This paper analyzes the implications of different designs of policies, which aim to contain the diffusion of the SARS-CoV-2 virus, with respect to induced economic loss and virus mortality. An agent-based simulation model is implemented and calibrated with German data, which combines the representation of a simple multi-sectoral closed economy with the explicit incorporation of virus transmission channels at the workplace, during shopping activities and other social contacts. It is demonstrated that under a policy resembling German containment measures the model closely reproduces the dynamics of pandemic and economic variables in the aftermath of the COVID-19 outbreak in Germany. Exploring alternative policy designs shows that any efficient policy should impose a low threshold of newly infected for moving from the lock-down to the opening-up stage and in the opening-up stage all restrictions on economic activity should be lifted. With respect to the reduction of consumption activities during the lock-down a trade-off between the induced GDP loss and the resulting mortality emerges. Regardless of the chosen design of the containment measures, the introduction of complementary economic support measures substantially reduces the induced GDP loss and leads to a reduction of the public debt accumulated during the considered time interval. The efficient design of containment policies changes substantially if lifting economic restrictions during the opening-up stage also results in reduced effectiveness of the individual prevention measures by agents.