This is an update of an earlier postwhere I described the results of a study I did with others on a flu pandemic. (It also appeared in a VoxEU ebook.) I have had a lot of enquiries from journalists in Europe about this blog and the paper, and of course what everyone wants to know is how similar is our modelling to the impact of this coronavirus. The features of this pandemic are now clearer than when I wrote the post, so I can say a few things on this issue. Our study looked at two cases: a base case and a severe case. The key parameters were the attack rate (how many get the virus) and the fatality rate (how many who get the virus eventually die). The base case is far milder than this coronavirus, so I’m going to focus on the severe scenario we modelled. That had an attack rate of 50%,
Simon Wren-lewis considers the following as important: coronavirus
This could be interesting, too:
Simon Wren-lewis writes The political economy and psychology of COVID rebounds
Jeffrey Frankel writes The Impact of the Pandemic on Developing Countries
Simon Wren-lewis writes The reality behind fiscal scare stories, or mediamacro is alive and well
Simon Wren-lewis writes Sabotaging the recovery
This is an update of an earlier postwhere I described the results of a study I did with others on a flu pandemic. (It also appeared in a VoxEU ebook.) I have had a lot of enquiries from journalists in Europe about this blog and the paper, and of course what everyone wants to know is how similar is our modelling to the impact of this coronavirus. The features of this pandemic are now clearer than when I wrote the post, so I can say a few things on this issue.
Our study looked at two cases: a base case and a severe case. The key parameters were the attack rate (how many get the virus) and the fatality rate (how many who get the virus eventually die). The base case is far milder than this coronavirus, so I’m going to focus on the severe scenario we modelled. That had an attack rate of 50%, and a fatality rate of 2.5%.
How does this coronavirus compare to that? The attack rate is almost certainly higher: we are only beginning to know the number of people who get the virus without even knowing it. The fatality rate of this coronavirus is unknown, partly because of the uncertainty about the attack rate. At first it was thought to be above the 2.5% number in our severe scenario, but it now looks like it is lower. However this number is not critical for the direct economic impact of coronavirus.
The reason is obvious once you think about it. The fatality rate for coronavirus is mainly among those who are retired, so there will be little impact on the size of the long term workforce. However there is a huge indirect effect of a high mortality rate, and that is on how the government reacts to the pandemic. I will talk about that below.
So the severe case seems, so far, a useful guide. In our paper we also looked at the impact of a large fall in ‘social consumption’ (pubs and restaurants, travel etc). We looked at a pandemic with or without this, partly because our base case was mild, and partly because some previous studies had not looked at this. However it is already clear that we were absolutely right to look at this issue. Airline industries around the world are mothballing most of their planes, restaurant bookings were way down before they were ordered to close, and so on.
One final point, that I will also discuss below, is that we assumed the pandemic was just a three month affair. If we look at our severe pandemic case including falling social consumption, we had GDP in the pandemic quarter falling by 30%. There was a similar fall in consumption. However, because our severe pandemic lasted for just one quarter, GDP for the year as a whole fell by only 6%. So how good a guide - in rough orders of magnitude - are those numbers to this coronavirus pandemic?
Let us look at the first quarter impact first, because that we know most about. Pretty much all Western countries have a suppression regime in place, which will probably last around three months while the virus numbers are brought right down. This reduces GDP from both the supply and demand side, and this combination makes it tricky to model with any degree of accuracy. The reason the fall in demand and supply don’t completely complement each other is that the fall in demand is going to be concentrated in sectors, while the supply side impact is more widespread.
Our study suggested the main supply side impact comes from school closures. On the reasonable assumption that these last until the summer holidays, that means a large section of the workforce will have to stay at home rather than work. (It will be interesting how the ONS measure output from working at home!) In contrast those who get sick will be off work for only two weeks. Furthermore, and unlike our exercise, workers will be reluctant to use grandparents if they get a choice.
On the demand side the drop in social consumption that we assumed would happen through individual choice will now be made by government advice/instruction. You can do your own calculation by looking at a breakdown of consumption. There is a limit to how far GDP can fall because we will not eat less, and we will not spend less on housing or heating. Expenditure on clothing and particularly durables may be delayed to some extent, as people avoid personal contact, but online purchases should continue.
Given all this, our estimate of a fall in GDP in the first quarter of the outbreak of -30%, although huge, does not look obviously wrong and may well be an underestimate. But of course this is not a precise figure at all, and it will vary from country to country depending on the strength of social distancing controls, the degree of business and worker support from governments and whether governments can relax social distancing before 3 months are up. The GDP fall could easily be 20% or 40%. Because the crisis began in most Western countries at the end of the first quarter, the Q1 GDP fall will be a lot less than this, and the Q2 fall relative to Q1 will be slightly less.
Where our study is less helpful is in thinking about what happens after the strict control period comes to an end because new cases come right down. We assumed there would be no attempt at suppression beyond school closures (because our main case was much milder), so our pandemic ended after three months. This coronavirus pandemic will not be a one quarter affair, because governments quite rightly have not been prepared to see a short sharp peak where their health services will be overwhelmed.
Once the number of cases are brought right down, it is likely governments will do what China is currently doing, and move to a strict contain regime. This involves a very stringent regime to test those who might still get the virus and the isolation of all known contacts, combined with some continuation of social distancing controls. China even has an app that helps do the tracing. I think this will inevitably be how other countries deal with the virus once numbers are down. Some will do it well, and others may not, leading to controls being reintroduced.
Containment is, of course, what all Western countries tried to do at first, but in nearly every case they moved too slowly. (The UK even tried the herd immunity route, despite what ministers might say, and seemnot to have done the basic sums of what that meant for the NHS and deaths.) This general initial failure was not inevitable: a few countries like Singapore have been much more successful. In short, nearly all Western governments underestimated how quickly the virus would spread. But it is easier to control the virus by relaxing controls than creating such controls from scratch, and hopefully lessons have been learnt. In addition, governments will have had time to ramp up testing capacity, and think about better tracing.
The key issue for the economy once numbers come down is how many controls can be reduced or eliminated while keeping a lid on new case numbers. Once again on the supply side whether schools reopen is key, and on the demand side it is what parts of social consumption can be made safer. Because some relaxation will almost certainly be possible, then GDP growth will partially bounce back, but how much they will bounce back is very unclear at present. Here our study, which had GDP being above the no-pandemic case in the second quarter, does not apply to the pandemic we are now in. As a result, the first year GDP impact of -6% in our study is much too small. It is impossible to predict what the full year impact will be until we know what controls are essential and what controls can be relaxed while maintaining an effective test/trace/isolate regime. We will get some ideafrom China.
I want to add a little, much more speculatively, on political consequences. There has been some talk about this pandemic provoking a scaling back of globalisation. You can see that happeningif talk about this being a Chinese virus brought in by foreigners is given air time. But it is equally possible that the opposite will happen as a result of the virus.
As we relax controls, governments will want to know what countries it can open up travel to. That will depend on how effective their test/trace/isolate regime is. As we have already seen with Macron phoning Johnson to urge stronger controls, international pressure of this kind will happen. As this goes on, there will therefore be a huge incentive for countries to copy successful measures. At the start of this pandemic we saw a complete lack of global coordination, despite WHO’s best efforts. Subsequently we should see much more.