almost) everyone to stay at home will bring the epidemic under control. Secondly, a lockdown does not stop the epidemic immediately. In Hubei the total number of infections increased roughly sixfold after enforcement of the lockdown. It is reasonable to expect a roughly (!) similar kind of increase elsewhere. Thirdly, it takes a month after the start of lockdown before case growth starts to decline significantly and two months to bring the outbreak under control. The less strict the lockdown is, the longer it will take to bring the epidemic under control and the more it will grow before it is completely under control.

Several European governments have decided for a lockdown. For example, a lockdown started officially in France on March 17 and is supposed to last for two weeks. That is obviously way too short, especially considering that many French people didn’t take the lockdown seriously and it still is relatively easy to evade. If the lockdown is kept in effect, then it will probably eventually bring the epidemic under control, but due to its laxity it will take (much?) more than two months and the decline of growth will be slower. The latter, obviously, implies that the total number of infections will increase more than what was the case in Hubei.

France had 7730 identified cases on March 17 (ten times the number in Hubei when that region went in lockdown!). Today (March 23) the number of deaths is 674. If the same mathematics apply to France as to Hubei and we take into account that it is only one week after the lockdown went into effect rather than two, then this would suggest that the total number of infections on March 17 was at least 80,000, but may have been much higher. It is, however, possible that the South-Korean mortality rate doesn’t apply to France due to differences in health, diet, and/or other circumstances, and for this reason, it is also possible that the total number of infections was actually a bit lower. If we use 80,000 as a rough estimate, and assume the same sixfold increase seen in Hubei (ignoring the fact that the lax lockdown in France implies a larger increase), then the epidemic will peak at somewhere around 240,000 infections. This number does, however, depend on some rather optimistic assumptions. More likely it will be significantly higher. I’m inclined to say that a million infections and (at least) 10,000 deaths are more probable numbers, but these are very rough estimates. (And don’t forget that these number assume a lockdown that lasts for months.)

Of course, all of this assumes that France continues its lockdown until the epidemic is under control, rather than the two weeks announced. If it indeed lifts the lockdown in a week from now, then its effects will be negligible and the epidemic will keep spreading.

In the meanwhile, economic and other effects have started to become clearer as well. The ILO has estimated that the COVID-19 pandemic could put 25 million people out of work. I doubt that it will be as few as that. Of course, most governments have thus far been busy propping up stock markets and stuffing the pockets of rich people who lost a few bucks due to the corona fall-out, but have mostly ignored the financial effects on “normal” people. (Fortunately, there are some exceptions, but it is nowhere near sufficient. The focus remains on bailing out large corporations to make sure that rich people can continue to be rich people.) There will be a huge wave of bankruptcies among the self-employed and small businesses in addition to the millions of people losing their jobs. Many of these people will look for scapegoats and considering that the rich control the media, the rich will make sure that that scapegoat will be harmless to them. Nationalism and racism will be used to deflect anger. (Trump is already doing this.)

The damage won’t just be economic (and political), however. Extended lockdown and stress will destroy many marriages and lead to an increase of domestic violence and child abuse. Especially for people living alone, social isolation can easily become intolerable, and this will almost certainly lead to significant increases in depression, suicide, and other mental problems and illnesses. And economic/financial stress and uncertainty will further increase these.

Problems like these are likely to create strong pressures on governments to lift lockdowns too soon. Whether some governments will be able to resist that pressure is hard to predict, but those that bend will create greater problems further down the road: more infections, collapsing medical facilities, and more deaths. There is no easy way out of this mess.

I suppose that some people are wondering what got us into this mess in the first place. Well, … we did. The destruction of nature and globalization are most to blame for the current crisis, and epidemiologists and other experts have been warning that something like this was going to happen sooner or later (and will happen again!) for many years. (In that sense, it isn’t really a “black swan”, although the date of the outbreak and the characteristics of the virus were almost completely unpredictable.) You can read more about this in a recent Guardian article.

(March 24) — In yesterday’s update, I used France as an example and gave a rough estimate of the number of infections and deaths in case of a lockdown lasting at least several months. Recently some politicians and others have voiced concern about such lockdowns – their economic effects might be worse than the effects of the epidemic themselves, they worry. In case of the US, I have already seen an estimate of 30% unemployment, for example.

