Now that there may be a light at the end of the tunnel, with a promising vaccine for the coronavirus, we can step back and see how the pandemic hit various parts of the country. Much of the data has been presented on a state-by-state basis, but this hides the wide variation within states. We can also look at the data on a city-by-city basis.
The Opportunity Insights tracker at Harvard publishes case and death data for Covid-19 for major cities. Case data is not terribly helpful, since it depends on testing, which has not been uniform, over either time or space. There were likely far more cases in the early days of the pandemic when testing was rare and people with mild symptoms were told just to stay home and ride it out. Now, testing is widely available, so marginal cases are being reported.
Death data is more useful, since medical authorities have been pretty good about flagging Covid when assigning causes of death. But since treatments have improved greatly since the beginning of the pandemic, and since the disease is spreading more in young people, for whom mortality is rare, deaths don’t tell us as much about trends in spread of the pandemic.
Deaths can allow us to compare various parts of the country, however. Figure 1 shows the cumulative death rate for the largest cities in the country, expressed as deaths per 100,000 population. The data is tracked by county, and Opportunity Insights defines each of the cities by the county that contains the city (e.g. Seattle is defined as King County) or the total of the collection of counties that have any part of the central city in them (e.g. Atlanta extends across more than one county).
The variation in death rates across cities is quite large. The death rates in Detroit and Boston are about eight times the rates in Portland and San Francisco. New York has been an outlier all along, but when considering rates, it is not that far away from the next hardest his city, Detroit.
If the curve traced by the ends of the bars in Figure 1 seems suspiciously smooth, it is because death rates are following a power curve. Figure 2 places this data on a plot with a logarithmic scale on the death rates.
The correlation of 96 along that trendline is quite remarkable. The impact of coronavirus among the larger cities in the country has not been linear at all. It is especially puzzling that the two cities in either end of the curve, San Francisco and New York City, are both dense, cosmopolitan, expensive places that serve as cultural capitals. One got hammered and the other got away relatively easily.
In general, the West Coast fared better than the rest of the country. The hardest hit city on the West Coast, Los Angeles, had a death rate one fourth that of New York City, the only larger city in the country. Figure 3 shows the cumulative death rate for six West Coast cities between March 1 and November 7. The data is presented as a seven-day trailing average.
Seattle comes out quickly, with a high initial death rate. Remember, this data covers all of King County, including Kirkland, where the first cases were discovered. But by June Seattle’s rate had leveled off, and deaths were increasing at rates similar to those in Portland, San Diego and San Francisco. Los Angeles has not seen a leveling off, and Sacramento saw a big jump in death rates in late summer.
In trying to explain the variation in death rates across major cities we can imagine a number of factors at work. Climate may play a part. The West Coast uses far less air conditioning and homes and businesses tend to cool themselves down with open windows and moving air. Summers on the West Coast are more conducive to outdoor activities, which have been found to be far less risky. But this does not explain the relatively low death rates in Dallas and Houston, both heavily air conditioned and not as conducive to outdoor summer activities.
Public policy was initially more aggressive on the West Coast, with early lockdowns in Washington, Oregon and California. And with the strange way that mask wearing became politicized, with those on the left tending to wearing masks more often and some of those on the right eschewing them, the politics of the Left Coast may have played a part. But, again, this does not help explain the experiences in the redder cities of Texas.
The coronavirus pandemic of 2020-2021 will provide fodder for social science researchers for decades to come: a massive, multi-dimensional natural experiment. Vast amounts of data have been collected and the variation in the timing and intensity of policy and responses around the country will provide the useful discontinuities that researchers love. Those studying human behavior in stressful settings will have limitless research opportunities and they may eventually figure out why San Francisco and New York City had such different outcomes.