StubbleJumper

Yep, if you are so inclined, it's a pretty basic exercise to estimate the potential number of deaths in the US that would be associated with the virus running its course. If you believe that herd immunity kicks in when ~200m have already had the virus (ie, ~60% of 330m population), and if you believe that the IFR is ~0.4%, you'd be looking at perhaps 800k covid deaths in the US over the 18 month epidemic. As a check of that arithmetic, the US is currently at ~193k deaths right now, which mainly reflects cases diagnosed up to about 3 or 4 weeks ago (ie, over the past 28 days there are about 1m officially diagnosed cases and probably about 20k more deaths in the pipeline for those ~1m infections that already exist today). The numbers are large, but none of that is particularly new. Larger numbers have been trotted out from various models over the past six months, but public policy in the US has evolved in its own particular way irrespective of the potential magnitude. SJ

Also disagree. Herd immunity requires the majority of the population to be infected which would be a terrible health outcome when there is a good chance by middle of next year an effective vaccine will be wildly available. There is an obvious middle ground of social distancing and mask wearing where social distancing is not possible. And after the hard lockdown was eased it was the perfect opportunity to try it out. But it hasn't been given a proper chance because there has been very little attempt to enforce these measures and compliance has been piss poor. In the UK for example people have swarmed to overcrowded beaches, fill the streets outside bars with drinks in hand, attend mass rallies, and walk side by side with their friends. Perhaps as a result cases have gone from <1,000 a day in the middle of the summer to over 3,000 this week. Of course there is more testing so the results aren't directly comparable. But there is no doubt that cases are rising again and it is gonna be a long winter. The government is already starting to impose new restrictions and if cases continue to rise the restrictions will get more and more onerous until we are back in full lockdown. And in Europe the same scenario will play out. I am not so familiar with the US politics and the US government seems comfortable with a much higher infection rate but I am sure there is still a threshold at which point their hand will be forced especially if other countries in the world are starting to lock down their economies. The question is: In the United States, which will come first, herd immunity by virtue of a vaccine, or herd immunity by virtue of the virus having already run its course? We are now roughly six months into this pandemic in the US and there have been 6.5 million officially diagnosed cases, but those are only the officially diagnosed. How many cases have there truly been? Maybe 60 million? Perhaps as many as 90 million? Maybe fewer than 60m or 90m? If you believe that herd immunity occurs once ~60% of the population is resistant, you would need roughly 200 million people in the US who are resistant. So are we one-third of the way there right now? Half way there? Less? If the vaccine doesn't show up on the scene before the middle of next year, as you suggested, it is entirely possible by that time that new cases will be tailing-off as the US asymptotically approaches herd immunity due to the spread over the previous 12 or 15 months. SJ

If FFH had unlimited capital, you'd definitely engage in an aggressive buyback when shares are trading at such a significant discount to BV. But, FFH does not have unlimited capital, so you need to think about the possible alternate uses for it. Suppose you had $1m of extra capital sitting at Crum and Forster. You have a couple of options: 1) Dividend the money to FFH holdco, then repurchase FFH shares: Continuing FFH shareholders get an immediate, one-time benefit equal to the prevailing discount to intrinsic value. So, let us suppose that FFH shares are currently 40% undervalued, then you'd get an immediate $400k benefit to continuing shareholders. 2) Expand C&F's book: Assuming that there's no shortage of business available, with $1m of "extra unused capital", C&F could easily write $1.5m of premiums. If you believe that C&F can write a 95 CR, that would be $75k of underwriting profit, and if you believe that Hamblin-Watsa can get a 2% investment return on the float that would be another $30k of investment return, for a total return of $105k on your $1m of extra capital in 2020. If you believe that this hard market will prevail until the end of 2022, then you'd also get that $105k in both of 2021 and 2022 when you renew the insurance policies and reinvest your float. In my example, one alternative is quantitatively better and carries less risk, but both alternatives are very economically attractive. If you believe that the accident year CRs are more like 90 instead of 95, it's a no-brainer to keep capital in your insurance subs during this hard market rather than conducting the buyback. On the other hand, if you are more pessimistic and believe that the accident year CR is more like 98, then FFH is wasting its time by trying to expand its book. I would say that it's not entirely clear which alternative is better, but I don't mind Prem's strategy of retaining capital in the subs because it is very likely to work out reasonably well for shareholders. SJ

