COVID-19’S PUZZLING PATTERNS.

Quick now, what’s your gut reaction? Looking across U.S. counties (all 3,100 of them) and controlling for a bunch of other socioeconomic characteristics, which counties are more likely to have more COVID-19 cases: counties with lower median incomes or countries with higher median incomes?

The simple fact I’m asking that question probably clues you in that the answer isn’t what common sense would predict, namely, that lower-income areas have a rougher time with COVID than upper-income areas. And your inference would be correct: a new study by Caitlin S. Brown of the Central European University and Martin Ravallion of Georgetown University reveals — but then explains — this counterintuitive result.

Just to heighten the mystery a little: the higher number of cases in higher median-income counties occurs despite the fact that such counties have generally increased their social-distancing more, according to Big Data compilations of mobility by Google and Unacast. (Google reports data for whether people are at “groceries and pharmacy, retail and recreation, transit stations, workplaces and residential locations.” Makes me glad I keep forgetting to charge my cellphone.)

The first sentence of Brown and Ravallion’s abstract gives a flavour of their theme: “Not much is obvious about how socioeconomic inequalities impact the spread of infectious diseases once one considers behavioural responses, correlations among multiple covariates and the likely non-linearities and dynamics involved.” That’s not what we typically get in news reports of disproportionate numbers of COVID cases or deaths among different groups. (Being abnormally tall, I was disturbed to read recently that a disproportionate number of COVID cases are of people over six feet, though the sample size in the study reporting this result was only 2,000.)

To solve these puzzles, economics gallops to the rescue, for, as the authors put it “compliance and behaviours regarding social distancing are … personal choices,” and that’s the domain of economics. People presumably weigh the cost of adjusting their behaviour — or behaviours, if you must — against their “expected future loss from infection.”

In this regard, poorer people may face tougher constraints. They presumably have fewer choices about whether to work, how to get there, how often and where to buy groceries, and so on. So maybe on balance they’re less able to adjust than richer people, who presumably have greater flexibility to respond. On the other hand, richer people may have more responding to do, in the sense that they have more social connections and greater exposure to visitors from afar. Social isolation is a classic feature of poverty, after all, while not many poor people are likely to have visited China in January or February or have friends who did.

That’s the solution to the puzzle about median-income counties, by the way. Yes, higher median-income counties showed more of an increase in social distancing after the pandemic hit. But they were starting at lower levels of social distancing to begin with, so it still stands to reason that they should experience more COVID cases on average than lower median-income counties.

There’s a similar effect regarding older folk. Many of us who are retired have relatively low levels of social interaction. (Apart from learning to wear a mask, the main COVID effect on my own life was that I had to stand farther away from the other owners in the local dog park. Until the town closed the dog park, that is.) On the other hand, if we elders do get the virus, the consequences are harsher — though it’s not yet completely clear whether that’s from age itself or the fact that we tend to have higher co-morbidities (e.g., diabetes, asthma, hypertension and lung disease). On balance, the share of its population that’s 65 or older doesn’t have much effect on a county’s death rate from COVID. Our higher chance of dying once we catch COVID is roughly offset by our lower chance of becoming infected, given our generally fewer social interactions.

The study’s main goal is to see if income inequality has a separate and distinct effect on social distancing and COVID infections and deaths. Here, too, the results are ambiguous. Across counties with roughly the same median incomes, those with higher poverty rates do have higher infection rates. So inequality does seem to matter, independently of income levels. That suggests anti-poverty policy may “complement … health policy in combating this infectious disease.” Of course, the disease is a problem here and now while anti-poverty policy is usually long-term — though continuing cash benefits for low-income people might make their personal COVID choices a little less cruel.

But — another puzzle — when the researchers control for race the effects of poverty and income largely disappear: “The effects of income inequality and poverty within counties largely vanish when one controls for the Black American population share, indicating that the directly relevant factor is race not income inequality or poverty per se.”

In the end, however, even after imposing all these socioeconomic controls, including race, richer counties still have higher infection rates. Which suggests the authors’ opening statement was apt: Not much about COVID is obvious.