As the COVID-19 pandemic continues to ricochet across the country, the need for a coordinated response has never been more urgent. But since there is no coordination at the national level, it falls to governors to create at least a coordinated regional response to tamp down the spread of the coronavirus.
The latest research by our team in the Social Analytics Lab at the MIT Initiative on the Digital Economy shows why coordination across large geographic areas is critically important. Our analysis found that the mobility patterns of people in one state are substantially affected by the policies and behavior of people in other, sometimes distant states.
The coronavirus is not limited by geography, and decisions made in a region will travel when people travel and carry the virus with them. Adjacency is not necessary for the health policies of one place — city, county or state — to affect people in other places because the effects are also transmitted through the influence of social media.
We take cues from our social media connections on what to believe and how to act. When we see our friends having barbecues on Instagram, we begin to think it might be safe to gather with friends. When we speak to our parents over Zoom in distant states, we may tell them to wear masks and stay home.
At scale, these digital connections move the needle — even more than local policy or geographically proximate travel. When just one-third of a state's social and geographic peer states adopt shelter-in-place policies, it creates a reduction in mobility equal to that state's own policy.
To study the spillover effects of uncoordinated policies, my co-authors and I combined daily, county-level data on all the shelter-in-place and business closure policies (whether strict, lax or nonexistent) in every state in March and April, with movement data from more than 27 million mobile devices, social network connections among more than 220 million Facebook users, as well as daily temperature and precipitation data from 62,000 weather stations, and county-level census data on population demographics to estimate the geographic and social network spillovers created by regional policies across the United States.
We collected the mobility data through a collaboration with Facebook’s Data for Good initiative and SafeGraph’s COVID mobility research program. We measured connections over social media using an index of the degree to which different U.S. counties are socially connected on Facebook. We then measured social and geographic spillovers by estimating how the initiation of a shelter-in-place order in one state created changes in the relative movement of mobile phones in other states, both through travel and through social influence over Facebook. We used the weather data to estimate causal relationships between policies in one state and mobility in faraway states — if inclement weather in one state is correlated with reductions in mobility in distant states that don’t experience that weather, social influence is at work.
Our research suggests that the coronavirus will bounce back and forth across the country unless states coordinate their travel and closure policies more closely.
After Georgia reopened and let its shelter-in-place policy lapse on May 1, for example, half a million people flocked to the state — a 13% increase in out-of-state visitors or about 62,000 additional visitors a day in the week after the state reopened. Ten days later, beginning May 10, Georgia experienced a marked uptick in new coronavirus cases that lasted for almost two weeks.
Myrtle Beach, S.C., reopened in mid-May and spent millions to promote tourism with advertisements and marketing campaigns targeting states like Ohio, Georgia, New York and New Jersey. Just two weeks later, the town and the surrounding county started seeing a surge in new coronavirus cases.
Our work also gives governors a way to coordinate their policies in the absence of national guidance. We created coordination maps for all 50 states that show which states affect each other most and which states would benefit from coordinating their policies more closely. We can see in these maps strong connections not just between geographically proximate states, but also between socially connected, geographically distant states.
For example, while Georgia is mostly influenced by neighboring states through travel, Florida is most affected by New York, through travel and digital social influence, though the two states are nearly 1,000 miles apart. New Hampshire has a strong influence on adjacent Massachusetts, despite being a small state, while Texas is most influenced by California, despite not being adjacent.
When a state reopens while its peer state remains closed, travel spikes from the closed state into the open state. Only when both states adopt similar shelter-in-place policies does travel between the states diminish.
New York and California provide stark examples of different approaches. In New York, Gov. Andrew Cuomo instituted strict travel restrictions and 14-day quarantine mandates for 36 states, with a $10,000 fine for not quarantining and a $2,000 fine for not filling out travel forms upon arrival in New York. Gov. Gavin Newsom of California, by contrast, has not imposed any travel restrictions since the pandemic began. New cases in New York have remained steady and low since June, while California is still in the grip of a coronavirus outbreak that started in late May.
It’s quite clear that what governors do directly affects the health and lives of millions beyond their borders. They have to exercise leadership to protect not only their own residents, but those in adjacent and even distant states.
Since the beginning of the pandemic, we’ve often heard that we are “all in this together.” Now more than ever, it is time to start acting like it.
Sinan Aral is the David Austin Professor of Management at the MIT Sloan School of Management and director of the MIT Initiative on the Digital Economy. He is the author of the forthcoming book “The Hype Machine.”