Lol. Deflect, Deflect, Deflect. I'm just here to enjoy the mental gymnasts at work--surely politicizing masks, downplaying the threat, and holding indoor gatherings in no way affected R0 in the U.S.A.! There are no levers that Trump pulled the wrong way, surely! Oh, I don't doubt that Trump pulled every lever that he had as badly as it could possibly be pulled. Don't get me wrong about that. But, his levers were far from the most influential. It's funny that the right wing rails about how some of the northern state governors are dragging their feet on re-opening, while the left wing rails about the neglect of the southern state governors in implementing control measures in their state, and then the incurably partisan participants blame the whole works on the president, despite an obvious difference in the outcomes between the two groups of states. SJ

Hey muscle--no one on here understands this virus better than you--that's for sure! I grant you an honorary doctorate. Where that herd immunity at? Still waiting... :-X We are still waiting for you to explain how effective test-and-traceback is for covid, given the numbers in Europe. Do you remember what you were writing in March? Test, test, test, right? SJ Yes--clearly I am wrong--we should stop testing altogether--that will make things better: no tests, no cases! You completely ignored the observations from a variety of interlocutors that pre-symptomatic and asymptomatic transmission meant that testing would have a very limited impact on aggregate numbers. So, what is happening right now in Germany and the Netherlands? What is going on in Canada and Italy? Lots of tests, but transmission merrily persists for the exact reasons that you were told in March. SJ Cases are up, but so are tests. The positivity rate in those countries is much lower than what it's been in the U.S. It will be much easier to control for them and to top it all off, they are not run by leaders who are downplaying the threat, mocking masks, and holding indoor rallies. So, once again, explain how your test and traceback assertions from March are working out? SJ

Well, you could list the actual levers available to the federal government. They include international travel restrictions, immigration control, decisions for the deployment of military forces, other? And then you could list the levers available to the state governments. They could include things like limiting the size of group gatherings, controlling the conduct of commerce within the state, preparing the health care system and managing the long-term care system. So which are the big levers? Closing shopping malls, movie theatres, stores and workplaces, gyms and sporting events? Or stopping international travellers from visiting? It's pretty obvious which are/were the levers that can have a meaningful impact on the transmission of covid. SJ

Hey muscle--no one on here understands this virus better than you--that's for sure! I grant you an honorary doctorate. Where that herd immunity at? Still waiting... :-X We are still waiting for you to explain how effective test-and-traceback is for covid, given the numbers in Europe. Do you remember what you were writing in March? Test, test, test, right? SJ Yes--clearly I am wrong--we should stop testing altogether--that will make things better: no tests, no cases! You completely ignored the observations from a variety of interlocutors that pre-symptomatic and asymptomatic transmission meant that testing would have a very limited impact on aggregate numbers. So, what is happening right now in Germany and the Netherlands? What is going on in Canada and Italy? Lots of tests, but transmission merrily persists for the exact reasons that you were told in March. SJ

Hey muscle--no one on here understands this virus better than you--that's for sure! I grant you an honorary doctorate. Where that herd immunity at? Still waiting... :-X We are still waiting for you to explain how effective test-and-traceback is for covid, given the numbers in Europe. Do you remember what you were writing in March? Test, test, test, right? SJ

The real question is whether the state governments were relying on what Trump was saying. Were the governments of NY, Michigan and Illinois so complacent that they were waiting for insight from Trump before taking decisions on social distancing in their state? Seriously, RB knows very well that both Canada and the US are federations with a vertical separation of powers. So, what exactly were powers available to the states' (provinces') and to what extent were those sub-national jurisdictions reliant on public statements from the federal government? Who holds the real levers of power to manage the pandemic (hint: in Canada it is not the Prime Minister)? SJ

Things have gone well in Italy over the summer. But, we are in what, the 3rd inning of this ball game? Frankly, with covid, it's astounding how quickly things change, for the better and for the worse. Take a look at Sweden's numbers in mid-June and France's. Or for that matter, look at the US numbers from 2 months ago. SJ

