The fifty states with their different laws and governments are natural laboratories of democracy. This year, those laboratories of democracy have held an epidemiological experiment. Preliminary results from that experiment are in: Elections are a matter of life and death.
For an American in the year 2020, the risk of death from COVID-19 has varied quite significantly from state by state. Part of the pattern is geographic; population density accounts for a majority of the variation in risk thus far. Part of the pattern can be seen in when governors are elected.
For good or ill, it is governors that have led the response to COVID-19. Not all governors have led equally capable; and not all state governments have been as competent and able as others. Competence and capability varies significantly with different election schedules; states where governors are elected in mid-term or off-year elections tend to have more corrupt state governments and lower gubernatorial approval ratings.
Death rates have been substantially higher in states whose governors are not up for re-election or replacement in 2020. This difference has been consistently visible in the data for most of the course of the pandemic. Washington, a state with 2020 gubernatorial elections, had an early lead in deaths, but this phase passed quickly once the disease spread nationally.
This difference is not likely to be an artifact of selective underreporting. The differences in transparency and accuracy of government reporting follow a similar pattern. States without 2020 gubernatorial elections, such as Florida, seem more likely to under-report their COVID-19 deaths. The actual gap in deaths is probably larger than shown in official data, rather than smaller.
States with 2020 gubernatorial elections, such as West Virginia, seem to have also done a better job at ramping up testing ahead of the spread of the virus as well as bringing outbreaks under control.
States with off-year elections — such as safely “blue” New Jersey and safely “red” Mississippi — have done particularly badly. At least 1 in 1,000 people in states with off-year elections died of COVID-19. On the other end of the scale, Vermont and New Hampshire elect governors every two years, and have experienced a quarter as many deaths.
Higher turnout means more accountability
Between off-year, midterm, and presidential year elections, the most notable difference is voter turnout. Presidential year elections tend to have the highest turnout; off-year elections tend to have the lowest turnout. The notorious political corruption problems in Louisiana and New Jersey have flourished in an environment where turnout in state elections is abysmally low. Special interests can more easily influence elections with low turnout.
The state with off-year elections that has done the best in avoiding COVID-19 deaths has been Kentucky. Perhaps not coincidentally, Kentucky’s last gubernatorial election cycle featured exceptionally high turnout. In general, the states with the worst COVID-19 death toll have generally been states with low turnout in gubernatorial elections.
While the relationship between voter turnout in the last gubernatorial election and current cumulative COVID-19 deaths is far from perfect, it is striking, negative, and in a range usually considered statistically significant.
This holds true whether we consider the correlation by itself or control for urbanization, population density, and party affiliation of the current governor. Additionally, this relationship has been clearly visible in every snapshot of the data since the beginning of April; the strength of the relationship based on a single snapshot is an understatement of the strength of the relationship observed over time.
Update from 12/23/2020: When this article was first published in October, p values were 0.007 and 0.018, respectively; as of 12/23/2020, the p values are 0.057 and 0.023. While the relationship has weakened slightly in the most recent snapshot, partly due to the enormous spike in cases in North Dakota, this is still quite significant.
Competence matters more than party affiliation
As noted earlier the difference in risk based on election schedule has been both larger and more enduring than the difference in risk based on partisanship. While there was a large initial gap between “red” and “blue” states at first, the gap has closed.
The reason why “blue” states were ahead in deaths early on is simple. The earliest major outbreaks were linked to international travel and major urban centers, beginning in three key Democrat-led states: Washington, California, and New York. These states had a head start, in other words. Now that the disease has spread widely across the country, both the death toll and the death rate are quite similar for states with Democratic or Republican governors.
This is quite remarkable, since the COVID-19 response has been highly politicized. Symbolic acts and rhetoric, however, don’t matter as much as competence and responsiveness. Higher-turnout elections on a more accessible schedule means a more competent and responsive government.
The results are in
In the 20th century, the United States started a long-running natural experiment. Almost all states lengthened their governors’ terms in office to four years. At the same time, many (though not all) shielded governors from high-turnout presidential elections by moving to a midterm gubernatorial election schedule. The COVID-19 pandemic has provided another test of that natural experiment; midterm gubernatorial elections failed.
Elections are a matter of life and death. It is critically important to elect leaders who are competent and accountable to the people at large. For a governor who serves four years in office, this is best done during presidential election years. Practices that lower voter turnout and therefore limit accountability, such as holding gubernatorial elections in off-years, have real negative consequences.
A governor serving a four-year term should be elected during presidential election years, when voters are most engaged. Governors who don’t face election during high-turnout years tend to have low approval ratings and helm state governments that are more corrupt and less competent.