A thread on new work Syverson and I just released looking in depth at the impact shutdown orders (and repeals). Short version: they don't do much. Of the 60% drop in consumer activity, only 7 came from shutdown orders. Fear of the virus is the main thing. https://bfi.uchicago.edu/working-paper/2020-80/
We have consumer visits to 2.3m businesses in 110 industries through the crisis. We tracked down the county level shutdown orders and can compare across borders in the same metro area where the policy differs. Lockdown orders don't matter much. Visits plunge on both sides.
If you just look at state level policy but don't control for overall fear with time effects, it seems like shutdown orders matter a lot--50%+ reductions. Once you control for what's going on in the area during that week, though, the coef is 1/10 the size. Policy isn't the driver.
Evidence of fear as the driver: 1) more covid deaths in your county drive down economic activity signif. even including metro-week dummies, 2) people heavily shift visits away fr/ larger/busier stores to smaller/less busy ones in the same industry (& esp. if lots of local deaths)
We have data up to late May and include some states ending their shutdown orders. The increase in economic activity is just as modest coming out as it was going in. Policy itself isn't the driver.
But there is 1 way policy matters: diversion fr/one kind of business to another. We have essential/non-essential defn in each place. Non-esstl biz collapses. Esstl biz soars. Restaurant/bar orders cause massive hit there but an equally big increase to grocery and food stores.
In the end, the results from March-May suggest that the fact that cases are back on the rise is very ominous not just for public health but for the economy. If people get scared again, a lot of activity may start to tank.