The original strategy of the UK and it seems the current preference of Trump is that it is better to allow people to behave as normal. There will be more deaths than the lockdown approach, but this it is argued that this is better than the alternative of a huge economic hit.
There is a similar and related idea that we might “shield” the elderly and vulnerable and allow everyone below the age of 65 to go back to work. The economic/health trade-offs may be different at different ages.
I wanted to make some estimate of what the world might look like. Please bear in mind that I have no expertise in this area and I very much hope I am very wrong, but my initial impression is that the results could be catastrophic.
The recent report from Imperial has been very influential in changing the UK government’s plan away from “herd immunity” (also known as “let the old people die”).
A key data table is below showing that mortality rates for people below 60 are very low. This initially supports the idea that targeted isolation could be a sensible approach.
Their summary of the implication of the 2 paths is below:
- if all patients were able to be treated, we predict there would still be in the order of 250,000 deaths in GB, and 1.1-1.2 million in the US.
- In an unmitigated epidemic, we would predict approximately 510,000 deaths in GB and 2.2 million in the US, not accounting for the potential negative effects of health systems being overwhelmed on mortality.
This to me seems quite compelling and explains the shift towards the “flatten the curve” approach the rest of the world is trying.
As an aside, a paper by an Oxford academic has got a lot of attention today as it suggests that half of the UK population might already have the virus and all will be ok. I think this should not be taken too seriously. Looking at the paper, it is a mathematical approach to build a parametric model where we have no idea what the parameters are. This means that you can assume pretty much anything and thus get pretty much any answer. Their description in the paper itself summarises why we should take no notice:
“Our overall approach rests on the assumption that only a very small proportion of the population is at risk of hospitalisable illness. “
If you have a mathematical model which starts by assuming that no one gets sick, it is not too surprising that the output result is that no one gets sick. This type of analysis reminds me of mathematical economists during the 2008 crisis. Utterly useless.
I wanted to take a different approach and go a little further and examine the age breakdown of the deaths, and possibly how we could estimate how the mortality rate changes if we do not flatten the curve. It is here that I found the results most surprising.
I took the recent CDC study in the US, assumed a third of the population become symptomatic, and came up with a couple of ways of thinking about mortality rates in the population
- Total deaths – version 1 (commonly seen in the press)
Take the distribution of population by age and multiply by the current mortality rate of around 1% plus some estimate of how many people are infected - Total deaths – version 2
Instead of the current mortality rate, estimate the mortality rate we would see in a crisis with the health service being overwhelmed. For this, I took the number of patients requiring ICU as a proxy.
The huge assumption I make in this very simplistic model is that, if the health service is overwhelmed, no one has access to ICU and all patients that required it will die (without intubation you cannot breathe). This is obviously overly pessimistic as some patients will still have access to ICU and intubations will be taught to many more non-ICU doctors and nurses. One the other hand my estimate of a third of the population becoming symptomatic I do not think is extreme. The high resulting mortality rate is not something I would expect but it provides a useful examination of a worst-case scenario and we can explore the implications.
The number of deaths is shocking at 5.8 million
But it was the age distribution that initially surprised me because I had been fed so many numbers which are mortality RATES I had not been thinking of the absolute numbers. Although the mortality rate of the old is much higher, the absolute number of old people is of course far lower. Furthermore, it is the younger people who respond better to treatment (ie many more younger people are hospitalised but far fewer die. If they cannot go to a hospital, we should not expect the same outcomes.
In this model, we have over 3 million Americans under the age of 65 dying from Coronavirus.
To be clear this is not a prediction. This is not that sort of model. It is an examination of the sort of impact we might see if we stop social distancing and go back to life as normal.
Stay home – stay safe
Applied Macro 
While the table makes sense, I don’t understand the chart and/or it doesn’t match the table. In the chart, it looks like the absolute number of deaths in the over 85 bucket is larger than the other ones.
Otherwise, I agree with your analysis, given the assumptions that make it a worst case.
LikeLike