Yeah, some people are focussed on politics. Strangely, nobody on this forum is talking about France. Yesterday, France had 6,500 new cases for a country of ~65 million people. Yesterday the United States had 29k new cases for a population of ~325m people. Over the past week, looks like France took over the lead for the most significant covid outbreak amongst developed countries, relative to its population? Things seem pretty manageable in Sweden these days... So, shall we engage in a series of venomous statements about President Macron, now that France is "winning?" Is this MFGA? SJ

Yep, the BB shareholders are up in arms. I guess their underlying assumption was that either BB doesn't need the liquidity or that there is a long line-up of potential lenders who would be prepared to lend a half-billion at 3.75% with no conversion privilege. I am from the school of thought that FFH's last note flotation was at 4 5/8%, so if that's what FFH pays for debt, what should a riskier outfit like BB pay? Maybe 7%? Seriously, a 15 minute walk through their financials for the past 3 or 4 years is enough to make a guy want to puke. Maybe there is a long line-up of outfits wanting to lend money to companies that have drastically transformed their business and are cashflow negative? I don't see it, but I've been wrong plenty of times before... SJ Blackberry shareholders are still up in arms about this: https://www.newswire.ca/news-releases/concerned-shareholder-objects-to-blackberry-s-related-party-transactions-with-fairfax-851354230.html SJ And the story continues to evolve. Now it looks like BB and FFH have downsized the convertible debenture issue: https://www.newswire.ca/news-releases/blackberry-announces-redemption-of-existing-convertible-debentures-and-provides-update-on-issuance-of-new-convertible-debentures-832813324.html I guess the downside to this for FFH shareholders is that BB traded at US$5.23 yesterday, so the option to convert has become a bit more valuable over the past 6 weeks... SJ

For those who subscribe to the Globe (or those who know how to get around the paywall), there is an interesting article which provides BB's explanation of why the complaints about the convertible debs are unfounded: https://www.theglobeandmail.com/business/article-blackberry-shareholder-asks-regulators-to-order-vote-on-refinancing/ SJ

Yep, the BB shareholders are up in arms. I guess their underlying assumption was that either BB doesn't need the liquidity or that there is a long line-up of potential lenders who would be prepared to lend a half-billion at 3.75% with no conversion privilege. I am from the school of thought that FFH's last note flotation was at 4 5/8%, so if that's what FFH pays for debt, what should a riskier outfit like BB pay? Maybe 7%? Seriously, a 15 minute walk through their financials for the past 3 or 4 years is enough to make a guy want to puke. Maybe there is a long line-up of outfits wanting to lend money to companies that have drastically transformed their business and are cashflow negative? I don't see it, but I've been wrong plenty of times before... SJ Blackberry shareholders are still up in arms about this: https://www.newswire.ca/news-releases/concerned-shareholder-objects-to-blackberry-s-related-party-transactions-with-fairfax-851354230.html SJ

