Category Archives: Uncategorized

Covid-19 in the US and Canada – deaths and unemployment

A few month ago I drew a comparison between US and Canada on two dimensions: Covid-19 deaths and employment losses since January. The data I used told a simple story: Canada has seen less death and roughly similar employment losses. This suggests that Canada is doing something (relatively) right: compared to the US, deaths are lower at no obvious economic cost.

Well, I have updated these data and in the process accessed some more reliable labour force information from the US and Canada. The sources for employment data:

Canada: Labour Force Survey, Statistics Canada

US: US Bureau of Labor Statistics

For Covid-19 data, I went to national government sources. All of these numbers contain some error — there is a lag in reporting deaths, and there are all sorts of ways that labour force data can be wrong. However if the data are anywhere close to correct, they tell an interesting story:

Note first that the employment recovery in Canada is well ahead of that in the US, and it has been trending that way since June. Second, note that the risk of Covid-19 death in the US is about 2.5 times higher than it is in Canada, but that this difference is particularly striking since July, at which point mortality risk began to diverge considerably. As of September 2020, the probability of dying from Covid-19 in the US is 16 times higher than in Canada.

In terms of attributable risk, in Canada, more than 14,000 lives were saved by not having the same risk profile as the US; conversely, over 120,000 American lives would have been saved if the US had Canada’s risk profile.

Take a look at the calculations yourself. Feel free to tell me if there are any necessary corrections!

Update to Covid-19 death comparison between Canada and the US

With a month of new Covid-19 data and employment data, the evidence I made reference to in my previous post can be updated. I’ll make a quick note here about it.

In August, the US saw more than 30,000 Covid-19 deaths. Canada saw 206. This amounts to a near 17 times difference in the rate of death–meaning that the average American was 17 times more likely to die from Covid-19 than the average Canadian in the month of August. I am not aware of any other cause of death, disease or injury that shows an equivalently striking difference between these two countries 1. For reference, the homicide rate in the US is only 2.8 times that of Canada.

The trend in employment growth remains about the same as in previous months. Canada saw a larger percent improvement in employment growth in August at 1.28% when compared to 1.08% in the US, but Canada is still slightly behind the overall return to normal (Canada is 5.57% below January employment and the US is 5.24% below January).

So far the message is clear: the behaviour of Canadians has saved lives that would not have been saved if Canadians adopted the US response to Covid-19. Let’s keep at it. Whatever we are doing seems to be working fairly well in comparison to the most available alternative approach.

_________________________________________________________________________

  1. I did a quick and dirty estimate of curling related injuries based on some research and Internet-based data. Google says there are around 436,000 regular curlers Canada. In the US, Google says there are 25,000 registered curlers. The injury rate (per curling exposure) has been estimated at around 2 per 1000. Putting these numbers together suggests that the average Canadian has a 17 times greater risk of curling injury than the average American.

Lives saved in Canada during the Covid-19 crisis

Here we are in August 2020 and the pandemic seems to be firmly attached to our short and medium term future. Many of us are holding out hope that vaccines will bring everything back to normal in short order. Yet questions remain. Will we see effective vaccines soon? Will they slow the spread of infections? Will they be safe? Will they be accessible?

Other questions–about the safety of returning to school, the risk of a second wave, and the long term impact on the economy add yet further uncertainty about our future.

The uncertainly may leave you feeling a bit depressed. But there are reasons to feel optimistic. While the future is uncertain, in Canada (and many other countries) there is growing evidence that this infection is manageable (masks, physical distancing and hygiene all seem to work), and returning to quasi-normal life could happen before an effective vaccine is widely available.

The purpose of this short essay is to make an argument that things could be much worse and that Canadians should take some satisfaction in the how we’ve handled the Covid-19 situation so far. I make this argument simply by drawing a comparison between Canada and our closest counterfactual: The United States of America.

Part 1. The epidemiological argument

In between March 1 and July 31st 2020, more than 150,000 Americans died from SARS-CoV-2 infection. During the same period, roughly 9,000 Canadians died from SARS-CoV-2. In per-capita terms, the US saw about twice the rate of death than Canada. This is widely known, and on the face of it suggests that Canada did something right that the US did not do. However, this difference is far more striking when viewed on a monthly basis. Take a look at the table below:

Looking at deaths per 100,000 for April, May and June, the ratio of rates hovers around 2 to 1 (the US has roughly twice the death rate of Canada). Who knows what accounts for this–is it access to health care? Population vulnerability? Environmental differences? It’s hard to say. What is clear is that in July, the ratio of the Covid-19 death rates changes dramatically: in the US, the rate is 7 times higher than in Canada. Such a short term change is almost certainly not due to some structural difference between the countries (like health care access or underlying health status), but due to short-term policy decisions and behaviours in the US that occurred in June and July. What specific decisions and behaviours, who can say. What is clear is that a country to country comparison strongly suggests that something right happened in Canada and/or something wrong happened in the US.

