Andrew Atkeson, Karen Kopecky, and Tao Zha write,
Relatively slow growth or even shrinkage of daily deaths from the disease was observed in every location that we study 20-30 days after that location first experienced 25 cumulative deaths, and the dispersion in growth rates of daily deaths across locations fell even more rapidly
They argue that the uniformity of this pattern across different locations implies that differences in lockdown policies are not important. This could be because the private responses to the pandemic are more uniform, or the network structure of social interaction leads to a rapid spread followed by a slower spared, or there is some biological factor at work. On the latter point, they cite another paper which reports that
of eight major influenza pandemics that have occurred since the early 1700’s (including the Spanish Flu of 1918-19), seven had an early peak that disappeared over the course of a few months without significant human intervention. Unfortunately, each of those seven had a second substantial peak approximately six months after the first.
I recommend the whole paper.
Meanwhile, Greg Ip writes,
Five months later, the evidence suggests lockdowns were an overly blunt and economically costly tool. They are politically difficult to keep in place for long enough to stamp out the virus. The evidence also points to alternative strategies that could slow the spread of the epidemic at much less cost. As cases flare up throughout the U.S., some experts are urging policy makers to pursue these more targeted restrictions and interventions rather than another crippling round of lockdowns.
My sense is that the best science says that we don’t know what works. Lockdowns are certainly the most theatrical policy, though. They allow politicians to show that they care. They are the equivalent of Hansonian medicine.
Note how well what I thought were outlandish predictions three months ago are holding up.