The federal response to the crises brought on by the COVID-19 pandemic stands as a dramatic about-face to the century-long trend towards federal government empowerment. The federal government’s initial reaction to the looming pandemic crisis was disjointed and at times contradictory, a response that stood in stark contrast to the strong and highly coordinated federal responses crises had elicited in decades past.
State governments quickly became pivotal in managing the health and economic crises, with President Trump telling governors “You’re going to call your own shots.” How did states respond to the public health crisis, and what did they do with their newfound authority? We summarize data compiled by the COVID-19 U.S. State Policy Database, that chronicles measures taken by states to respond to the twin health and economic crises spurred by the pandemic. We then compare COVID-19 incident rate (infections per 100,000 people) across states as of July 31st and find great variation between states’ responses. We suggest some of this variance can be explained with reference to population density: with early COVID-19 outbreaks constrained to the costal urban centers, sparsely populated interior states did not face the same pressure to impose restrictive measures. Yet a clear political pattern to states’ policies is also evident: while states that adopted restrictions were governed by leaders of both major parties, Republican affiliated governors tended to impose far fewer COVID-19 related restrictions.
As described above, all states, territories and several major cities declared formal states of emergency, allowing executives to mobilize resources to combat the coronavirus. A related policy, taken up by thirty-nine states and the District of Columbia (DC) were ‘stay-at-home’ or ‘shelter-in-place’ orders, which discouraged residents from leaving their homes for any reasons other than doctors’ visits, grocery store or pharmaceutical shopping. This would coincide with the mandated closure of restaurants, bars, and non-essential retail operations in 49 states and the District of Colombia.
The shuttering of the world’s largest economy brought with it an unprecedented economic collapse that eclipsed previous crises by orders of magnitudes. With the help of the federal stimulus package, the states became the conduit for relief provisions and support for millions of newly unemployed Americans. Only five states declined to provide some sort of eviction relief—either by suspending judicial proceedings or formally loosening enforcement mechanisms. Thirty-four states prohibited shutting off utilities or gas to homes for COVID-19 related claims. Finally, there was an across-the-board expansion of social safety net investment by the states: every state (and DC) increased access to food security programs and healthcare for lower-income families (Medicare), and many also loosened restrictions for extended access to unemployment benefits.
One viral mitigation policy that spurred intense political debate was mandatory mask-wearing requirements, which we show in Figure 1. Although the CDC advised that masks are effective in mitigating the viral spread in early April, mask mandates were quickly politicized, with adherents and opposition coinciding neatly with partisan affiliation. Opponents to mask-wearing ordinances claim that these measures are an unconstitutional restriction on freedom, and President Trump expressed ambivalence: “you can do it. You don’t have to do it. I am choosing not to do it. It may be good. It is only a recommendation.” Again, this complacency further undermined state and local governmental efforts to normalize mask-wearing, and publicly disincentivized compliance with mask-wearing rules.
Figure 1. Mask Mandates by State
Italicized state names Republican governors. **Denotes states above the median of population density. Puerto Rico and Washington DC are unranked by population density.
As shown in Figure 1, an absolute majority of states and territories adopted mask mandates. The role of population density is an evident factor at play: sparsely populated states such as Alaska and Wyoming declined to impose restrictive mask-wearing rules, whereas densely populated states (Hawaii) and states of major metropolitan areas (New York) imposed some of the strictest measures. Yet a clear pattern of partisanship also emerges, reflecting the politicized nature of the mask-wearing recommendations. Of the twenty-six states with Republican governors, only 38% (10) would adopt statewide mandates for mask wearing; amongst non-Republican states and territories, 96% of them required masks. Indeed, of the 17 states that declined to adopt a mask-wearing requirement, only Nevada was led by a Democratic governor.
We now examine one possible measure of effectiveness of COVID-19 restrictions to consider the closure of non-essential businesses across states (see Figure 2). The x-axis represents non-essential business closures in days, while the y-axis represents the number of states reported COVID-19 cases per 100,000 people on July 1. Non-essential business closures ranged from twenty to eighty days. An exception is South Dakota, which never officially closed. States with the lowest case counts are Hawaii, Montana, and Alaska, whose businesses were closed for 43, 30, and 27 days, respectively. States with the highest case rates include Illinois, New York, New Jersey, Pennsylvania, and Washington, D.C. Except for California, Florida, and Massachusetts, these states are home to some of the largest – and most densely populated – urban centers.
Figure 2. COVID-19 Cases Per 100K on July 1, by Days of Non-Essential Business Closure
There are several interesting trends of note from Figure 2. First, the bivariate relationship between non-essential business closure and case count shows a modest, albeit positive correlation (r = 0.25). This implies that those states that were shut down the longest also saw the highest rates of infection come July 1. This is unsurprising given the states with the longest duration of shutdowns are also those with major urban centers. Relatedly, the correlation between population density and case count is a positive, but still modest, r=0.35. Finally, when we remove the ten greatest outlier states from our data, the correlation is r=-0.09, which is statistically indifferentiable from zero. This correlation would imply that for the 40 ‘typical’ states, longer shutdowns may have yielded a reduction in the infection rate. This is only conjecture, and if it were true the effect would be minor.
Critically however, with only bivariate correlations, we cannot draw causal conclusions, and the weak correlations we do find demand even more inferential restraint. We do not control for confounding variables, including those that would be critical in explaining incident rates, such as population density (which impacts transmissibility) or breadth and accuracy of testing (which impacts detection). These caveats aside, the relatively weak nature of our correlations point to two potential conclusions. First, whereas the states that closed the longest are also home to populations where transmission was more readily spread (due to high population density), then our weak correlations might indicate the successful mitigation of COVID-19: in these densely populated states and areas, case counts could have been far worse were it not for the state mandates to limit public interaction. Second, our weak correlations might suggest that the states are too diverse to adopt a standardized approach to non-essential business closures and related mitigation strategies. Future research will no doubt interrogate these possibilities in more depth.
Source for featured image: https://www.pexels.com/photo/hands-with-latex-gloves-holding-a-globe-with-a-face-mask-4167544/