Throughout the analysis provided in this website, we aim to estimate the impact of the implementation of different containment policies on outcomes such as the growth rate of the number of deaths/cases related to Covid-19. While doing so, we implicitly rely on the hypothesis that the containment policies were respected and led to strong reductions of social contact and mobility. In this post, we check if this hypothesis holds by analysing data from the Google Mobility report (found here).

As explained in the data section, the Google Mobility report dataset shows how the daily number of visits and length of stay at different places changed relative to a baseline value defined prior to the start of the pandemic. In particular, the baseline value for each day of the week corresponds to the median mobility index in that day during the 5-week period bweteen Jan 3–Feb 6, 2020. Changes in the mobility index for each day after the start of the pandemic are expressed relative to the baseline value.

The dataset computes changes of mobility for the 6 following categories/places:

• Grocery/pharmacies

• Parks: local and national ones, public gardens, beaches

• Transit station: such as subway, train and bus stations

• Retail and recreation: restaurants, cafés and bars, museum, movie theaters, ...

• Residential areas

• Workplace

The definition of these categories/places vary from region to region, so the following analysis should be read with that caveat in mind.

First, we report changes in the mobility index from 15th of February to mid May in five broad geographical areas: Africa, Asia, Europe, Latin America and the Caribbean (LAC), North America (USA and Canada), and Oceania.

Except residential areas, all other categories show a strong decline in mobility around mid-march. However, each category shows a different pattern:

• Workplace areas: In the six continents, we observe a strong decline of the mobility index around mid-March (which correspond to the date of lock-down in most countries that have implemented this type of policies). From mid-April onwards, mobility seems to slowly return to baseline level. This can be explained by the gradual reopening of some sectors of the economy (such as construction for instance).

• Residential areas: We see a strong increase in the mobility index for residential areas from mid-March, which corresponds to the implementation of lock-down policies. This index appears to slowly decrease from the beginning of May.

• Retail/recreation areas: This category is the one that has suffered the biggest drop. In Europe ,the reduction of mobility was close to 80% relative to baseline.

• Grocery/pharmacies: There is an increase of the mobility index for this category around mid-March (at the beginning of the lock-down). This might be explained by the initial panic and the desire of people to stock up. After the peak, the mobility index falls below the baseline, but the decline is lower relative to other categories. Groceries and pharmaceutical products are necessary goods during the lock-down.

• Parks: the figure shows large discrepancies across continents. In North America and Europe there is a slight and temporary drop in mobilty, while in LAC and Asia the fall is sharp and sustained. Seasonal effects migh be partly responsible for the trends.

• Transportation: The mobility index in transit stations appears to follow the same pattern as the mobility index in the workplace. This is expected since the vast majority of the population use public transportation to go to work.

We now analyze if there is a link between the timing of the implementation of the containment policies and changes in mobility. Our goal is to check if changes of mobility are mostly due to the implementation of containment policies. The analysis is restricted to the categories that show a similar pattern across continents: workplace, residential areas, retail/recreation and transit stations.

The impact of the stay-at-home policy on the mobility index of the 4 categories is clear: there si a significant decline (or rise in the case of the residential areas) immediately after the implementation of a required stay-at-home policy. We also observe a slight rise of the mobility index (or decrease in the case of residential areas) after 35 days of lock-down. This can be explained, as stated above, by the partial and progressive reopening of some economic sectors.

The second observation is that the reduction of mobility began before the implementation of the stay-at-home policy. The mobility index had a strong decline (or increased for residential areas) during the two weeks preceding the implementation of the policy. The patterns suggests that these type of policies are not uniquely responsible for the observed decline in mobility. Social awareness of the epidemic because of the rapid spread of information led people to start social distancing even before mandatory lock-downs.