Since the start of the COVID-19 pandemic, researchers have been working on ways to model the spread of the novel coronavirus—and its effects on the economy. They’ve used temperature data to predict hotspots, cough simulations to model the virus’s aerosol spread, and location data to model the risk of exposure. The issue, though, is that there are so many factors at play: whether people are working from home or not, how they’re shopping, and if they’re wearing masks all have an impact on not just the public health of a city, but its businesses, as well. And no two cities are the same.
A new model developed by Stanford researchers tries to account for all those aspects, and show how different reopening policies—whether alternating schedules for workers and students or keeping everyone but essential workers home; whether mask usage is widespread and required, or rare and voluntary—affects both the employment and health outcomes in a city. An interactive website based on their research lets you try out different policies and instantly see the changes in health and economic outcomes.
On the website, you can pick your own policy combinations, deciding who can go to their workplaces, how the community will interact at stores (such as using curbside pickup or going in person), whether those 60 years and older are advised to stay home or not, and how prevalent mask wearing and social distancing are. Selecting an option for each category changes how many approximate deaths and lost work days are expected, and that final result also changes by city. A full lockdown—essential workers only, with necessity-only shopping—with widespread mask usage would lead to about 49,000 deaths and 520 million lost work days in New York City, but only 99 deaths and 35 million lost work days in Sacramento.
The website isn’t predicting what will happen next as cities try to reopen without increasing COVID-19 deaths, but instead showing a snapshot of what would have happened if these policies were in place from May to August. Even at a best case scenario, for instance, the model does show a higher death count for New York City than the current total—which is about 24,000 deaths—in part because the data is looking at the New York Metropolitan Area rather than just the city, and because the model is calibrated using mortality rates from around April and May. Since then, the researchers explain, treatments have improved, mortality rates have decreased, and mask usage has become a bit more widespread.
To create this model, the researchers, led by Stanford economists Mohammad Akbarpour and Shoshana Vasserman, used data from Replica, which uses cell phone location data to figure out where people go and how long they spend there, and then creates an example of a typical day of movement. It’s not an actual map of your movements, the researchers explain, but a simulated population based on behavior patterns. This data allowed the researchers to see how many people on an average day were in the same place at the same time—at a coffee shop before the workday starts, for example—which they used to estimate infection rates. Each variation of reopening policies on the website changes how many people would be in the same place in a city at the same time.
The reason for those differences in New York versus Sacramento, even with the same lockdown policies, lies in the structure of each city’s network. New York is more dense, with more points at which people can come into contact with each other while going to work or running errands, than in a city like Sacramento. “If you have more contacts, then every interaction is more impactful in some sense,” Vasserman says. The relationship between the number of contacts that you have and the spread of the virus is not linear, she adds, meaning that interacting with two more people in a day is not equivalent to just a 2% higher chance of infection. Every interaction exponentially increases your chance of COVID-19 exposure.
Density isn’t the only factor, though. How a city fares with the pandemic, both from a health standpoint and economic standpoint, also depends on the distribution of comorbidities—the presence of one or more health conditions that may worsen someone’s COVID-19 case—and what kind of employment sectors are most common in a city.
If there are a lot of people who cannot do their jobs while working from home, like restaurant workers, for example, then a policy that shutters businesses and requires remote work will have a more harmful effect on the economy in that city than in a place where more people can switch to remote work. “Similarly, if the kinds of people who are exposed when you send them to work have more comorbidities, then they’re more likely to die if they get infected,” Vasserman says. (In New York City, 88% of hospitalized COVID-19 patients had more than one comorbidity, according to an April study.)
By showing how each policy affects each city, the researchers hope it helps people understand why there were differences in how cities across the country were affected by the pandemic. In April and May, Vasserman says, there was a lot of broad talk about “science says this is how you reopen,” she says. “There was a lot of resistance among especially smaller cities whose leadership said the policies didn’t make sense to them. Part of what we wanted to do was add some language and a rigorous way to talk about the reasons that there might actually be differences between different kinds of municipalities, and what’s the best approach for them.”
No one reopening policy is perfect for every city, but the researchers did note that higher mask usage and allowing those who can work from home to do so tended to lead the best health and employment outcomes across their model. Vasserman and Akbarpour do hope that policy makers can use this to figure out the best reopening strategy for their own cities, but they recommend that leaders reach out to them for a more detailed, customized version of the model tailored to each city. They also hope it will help city leaders realize all the things they need to consider when it comes to being prepared for any pandemic. “The details of these lockdown policies matter a lot,” Akbarpour says. “In some sense, the big impacts [of this research] might come in the next pandemic.”