BIS: Macro Spillover Effects of Pandemic

BIS: Macro Spillover Effects of Pandemic

The Bank for International Settlements (BIS) released a BIS Bulletin today written by staff members Emanuel Kohlscheen, Benoît Mojon and Daniel Rees, which predicts a total GDP shortfall that could be as much as twice that implied by the direct initial effects of confinement in the face of the global coronavirus pandemic.

Given the historical persistence of economic activity, the reduction of GDP due to confinement measures is likely to drag on over several quarters, the authors say, noting that the views expressed are those of the authors and not necessarily the views of the BIS.

This persistence reflects in part two types of spillovers across countries, the report says. One is due to the risk that uncoordinated confinements lead to repeated virus outbreaks and confinements across the globe. Another is the more traditional trade and financial integration interlinkages.

Economic spillovers and spillbacks across the major economic blocs are large, they note. There is no immunity from the economic effects if the epidemic is controlled in only one or two regions. Countries should adopt confinement, border control and macroeconomic policies that internalise these global considerations, the authors posit.

Noting that the spread of the coronavirus (Covid-19) is an unprecedented shock to the global economy, the authors point out that there is a risk that, until a vaccine or an efficient treatment is developed and distributed globally, countries may undergo additional waves of contagion after they recover from the initial effects of the pandemic.

Simulating the propagation of the Covid-19-induced slowdown with a simple model of the global economy, the model considers five major economic blocs: the US, China, the Euro area, “other advanced economies” (OAE) and “other EMEs” (OEM), noting that these economies interact, with each causing spillovers to, and responds to spillbacks from, the other four blocs.

The economic consequences of the virus will depend on a number of factors, including: the direct effects of confinement measures to limit its spread; the required duration of these measures; the extent to which the direct economic effects persist and magnify; and the size of spillovers and spillbacks across regions.

Rather than forecast the precise size of the slowdown, the authors present estimates for a range of plausible assumptions about the direct impact of confinement measures on GDP, and consider scenarios where the initial measures cannot prevent a second wave of infections.

An economic simulation of a global pandemic

To simulate the economic effects of a global pandemic, the authors look at the effects of an initial “shock” that in isolation would lower GDP in every region of the world. This represents the direct effects of confinement policies to limit the spread of the virus. After the initial impulse, they allow the model to trace out the subsequent evolution of economic activity, reflecting the historical relationships between economic variables observed in the data, including the spillovers and spillbacks between regions of the world.

Because the direct effect of confinement measures is uncertain, they report simulations for four alternative scenarios. These four scenarios combine two assumptions on the initial reduction of GDP and its shape. Size-wise, the “less severe” scenario is one where the direct effect of confinement measures lowers GDP by 2.5%. The “more severe” scenario doubles the effect of the confinement measures, so that the initial direct hit to GDP is 5%. This range is broadly consistent with external estimates of the effects of these measures for confinements that last one to two months, even if there is enormous uncertainty about the estimates, the authors say.

Turning to the shape of the scenario, the authors also consider two alternative profiles for the effectiveness of the containment measures. The first one, called the V-type shock scenario, shows a ‘best case scenario’ in which a single wave of confinement measures is sufficient to contain the virus. Consistent with what they observed so far, they stagger the confinement measures across countries and regions as follows:

  • In China, confinement hits GDP exclusively in Q1 2020
  • In the Euro area and OAE, confinement is spread equally in Q1 and Q2 2020
  • In the US and OEM, 25% of the confinement takes place in Q1 2020 and 75% in Q2 2020

“Unfortunately, in the absence of a vaccine or a treatment, we cannot rule out the possibility that the virus will re-emerge after the confinements are unwound,” they say. “To account for this, we also consider a double wave pandemic (a W-type shock scenario). In this scenario, we assume that a second wave of confinement follows two quarters after the first wave. The exogenous effects of the second wave on domestic GDP are, however, only half as large as the first round. This reflects the possibility that countries ‘learn’ and ‘calibrate’ containment measures, so as to damage economic activity less than in their reaction to the first wave of the virus. Graph 1 shows the sequences of shocks for each of the four scenarios.

