Inequities of suffering, relief, and greed have accompanied the extraordinary 2020 COVID-19 pandemic in the United States. While several analysts have reported on these inequalities separately, the full horror of the situation can’t be grasped unless we examine them side-by-side. By focusing on New York, the U.S. epicenter, we can better understand the root inequities that have contributed to the disease’s dramatic growth and lethality.
Our basic finding is that there has not been just one pandemic. This is a tale of “three cities:” rich and poor, whites and people of color, and old and the young. The distribution of suffering and relief, as well as of profiteering and greed, have been highly skewed by class, race, and age.
vieBy far the most devastating effects have been felt among those who were the least well off and empowered to begin with. The poor, people of color, homeless, aged, or undocumented have experienced much higher rates of infection, fatality, and economic hardship than their middle-class and upper-class neighbors. This is especially true in densely populated urban areas, such as New York City, Detroit, Chelsea/Boston, and Chicago, where the incidence of disease and death has been at least ten times what it has been elsewhere. (See Chart 1.)
America is a country that prides itself on being a meritocracy, yet it has allowed the chances of survival and the distribution of hardship from this insidious disease to be determined by largely inherited characteristics, especially class background as well as the color of one’s skin. As an emergency room nurse in New Jersey told us:
I don’t even know where to start. The unimaginable is happening. Yes, they get on vent, but they pretty much have zero help. If they [go into] cardiac arrest, they get soft resuscitation, a few minutes of soft CPR, and one Epi, and that’s it. Bag them up and put them in a semi-truck. There’s no supplies. Crash carts don’t get restocked. We are lucky [to] have one push dose epi. Only point assessments are every 12 hours. I have nine drips today. Maxed out on Epi, Norepi, Vasopressin. It’s unreal! Not just sepsis. It’s much worse. . . . They are really scared to be intubated, lying in their own stools, no family present, maybe just a Skype session . . . no advocates, dying alone.
On top of this, as we’ll explore in the second part of this essay, these inequities have only been aggravated by our relief efforts. Even though the government has shut down the entire economy and enacted unprecedented fiscal programs in response to the pandemic, it has failed to get relief to many of those who need it most. Meanwhile, those at the other end of the economic ladder have received a disproportionate share of the trillions of dollars in “stimulus” relief.
Overall, the U.S. response to this pandemic has been deeply shaped by the reality that, especially in this crucial election year, a huge share of the burden is being born by poor people of color in blue states. If this comes as no surprise, that in itself is a pretty disturbing indicator of where things stand in America 2020.
One basic principle of investigative economics is to rely on multiple sources and methods. Our support for these stark conclusions is based on several different kinds of evidence, including data from multiple sources, a variety of statistical methods, and probative interviews with medical staff who have been on the front lines of the “COVID-19 war zone.” Initially we focused on New York, our home base, which has already recorded more than 313,000 reported cases and 24,198 deaths attributed to COVID-19 in the eight weeks since the first official New York fatality was reported on March 1. This is nearly 28 percent of all reported cases in the United States, 38 percent of all reported U.S. fatalities, and a tenth of all reported COVID-19 fatalities in the world. The epicenter has been New York City, with over three-fourths of all New York State fatalities.
In terms of relative impacts by racial and ethnic group, the latest official data from the New York State’s Department of Health speak very loudly, even before we allow for under-counting. As indicated below, as of late April 2020, blacks and Latinos in New York City accounted for 62 percent of all COVID-19 fatalities in New York City, compared with their 51 percent population share, for a “relative burden ratio” of 122 percent. (See Chart 2.)
Outside New York City, blacks, in particular, have experienced an even higher relative burden ratio. All told, statewide, at just 36 percent of New York’s population, as of late April blacks and Latinos have accounted for at least 55.1 percent of New York State’s reported COVID-19 fatalities—12,636 out of 22,912—for a relative burden ratio of 154 percent. These disparities are consistent with other disparities that are also associated with race and ethnicity—for example, median incomes, poverty, particular diseases like diabetes, and unemployment.
The disproportionate burdens suggested by this aggregate New York City and New York State data are also apparent when we look at the distribution of COVID-19 cases across New York City’s 183 zip codes. (See Chart 3.)