i think i see what you're getting at but i may have some trouble if you mean to imply that the risk is much ado about nothing. It's been widely recognized (Ontario also) that, due to testing 'strategy' and other various reasons, significant bias can be introduced when analyzing the healthcare worker risk. The main bias risk is the "healthy worker effect". Because testing is typically used proportionally more in frontline workers (mostly for evidence-based reasons and the spreader issue), there is a tendency to report higher absolute numbers of 'detected' cases and to underestimate (vs if that bias were not there, at least temporarily, compared to other population groups) the CFR. When the proportion of testing in the healthcare population vs total population is high, this can introduce bias at the population level CFR also. Having said that, given the use of PPE and procedure adjustments, it appears that most CV+ cases in the healthcare workers result from acquisition in the community but not all. i guess 'we' will find out more precisely over time. In terms of cost effective measures and keeping the precautionary principle in mind, it's probably a good idea for high risk workers to undress in the garage and take a shower before hugging family members even if community transfer in that setting has not been well documented. Interestingly, during the SARS episode (SARS didn't kill so softly so it was more obvious), there were clear and direct links between healthcare workers becoming sick and dying and taking care of SARS-infected patients. An interesting lesson from the SARS episode though was that it was possible to bring the risk of the frontline worker to the same level of the general population, given appropriate measures. From a recent report coming out of the great Alberta province: "It is important to note that evidence from 2003 SARS demonstrated that risk to HCWs could be mitigated by diligent hand hygiene and use of personal protective equipment (PPE)". Not everybody 'believed' that then, in the heat of the action. And some still say it was a hoax. No, I was not at all attempting to suggest that it was much ado about nothing. Just sharing one jurisdiction's actual numbers. Every jurisdiction seems to be good at capturing some data series, but are often terrible at capturing other series. And, for Ontario, the healthcare worker data is actually published. My point about the calculated infection fatality rate is that, setting aside the risks associated with small sample size, the fatality rate of those infected does not appear to be much different for healthcare workers than for the broader population in the age groups that would be typically employed in a health care setting. The implication is that the calculated IFR would not provide evidence of an assertion that health care workers are healthier than the average population, nor would it provide evidence of favourable treatment in the healthcare system (queue jumping), nor would it provide evidence that the work environment delivers a heavier viral load resulting in an elevated fatality rate (all of those things are possible, but this data does not provide much in the way of evidentiary support for any of them). With respect to the heavier testing of healthcare workers, you are undoubtedly correct. I quite confidently took Ontario's nursing home worker deaths and divided by the officially diagnosed healthcare worker cases to calculate an IFR. I did that because almost all infections amongst nursing home employees were officially diagnosed due to a heavy testing program (ie, CFR=IFR for that group). But, you are correct that for the general population, the IFR estimate involves a bit of guesswork as a considerable portion of infections do not show up in the official statistics and need to be separately estimated (ie, CFR much higher than IFR for general population). As to the source of the health care worker infections, I'd say we'll probably never know. But, the logical and compelling assumption would be that they were mostly caught in the workplace due to unavailability of PPE early in the pandemic, failures of PPE or inconsistent usage of PPE. I doubt that we'll ever truly know for sure. SJ

Then, let me start with a related challenging question: In the US, if one lets the virus spread and not encourage or adopt appropriate population protection measures, "shouldn’t we expect more deaths" compared to other countries where such measures are applied? Answer: yes, a lot more :( Back to your specific question, the short answer AFAIK is 'we' don't know. There have been cases and deaths in healthcare workers having direct contact with patients but testing is said to be high and, at this point, it's hard to be precise. Eyeballing it (and using reference below), it seems that working directly with patients (even with precautions, protocols and equipment) means higher risk to catch the disease. In your country and various jurisdictions, criteria have been relaxed to accept the link between a Covid diagnosis and the job status as a healthcare worker (workers comp) which, it seems to me, is the reasonable thing to do at this point. It seems that mortality profile adjusted for risk factors is similar to the population in general except for the unusual cases early on (people not aware or accepting the risk anyways) and when protective equipment has been insufficient. i think your wife works in a neonatal ICU so this was not (per se) a high risk area but maybe she heard how people tend to avoid "red" zones in certain hospitals even if they say that it's no big deal when writing posts on the internet. There is no doubt that the viral loads involved in high-risk and repeated exposures without protection would have resulted in a healthcare workers' hecatomb in high risk areas. The key was the rapid (although disorganized and confused at first) definition of risk stratification and application of protocols (with appropriate level of equipment). Low risk required simple adjustments and high risk required complicated and often cumbersome requirements. Aerosol management has been a key aspect. Here's a video from China (they describe the protocol to put a tube down one's throat) that looks like it's coming from a bad movie but is quite representative of what has happened in various global emergency rooms, operating rooms and ICUs. https://www.thelancet.com/journals/lanpub/article/PIIS2468-2667(20)30164-X/fulltext This virus is not equipped with a shotgun. It uses a musket but it's awfully good at replicating and getting close to you (or your threshold), if you let it. The data for Ontario health care workers is found in Table 2 of this document: https://www.publichealthontario.ca/-/media/documents/ncov/epi/2020/covid-19-daily-epi-summary-report.pdf?la=en So, 2,600 infections and 8 deaths among health care workers, or an IFR of about 0.3%. Making all of the normal disclaimers about SSS, the IFR seems pretty consistent with a weighted average IFR for a group of people with ages evenly distributed from, say, age 25 to 60 (ie the likely age range of nursing home employees). SJ

