The easing of containment measures is already showing its first traces in the data. Both economic mobility and the number of confirmed COVID-19 cases have increased markedly in recent days. Our estimates suggest that the number of new infections in the coming weeks is likely to be 25% higher than the number of cases expected without the recent relaxations. However, the increase in economic mobility by more than ten percentage points also allows for more economic and social activity, so that the negative consequences of the lockdown are mitigated. The model employed in the study can be used to quantify this short-term trade-off; this provides policymakers with an important basis for decision-making. Because both continued lockdown and relaxation have significant negative consequences, it is even more important to allow further relaxation through better testing and quarantine strategies and by increasing the pace of vaccination without compromising human health.

To estimate the relationship between economic mobility, infected persons and deaths, a Bayesian panel model in the form of a structural vector autoregression is used (Camehl and Rieth 2021). It includes five variables: cumulative COVID-19 deaths, cumulative COVID-19 infections, a containment index, an economic mobility index, and a small business stock price index. All variables except the mobility index are in logarithms and enter the model in levels. There are 14 lags used in the model. The high number of lags allows for reliable forecasts for several months. To have enough observations to estimate the high number of model parameters, data for 44 countries are used. These countries represent about 80% of the world's infected and deaths and three quarters of global production.

The structural containment policy shock is identified by using sign restrictions (Arias et al. 2018)⁷. It is assumed that an unexpected relaxation of containment measures lowers the stringency index in the model. Furthermore, this relaxation is assumed to lead to an increase in economic mobility and an increase in the number of infected people after seven days at the latest. However, the exact dynamics and magnitude of these increases emerge from the estimation and depend on the size of the relaxation, i.e., the size of the shock.