We find that shocks to unemployment are followed by statistically significant increases in mortality rates and declines in life expectancy. We use our results to assess the long-run effects of the COVID-19 economic recession on mortality and life expectancy. We estimate the size of the COVID-19-related unemployment to be between 2 and 5 times larger than the typical unemployment shock, depending on race/gender, resulting in a 3.0% increase in mortality rate and a 0.5% drop in life expectancy over the next 15 years for the overall American population. We also predict that the shock will disproportionately affect African-Americans and women, over a short horizon, while white men might suffer large consequences over longer horizons. These figures translate in [to] a staggering 0.89 million additional deaths over the next 15 years.

This is from Francesco Bianchi, Giada Bianchi, and Dongho Song, “The Long-Term Impact of the COVID-19 Unemployment Shock on Life Expectancy and Mortality Rates,” NBER Working Paper No. 28304, December 2020.

An excerpt:

For the overall population, the increase in the death rate following the COVID-19 pandemic implies a staggering 0.89 and 1.37 million excess deaths over the next 15 and 20 years, respectively. These numbers correspond to 0.24% and 0.37% of the projected US population at the 15- and 20-year horizons, respectively. For African- Americans, we estimate 180 thousand and 270 thousand excess deaths over the next 15 and 20 years, respectively. These numbers correspond to 0.34% and 0.49% of the projected African- American population at the 15- and 20-year horizons, respectively. For Whites, we estimate 0.82 and 1.21 million excess deaths over the next 15 and 20 years, respectively. These numbers correspond to 0.30% and 0.44% of the projected White population at the 15- and 20-year horizons, respectively. These numbers are roughly equally split between men and women.

Francesco Bianchi is an economist at Duke University, Giada Bianchi is an MD in the Division of Hematology, Department of Medicine, Brigham and Women’s Hospital Harvard Medical School, and Dongho Song is an economist at the Johns Hopkins University’s Carey Business School.

The authors write:

We interpret these results as a strong indication that policymakers should take into consideration the severe, long-run implications of such a large economic recession on people’s lives when deliberating on COVID-19 recovery and containment measures. Without any doubt, lockdowns save lives, but they also contribute to the decline in real activity that can have severe consequences on health.

I’m not sure why they are confident that there is zero doubt that lockdowns save lives. They admit in the last quoted sentence above that lockdowns “contribute to the decline in real activity that can have severe consequences on health.” What if lockdowns are responsible for half of the bad unemployment consequences, and voluntary actions in response to the fear of getting the virus are responsible for the other half? Then, assuming a linear relationship between unemployment and fatalities, the lockdowns would be responsible for half of 0.89 million to 1.37 million deaths, which translates to between 450,000 deaths and 685,000 deaths. Can they really be confident that lockdowns saved at least 450,000 lives?