Unfortunately, fatality data is not available by ZIP code; nor has New York City released demographic data on confirmed COVID-19 cases by ZIP code. But we do have total confirmed cases by ZIP code that can be compared with their demographics.
Identifying “outliers” in this data reveals some striking contrasts among ZIP codes at the opposite ends of the social order. (See Chart 3.)
As of late April, Queens had the second highest rate of reported COVID-19 cases per capita of New York City’s five boroughs, and the highest number of total reported fatalities. East Elmhurst has become “the epicenter of the epicenter,” with the city’s highest incidence. Elmhurst has long been one of Queens’ most diverse, vibrant neighborhoods; the New York City Health Department has had to distribute COVID-19 fact sheets in 15 languages here. And just a mile from Elmhurst is the aptly named Corona Queens, Louis Armstrong’s former home. Silent but for an occasional ambulance siren, there is no jazz here now. As of late April the Bronx had the highest COVID-19 incidence per capita of New York’s five boroughs. In Coop City Bronx, 14 miles from Elmhurst, 86 percent of residents are black or Latino, and a fifth are seniors. Right next door in Clason Point, Bronx is a traditional working-class neighborhood where 94.3 percent of residents are black or Latino, the poverty rate is 22.7 percent, the median household income for a family of four is $41,638, and the share of households with incomes great than $200,000 is only 2.7 percent. As of late April, Clason Point’s reported COVID-19 infection rate as 23.25 per thousand.
At the other end of the spectrum, things are a bit calmer in Manhattan, New York City’s most affluent borough, at least for those residents who haven’t fled to the Hamptons. Manhattan reportedly has by far the lowest COVID-19 incidence per capita and the lowest number of reported fatalities of New York City’s five boroughs, at 12.2 cases per 1,000 and 1,600 fatalities.
In the City’s posh 10280 ZIP code in Lower Manhattan, 12 miles from Elmhurst, 41 percent of households have incomes greater than $200,000, and just 10.6 percent of its residents are black or Latino. There is almost no poverty here. As of late April, the COVID-19 infection rate was 3.52 per 1,000, an order of magnitude lower than the rate in East Elmhurst, but just a twenty minute subway ride away. About 8.4 miles from Elmhurst, Wall Street ZIP code 10006 is home to the New York Stock Exchange. The median income is $176,250 per year. As of late April 2020, the incidence of COVID-19 was 4.3 per 1,000. In fashionable Tribeca, 9.2 miles from Elmhurst, the median household income is $246,813 and nearly two thirds of residents have incomes greater than $200,000. As of late April, Tribeca’s official COVID-19 infection rate was six per 1,000. The city’s Upper West Side, a cultural touchstone, only six miles from Elmhurst, is an old upper-middle class neighborhood. About 13 percent of its residents are seniors, its median income is $131,668, and 14 percent of its residents are black or Latino. As of late April, its infection rate was just 8.3 per 1,000, a quarter of Elmhurst’s rate. We know many of this area’s residents have left the city for second homes, or simply to be somewhere safer. (See Chart 5, above, and Chart 6, below.)
Overall, therefore, this “outlier” analysis provides some striking qualitative support for our hypotheses about the relationships between social inequality and the incidence of this disease. When we expand the number of ZIP codes to include all 177 in New York City, these relationships also show up clearly. The influence of class—as captured by the share of households with incomes greater than $200,000 per year—is especially striking (see Chart 6). A disproportionate number of poorer people of color have caught this disease. But, at least in New York City, the epicenter of this disease in the United States, if you really want to be safe, live in a wealthy neighborhood.
We have also developed a statistical regression model to explain COVID-19 incidence across New York City’s ZIP codes as a function of several variables that are a proxy for the influences of class, race and ethnicity, age, and population density. This model fits the data pretty well, with expected signs and statistical significance for the key variables, especially class. Our basic finding is that class, especially, as well as race, help to account for the nearly ten-to-one disparity in COVID-19 incidence that we see across New York City ZIP codes. Of course all such statistical analysis is subject to caveats. The key question is their net effect. These caveats imply that the relationships we have identified need to be tested against more complete data.
