Modelling the economic impact of coronavirus
| 15 Apr 2020
The outbreak of the coronavirus pandemic has had a devastating human impact, respecting no borders, nationalities, professions, sex or age. At Sarasin & Partners we have warned for some time about the impact on the Chinese economy from both the enormity of the health crisis unfolding and the government’s lockdown measures.
The Chinese response, based on suppression of the spread of the virus, has now been replicated across much of the world. In quick succession, countries have implemented nationwide lockdowns to slow transmission of the virus, and travel restrictions to prevent imported cases.
Meanwhile, financial markets have crashed into bear territory, at a pace never seen before. Central banks and governments have stepped in to provide monetary and fiscal backstop to the economic devastation that the crisis will inevitably bring forth. Yet, with economic data released with a lag, and lockdown policies dependent on health outcomes, how does one begin to estimate the economic loss and provide a reasonable guide for the global economic outlook?
Clearly, the art of economic forecasting is difficult at the best of times. Uncertainty is sky-high, and making predictions may appear to be a fool’s game. Yet understanding the methodology, underlying assumptions, and the various moving parts can prove to be a useful guide. And is a spoonful of monetary and fiscal sugar, best enjoyed whilst under house-arrest, the right medicine?
Adopting a back-to-first-principles approach we try to estimate the global economic cost using the GDP framework. GDP is measured in three ways, by expenditure, income or sectoral production, all of which produce an equivalent estimate. We adopt the production approach as the economic effects of the pandemic are more easily calculated on the service sector owing to government restrictions on mobility, travel, accommodation and retail. The service sector accounts for around 65% of our $86 trillion global economy, but the percentage ranges from 52% in China to 75% in the US. Second, we look at the timeline of government lockdown measures implemented, starting with China in the last week of January, followed by (parts of the) euro area and Japan in February, and the US and UK in March. Accordingly, the output of the service sector is haircut by the number of weeks that countries are subject to lockdown policies in each quarter – in China this equates to a reduction of 35% in Q1, 10% in Q2 and 5% in Q3. Other countries and regions experience a similar magnitude of shock, around 30% but starting later in Q2. We also assume that the health crisis does not resurface later in the year, such that all restrictions will be gradually removed and all countries return to full capacity by Q4.
This exercise generates a range of estimates for the impending global shock. If we consider that activity in the agricultural and manufacturing sectors continues unharmed at pre-crisis rates, we arrive at the optimistic range of our base-case estimate (-2.7%). Alternatively, if activity in those sectors comes to a standstill, global economic growth could be at the lower end of our base-case estimates (-3.7%). Inevitably, the key message that emerges is that a large global recession is inevitable, and the shock will be almost twice as large as the 2008-09 global financial crisis (figure 1).
By country, the estimates vary depending on the importance of the service sector, as well as the government’s fiscal response, which helps offset some of the impact. During times of recession governments have historically responded with a stimulus of 2% of GDP on average. This forms our assumption for most regions, other than the US which at the time of modelling was proposing a $1 trillion fiscal package (or 5% GDP).
The other key underlying assumption is the question of when economic activity can be fully resumed. Given uncertainty over the duration of the health crisis and the prospect for vaccine availability and therapeutics, we consider alternative scenarios for when lockdown measures will be removed. In the base-case scenario, we assume that activity fully resumes to pre-crisis levels in Q4, while the bear and bull case assume a slower and faster return to activity respectively.
What can change these estimates?
Firstly, we have not accounted for feedback loops, which are both important, but exceedingly complex to model. These dynamic second-order effects include supply-chain disruptions across and within countries, financial market effects, and currency volatility. For instance, the closure of factories in China in January quickly led to the shortage of car parts across the world. Temporary car plant closures were announced by Hyundai in South Korea, and Nissan in Japan, while Jaguar Land resorted to flying car parts in suitcases out of China to prevent plant closures in the UK.
Second, policy measures can help arrest some of the projected decline in activity. Central banks have acted early and decisively to boost sentiment by injecting unprecedented amounts of liquidity into the financial system to prevent an economic crisis turning into a financial one. Government fiscal measures have also exceeded expectations, with the US now implementing a 10% fiscal package focusing on small and medium-sized industries, airlines and the healthcare industry, as well as direct cash handouts to households. Even in Germany where fiscal deficits are outlawed by the constitution in normal times, the government has unleashed historical stimulus measures. Job retention schemes have formed a key part of stimulus measures to prevent mass unemployment across countries. In the UK, Germany and Australia, governments have committed to paying firms up to 80% of their employee wage bills for 3-6 months. Undoubtedly, government budget deficits will rise sharply, and reach beyond 10% in some countries. With capitalism not to blame for the health crisis, it appears that fiscal prudence has given way to fiscal decadence – at least in developed economies.
Third, the duration of government lockdown measures is impossible to predict. While we assume that activity will return gradually in Q3, and normalise by Q4, this is contingent on health outcomes. Slowing infection rates, in turn, are driven by and large by compliance with social distancing measures. This is in the absence of long-term solutions such as a vaccine development – which is deemed to be 12-18 months away – or best a cure. In China, where compliance with social distancing measures was enforced through strict government surveillance, the economy has now resumed to around 70-90% capacity after measures were first imposed in mid-January. This suggests that social distancing measures need to be in place for a minimum of eight weeks.
All things considered, and forecasting errors aside, the economic costs from the pandemic are estimated to be astonishingly severe. This will represent a large, permanent cost to the global economy, which cannot be resolved by accelerated catch-up growth in the subsequent quarters.
Yet, as countries look ahead and assess the path back to normal, we ask can the return to normal be managed in less costly terms once the peak in infection rates has passed? Social distancing, while proven effective in the absence of other solutions, is a blunt, indiscriminate option, with the objective of flattening the epidemiological curve to reduce pressure on overburdened healthcare systems. However, as shown, it is also very costly for the economy, and the long-term effects of extraordinary stimulus are yet to be seen.
South Korea and Germany have shown that diagnostic testing is important – particularly given high asymptomatic cases. Both countries have tested more than half a percent of the population – greater than other countries – and reported significantly lower mortality rates. Greater testing of the population for current infection, as well as blood tests to check for the presence of antibodies and thereby immunity, can be an important development to correctly identify those who need to be quarantined and those that can safely return to normal economic activity. Better understanding of the health risks can result in better decision making.
Another option, albeit controversial, may be to employ digital surveillance technologies to enforce quarantine measures and ease the burden of contact tracing. While running against the liberal ideologies of the west, there is an active debate about whether the merits of greater surveillance measures are justified. These measures can be wide ranging, but have thus far ranged from the installation of CCTV cameras in China to alert authorities of quarantine infringements, to a government app in Singapore, to the wearing of digital wristbands in Hong Kong, and smartphone location data and location mapping in South Korea. Yet, as the US has shown with the introduction of the Patriot Act after the 9/11 Terrorist Act, surveillance laws are often difficult to wind back once put in place.
Lastly, there is also another option, much neglected in recent years: greater global cooperation. While developed countries may be able to help offset the health crisis response with monetary means, developing countries will struggle. Their choices will be difficult; lockdown measures will deny many of their livelihood and may cause mass famine, yet allowing the virus to spread unsuppressed could cause particularly high fatality rates given poor healthcare systems. India has chosen the route of suppression. Yet, as we have learnt from this experience, we live in a closely connected global village. The sharing of information, best practices, and indeed resources can improve outcomes. Cooperation can deliver outcomes better than the sum of the parts, both help ease the pain in the east, and also prevent subsequent waves returning to the west.