Graph 2 (left-hand panel) shows the implied evolution of the level of GDP during the first eight quarters for each of the regions in our model. In the case of the more severe V-type scenario, the initial – 5% impulse to output leads to a trough in the level of GDP of between 8.5% and 11% (below its baseline, ie its predicted level in the absence of the shock) in the second quarter of 2020. For the less severe scenario, the decline in GDP is half as large. In both cases, the decline in GDP is eventually around twice as large as the initial impulse, highlighting the powerful effect of multipliers and spillovers in propagating contractions in activity within and across economies.

Even though the authors assume that containment measures are relaxed in the second half of 2020, the output losses for the V-type scenarios are protracted, and in all regions output in Q4 2020 is below its Q1 2020 level. On average, the full-year GDP loss due to the V-type shock would be between 1.5 and 2 times the initial impulse. The declines would be concentrated in the first half of the year, while the second half would be characterised by a recovery in growth rates. But the recovery is modest, and even in Q4 2021 the level of GDP in all regions would still be below what it would have been had the pandemic not occurred.

In the W-type scenarios, the weakness in economic activity persists for even longer. In most regions, GDP growth is negative throughout the 2020 calendar year and a sustained recovery would not begin until 2021, or around six months later than in the V-type scenarios.

The results assume that the severity of the virus is similar throughout the world. That is, the direct initial shock is of similar size everywhere, although distributed differently between Q1 and Q2. An important question is how much of the decline in GDP is due to the domestic effects of the virus on economic activity, and how much reflects spillovers and spillbacks from weaker economic conditions overseas? Relatedly, how much smaller might the economic consequences be for a region that successfully limits the spread of the virus domestically, if other economies still experience a severe outbreak?

The authors address these questions in two steps. “First, we simulate the model assuming (the counterfactual) that the virus directly affects only China and OEM, while advanced economies are spared. For a ‘more severe’ scenario.6 The red bars in Graph 3 show the output losses in Q4 2020 if the virus affects only EMEs. These show the direct effect of the virus on the EMEs, and the spillovers from EMEs to advanced economies. The blue bars show the difference between this exercise and the estimates for the “more severe” scenarios shown in Graph 2.

The results point to large spillovers in both directions. In case of a W-type shock, for instance, the loss of output (relative to the benchmark without the virus shock) in EMEs would be only between half-to-two- thirds as large if only EMEs are were affected by the virus. For advanced economies, spillovers from EMEs account for between 25 and 30% of the GDP shortfall in Q4 2020. These spillovers would be larger for the euro area, for which exports account for a larger share of GDP, than for the United States.

In the second step, we simulate another scenario to illustrate the extent to which an economy can insulate itself from the economic consequences of the virus in the absence of effective containment measures abroad. In this scenario, we consider that the effects of containment measures on euro area GDP are halved, ie, the domestic GDP shock is -2.5% instead of -5%. But the shock still hits other regions in the model by -5%. Graph 4 shows the results of this exercise. Once again, spillovers are substantial. In particular, for the euro area, the decline in GDP after four quarters would still be 6.5% relative to the benchmark without the negative demand shock in the case of a V-type scenario, and 9.9% for a W-type scenario, even though the initial impulse to domestic GDP is only -2.5%. These relatively small improvements in economic outcomes reinforce the importance of international cooperation in designing policies to limit the spread of the virus and combat its economic consequences.

The key message from the simulations presented here is that the economic spillovers and spillbacks of pandemic-type recessions are very large. International coordination of macroeconomic policies is crucial at two levels. First, uncoordinated confinements raise the possibility that the virus will re-emerge sequentially across the globe. This would mean repeated confinements and their associated heavy toll on economic activity. Second, even a country that engineers a domestic policy package that successfully limits its domestic slowdown will not be immune from insufficient or ineffective policies put in place in other parts of the world. No-one can hide from the consequences of a pandemic, and unilateral macroeconomic policies are doomed to fail.

On a more positive note, the authors’ model reflects only the ‘average’ endogenous monetary and fiscal responses in each country as estimated from 1997 to 2019. Policy packages unveiled so far are much larger than those. Additionally, some countries have also taken crucial actions to preserve existing employment relations and avoid widespread firm closures and bankruptcies. This, they say, should help to limit the downturn in aggregate demand and assist firms to restart production rapidly. “That said, as we are in uncharted waters, it remains to be seen how spending responds when economies are still under lockdown,” the authors conclude.

The full report and accompanying graphs may be found at:

Julie Ros
Written by

Julie Ros

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *