Measuring the impact of COVID-19 confinement measures on human mobility using mobile positioning data. A European regional analysis

Abstract

This work presents a mobility indicator derived from fully anonymised and aggregated mobile positioning data. Even though the indicator does not provide information about the behaviour of individuals, it captures valuable insights into the mobility patterns of the population in the EU and it is expected to inform responses against the COVID-19 pandemic. Spatio-temporal harmonisation is carried out so that the indicator can provide mobility estimates comparable across European countries. The indicators are provided at a high spatial granularity (up to NUTS3). As an application, the indicator is used to study the impact of COVID-19 confinement measure on mobility in Europe. It is found that a large proportion of the change in mobility patterns can be explained by these measures. The paper also presents a comparative analysis between mobility and the infection reproduction number $R_t$ over time. These findings will support policymakers in formulating the best data-driven approaches for coming out of confinement, mapping the socio-economic effects of the lockdown measures and building future scenarios in case of new outbreaks.

Keywords: COVID-19; Confinement measures; Coronavirus; Human mobility; Mobility data.

Fig. 1

Left: Mobility indicator construction. The matrix is the original ODM at higher level of spatial granularity than the indicator. The values of the orange cells are aggregated to obtain the internal mobility indicator, the blue cells are aggregated into the outwards indicator, and the green cells into the inwards indicator. Right: Mobility indicator construction when no values for the main diagonal of the ODM are available. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Mobility indicator for four NUTS3 areas in Spain (columns). The rows show (from top to bottom): $\mathit{internal},\mathit{inwards},\mathit{outwards}$, and total indicator. Read vertical line marks 15 March, when the national lockdown became effective. The red dots indicate Sundays. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

Change in mobility between 28 February and 3 April (left) and between 28 February and 29 May (right).

Fig. 4

Normalised total mobility indicator at country level. The indicator has been smoothed with a 7-day moving average.

Fig. 5

Normalised total mobility indicator for Italy (7-day moving average) with key confinement measures marked.

Fig. 6

Normalised total mobility indicator for France (7-day moving average) with key confinement measures marked.

Fig. 7

Adjusted $R^2$ for the models evaluated (horizontally). One boxplot is computed for each model based on the distribution of $R^2$ values for the different countries (coloured dots). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Fig. 8

Comparison of the mobility indicator (dashed line, left hand axis) in Italy and four different estimates of $R_t$ (solid lines, right hand axis).