Stringency Index Is Not Associated With COVID-19 Death Rate Across U.S. States, but IS Associated With Higher Unemployment

Some people oppose lockdown on principle, arguing that the government should never infringe on fundamental liberties like the freedom to leave our home or open our business, regardless of the impact this may have on disease transmission.

It’s a reasonable position, but I’m more drawn to the consequentialist case against lockdowns. This can be summed up as “benefits small, costs large”. In other words, even if lockdowns do reduce mortality from COVID-19 (under some circumstances), they don’t do so by anywhere near enough to justify their costs.

As I noted recently, several cost-benefit analyses of the U.K.’s lockdowns have been published, and each one concluded that the costs almost certainly outweighed the benefits. (Which may explain why the Government has thus far refrained from publishing any estimates itself.)

A rather elegant demonstration of the consequentialist case against lockdown was provided back in May, in the form of a Twitter thread by the data scientist Youyang Gu.

Comparing the 50 U.S. states, Gu obtained data on the COVID-19 death rate, the change in unemployment rate, and the average Government Stringency Index. The latter is a measure of the number and severity of restrictions imposed during the course of the pandemic (school closures, stay-at-home orders, etc.). Gu’s two main charts are shown below:

He found that the Stringency Index was not associated with the COVID-19 death rate (left-hand chart), but was strongly associated with an increase in unemployment (right-hand chart). In other words, U.S. states with longer and more stringent lockdowns haven’t had fewer COVID-19 deaths, but they have seen higher unemployment.

In the replies to Gu’s thread, some critics argued that restrictions were often imposed in response to large outbreaks, so you can’t assume that causation only goes from restrictions to deaths and unemployment. However, Gu points out that the relative ordering of restriction levels is fairly constant over time, so this is unlikely to be a major issue.

His analysis adds to a large body of evidence indicating that – for the vast majority of Western states – the benefits of lockdown were small, but the costs were very large. Gu’s thread is worth reading in full.