Nevertheless, we believe that these data send out strong signals that the qualitative relationships we have postulated are real. Indeed, if anything, our findings may actually turn out to understate the influence of class and race on the incidence and fatality of COVID-19 in New York City neighborhoods.
Consistent with this, reported fatality data for New York City as a whole does show that African-Americans and Latinos have been dying from COVID-19 at rates of at least 92.3 per 100,000 and 74.3 per 100,000, respectively—at least twice the average rates of 45.2 per 100,000 for whites and 34.5 per 100,000 for Asians, even before adjustments for undercounting.
Obviously the conditions under which many black and Latino New Yorkers live, work, and commute are much more conducive to contracting and propagating the disease. They have far fewer opportunities to avoid the disease to begin with, let alone to escape to the Hamptons. If they do get sick, they are also probably more likely to engage in self-help treatment at home, despite its perils.
But it is also clear that the risk of contracting COVID-19 is not determined by race alone. Indeed, our ZIP code analysis suggests that class may be at least as important as race in predicting COVID-19 incidence, as opposed to fatality. In other words, given a choice between “being black in a rich neighborhood” and “being white in a poor neighborhood” in New York City, our results indicate that class background, although of course related to race, is worth distinguishing as a causal factor. It is very important to examine both kinds of disparity, because they may have different roots, consequences, and solutions.
Since very different factors may be at work on infection rates and fatality rates, it is essential to gain access to the precise demographics of COVID-19 cases per ZIP code. So far both New York State and New York City have declined to reveal publicly such demographic data for COVID-19 cases (as opposed to similar data for fatalities)–joining just seven other states in refusing to do so. Their argument is that testing is still too incomplete, and that fatality data is intrinsically more accurate. We’ll see below that the latter claim is easily overstated, given the omission of numerous of deaths that are appropriately attributable to COVID-19. In any case, in the interests of transparency, we believe that it is not too early to see what the case data say, even while we labor night and day to improve it.
When these states and localities do get around to releasing case data demographics by locality, we expect that it will show that African-Americans and Latinos have indeed suffered a disproportionate share of infections as well as fatalities, but also that class may compensate for racial disparities. Parsing out the respective roles of race and class remains an important part of our research.
One might have expected the Centers for Disease Control and Prevention (CDC), as the leading national authority on health statistics and pandemics, to have contributed to our understanding of this issue. But it turns out that the CDC has agendas. For example, rather than rely on New York State/New York City figures for the distribution of COVID-19 fatalities by subgroup, the CDC’s National Center for Health Statistics (NCHS) produces its own independent figures, based on “a current flow of mortality data in the National Vital Statistics System.” The CDC admits that these data are subject to a delay of “one to eight weeks or more” between the time when deaths occurred and death certificates were completed, submitted to the NCHS, and processed. But it has continued to publish undercounts in the interim.
It is clear that these lags—or other factors—are yielding distortions and understatements. For example, the CDC estimates that by the end of April 2020, the share of COVID-19 fatalities among blacks and Latinos in New York City was 24.8 percent and 26.4 percent respectively. This was nearly 20 percent less than the latest New York State/New York City estimates of 28 percent and 34 percent, respectively. For the region outside New York City, the CDC estimates that the black share of fatalities was only 14.8 percent, compared with New York State/New York City’s 18 percent estimate. On the other hand, the CDC attributes 8.9 percent of New York City fatalities to “Asians” (the latest New York State/New York City estimate is 7 percent), 29.3 percent to “whites” (New York State/New York City: 27 percent) and 10.4 percent to “all other” (New York State/New York City: 4 percent). For the portion of the state outside New York City, it attributes 66.5 percent of fatalities to “whites” (New York State/New York City: 60 percent). All told, as of late April, the CDC’s aggregate COVID-19 fatality data for New York State and New York City was missing 2,077 fatalities among blacks and Latinos—16 percent of the official body count. And this is before we even get to the much larger problems of fatality undercounting.
