Pandemics Depress the Economy, Public Health Interventions Do Not: Evidence from the 1918 Flu

Jiageng Liu, Research Assistant, MIT Sloan Finance Group

(Consistent with the GCFP’s mission of non-partisanship and offering analysis but not policy advice, any recommendations contained in GCFP blog posts are those of the authors and do not represent the views of the GCFP.)

The COVID-19 pandemic damages the U.S. and the world economy and puts policymakers in an unfamiliar dilemma. Non-pharmaceutical interventions (NPIs) such as social distancing slow the spread of the disease but could also put millions of people out of work. As the dire situation continues to rise, there’s an urgent need to estimate the economic consequence of the pandemic and calculate the optimal policy response. Researchers studied the 1918 flu and found that while the pandemic reduces the manufacturing output by 18%, regions that intervened earlier and more aggressively not only ended up with fewer deaths, but also recovered faster economically after the pandemic.

Prof. Emil Verner at the MIT Sloan School of Management, along with Dr. Sergio Correia and Dr. Stephan Luck from the Federal Reserve, studied the policy response to and the aftermath of the 1918 Flu pandemic in major U.S. states and cities. The pandemic lasted from January 1918 to December 1920 and infected around one-third of the world population. It killed 50 million people worldwide, including around 600 thousand in the U.S. As the disease spread, the U.S. economy experienced a deep recession and serious deflation until July 1921.

The researchers look into states’ exposure to the flu and their local economy and summarize their findings in Figure 1. Cities that suffered from the highest level of the 1918 Flu mortality also saw the largest decline in manufacturing employment. The default rate among businesses and households rose, while bank assets and consumer durables declined relative to the less affected area. They also find the relative declines to be persistent. Areas more inflicted remained more depressed from 1919 through 1923.

The 1918 Flu first arrived in the eastern U.S., where cities like Philadelphia did not know much about the disease and suffered from a high death toll. The rest of the country was then alerted, and western cities like Seattle reacted with harsher measures as soon as their first cases emerged. The government of that time enforced similar NPIs as today: businesses and schools were closed, public gatherings are banned, and people were asked to stay home if they felt sick. Arguably, these NPI orders saved lives at the cost of accelerating the economy shutdown, as both supply and demand are hurt.

Instead, the authors show that early and forceful NPIs did not worsen the downturn. Cities that intervened earlier and more aggressively would have a relative increase in multiple economic indicators after the end of the pandemic. Cities that reacted 10 days earlier after the first case increased manufacturing employment by around 5% after the pandemic. Likewise, implementing NPIs for an additional 50 days increased manufacturing employment by 6.5% in the post-period. Figure 1 shows that cities that enforced NPIs for the time length above the median (in green) have a better outcome in both mortality and economy than those below the median (in red).

The authors note that several issues could hamper their arguments, such as the simultaneous economic shock from the end of World War I and the lack of high-frequency data to study the detailed mechanism. They also warned that the evidence cannot directly translate to today’s decisions. The 1918 Flu was deadlier, especially among young people. Compared with the industrial U.S. economy in 1918, today’s economy is more globalized and service-based, and communication technology enables people to work from home. 

Still, supported by epidemiological evidence and robustness checks, the researchers conclude that although pandemics are “highly disruptive for economic activity,” the quick response of NPIs “can reduce mortality while at the same time being economically beneficial.”

Read the full working paper by Sergio Correia, Stephan Luck, and Emil Verner here.