AdobeStock_335404311_0.jpeg
VoxEU Column COVID-19

Staying at home: The mobility effects of COVID-19

Lockdown policies are used to ‘flatten the curve’, but their success rate remains uncertain. This column uses GPS data from mobile phones in the US to show that stay-at-home orders do reduce mobility. However, voluntary reductions are also important, regardless of stay-at-home orders. Counties with a higher share of older people or a lower share of Republican votes are more responsive to lockdown measures. Further, counties with a larger share of jobs that are ‘teleworkable’, a higher median income, or a lower use of public transit are also more responsive, suggesting that multiple factors must be considered.  

An essential task in controlling the outbreak of the novel coronavirus is to reduce the reproduction rate (the number of subsequent infections spread by an infected individual). Economists have responded extensively, arguing for swift policy action in response to COVID-19 (Baldwin and di Mauro 2020). ‘Stay-at-home’ orders have emerged as a crucial policy tool for governments to encourage reductions in person-to-person contact. However, even without stay-at-home policies, individuals may still choose to decrease mobility in response to the threat of local case prevalence. Moreover, empirical analyses of reductions in person-to-person contact are important for informing both epidemiological work (Wang et al. 2020), as well as in wider economic research (Kaplan et al. 2020). 

Mobility data

Due to the proliferation of smartphones, researchers now have access to novel GPS location data which have already proven to be invaluable in monitoring the effects of COVID-19, as well as the associated social distancing policies. A number of economists have leveraged this data to find that political orientation, belief in science, and poverty status influence how readily people respond to social distancing orders (Painter and Qiu 2020, Brzezinski et al. 2020, Wright et al. 2020).

From cell phone GPS data, average distance traveled by individuals in a county can be measured and compared to pre-COVID-19 levels. The GPS data in this column is generously provided by Unacast. Figure 1 plots the change in average distance traveled relative to the same weekday pre-COVID-19 (on 24 February 2020). The overall light colour in the figure indicates that at the beginning of the outbreak (when there were very few confirmed cases) there was little change in mobility. 

As shown in Figure 2, one month later (on 23 March 2020) the average distance traveled decreased significantly in most counties across the US, with a particularly large drop in New York, California, Colorado, and Florida. It is worth noting that at the time, these were areas with relatively high numbers of COVID-19 cases.

Figure 1 Change in distance traveled relative to the same weekday pre-COVID-19, 24 February 2020

Figure 2 Change in distance traveled relative to the same weekday pre-COVID-19, 23 March 2020

Figure 3 shows the 10th quantile, median, and 90th quantile of the changes in average distance traveled at the county level. Mobility starts to decrease for median counties at around 10 March 2020, well before the announcement of restriction orders. There is also a distinct pattern where mobility drops further on weekends relative to weekdays which may indicate that occupation requirements are important factors in how much people are able to limit their mobility.                                                                        

Figure 3 Quantiles of changes in average distance traveled

Stay-at-home orders

The vast majority of states in the US implemented stay-at-home orders. Figure 4 shows the percentage of the contiguous US population under restriction orders by date. A county is counted as being restricted if either the county or state government has implemented a stay-at-home order (or equivalent). The first order occurred on 19 March 2020, with the last being announced on 18 April 2020. In total, over 95% of the population in the contiguous US fell under a restriction order.

Figure 4 Share of contiguous U.S. population under a stay-at-home order

Stay-at-home orders are a somewhat blunt policy instrument and may have many side-effects outside of their intended epidemiological goals. For example, lockdown orders have been shown to significantly affect household incomes and expenditures, with heterogeneous effects across age and income levels (Chronopoulos et al. 2020). Additionally, there are important political economy effects. Government lockdown orders have been shown to engender a ‘rally effect’, where individuals show more support for incumbent leaders and trust in government than before lockdown measures were introduced (Blais et al. 2020). Concern over certain side-effects and additional cultural factors may explain why certain countries such as Sweden have resisted lockdowns thus far (Karlson et al. 2020).

Mobility effects of COVID-19

In a new paper (Engle et al. 2020), we seek to quantify the mobility response to local COVID-19 prevalence and stay-at-home orders. The results presented in this column have been updated to incorporate up-to-date data (the sample runs from 24 February 2020 to 27 April 2020).

We find that in a county with median characteristics, a stay-at-home order reduces mobility by 7.27%. This effect is stronger for counties with higher median incomes, or where there are fewer votes for the republican party in the 2016 presidential election. The effect is also stronger in counties where there are more ‘teleworkable’ jobs, or there is a larger share of at-risk population (aged over 65). The effect is slightly weaker in areas that rely on public transit for commuting. To get a sense of the relative importance of these demographic factors, we perturb the characteristics one at a time, and compare the effect of a stay-at-home order to the reported baseline number, 7.27%. These results are shown in Table 1.

Table 1 Demographic variables influence the effectiveness of stay-at-home orders

Note: ‘pp’ stands for ‘percentage points’

We also find that local case levels (as well as case levels in neighbouring counties) have negative effects on mobility. For example, consider caseloads in King County, Washington from pre-COVID-19 (14 February 2020) to the day on which the stay-at-home order became effective (23 March 2020). Reported caseloads increase from zero to 1,166 (0.0006% of the population). This increase is associated with a decrease in mobility of 0.89%. During the same period, increase in disease prevalence in neighbouring counties decreases mobility by 0.20%. The relative effects of local cases versus neighbouring cases shows that local cases matter more. However, caseloads in neighbouring counties do still affect behaviour. Lastly, we note that after stay-at-home orders are implemented, the marginal effects of caseloads are significantly weakened. This suggests that stay-at-home orders may crowd-out natural reductions in mobility due to local prevalence.

References

Baldwin, R and B di Mauro (2020), Mitigating the COVID economic crisis: Act fast and do whatever it takes, VoxEU.org eBook, London: CEPR press.

Blais, A, D Bol, M Giani and P J Loewen (2020), “COVID-19 lockdowns have increased support for incumbents, trust in government, and satisfaction with democracy”, VoxEU.org, 07 May.

Brzezinski, A, V Kecht, D Van Dijcke and A Wright (2020), “Belief in Science Influences Physical Distancing in Response to COVID-19 Lockdown Policies”, Becker Friedman Institute (preprint), posted 30 April.

Chronopoulos, D K, M Lukas and J Wilson (2020), “Real-time consumer spending responses to the COVID-19 crisis and government lockdown”, VoxEU.org, 06 May.

Engle, S, J Stromme and A Zhou (2020), “Staying at Home: Mobility Effects of Covid-19”, Covid Economics 4. CEPR.

Kaplan, G, B Moll and G Violante (2020), “Pandemics According to Hank”, (preprint).

Painter, M and T Qiu (2020), “Political Beliefs Affect Compliance with COVID-19 Social Distancing Orders”, Covid Economics 4. CEPR.

Karlson, N, C Stern and D Klein (2020), “The underpinnings of Sweden’s permissive COVID regime”, VoxEU.org, 20 April.

Wang, H, Z Wang, Y Dong, R Chang, C Xu, X Yu, S Zhang, L Tsamlag, M Shang, J Huang (2020), “Phase-Adjusted Estimation of the Number of Coronavirus Disease 2019 Cases in Wuhan, China”, Cell Discovery 6(1): 1-8.

Wright, A, K Sonin, J Driscoll and J Wilson (2020), “Poverty and Economic Dislocation Reduce Compliance with COVID-19 Shelter-in-Place Protocols”, Becker Friedman Institute (preprint), posted 29 April.

4,724 Reads