In an otherwise excellent blog post that I recommend reading for its own sake, Cato Institute health economist Michael Cannon writes:

Similar to health insurers, auto insurers are saving money because people are driving less. The Wall Street Journal reports, “Allstate Chief Executive Tom Wilson said mileage was down ‘an unprecedented’ 35 percent to 50 percent across the U.S. since mid‐​March, including in states without shelter‐​in‐​place restrictions.” (That latter part provides a helpful reminder that government affects pandemic response only at the margin.) Less driving means fewer accidents, so claims have likewise fallen as much as 40 percent.

His parenthetical statement is almost certainly wrong. The fact that driving is down by about the same percentage in states without SIP restrictions as in states with such restrictions means that government affects driving at the margin.

Government affects pandemic response in a much greater way. States without such restrictions have not forced people to quit working. And as a result, many people in various businesses in, say, South Dakota, have kept working while people in states with restrictions have been forcibly prevented from working. Both the means and the effects are a huge difference.

By the way, Michael’s mistake is similar to the one that Justin Wolfers made in our debate when he argued that if voluntary social distancing achieved 90 percent of what compulsory lockdowns achieved, they would also entail 90 percent of the costs.