These discrepancies may sound small, but they all tend to give the impression that the CDC has basically been bending over backwards to “correct the perception” that this pandemic is having a disproportionate effect on people of color. For example, in addition to using the lower fatality share estimates for people of color just described, the CDC has also decided to promote a measure that it calls the “weighted population distribution.” In effect, this artificially inflates the share of population for black and Latinos and slashes the disproportions. The CDC claims it needs to make this adjustment in order to make the statistics more reflective of the actual concentration of the disease in particular geographies. From our standpoint that begs the question of why so many people are living in these conditions in these areas, and why the CDC would go out of its way to adjust the statistic in such a distorted way. At this point we do not have a clear answer to that question.
To pick one glaring example: In the case of Wisconsin, the unadjusted black share of the state’s population is 6.4 percent and the Latino share was 6.9 percent. As of late April the black share of COVID-19 fatalities in Wisconsin was 31.9 percent and the Latino share was “unavailable.” The CDC’s “weighted population shares” for these groups adjusted their population shares sharply upward by excluding any counties in the state that had no reported COVID-19 cases. This arithmetic wangle nearly tripled the share of population in Wisconsin attributed to African-Americans, from 6.4 percent to 21.8 percent. (It also nearly doubled the Latino share from 6.9 percent to 13.6 percent.) As a result, after we substitute these higher population shares for the true ones, COVID-19 no longer appears to be much of a “disproportionate” a killer of African-Americans in Wisconsin.
Similarly, according to the CDC, for the United States as a whole, as of the end of April, African-Americans accounted for 21.2 percent of COVID-19 reported deaths, twice their 12.6 percent “unweighted” share of the population. Before the CDC applied its magic, 19 of the 26 states that reported such fatality statistics had unadjusted burden ratios greater than one—for an average ratio of 1.7. In other words, the share of black fatalities exceeded the unadjusted population shares by 70 percent. The CDC’s adjusted population shares slash this ratio from 1.7 to 1.15. For blacks in Florida, Illinois, Michigan, Missouri, New York City, New Jersey, Pennsylvania, Ohio, Tennessee, and Washington, as well as for Latinos in New York State, the CDC’s adjustments have virtually eliminated racial disparities.
A simple hypothetical case captures the shortcomings of this CDC adjustment procedure. Imagine that New York City’s black and Latino population accounts for 100 percent of all COVID-19 fatalities in New York State but just 20 percent of the state’s “unweighted” population. The CDC’s “adjusted population share” weights New York City by its share of COVID-19 fatalities (in our hypothetical, 100 percent) but excludes the entire rest of the state from its population base, because it has no observed fatalities. Thus the CDC’s approach would magically slash the relative burden ratio for blacks and Latinos from five (100 percent divided by 20 percent) to one. Houdini couldn’t do it any better. Extreme inequality in the incidence of COVID-19 vanishes in a puff of statistical hocus-pocus. In effect, the social group that in our hypothetical example actually bears 100 percent of the COVID-19 burden would bear no “disproportionate” burden at all according to the CDC’s measure. Perhaps this should be called the Bantustan adjustment, because it reminds us of similar procedures that were employed by statisticians in pre-apartheid South Africa to argue that blacks who had been removed to Bantustans were really quite well off, relative to other blacks.
As we’ve seen, the real story of COVID-19-related suffering is indeed about concentration and community health—the concentrated suffering that many of our poorest, most vulnerable neighbors are enduring right now, in communities that lack even the most basic health care and economic resources needed to care for them.
This is a problem that data analysts who are employed by well-endowed Federal agencies should be exploring more closely right now, not minimizing, intentionally or not. In principle, the CDC should be leading the way in helping to improving COVID-19 statistics, not inventing dodgy measures that undermine state and local health departments. It should also be making new impact measures available—for example, the incidence of COVID-19 fatalities by income group would be very useful. Unfortunately, this turns out to be just one in a series of recent episodes in which, to put it mildly, it is very hard to make the CDC’s track record look like an achievement.
Evidence from Other States
Our conclusions from the analysis of New York State data and New York City ZIP code data are also supported by a growing number of reports from other U.S. states and localities that show that poorer blacks, Latinos, and Native Americans have been bearing a disproportionate share of COVID-19 cases and deaths. Of course, not all states and localities are eager to publish such statistics. Furthermore, as of the end of April 2020, only two states, Kansas and Illinois, had published the extent of testing by racial and ethnic group. However, a disturbing pattern is already clear:
- Alabama: African-Americans account for 37 percent of official COVID-19 cases, 45 percent of reported deaths, and 26.5 percent of the state’s population.
