Introduction

The SARS-CoV-2 pandemic confronts the world with a rapid spread of infections and deaths associated with COVID-19. Several governments have used or are still using non-pharmaceutical interventions (NPIs) such as social distancing regulation, prohibition of public events, school closures, or restrictions of business activity to slow down and contain the pandemic. Evidence suggests that these measures indeed reduce the number of infections [1,2,3]. At the same time, the pandemic and shutdown measures give rise to substantial economic costs [4, 5].

In the public debate, interests of public health and the economy are often presented as being in conflict [6, 7]. Although this trade-off view may seem intuitive, evidence on medium- and long-run economic consequences of past epidemics suggests that an unregulated spread of a virus with larger disease burden can also have adverse effects on the economy [8,9,10]. New infection waves, e.g., due to accelerated loosening of restrictions, could cause a large rise in absenteeism from work due to illness and could reduce trust of consumers and investors. As consequence, companies would have to shut down or to reduce their business activities again—regardless of government regulations—resulting in considerable further costs. Conversely, stricter regulations may also give rise to indirect disease burden in other areas [11]. The aim is to make the fight against the pandemic sustainable and to reconcile public health and economic objectives [12].

An increasing number of studies on NPI strategies concludes that immediate shutdowns and health policy interventions is the most favorable strategy [10, 13,14,15,16]. A separate question in the public debate, however, is about the optimal shutdown duration, and the timing and speed of the phasing-out of NPIs [17, 18]. A conflict between health protection and economic interests arises if a strategy with lower economic costs leads to significantly higher death numbers. Such a conflict would be particularly challenging if the reduction of economic costs requires a rapid opening process. Yet, previous studies using integrated macroeconomic and epidemiological models conclude that limiting the spread of the virus is the economically optimal reopening policy [19,20,21,\(R_\mathrm{t}=0.53\)), costs decrease in a strategy of a slight loosening of restrictions. The long-term economic costs are minimal if the reproduction number is around 0.75.

Common interest of economy and health

We cannot identify a conflict between the economy and health protection in relation to a strong relaxation—the costs would be higher in both dimensions. Accelerated opening leads to substantially more COVID-19 deaths and increased economic costs. Our findings clearly challenge statements which suggest exit strategies with \(R_\mathrm{t}\) values close to one to be economically preferable [21]. While strong opening policies would allow for more economic activity in the short term, our simulations suggest that the long duration of remaining restrictions would increase relative economic costs compared to alternative gradual opening strategies.

Our results suggest that a balanced strategy is in the common interest of health protection and the economy. The scenario calculations show that a slight, gradual lifting of shutdown restrictions which keeps reproduction numbers at an intermediate level and which allows to further reduce infection numbers in a significant manner is suitable to reduce the economic losses without jeopardizing medical objectives.

Fig. 4
figure 4

A Overall economic activity over time for three baseline policy scenarios (denoted by their respective reproduction numbers, 0.1, 0.5, and 1.0). Pre-crisis economic activity is normalized to 100. B Relative costs for each policy scenario, in percentage difference to the reference scenario (\(R_\mathrm{t} = 0.53\)). Economic costs are given as the aggregated loss of activity occurring as a result of the shutdown and recovery phase. The bold line indicates the baseline scenarios; the shaded grey lines indicate the results of the robustness tests. The numeric values can be found in the SI Appendix, Tab. S3

Full size image

Robustness of results suggests general applicability

Clearly, generalization of our results beyond Germany and across time is limited to comparable regions, situations and given NPIs. The relationship between economic activity and the reproduction number might not be the same across world regions. Moreover, the shutdown duration and final death toll are influenced by the number of new infections at the point of entering or changing shutdown measures. However, our results are robust to several sensitivity tests in assumptions regarding the relationship between the shutdown severity and economic activity, affected industries of exogenous shutdown restrictions, the duration of economic recovery, and the number of daily new cases that needs to be reached to control the epidemic. We also tested the sensitivity of our assumption on the time of large-scale availability of a vaccine (see SI Appendix, Tab. S5).

The assumption of a linear relationship between shutdown levels and economic activities is clearly a simplification in our simulation model, although the slope of our linear relationship is based on observed data. Our robustness tests include simulations with (non-linear) isocost-curves that indicate how severely the linear assumption needs to be violated for our results to no longer be valid. The results show that it would require implausible assumptions of extreme non-linearities to invalidate our findings (see SI Appendix, Fig. S6). All robustness tests can be found in the supplement (SI Appendix, Tab. S5). Our inferences do not change. Minima of relative economic costs are between \(R_\mathrm{t}\) values of 0.7 and 0.8 in all sensitivity tests (see light-grey lines in Fig. 4B).

Discussion

We consider the qualitative statement of our results to be robust and of general nature. It is in common interest of health and the economy to implement opening policies with prudent steps and to closely monitor the respective reaction of the infection figures. Our conclusion is in line with retrospective studies of the influenza epidemic in 1918 in the USA [13, 30], and current economic studies supporting a strategy to manage the COVID-19 pandemic [19, 20]. We show that it is also in the interest of the economy to balance non-pharmaceutical interventions in a manner that further reduces the incidence of infections. By contrast, NPI policies that are too loose could cause higher economic costs in the long term. We provide an additional guideline for policy-makers whether extending or easing restrictions minimizes long-term economic costs once the effective reproduction number is already below one. Using counteracting measures—such as face masks, behavioral rules, improved trace and isolation techniques, new technologies and increased testing—may limit the spread of the virus or even may help to contain the pandemic [16, 31,32,33,34,35,36] and thus creates leeway for larger opening and economic recovery. The level of economic restrictions thus depends to a large extent on technical improvements and behavioral adjustments of the population until a vaccine or effective medical treatment is available at large scale for all in need.