UNC economist notes governors’ partisan split on COVID-19 lockdowns

Economist Martin Zelder of UNC-Chapel Hill has this to say about his latest column:

Our governor, Roy Cooper, has pronounced that “Pandemics cannot be partisan.” But — according to my latest analysis — the policy response to them seems to be. I found that Democrat governors are more likely (in general) to lock down than Republicans, and more likely specifically to do so as a state’s black population is greater.

Here’s the full column:

WHY DO GOVERNORS LOCK STATES DOWN?

You might wonder why you have spent so much time under coronavirus lockdown lately. In North Carolina, for example, you were under complete lockdown for 52 days, and only liberated (with limitations) on May 22 when restaurants were permitted to reopen. Fifty-two days is a long time to think, and analyze, or at least it was for me. During this time, I looked at what factors made states likely to have longer lockdowns — in particular, comparing states that on May 13 were still locked down (in terms of restaurant dining, one indicator of a lockdown’s extent) compared with those who weren’t. There are 17 states who liberated diners (and others) before May13: Alabama, Arizona, Arkansas, Colorado, Florida, Georgia, Idaho, Indiana, Kansas, Maine, Mississippi, Missouri, Montana, South Carolina, Tennessee, Texas, and West Virginia.

What factors distinguished the other 33 states (call them ‘lockers’) from these 17 others (call them ‘liberators’)? I looked at what percentage of different groups of states were ‘liberators.’ Two obvious differences were the governor’s political party and geography. The most evident was that 66% of Republican governors were liberators, compared to 19% of Democrat governors.

A second significant difference was geopolitical, namely whether the state had been a member of the Confederacy or the Union. While 38% of former Union states are liberators, 73% of former Confederate states are. But former Confederate states governed by Democrats look dramatically different from former Confederate states governed by Republicans. Specifically, all (100%) of the Confederate liberators are Republican-governed, and all (100%) of the Confederate ‘lockers’ are Democrat-governed.

To put it another way, 0% of former Confederate states with Democrat governors are liberators, and 49% of all other states (former Confederate states with Republican governors, plus former Union states with either Democrat or Republican governors) are liberators. So, for example, North Carolina fits this pattern.

I then wondered if Democrats governing former Confederate states faced different conditions that might explain their propensity to lock down their states. In the simplest (univariate) analysis, I did not find any significant differences between Confederate states that are Democrat-governed compared to those that are Republican-governed. That is, they were statistically indistinguishable (in this simplest analysis) in terms of: population density, percentage of black residents, percent elderly, rate of coronavirus infections at an early date (March 20), percent of elderly receiving pneumonia vaccinations, percent in poverty, and income per capita.

So, this suggested a more sophisticated analysis (multivariate regression), where I simultaneously tested whether these factors were significant predictors of whether a state was liberated or locked. (One does this because sometimes patterns not found in the simpler analysis are revealed when one does this sort of simultaneous testing, which is very standard in the social sciences.) This analysis did uncover factors that were common to the ‘lockers’ compared to the ‘liberators.’ Specifically, states that were still locked down on May 13 had higher population density. This makes epidemiological sense in that denser states are presumably more vulnerable to infection-spreading and its consequences.

The analysis also found a somewhat complicated but interesting relationship between lockers and the size of their black populations. States with higher black populations are, on average, much less likely to be lockers. But states with higher black populations and a Democrat governor are only slightly less likely to be lockers. In other words, states with higher black populations are more likely to be lockers if they have a Democrat rather than a Republican governor.

When North Carolina Governor Roy Cooper announced his deliberate phased outline for removing the lockdown, he declared that “Pandemics cannot be partisan.” Someone should ask him why the response to our pandemic does seem to be partisan, with seemingly disparate impact on black people.