COVID-Related Joblessness May Pose Long-Term Challenge to Public Health, Study Finds

A new working paper co-authored by Johns Hopkins Carey Business School Assistant Professor Dongho Song explores the long-run effects of the COVID-19-induced economic recession on mortality and life expectancy.

Because of the size of this sharp and deep recession, Song and his colleagues find that the long-run mortality effects of the recession could far exceed mortality tied to the virus.

According to preliminary analysis in the study, the public health effects of the pandemic-induced unemployment shock are estimated to lead during the next two decades to the deaths of nearly 1.4 million Americans—these are additional deaths that wouldn’t have occurred if the pandemic hadn’t happened.

The researchers modeled their work on U.S. data from 1950-2017, which showed that “shocks to unemployment are followed by statistically significant increases in mortality rates and declines in life expectancy.”

The paper has appeared on the website of the National Bureau of Economic Research. Song’s co-authors are Francesco Bianchi, a research associate at Duke University, and Giada Bianchi, an oncologist at Harvard Medical School.

In the following Q&A, Song discusses some of the findings from the study.

QUESTION: In your study, you separated the effects of the 2020 unemployment spike from other factors that typically affect mortality and life expectancy. How did you achieve this?

DONGHO SONG: We use historical data and a statistical model to infer the typical effect of an increase in unemployment on mortality and life expectancy. We then use the same assumptions to infer the size of the COVID-19 unemployment shock. So far, our results are robust to different assumptions about the identification of the unemployment shock.

Your findings indicate that limitations on access to health care during the pandemic are the driving force behind this projected increase in the mortality rate. Can you expand on that?

Our analysis is based on an historical relationship between unemployment and well-being. Here, unemployment can be considered a variable summarizing how well the economy is doing. Low unemployment means that more people are receiving wages and have good health insurance. Thus, it is not just a matter of access to health care. We discuss factors that can make things better or worse with respect to the past. One factor that could make things worse is that people might be reluctant to go to the hospital or see a doctor because of fear of getting sick with COVID.

You state in the paper that deaths resulting from the unemployment spike “will disproportionately affect African-Americans and women, over a short horizon, while white men might suffer large consequences over longer horizons.” Why is that the case?

This conclusion is based on the historical relations that we uncover. The interpretation of the result would arguably require us to conduct a more in-depth analysis of the causes of death. For this, we would need to access more data than what we currently have.

You and your colleagues take pains to note that you don’t oppose lockdowns as an important public health response. However, you add, policy makers should find additional ways to fight the pandemic that don’t cause so much distress to the economy – to employment in particular. What might those additional measures include?

These include wearing masks, social distancing, and widespread testing. We have a much better understanding of how the virus is transmitted. We then also discuss how we should provide as much support as possible to those in need.

Have the 2020 CARES Act and other government aid provisions done much to mitigate the economic pain that you address in the paper?

As mentioned before, we use past data to try to predict what will happen in the future. Based on 1950-2017 data, we estimate the historical relation between unemployment, life expectancy, and mortality. Then, we back out the COVID-19 unemployment shock given the structure and project excess deaths based on the estimated “historical relation.” It is possible that ACA will reduce the impact of economic distress on mortality rates. However, there are other factors (like fear of contagion) that work in the opposite direction. And we do not have enough data to understand how much all these factors will modify the final outcome.

As you observe in the study, this is the first major American recession since the passage of the Affordable Care Act. Have ACA programs provided much of a safety net for the people feeling the worst effects of the unemployment spike? That is, would your projections of the death rate have been worse if the ACA didn’t exist?

It is too early to say. Several studies have reported a reduction in preventive care and health interventions not related to COVID. Thus, unfortunately, we do not believe that ACA by itself can fully solve the problem.

We’re not out of the woods yet with the pandemic. Do you foresee possibly having to revise your projections of the death rate if the crisis drags on for another year or so? That is, revise upward?

Yes, if unemployment were to increase again, we would have to revise our projections upward. I hope that there will not be the need of doing so. We are already very concerned about the current outlook.