To make the difference tangible, I’ve also calculated attributable risk seen in the last two columns in the table above. The attributable deaths in the US are the deaths that occurred as a result of the excess mortality rate in the US. In simple terms, these are deaths that would not have occurred if the US had the same risk profile as Canada. In the last column these are deaths that would have happened if Canada had the US mortality rate. We can think of these as lives saved due to our decisions and behaviour that were different from those in the US.

Again, the July data are most striking. If Canadians had the same risk profile as the US, an additional 2336 people would have died in July alone. In the US, more than 20,000 people died in July that would not have died if the they had Canada’s risk profile.

Part 2. The economic argument

Some have suggested that the US made a calculated trade off. The argument is that unemployment causes death too. By shutting down economies people lose jobs, which will cause more death as a result of unemployment and economic desperation. By returning to normal faster, there was more Covid-19 death in the US, but less unemployment-related-death.

What do the unemployment numbers say? Take a look for yourself:

In total, the US has slightly fewer net job losses than Canada for 2020–by about half a percent. However, the data also show that the timing of job losses differ between these countries. In Canada, there was a rapid drop in employment and a slightly lagged recovery, but more job growth in June and July. Once the August numbers come out, Canada may yet end up better off overall. In any case, at the moment, the difference in Covid-19 related employment impacts between these two countries seems pretty small.

It’s hard to say definitively if the small difference in employment recovery makes up for the excess loss of life due to Covid-19. In the US, half a percent of the workforce corresponds to about 700,000 workers. This means that if the US had Canada’s employment profile, 700,000 fewer Americans would have a job at the end of July. Are 700,000 jobs a worthwhile trade-off in exchange for the lives of the 72,000 excess deaths due to Covid-19? It’s hard to judge. However, the relationship between unemployment and mortality risk does depend on time; short term unemployment is more weakly associated with increased mortality than long term unemployment. Many of the job losses thus far have been short lived. Moreover, the wealth transfers (the CERB in Canada and whatever they have in the US) would probably offset some of the trauma of losing a job. So it seems unlikely that unemployment due to Covid-19 will have lead to a spike in mortality similar to equivalent unemployment growth in ‘normal’ times.

Conclusion

The US and Canada are suitable for comparison because of the timing of infection, and the similarities in culture and population vulnerability. Comparisons between these two countries are more meaningful than comparisons between either of these countries and any other country in the world. In this comparison, Canada comes out clearly ahead.

Many countries have lower mortality than the US and Canada, though Canada’s mortality rate is (as of July) low by any standard of comparison. Lots of things went wrong in Canada, and maybe thousands of deaths could have been prevented. However, in comparison to the most natural alternative (the US) Canadians should be confident that some of the actions we took clearly saved lives, and that if we continue on this course, we can prevent many more unnecessary deaths.

A paradox of consensus?

Consider the following.  After years of study, researchers estimate with a high degree of certainty that there is a 60% chance of a particular event, (call it A), happening.  When asked to make a discrete prediction of whether or not A will actually happen at a moment in time, 100 out of 100 experts independently conclude that the event will happen.

Now consider this.  After years of study, researchers estimate with a high degree of certainty that there is a 50% chance of a particular event, (call it B), happening.  When asked to make a discrete prediction of whether or not B will actually happen at a moment in time, 50 out of 100 experts independently conclude that the event will happen.

The expert predictions in both of these scenarios are perfectly rational.  These independent expert predictions provide the most accurate long-run information about the whether or not A and B will happen.  However, in the second scenario the aggregate prediction (e.g., by taking the average) is precisely correct, and the first scenario the aggregate prediction is infinitely wrong.

If you want to see a real world example, take a look at the predictions of 18 experts on the NHL post season for 2020:

https://www.sportsnet.ca/hockey/nhl/sportsnet-nhl-insiders-2020-stanley-cup-playoffs-predictions/

All 18 experts predict that Pittsburgh is going to win their playoff series.  For each expert this prediction makes sense–by most measures, Pittsburgh is the better team.  However, this information does not give me a realistic representation of the actual probability that Pittsburgh will win.  As bad as Montreal is, they have a better than 0% chance of winning the series.

Attack falters again as Canadiens fall to Penguins - TSN.ca
Used without permission.
https://www.tsn.ca/attack-falters-again-as-canadiens-fall-to-penguins-1.653364

In contrast, if we sum the total number of experts predicting New York will win and divide it by the total number of predictions, New York is given a 56% chance of winning their series.  This number is probably a pretty good long-run estimate of the probability that New York will win the series.  There is no consensus, and that actually yields a more realistic aggregate prediction!

What this quasi-paradox suggests is that the closer experts are to a consensus about an event, the more likely we are to get a bad aggregate prediction of the true probability of an event.  If we combine the expert predictions, we will think that the event is more (or less) probable than it actually is.

This is a reminder of why when consulting an expert, we should not ask if something will happen, but instead ask about the probability that something will happen. Among other things, this probability is something we can average across experts to get a sort of ‘meta’ prediction.

It is also a reminder not to mistake an expert consensus about an event as equivalent to a guarantee that the event will happen.