- Arizona: Native Americans are 16 percent of reported COVID-19 fatalities but 4.6 percent of the state’s population.
- Illinois: African-Americans are 38.1 percent of COVID-19 deaths, 24.2 percent of confirmed cases, and 16 percent of the state’s population. As of late April, 13.2 percent of those tested were black.
- Louisiana: African-Americans are 59.9 percent of 1,767 reported COVID-19 deaths and 34.9 percent of the state’s population.
- Maryland: African-Americans are 47 percent of reported COVID-19 deaths and 31 percent of the population. Journalists have found similar patterns in particular heavily black areas like Prince George’s County.
- Massachusetts: A prominent Boston doctors tells us that “Chelsea [Boston’s “city of the working Latino immigrant”] is on fire with COVID-19.”
- Mississippi: In the state with the highest poverty rate in the United States, 60 percent of COVID-19-related deaths and 52 percent of reported COVID-19 cases are African-American, compared with 37 percent of the state’s population.
- Washington, DC: African-Americans are 45 percent of the capital’s population but make up as much as 80 percent of COVID-19 deaths.
At the micro level, many different kinds of inequality traffic patterns have contributed to this outcome, although it is hard to divvy up each pattern’s share of blame, especially given the abominable level of generic testing. As of our writing, only Kansas and Illinois had published data on testing by race. All other states either have not kept records of this important statistic, or chose not to share the results. COVID-19’s R0 number, or basic reproduction number (the expected number of infections that any given infected person passes on to others), is undoubtedly much higher in poorer communities because of factors like work exposure, denser housing and street traffic in urban areas, dependence on public transit, the relative frequency of public events like church meetings and sporting events, and the prevalence in jails and prisons. Basic information about safe public health practices is also harder to come by—although most Americans have had trouble keeping up with the CDC’s oscillating recommendations about social distancing, masks, and attendance of public events like Mardi Gras, Spring Break, and religious gatherings.
Among people of color, the high prevalence of “comorbidities” with COVID-19, like diabetes, obesity, and cardiovascular disease, also contributes decisively to high mortality rates. To a great extent their prevalence, in turn, is explained by unhealthy patterns of work, diet, and lifestyle “choices” that are strongly correlated with class and race.
Of course the striking maldistribution of COVID-19 fatalities that we have found is hardly news to many of these communities or the medical practitioners who serve them. Nor is this the only injustice that they suffer. In addition to the comorbidities just noted, for example, New York City’s black and Latino communities also experience relatively high rates of poverty, unemployment, non-documentation, crime, malnutrition, and addiction, as well as persistent shortages of decent hospital care, housing, education, security, quality policing, legal services, and many other amenities.
From this angle, the real “news” here might have been to find that poorer, blacker communities had somehow managed to avoid even greater suffering at the hands of this pandemic. But most of these other long-standing afflictions—and the chronic institutional violence they embody, which is generating many more casualties over time than COVID-19 is ever likely to—have long since dissolved into the background of social ills that most of us take for granted. Consider the 1.3 million people around the globe who will probably die of tuberculosis this year, or the 1.4 million children who will probably die of diarrhea from dirty water—this year, next year, and the year after that, but mainly in distant lands.
Compared with these other afflictions, the interesting thing about COVID-19 is not only its novelty and its explosive growth. It is also that, while there is little doubt that its health and economic effects will turn out to have been very unequally distributed, at least for the time being many of us have been forced to acknowledge that the pandemic is indeed a community-wide problem—after all, even Prime Ministers and Kings and hedge fund hot shots are vulnerable to it. Our society has not quite yet been reduced to becoming a loosely affiliated archipelago of occasionally connected but more or less self-sufficient private compounds. While some yacht- and island-owning billionaires may have become “islands unto themselves,” the pandemic reminds us that most of us still do live in some kind of community.
Our last key finding with respect to the distribution of suffering and grief caused by this pandemic is that the rates of COVID-19 infection and fatality have been seriously undercounted, especially in poorer neighborhoods.
There are several reasons for this. First, as noted earlier, “reconnaissance testing” of the general population has been slow to get off the ground, even in New York City, which has led the way.