Have you considered the possibility that the reason why things are going okay in Lombardy these days is because they actually made significant progress towards herd immunity during their atrocious outbreak in the spring? Lombardy is a region where the official numbers state that there have been 97k diagnosed cases and 17k deaths recorded for a region of 10m people. With a ~17% calculated CFR, clearly the official statistics are drastically under-counting the true number of cases in that region. So, run the calculation backwards, beginning with the number of deaths to infer a plausible number of infections in the region. If you are in the camp that believes that the IFR is likely around 0.5% or 0.6%, then the 17k deaths implies about 3 million infections. A region of 10m people with ~3m infections could be quite far advanced along the path to herd immunity. So, are the current daily numbers in Lombardy indicative of good management in the present, or poor management in the past? SJ Specifically around this exchange about herd immunity, in early June, the Bergamo province reported population antibody levels at 57% which would clearly not support the notion that much lower antibody levels prevalence would be sufficient herd immunity under any circumstances. However, the numbers suggest that herd immunity had become a contributing variable going forward, at least for those who could mount one (immune response). ----- Bergamo has meaning here because it's a nice city perched on a hill that marries well the ancient with the modern and is a great example of the mixed human nature of the Piedmont region. Its geographical features also make it an ideal place for the start or the finish of cycling race. It's also relevant because this CV threat, for me, became really real when first accounts (early March when choices had to be made for survival and when army trucks were handling coffins) of what was happening in hospitals made it to my email box. This is when i decided to establish a scorecard looking at the effectiveness of various national public health organizations and leaders across the world in dealing with this phenomenon.. The Bergamo province and Northern Italy had a similar experience compared to various areas of Spain, New York and my jurisdiction with a large spread occurring before the actual implementation of measures (spontaneous, encouraged and imposed). In hindsight, whatever the causes (close to high volume international airports, large events as super-spreading catalysts, older and more fragile population etc), it looks like every day counted and it is reasonable to suggest that the outcome would have been a 100x better if measures that were actually applied had been applied about 2 weeks earlier (with obviously a much lower cost). It's been shown that the viral load is highly determinant as to whether one catches the disease or not and how sick or dead one can become. The viral load concept can be applied to the population level and is one more argument suggesting that herd immunity is a dynamic concept. Bergamo showed that a high population viral load will drive up the population herd immunity required to contribute to flattening the curve. There are several quantifiable and sophisticated ways to report on what happened in Bergamo. Here's one example (and i'm sorry to say a preventable one): Note: to understand, there is no need to speak Italian or for fancy statistical knowledge. ----- Even though the value in the hotspot is 57%, Italy as a whole is far from herd immunity with only 2.5% antibodies overall. https://medicalxpress.com/news/2020-08-italians-covid-north.html Another tidbit, for whatever it’s worth. Italiy’sVOVID-19 fatality rate is 586/1M. When the US reaches 198k death (we are currently at 177k and increasing by almost 1k/day) we will be just as bad as Italy. That should occur around mid September at latest. I recall folks talking on Italy’s incompetence back in February. Well, it took a while but we have beaten them to it. That's what the serology testing might suggest. But, if you believe that the serology results are true, then you must also believe that the IFR for Italy is a shade higher than 2%. Never say never, because the elderly population was hit particularly hard in Italy. But, generally, it seems that IFR estimates are mostly coming in at about 0.6%, which would suggest a much larger portion of the population has probably already had covid (maybe about 6% of the population). Or maybe some of Italy's 36k deaths have been misclassified as covid deaths? SJ

Yes, to the extent that it is possible, it is a worthwhile exercise to adjust the BV to reflect fair value, and you probably should also attempt to build a pro-forma income statement that strips out the non-recurring gains/losses and the extraordinary non-recurring cats (cats are a fact of life, but some years they are ridiculous). SJ

They don’t have the capital. I was surprised that they even bought back a few shares last Q. Given their situation where they are a bit reliant on that revolver, I would have thought that they'd be doing everything possible to keep the ratios that are used for the covenants as low as possible. SJ

Have you considered the possibility that the reason why things are going okay in Lombardy these days is because they actually made significant progress towards herd immunity during their atrocious outbreak in the spring? Lombardy is a region where the official numbers state that there have been 97k diagnosed cases and 17k deaths recorded for a region of 10m people. With a ~17% calculated CFR, clearly the official statistics are drastically under-counting the true number of cases in that region. So, run the calculation backwards, beginning with the number of deaths to infer a plausible number of infections in the region. If you are in the camp that believes that the IFR is likely around 0.5% or 0.6%, then the 17k deaths implies about 3 million infections. A region of 10m people with ~3m infections could be quite far advanced along the path to herd immunity. So, are the current daily numbers in Lombardy indicative of good management in the present, or poor management in the past? SJ

