[🚨NEW WORKING PAPER🚨] As local governments reopen, we study how policies in one region affect mobility & social distancing in other regions & the consequences of uncoordinated policy responses to #Covid_19. 1/
bit.ly/2XgLV55
bit.ly/2XgLV55
The key takeaway is a state or county's policies significantly affect mobility in other states & counties -- not just in geographically proximate states but often a great distance through long distance travel & influence over social media and other communication technology. 2/
We combined daily, county-level policy data w/ movement from over 27M mobile devices, social networks among 220M FB users, daily weather from 62K weather stations & county-level census data to estimate geographic & social network spillovers in regional policies across the US. 3/
We first estimate the diff-in-differences of direct county-level effects of social distancing policies, but not geographic or social spillovers. Shelter-in-place (SiP) policies led to a 3.2% decrease in the % devices leaving home & a 6% decrease in the # of locations visited. 4/
We then accounted for geographic & social spillovers using @Safegraph data on visits between counties and the @Facebook connections between counties calculated by combining FB's Social Connectedness Index & county-level Census data. 5/
Accounting for geographic but not social spillovers, a county's SiP policy reduced the # locations visited by 4.0% & the % leaving home by 2%. But when half of the county’s geographic peers also adopted SiP, it further reduced visits by 2.3% & the % leaving home by 1.4%. 6/
When only one county in a pair implemented SiP, travel from that county to the non-SiP county increased by 0.55%, while travel from the non-SiP county to the SiP county decreased by 1.2%. When both implemented SiP, travel between them decreased by 0.51%. 7/
The geo results highlight the need to coordinate proximate regions but miss social spillovers. When considering both social & geographic spillovers, the effect of all peer states implementing SiP caused a 13% decrease in locations visited & a 9.1% decrease in % leaving home. 8/
This means it's not only geographically proximate states that affect outcomes, but also the communities to which a state is digitally, socially connected. 36% of a state’s geo & social peer states implementing SiP is as effective at reducing mobility as the state's own policy. 9/
We then used shifts in weather & industry visit "shift share" as natural experiments. A 3% decrease in peer locations visited caused a 5.6% decrease in # locations visited in a focal county & a 1.5% decrease in # peers leaving home caused a 2.4% decrease in the # leaving home 10/
This means social spillovers are substantially mediated by peer behavior. In other words, people in a focal state are significantly influenced by the behavior of their peers in other states when calibrating their own social distancing behaviors and choices. 11/
Estimates of spillovers between regions suggest the Northeast (NE) is most influenced by the NE & the South (SO). The Midwest (MW) is most influenced by the MW & SO. The SO is mainly influenced by the SO. And the West (WS) is most influenced by the WS & SO. 12/
Our work gives governors maps to coordinate in the absence of national guidance by showing which states affect each other the most. A state's success is impacted not just by geographically proximate states, but also by socially connected, but geographically distant states. 13/
GA is mostly influenced by neighboring states through travel. But, FL is most affected by NY, through travel and digital social influence, despite the states being distant. NH has a strong influence on adjacent MA, despite being a small state. We show maps for all 50 states. 14/
We used our data to calibrate a model of the "Loss from Anarchy" created by states failing to coordinate. Utility is up to 69% lower when states fail to coordinate in the presence of geographic and social network spillovers of the sizes detected in our empirical analyses. 15/
When spillovers are high, states’ policies diverge, with one state, for example, having to compensate for the neglect of another state’s loose restrictions by imposing even stricter, more costly policies than necessary to achieve their desired immobility targets. 16/
But, when states coordinate, social & geographic spillovers *help* them achieve targets more efficiently by providing “free treatments” as cooperative behaviors of peer states positively influence distancing behaviors in focal states, highlighting benefits to coordination. 17/
Our findings indicate that governments' decisions to lift SiP policies affect the behavioral and health outcomes of not only their own citizens, but also the citizens of geographically and socially proximate communities. 18/
We do not advocate a "one size fits all" policy and recognize the need for different localities, with different circumstances, to implement heterogeneous policy responses. However,... 19/
Our results suggest significant negative welfare effects from uncoordinated government policies & imply that federal governing bodies, like the US and EU, should coordinate policy action even when local policies should differ from region to region. 20/
Frankly, an amazing and miraculous amount of teamwork by @DaveHoltz, @mfzhao, @SBenzell, Cathy Cao, @aminrahimian, @jeremyzyang, Jennifer Allen, @avi_collis, Alex Moehring, Tara Sowrirajan, @ghoshd7, Yunhao Zhang, @dhillon_p, @CNicolaides and @deaneckles.
Tag your Governors here: [____]
