Income Inequality and COVID-19

It has been a very long time since I last made a post here. I am coming back with a post about the relationship between income inequality and COVID-19.

The latest issue of The Economist has an article on this topic, which led me to three recent studies about this relationship and an interesting Twitter thread. (Do watch out for the careless conflation of wealth with income in the second tweet in the thread.) I will say a few words for each of the three studies. Before I do that, I need to issue the disclaimer that I am not a statistician or an econometrician, therefore I cannot, and will not, claim to evaluate the appropriateness of the statistical modeling in these studies.

Let’s start with “Association Between Income Inequality and County-Level COVID-19 Cases and Deaths in the US”, by Annabel X. Tan, MPH; Jessica A. Hinman, MS; Hoda S. Abdel Magid, PhD; Lorene M. Nelson, PhD, MS; Michelle C. Odden, PhD, from JAMA Network Open, doi:10.1001/jamanetworkopen.2021.8799. The authors collected data on COVID cases and deaths for a year (2020-03-01 to 2021-02-28), as well income inequality (measured by the Gini coefficient) for 3220 counties in all 50 states plus Puerto Rico and DC. Their main finding was a positive correlation between income inequality and COVID cases and deaths, which was most pronounced in the summer of 2020. Several additional variables were included as controls, such as the poverty rate, age, race, mask use, crowding, educational level, urban versus rural population share, and availability of physicians. I find it remarkable that income inequality showed up as correlated with COVID cases and deaths in the presence of all these additional variables that one would expect to be more strongly correlated with COVID outcomes.

Next, we’ll talk about “COVID‐19 and income inequality in OECD countries” by John Wildman, The European Journal of Health Economics (2021) 22:455–462 The COVID variables are cumulative deaths per million and recorded daily cases per million in the early months of the pandemic. In the words of the author, “The results demonstrate a significant positive association between income inequality and COVID-19 cases and death per million in all estimated models. A 1% increase in the Gini coefficient is associated with an approximately 4% increase in cases per-million and an approximately 5% increase in deaths per-million.” The author proposes that income inequality is a proxy for other variables that correlate with bad COVID outcomes, such as “poor housing, smoking, obesity and pollution.”

Finally, let’s take a look at “The trouble with trust: Time-series analysis of social capital, income inequality, and COVID-19 deaths in 84 countries” by Frank J. Elgar, Anna Stefaniak, and Michael J.A. Wohl, Social Science & Medicine 263 (2020) 113365. Here is the abstract of the paper:

Can social contextual factors explain international differences in the spread of COVID-19? It is widely assumed that social cohesion, public confidence in government sources of health information and general concern for the welfare of others support health advisories during a pandemic and save lives. We tested this assumption through a time-series analysis of cross-national differences in COVID-19 mortality during an early phase of the pandemic. Country data on income inequality and four dimensions of social capital (trust, group affiliations, civic re- sponsibility and confidence in public institutions) were linked to data on COVID-19 deaths in 84 countries. Associations with deaths were examined using Poisson regression with population-averaged estimators. During a 30-day period after recording their tenth death, mortality was positively related to income inequality, trust and group affiliations and negatively related to social capital from civic engagement and confidence in state in- stitutions. These associations held in bivariate and mutually controlled regression models with controls for population size, age and wealth. The results indicate that societies that are more economically unequal and lack capacity in some dimensions of social capital experienced more COVID-19 deaths. Social trust and belonging to groups were associated with more deaths, possibly due to behavioural contagion and incongruence with physical distancing policy. Some countries require a more robust public health response to contain the spread and impact of COVID-19 due to economic and social divisions within them.

I find these papers extremely interesting, and I want to make them part of my economic inequality course. You could say that this post is my very rough first reaction, simply noting the main conclusions of this research, conclusions that point out a clear connection between income inequality and the COVID-19 pandemic outcomes.

A paper that simulates COVID-19 in a University

The paper “Simulating COVID-19 in a University Environment“, written by Philip T. Gressman and Jennifer R. Peck appeared in ArXiv on June 5, 2020. It contains a stochastic agent-based model that simulates the likely progress of COVID-19 disease transmission in a fictional U.S. University with 20,000 students and 2,500 faculty members that opens for a semester of 100 days in a world with certain (unknown) members of the general population infected with the disease. It also includes an analytical model that supports the main conclusions drawn from the simulations. It complements the paper “The small world network of college classes: Implications for epidemic spread on a university campus” by K. A. Weeden and B. Cornwell.