Not true. Ontario's data is much more recent than most of the serology reports in that thread. Based on his chart, he seems to by using NY serology from mid-June. But this is irrelevant since Ontario's epidemic was crushed by the end of June (a few deaths per day since). There is no problem with the Ontario analysis due to recency or lack of recency. Essentially, what the data show is that about 0.4% of Toronto residents were officially diagnosed as being infected with covid up to June 30, but about 1.5% of people in Toronto who submitted a blood sample to the health system showed the presence of antibodies. On the face of it, the ratio between people carrying antibodies and those officially diagnosed is a bit low compared to other seroprevalence studies (ie, ~3.75:1), but on the other hand people who are young and healthy do not generally submit blood samples to health care system, so it is quite possible that the blood samples were biased to those who had existing co-morbidities and who were quite rightly taking their own social-distancing measures to avoid the virus. SJ Yes, since June 30th, in Ontario, standard measures such as daily new cases per million population have remained very low (even lower than British Columbia) and the positivity rate has stayed below 1% and has been declining. There is no evidence that community spread has continued to any significant degree, enough to materially alter the conclusions that could be obtained from using antibody levels during the study period, as an input. To be clear, the methodology used by the Ontario public health is not perfect but is quite robust: "Specimens tested to generate seroprevalence estimates were originally submitted to PHO Laboratory for clinical testing for antibodies to a variety of infectious diseases (but not COVID-19)". They had to make some adjustments for various reasons but their method is quite dependable. No, things have been quite impressive in Ontario, but that is largely due to the Draconian measures that were put in place by governments during the lockdown (some aspects were more severe than in Quebec, some less). Now, you are a medical practitioner, so you can tell us. How many perfectly healthy people are asked to provide blood samples? How many diabetics are asked to provide blood samples? How many people with coronary difficulties are asked to provide blood samples? It is a physician-selected group of people who *already* have interactions with the health care system, which generally would exclude the healthy population under age 50 and would particularly exclude the healthy population under age 40. Ontario does not need to apologize for using existing samples of blood, but let us not pretend that it was in any way random (remember the non-random sample of Manhattan denizens who were tested when they went out for groceries?). These were probably sicker than average people. So is this better or worse than using blood donors? The blood donors are probably healthier than average people. But, let us not pretend that they are random either. In any case, the recency or non-recency of the sample is not particularly problematic. The seroprevalence work in Ontario is directionally consistent with that done for other populations, but I remain hesitant to accept the magnitude. SJ From old and recent knowledge, i gather that the methods of these serosurveillance investigations done at the population level will depend on what you want to achieve and compromises have to be made between the 'investment' required and the precision and generalization of findings. You need to decide if the study will be cross-sectional (performed once) or longitudinal (follow-up). You also need to decide how to sample the population. The Ontario method used above and blood donors (convenience sampling) are not truly random but are relatively simple to design, to do and to analyze compared to fancy random selection of different population stratas. Blood donors tend to participate well but tend to be different from the general population (often healthier and other factors). The Ontario study design is interesting because it likely represents fairly well the general population. Random sampling is more complicated, more expensive and takes longer to perform so there is a price to pay for precision that may not be critical for population level measures that may be helpful during the acute decision making process. Also, people that get randomly sampled have to consent to the lab test etc and this is a source of efficiency and statistical problems. There is a more complex project going on: https://abcstudy.ca/about/ The seroconversion data at large has been a bit of a cold shower if the hope was to rapidly get to herd immunity (whichever form or definition that you adhere to) i would like to hear from cherzeca, yourself or others about how a such a low seroprevalence in Ontario (or Toronto) could be reconciled with herd immunity under any circumstances going forward. Hope this helps in your thought process. Well, if you truly believe the Toronto seroprevalence work, it cannot be reconciled with impending herd immunity in any realistic manner. If you are in the camp that there are only 1.5% of Torontonians who have antibodies, then it's a hell of a long haul to have this virus peter out *irrespective* of whether you think that a 15%-20% presence of antibodies will do the job, or whether you are in the traditional camp that holds the view that ~60% of the population with antibodies would be required. In the end, you need to question whether the methodology of Ontario's seroprevalence work results in a fair representation of the actual path of covid in the province, or whether the many other studies scattered amongst a wide variety of other jurisdictions are more representative of reality. If you believe the results, what it does suggest is that Ontario, despite having already had a catastrophic outbreak in its retirement homes might still be very susceptible to a second wave of covid (see what is currently happening in Spain, France, Belgium and the Netherlands). On that, I can offer the opinion that there will be much less willingness for a vigourous lockdown in a potential second wave than there was in the first. Like what is happening in Europe right now, there will likely be a much greater tolerance to the presence and spread of covid than what there was in April. It calls into question the wisdom and sustainability of a great many of the measures that governments imposed in March/April. Even in the "best" of cases where you would optimistically hold the view that there is a 15:1 ratio between people who hold immunity and officially diagnosed cases, Toronto would currently only be at 7.5% of the population with antibodies, right? If you think 15%-20% is required, there still remains much misery to occur, and if you are in the camp that 60% is required there might be years of misery in Toronto's future. SJ

