by Joseph Coletti
Senior Fellow, Fiscal Studies, John Locke Foundation
When I first compiled the Covid Misery Index to understand how states were managing the tradeoff between jobs and lives in their policy responses to the pandemic, I was surprised to see less of a pattern than I expected. States with high deaths sometimes had high job losses but not always, while states with relatively low job losses and deaths spread across the country. The picture has become murkier in the ensuing four months as states have gone from great to terrible and vice versa. What has become clearer to me is that the differences from state to state have more to do with individual choices than state mandates.
At the end of February, 23 states had a higher Covid death rate than New Jersey’s 1,510 per million people, or 0.15 percent, at the end of May 2020. Another four are approaching that level. New Jersey (2,618 deaths per million), Rhode Island (2,373), and Massachusetts (2,338) continue to have the highest proportion of deaths. Mississippi (2,245) and Arizona (2,195) have passed Connecticut (2,138) to round out the worst five. North Carolina’s 1,039 lives lost per million people is the 13th lowest in the country.
Three states had more people working at the end of February 2021 than a year earlier: Idaho, Kansas, and South Dakota. Revisions in January flipped the story for Alaska and Tennessee, and both now have job losses like the other 45 states. Nevada, which set the Index score of 100 in May 2020 with the worst job losses in the country at the time, has improved greatly since then, as have most states. Only Connecticut had worse job losses in February than at the start of the pandemic. North Carolina’s 16,690 jobs lost per million people is 23rd highest in the country, though still better than the national average.
To create a composite score that provides comparison across states over time, I set the deaths per million people in New Jersey as of May 31, 2020, and the jobs lost per million people in Nevada between February 2020 and May 2020 each at 100. Those were the highest number of deaths and largest job losses that month. As Covid deaths have risen since then, job losses have generally shrunk. When there are more people at work in a state than there were in February 2020, the state’s Job Loss score will be negative. Because deaths from Covid continue to increase, such job gains can only mitigate the Misery Index score but cannot reverse its growth over time.
Ten states have Misery Index scores above 140. They are New Jersey (200), Connecticut (194), Rhode Island (183), Massachusetts (171), New York (154.4), Louisiana (154.3), Arizona (152), Mississippi (151), Illinois (142.5) and Pennsylvania (142.0). Six of the nine states with a Misery Index score below 70 are in the west—Alaska (33), Oregon (37), Utah (42), Hawaii (51), Washington (59), and Idaho (68)—and the other three are the New England states of Maine (53), Vermont (55), and New Hampshire (64).
The variation over time in overall scores as states experience outbreaks and recover is evident in the 24 states that have ranked in the 10 best Misery Index scores in a given month, the 19 states that have ranked in the 10 worst Misery Index scores, and the four states—Arizona, Hawaii, New Hampshire, and North Dakota—that have been in both the top ten and bottom ten at different times.
North Carolina (82.1) is slightly better than Wyoming (82.3) with the 13th lowest Misery Index score. In April 2020, the Old North State had the 16th highest score.
Here are the Covid Misery Index charts. My analysis follows.
As one might expect, there has been a negative relationship between the number of jobs lost in a state and the number of lives lost, which fits the standard rule people have used to justify or condemn economic constraints: states could sacrifice jobs to save lives or could sacrifice lives to save jobs.
The relationship (with an R value of -0.22) is not nearly as large as the standard rhetoric would suggest, however, and almost no predictive ability (note: the R-squared statistic ranges between 0 and 1, and the closer to 1 is considered more predictive — but here the R-squared value is not quite 0.05). More surprisingly, the relationship is positive in any single month, and the strength of that relationship has weakened over time. In February, the correlation was 0.07 and R-squared was 0.01; in September, the correlation was 0.22 and R-squared was 0.05; and in June, the correlation was 0.54 and R-squared was 0.30.
Looking again at the overall relationship in light of the monthly relationship, two possible stories emerge. One possibility is that states with high death rates imposed stringent lockdowns, sacrificing jobs to prevent even more deaths. Another possibility, which need not contradict the first, is that states that allowed more activity preserved jobs but sacrificed lives later (November–January). This second possibility fits the original explanation of “flatten the curve”—the total number of deaths in a given population would be the same but spread over a longer period of time so as not to overwhelm the medical system’s capacity.
This analysis leaves unresolved the question of what impact Covid mitigation policies such as lockdowns, school closings, and legal restrictions on bars and many other businesses had on the spread of the disease. People who had cut off personal contact with the outside world struggled to leave their homes and apartments, even after being vaccinated, while people with higher risk tolerance have relocated to places more accommodating of their preferences. Few people still manage to wear any mask properly or even wear the right kind of mask, making mask mandates largely ineffective.
More importantly, by measuring only deaths from Covid and job losses, I leave unexamined other causes of death, mental health issues arising from isolation, learning losses in K-12 schools, increases in child abuse and domestic violence, and other costs from personal and policy choices over the past year that all weigh against the duration of extreme measures, however justified they may have been when little was known about the coronavirus.