Economic effects of long-term lockdowns are hard to estimate, but they will be severe indeed. But for a fair comparison, we need to know the likely costs in human lives in case no lockdown is implemented. One thing we need to know for that is what percentage of people in a population will get infected if there are no significant measures to prevent spread in effect, but we really don’t know that percentage. Early estimates by epidemiologists tended to be around two-thirds, but considering how infectious the virus is, it may be even more than that. Since we have no better estimate, let’s use two-thirds. In case of France (the same example as above), this would imply approximately 45 million infections.

Next we need to know the mortality rate. This won’t be the roughly 1% in case of the lockdown scenario, because if there are 45 million infections, medical facilities will be completely overrun and there will not be enough medical resources to treat even a small percentage of the severe cases. So, we need to know what percentage of cases is so severe that without hospitalization, drugs, and other relevant care, the patient would almost certainly die. That we don’t know either, however. The hospitalization rate appears to be at least five times as high as the fatality rate. Perhaps, we can use that as an estimate. In that case, total mortality (relative to infections) might also be roughly 5 times higher, and thus close to 5%. In the case of France, this could mean more than 2 million deaths.

So, a fair comparison between the lockdown and no-lockdown scenario for France has 10.000 deaths and massive economic damage on one side, and 2 million deaths and slightly less economic damage on the other side. “Slightly less“, because 45 million infections and 2 million deaths will also have severe economic effects. For other countries the numbers would be different, of course, but the relative difference between the two scenarios is everywhere approximately the same.

Of course, all of these numbers are very rough estimates, but they give a reasonable impression of what is at stake, and what the policy options and their implications are. I don’t expect politicians to make responsible decisions, however, as there is something more to take into account. The higher mortality in case of a no-lockdown scenario doesn’t apply to everyone. Regardless of how overrun medical services are, the rich and powerful will almost certainly get the best care, so for them the risk of the no-lockdown scenario isn’t as great as it is for the rest of us. For them, the economic damage might be a greater concern indeed. And some of them may even favor a culling of the disposables. Since governments almost always prioritize the interests of the rich (as they have been doing in their measures to “counter” the economic effects of COVID-19), we cannot trust that they have our best interests in mind.

Virus + class war = zombies. See You are a zombie about this equation. Some governments appear to take the zombie approach indeed: the rich and powerful inside their fortresses, mostly protected from the fall-out of the virus; the rest of us on the outside, surrounded by sickness, suffering, and death.

If you found this article and/or other articles in this blog useful or valuable, please consider making a small financial contribution to support this blog 𝐹=𝑚𝑎 and its author. You can find 𝐹=𝑚𝑎’s Patreon page here.


  1. There are some documented cases of asymptomatic spreaders.
  2. The Novel Coronavirus Pneumonia Emergency Response Epidemiology Team (2020). “The Epidemiological Characteristics of an Outbreak of 2019 Novel Coronavirus Diseases (COVID-19) — China, 2020”, China CDC Weekly 2.8: 113-122. Available online HERE.
  3. This is based on an estimate I read somewhere in the end of January, but for which I cannot provide a source, unfortunately.
  4. And the difference in diagnoses may be due to differences in symptoms. If young people who are infected show few symptoms, very few cases among young people would be detected.
  5. Can he do that? Well, yeah, he can do anything he like, because he has unconditional support from the Republican party and from the rich and powerful who control the US and its economy behind closed doors. Will he do that? No idea, but I wouldn’t be surprised.
  6. Qifang Bi et al. (2020), “Epidemiology and Transmission of COVID-19 in Shenzhen China: Analysis of 391 cases and 1,286 of their close contacts”. Prepint available HERE.
  7. This does not mean that all of these cases will become infectious. Most likely, the vast majority of them never developed any (noticeable) systems and never will either.
  8. Unless the epidemics in the two countries were dominated by different strains of the virus and the dominant strain in China was more deadly than the dominant strain in Korea. On the two strains identified this week, see: Xiaolu Tang et al. (2020), “On the origin and continuing evolution of SARS-CoV-2”, National Science Review. Available HERE. The conclusions in the study by Tang et al. are controversial, however. Some scientists dispute that there are multiple strains, and have pointed out other methodological errors. For a critical response to Tang et al., see THIS ARTICLE.