Not true. Ontario's data is much more recent than most of the serology reports in that thread. Based on his chart, he seems to by using NY serology from mid-June. But this is irrelevant since Ontario's epidemic was crushed by the end of June (a few deaths per day since). There is no problem with the Ontario analysis due to recency or lack of recency. Essentially, what the data show is that about 0.4% of Toronto residents were officially diagnosed as being infected with covid up to June 30, but about 1.5% of people in Toronto who submitted a blood sample to the health system showed the presence of antibodies. On the face of it, the ratio between people carrying antibodies and those officially diagnosed is a bit low compared to other seroprevalence studies (ie, ~3.75:1), but on the other hand people who are young and healthy do not generally submit blood samples to health care system, so it is quite possible that the blood samples were biased to those who had existing co-morbidities and who were quite rightly taking their own social-distancing measures to avoid the virus. SJ Yes, since June 30th, in Ontario, standard measures such as daily new cases per million population have remained very low (even lower than British Columbia) and the positivity rate has stayed below 1% and has been declining. There is no evidence that community spread has continued to any significant degree, enough to materially alter the conclusions that could be obtained from using antibody levels during the study period, as an input. To be clear, the methodology used by the Ontario public health is not perfect but is quite robust: "Specimens tested to generate seroprevalence estimates were originally submitted to PHO Laboratory for clinical testing for antibodies to a variety of infectious diseases (but not COVID-19)". They had to make some adjustments for various reasons but their method is quite dependable. No, things have been quite impressive in Ontario, but that is largely due to the Draconian measures that were put in place by governments during the lockdown (some aspects were more severe than in Quebec, some less). Now, you are a medical practitioner, so you can tell us. How many perfectly healthy people are asked to provide blood samples? How many diabetics are asked to provide blood samples? How many people with coronary difficulties are asked to provide blood samples? It is a physician-selected group of people who *already* have interactions with the health care system, which generally would exclude the healthy population under age 50 and would particularly exclude the healthy population under age 40. Ontario does not need to apologize for using existing samples of blood, but let us not pretend that it was in any way random (remember the non-random sample of Manhattan denizens who were tested when they went out for groceries?). These were probably sicker than average people. So is this better or worse than using blood donors? The blood donors are probably healthier than average people. But, let us not pretend that they are random either. In any case, the recency or non-recency of the sample is not particularly problematic. The seroprevalence work in Ontario is directionally consistent with that done for other populations, but I remain hesitant to accept the magnitude. SJ

Not true. Ontario's data is much more recent than most of the serology reports in that thread. Based on his chart, he seems to by using NY serology from mid-June. But this is irrelevant since Ontario's epidemic was crushed by the end of June (a few deaths per day since). There is no problem with the Ontario analysis due to recency or lack of recency. Essentially, what the data show is that about 0.4% of Toronto residents were officially diagnosed as being infected with covid up to June 30, but about 1.5% of people in Toronto who submitted a blood sample to the health system showed the presence of antibodies. On the face of it, the ratio between people carrying antibodies and those officially diagnosed is a bit low compared to other seroprevalence studies (ie, ~3.75:1), but on the other hand people who are young and healthy do not generally submit blood samples to health care system, so it is quite possible that the blood samples were biased to those who had existing co-morbidities and who were quite rightly taking their own social-distancing measures to avoid the virus. SJ

