Coronavirus: initial results from economic modelling
In Coronavirus: how to model the economic impacts of a pandemic Hector Pollitt, our Head of Modelling, wrote about the difficulty involved in using economic models to tackle challenges like the novel Coronavirus (COVID-19).
Here, we show the results of an initial modelling exercise that tries to understand and quantify effects in a specified case. As such it can serve as an indication of what can be expected.
Of course, economic considerations come second to protecting public health. However, economic implications are not only linked to health outcomes, but can stay with us after the health crisis is averted.
Therefore, for all the noted limitations of economic models and the uncertainty of outcomes, it is important to assess how the global economy may be affected by COVID-19 and any measures taken to limit its extent.
According to economist Simon Wren-Lewis it is accepted that pandemics have a two-fold effect on the economy. They create a supply-side shock, reducing working capacity (due to health effects and measures to contain the virus – e.g. quarantines), but they also generate a demand-side shock.
Just think how people are stocking up on medical supplies while cancelling flights and holiday plans. At the same time, social consumption (mainly services) may also suffer a fallback. The E3ME model provides a framework to assess both the demand and supply-side shocks associated with such a scenario.
Constructing a global pandemic scenario
There are various uncertainties about the COVID-19 virus that are yet to be pinned down by healthcare professionals and experts in infectious disease. The best we can do is to use numbers estimated by these experts and to base our assumptions on historical experience.
For this modelling exercise we have set up two scenarios: (1) a global pandemic scenario, where we assume that the virus will not be stopped effectively, infects a significant share of the global population, and where governments do not intervene successfully; and (2) a scenario where we account for government relief packages already announced by the European Commission, and national governments. The two scenarios are identical in most aspects, the main difference being the level of government response to the pandemic.
To model these scenarios in E3ME, we first need an assumption for the scope of the pandemic. Estimates of the potential extent of infection are mostly informal: Marc Lipsitch of Harvard has estimated a 40%-70% infection rate, while Gabriel Leung of Hong Kong University has suggested an infection rate of about 60%. Based on these numbers, we assume a 60% infection rate for this analysis.
We assume that there are several outcomes of the infection. Cases with severe symptoms are assumed to lead to a loss of working hours, as people take time to recover from the virus. We calculate the reduction in the productive capacity of the economy, taking account of expected mortality rates along with a loss of working hours due to the prevalence of severe symptoms. It is also assumed that there will be strict quarantines globally, for up to a month. To account for the loss of productivity due to remote working, the loss of working hours is discounted with sector-level remote working potential based on US data.
Demand-side effects are even more uncertain. Our estimates are largely based on past experience, including the magnitude of effects that we have already seen in China. Impacts that are already evident in the Chinese economy include the fallback of road transport, entertainment, tourism and air travel.
Our assumptions are summarised in the table below. We keep these assumptions fairly simple as uncertainties are high and, at this point, transparency is more useful than a high level of detail.
Modelling results – a blow to the economy with varied recovery
We compare results to a standard baseline (how the economy would have developed without the virus). We estimate a 2.3% loss of global GDP this year for the global pandemic scenario which is mitigated to a 0.8% loss of global GDP in the scenario with government intervention. At the beginning of March, the OECD published a forecast of 1.5% global GDP loss this year (in a Downside scenario). The difference in our results can be understood as (1) the effect of demand-side impacts, (2) the fact that we are considering the situation with more information (and we already start to see how the EU is reacting, for example).
The E3ME model also estimates what kind of recovery we can expect. In What Coronavirus Could Mean for the Global Economy BCG partners have shown that historically after pandemics the recovery is likely to follow a “V” shape. Our results support this – with government interventions there is likely to be a global rebound even if lost output is not recovered. But while global GDP can stabilize as early as the following year, employment – without fiscal interventions – could show a slower recovery.
Different consumption behaviour, different recovery
Sectoral effects are also of interest as sectors can produce varied recovery paths as noted by Bain. Tourism is likely to be one of the hardest hit sectors. Measures to contain the virus, and the virus itself, have already caused a major loss of tourism revenues. But what could be even more important is the nature of the purchases of tourism and ‘social consumption’ services.
In the case of consumer goods (for example, home electronics or apparel) if purchases are delayed, there will probably be an excess purchase when the crisis is over. But this is less likely to be the case for vacations and so the consumption (and output) is lost. Our Chairman Richard Lewney provides a further explanation of this in his explainer blog: The economics of the Coronavirus pandemic.
To some extent our modelling bears this out: the recovery in growth of sectors with ‘normal consumption’ (e.g. food, beverages, apparel) and with social consumption (e.g. entertainment, sports) – even if they suffer high losses in this year – can be relatively quick, but the tourism sector could potentially face a slower recovery.
The impacts in all sectors, as modelled in E3ME, are partly mitigated by price effects – the loss of demand results in lower prices in the economy, which then supports consumption a bit (albeit at the expense of profits for the producers).
The role of policy
The final uncertainty relates to potential policy responses. Our modelling of the scenario with government interventions includes some recently announced fiscal policies, “relief packages”, to counteract the economic slowdown.
We modelled these packages as a general increase in government spending over the next four quarters. Such packages will be needed to sustain the health of our economies. To maximise effectiveness, these packages will also need to be targeted to support those sectors, such as tourism, that are most likely to suffer substantial losses. As it can be seen from the figure above, our modelling indicates that these packages are important not only to decrease the losses of this year, but to mitigate medium-term economic decrease as well. Richard writes about some of these potential policy responses in his post.
Modelling exercises like this one could play an important role in identifying how sectors will ultimately be affected by the spread of a virus such as COVID-19, while modelling of potential policy actions could help us understand the effectiveness of government action.
 We have not included packages announced after 17/03, including the ECB’s EUR 750bn relief package.  A list of these measures is available here, our modelling is based on information available as of 17/03/2020.