As testing has accelerated, we are beginning to see that the segregated patterns by class and race that we identified earlier were, if anything, understated. For example, based on expanded testing in the Bronx, New York, Governor Andrew Cuomo’s May 1 press conference reported an estimated 17 percent COVID-19 incidence in the relatively low-income, predominantly black and Latino Bronx, home to tens of thousands of New York City’s 500,000 undocumented immigrants. This is much higher than the city-wide average.
Second, the system employed by most health departments for collecting data on COVID-19 fatalities has relied heavily on reporting by hospitals. But many COVID-19-related deaths aren’t taking place in hospitals. While New York State and New York City, for example, have recently taken steps to improve reporting on COVID19 fatalities in nursing homes, home deaths are still very poorly reported. These include quite a few even after discharge from the hospital, which raises serious questions about the meaning of the trends in “hospitalization” and “discharges” that have frequently been reported as success indicators by officials. Since March 2020, “first responders” (EMTs, police, and fire) in New York City have reported a dramatic rise in home deaths from heart attacks—70 per week now, compared with just 20-30 per week for the same period in 2019. New Jersey morticians also report that total fatalities per month have increased this spring by nearly 67 percent, from an average of 6,000 per month to more than 10,000. So far there have been no efforts to test these home cadavers for antibodies, so they have not been attributed to COVID-19.
Third, even when COVID-19-related deaths take place in a hospital, the “primary cause of death” listed by physicians or morticians on death certificates may not be COVID-19. For example, the head of New Jersey’s Association of Funeral Directors has said that the members of his association were instructed by state authorities—under “crisis guidance” from the CDC’s Vital Statistics office—to list derivative conditions like pneumonia, stroke, or heart attack as the primary causes of death on the top lines of death certificates. In practice, this would prevent such fatalities from being attributed to COVID-19. How widespread such practices are across states and localities is not clear.
Meanwhile, a growing number of analysts all over the world have reported a dramatic rise in total mortality from all causes this spring. A recent review of these reports by the Financial Times has concluded that the global COVID-19 death count may be understated by at least 60 percent. This is not only because of the testing, classification, and detection problems just noted, but also because, in epicenters from Wuhan to New York City, from Quito to Sao Paulo, from London to Milan, and from Tehran to Lahore, another key impact of the epidemic has been to “crowd out” treatment for other serious diseases and conditions. This means that exceptional increases in total mortality are a better measure of COVID-19’s impact than the kind of statistics published by the CDC or the New York City Health Department.
It is also likely that the several of these factors responsible for the undercount have much greater impact in poorer neighborhoods. Unfortunately, total mortality data is slow in coming for many jurisdictions, but they are also not readily available on a disaggregated basis by ZIP code, class, or color.
However, using the official New York City COVID-19 mortality data by group that we reviewed earlier, plus a plausible range for undercounting, it is possible to scale up the figures reported for official COVID-19 mortality to get a better estimate of the overall burden and its distribution, at least in New York City. This analysis indicates that as of May 1, 2020, New York City’s official COVID-19 reported mortality figures may have undercounted the number of black and Latino fatalities by at least 6,600 deaths and all other COVID-19 fatalities by 2,300, for a total undercount of nearly 9,000—including 60 percent of all reported black and Latino deaths.
In 1966 it was considered a major scandal to learn that African-Americans were accounting for over 20 percent of all U.S. combat deaths in Vietnam—almost twice their share of the U.S. population. In response to an outcry from civil rights leaders like Martin Luther King Jr., military deployments were altered. By the war’s end, the ratio had been reduced to 14 percent. Still, in addition to its many other problems, that war—the first one conducted by a post-civil rights America—was a striking demonstration of just how unequal such war-fighting burdens can be.
As of May 2020, the total number of African-Americans and Latinos who have already died in the COVID-19 “pandemic war” in New York State alone is already twice the number of those who died in the Vietnam War—and also twice their share of New York’s population.
Curiously, this time around, this disproportionate burden of death and suffering has not yet led to much of an outcry. As the German theologian and anti-fascist crusader Dietrich Bonhoeffer once said, “We have been silent witnesses of evil deeds.”