The twitter posts referenced compare a capital and most populous city in all of Scandinavia (Stockholm) to the average for all of NY State, much of which is very rural and sparsely populated. New Jersey, Rhode Island, Massachusetts, Connecticut, Maryland and Delaware all rank as more densely populated that NY State. This particular brand of false equivalency has been brought up before in this thread and seems to be a typical tactic for those who aren't really searching for the truth, but are pushing a political agenda. So we all might want to be on the look out for it before posting in the future, unless it is to point out misleading presentation of data. Stockholm and in fact all of Scandanavia are likely on average more homogeneous, and better educated than the USA. Plus they as a region have done very well at remaining cohesive by fighting disinformation and fake news, in fact I suspect these tweets would find a less receptive audience there than in the USA. Here is at least one relevant article regarding Finland, but there are many I could have chosen instead including more academic articles relevant to Sweden: https://www.theguardian.com/world/2020/jan/28/fact-from-fiction-finlands-new-lessons-in-combating-fake-news The real secret to Stockholm is likely education level, cohesiveness, and a pro-social culture. With a conscientious population, there is less need to threaten or cajole, especially if they see it as personal interests aligning with self-interest, which should be the case all around the world. Evidence that they acted individually without national mandate can be seen in the data indicating a decline in mobility in Sweden which indicates they achieved similar or better social distancing without dictates. As I have been saying since February, it's a "complex, adaptive system". Human beings have agency and that complicates policy decisions and analysis. Stockholm syndrome could be useful to study for anyone who feels they are repeatedly fall prey to bad information: https://en.wikipedia.org/wiki/Stockholm_syndrome Actually, I found that twitter thread to be fascinating. The participants seemed to be fixated on the difference between the population density of NYC vs Stockholm, and they dedicated virtually zero attention to the argument that the fellow was actually making, which was that it is possible that the Covid begins to peter out when a population hits 15%-20% presence of antibodies. I don't doubt that comparing the density of NYS to Stockholm is analytically weak, but that particular element was not crucial to his central hypothesis. Does anybody have any comment about his assertion that Covid peters out at 15%-20%? Looking at the United States at the national level, the daily new cases seems to be trending down. Official confirmed cases are 5.4m, and most seroprevalence studies show that there are 10x to 15x people with antibodies compared to the official number of cases. So, where is the US currently at from the perspective of prevalence of antibodies? It would not come as a surprise to me that the US might be in that 15%-20% range right now. As I noted, the daily new cases seem to be trending down, so as I asked last week, have we hit peak-virus in the US? The twitter guy's hypothesis would be yes. Or like the participants of the twitter thread, we could ignore the guy's core hypothesis and focus in on a peripheral analytical weakness. SJ

Of course there are many variables, but still, it's a data point that doesn't help support the approach. A few observations: 1) You need to look at both Q1 and Q2 GDP numbers because these are QoQ numbers, not YoY. Of particular importance, Denmark experienced mild economic contraction in Q1, while Sweden had mild economic growth in Q1. So, start from January 1 and look at the cumulative contraction; 2) Reducing the economic pain was only one of several objectives articulated in the Swedish approach. To a large extent, Sweden's contraction in GDP is one of the most favourable in the western world. But, equally important, the psycho-social benefits of maintaining organized schools, sports, and social activities is very important. The lockdowns have been psychologically very challenging for large swaths of the population in most western countries, but likely less so in Sweden; 3) Sustainability of response measures was a key objective. It's all fine and good to take heroic measures to suppress the transmission of the virus, but can those measures be sustained for a prolonged period until a vaccine or treatment are developed? A large swath of Europe took those heroic measures and managed to flatten the curve, but it's worth taking a look at where they are 12 weeks after having relaxed those heroic measures (I am talking about you, Spain, France, Belgium, Netherlands!). As I noted earlier in the week, Spain and France are one doubling away from having their daily new cases being at the very same levels that triggered their lockdown in March/April. This summer is merely the first intermission and there is still a couple of periods of hockey left to play. It will likely be a couple of years before we can come to a fulsome assessment of the relative success of the response strategies that countries adopted. SJ