How to open a University campus relatively safely and conduct instruction with minimal numbers of infection and the maximal possible effectiveness of instruction is keeping University and College administrators up at night, not to mention faculty members like myself, who dread the possibility of being forced to teach in a classroom at a risk to their health they deem too high. Papers like the Gressman and Peck paper are valuable contributions to administrators’ decision-making and I hope they are taken seriously by them.

The main conclusions of the paper can be summarized simply. I do so now and I discuss the assumptions of the paper at the end of this blog post. The two outcomes the simulations focus on to evaluate the effectiveness of various infection control measures are (1) the total number of infections and (2) peak quarantine population.

Disclaimer: I read with reasonable care the main body of the paper and glanced at the part of the appendix where the results of robustness testing are reported. I did not read the analytical model presented at the end of the appendix, Section 5.3.

Main conclusions of the simulations:

  • The rate that testing for COVID-19 yields false positive results is unexpectedly and massively important for the results. This is in the context of the regime that several Universities (including mine) have announced for Fall 2020, where there would be extensive testing of community members and tracing and isolation of contacts that test positive. Such tracing would likely result in quarantining 10 to 20 students for every student who tests positive, which imposes a high cost in terms of the number quarantined.
  • An almost certain way to guarantee widespread infection is to allow classes larger than 120 students to meet face to face. As part of the main intervention the authors consider, classes with over 30 students would meet online only.
  • It matters a lot that students refrain from “all contact outside of academic and residential settings” (page 2).
  • Instructors have to prepare for online delivery of instruction to quarantined students, expecting at least 10% of students in any class to be quarantined on a given week. The experience of these students would not be the same as those attending class in person.

I found the paper convincing and its conclusions credible. I do want to emphasize some limitations of the analysis of the paper stemming from its assumptions. These are mostly made clear by the authors but do tend to push in the direction of making the conclusions overoptimistic. I offer these critical comments as a caution for readers, especially should they be University administrators, and not in order to diminish the contribution of the authors; obviously, all analyses have their limitations.

  • The first limitation is prominent in my own calculations for my personal safety: exposure to infection from using public transportation to commute to campus is not considered in the paper. Numerous faculty and students can be expected to use public transportation and import infections to campus in this way.
  • Compliance of individuals with regulations is assumed throughout. What are the chances individuals aged 18 to 22, to speak of the traditional age students who are still a large proportion of most campuses, will act responsibly outside of class and dorm, not going to multiple parties without any physical distancing in place? It is one thing to require mask-wearing in the classroom and another thing to expect mature behavior outside the campus setting by young people who also realize that their personal risks for a serious and possible fatal infection are small.

I could offer more minor nitpicking comments on the assumptions of the analysis, but I am stopping here, after having listed my main thoughts about the limitations of the paper. I view it as a very good and interesting paper and I look forward to additional simulations along the lines of those it offers that expand the reach of the model with assumptions amended along the lines I outlined in my critique.

Same-sex marriage legalization linked to a decrease in teen suicide attempts

This article in The Guardian discusses in detail a paper just published in JAMA Pediatrics, co-authored by Julia Raifman, Ellen Moscoe, S. Bryn Austin, and Margaret McConnell.

The first link above leads to a nice explanation of the results of the paper, and the second leads to an extended abstract that readers who are adept in econometrics (and other statistics-savvy people) will want to read closely.

It seems to me that the marginal benefit of same-sex marriage legalization includes the saving of many lives. The marginal economic cost is negligible, if it is even positive, compared to such a marginal benefit. Individuals wishing to argue that the moral marginal cost outweighs the marginal benefit will find it very hard to convince me of their case.

UPDATE: I changed “suicides” to “suicide attempts” in the title of this post for higher accuracy.

Stress getting worse for adolescents

I just came across this piece by Vicki Abeles in the New York Times. It discusses the effect on young people of the stress that the rat race in schools creates. The numbers are alarming.

For an economist like myself, a big question arises from this regarding the (in)famous efficiency of “free”, “perfectly competitive” markets, which are of course a theoretical fiction used to exalt the effect of competition on human welfare. Pretty obviously, we need to be very careful to amend our treatment of welfare. This brings to mind a sequence of blog posts on interfluidity, starting with this one, which I have set aside on my browser to read carefully. Now that the holiday festivities have quieted down, I plan to do that and post my thoughts here. Consider this a promise.