The key role of mobility
We know for sure that if the average number of daily contacts between individuals declines by enough, then new COVID-19 infections can be brought down to manageable levels. We saw this clearly in the first infection wave in the spring when government restrictions, and increased caution on the part of individuals, caused a sharp decline in daily contacts that turned the tide on new infections. The lag between the imposition of lockdowns and new infections averaged around 16 days.
The theory behind this is fairly straightforward. The reproduction number – which determines whether new infections are rising, stable, or falling – is driven by three things. First, the average number of daily contacts between individuals in the population. Second, the number of days during which infected individuals can infect others. And third, the probability of infection when in contact with an infectious individual. The critical role of daily contacts – which we proxy by Google mobility data – can be seen clearly in the first COVID-19 infection wave, when a very sharp decline in Google mobility corresponded with a very dramatic decline in reproduction numbers across countries. (We use the retail and recreation component of the Google data as our proxy for daily contacts.)
Back in the spring, the initial decline in reproduction numbers was caused by the collapse in mobility. As time has passed, other developments have also weighed on reproduction numbers, such as mask wearing, increased hygiene, and test, trace, and isolate regimes. These developments allowed mobility to gradually recover over the late spring and early summer. Without these other developments, infection control would have required mobility to remain at the very depressed levels seen in the spring. Instead, there is an equilibrium level of mobility that, given all the other influences on the reproduction number, would keep the reproduction number at one or below. Clearly, European mobility exceeded that equilibrium during the summer, setting the scene for the second wave of infection.
Given this framework, the key question to be answered is how far mobility needs to decline to push the reproduction numbers across Europe back below one. Table 3 provides a way of answering that question for some of the Western European countries and the US. Due to the massive increase in testing in the first infection wave, we use hospitalizations to estimate the reproduction numbers rather than new infections. Unfortunately, this excludes France and Italy from the analysis due to insufficient data.
Table 3: Mobility and Reproduction numbers | | | | | | |
| Decline in mobility in first wave (%-pt) | Decline in R in first wave | Impact on R of a 10%-pt decline in mobility | Latest peak in R | Recent decline in mobility (%-pt) | Implied latest reading for R |
Germany | -63 | -2.2 | -0.35 | 1.49 | -27 | 0.55 |
Spain | -92 | -2.9 | -0.32 | 1.15 | -25 | 0.35 |
UK | -81 | -3.4 | -0.42 | 1.36 | -20 | 0.52 |
US | -49 | -1.7 | -0.35 | 1.30 | -4 | 1.16 |
Source: J.P. Morgan. 'R' refers to the reproduction number and is based on hospitalization data. | | | | | | |
| | | | | | |
The first three columns assess the impact on the reproduction numbers of the moves in Google mobility in the first infection wave. On average in the first wave, a 10%-pt move down in the Google mobility data depressed the reproduction number by 0.36pt. Meanwhile, the final three columns assess how much mobility needs to decline in the current infection wave to get the reproduction numbers back below one. It is important to recognize that reproduction numbers at the start of the second wave were lower than at the start of the first, due to a combination of mask wearing, increased hygiene, test, trace, and isolate regimes, increased self-shielding by the elderly, and some buildup of immunity in the population, so that mobility doesn’t have to fall as much as it did in the spring to gain control over the virus.
Our calculations suggest that mobility has already fallen enough in Germany, Spain, and the UK to get the reproduction numbers back below one. Interestingly all three countries have seen similar-sized declines in mobility. Given that the Google mobility data do not yet fully reflect the impact of the latest restrictions (they go through November 6), further falls in mobility are likely, which will put even more downward pressure on the reproduction numbers. Due to lags in the publication of the Google mobility data, it will take another week or so to see the full impact of the current measures. With reproduction numbers back below one, new infections across Europe should peak very soon.
In contrast to Western Europe, the reproduction number in the US appears still to be above one, and mobility has hardly declined at all in recent weeks. This suggests that new infections will continue to climb, with hospitalization pressure increasing and deaths rising further. It looks like a decline in mobility in the US is needed, which would have to come from either government restrictions or a shift in individual behavior. On the basis of our analysis, US mobility needs to fall by a further 10%-pts to push the reproduction number down to just below one. This cumulative move of 14%-pts would be around 29% of the decline seen in the spring. By comparison, the falls in mobility already seen as a proportion of the first wave move are 43% in Germany, 27% in Spain and 25% in the UK.
Evidence from the early movers
Our confidence in the view that Europe will ease lockdowns by early December rests partly on the theoretical framework outlined above, and partly on the experience of the early movers. Not everyone introduced restrictions at the same time and it is evident that some of the countries and regions that moved first have already seen a turn in new infections. On a regional basis, early moves were made in Liverpool and Manchester (UK), in Marseilles and Paris (France), and in Madrid (Spain). Among these, infections have already turned down in Liverpool, Manchester, and Madrid. In terms of countries, early movers such as Ireland, the Netherlands, and Slovakia have also seen new infections turn lower (Figure 3).
View attachment 238503
These experiences suggest that the restrictions put in place across Europe in recent weeks will be sufficient to reassert control over the virus, at least for now. Unless a vaccine is rolled out relatively quickly, further lockdowns may be needed next spring. In contrast to developments in Europe, it does not look like mobility has fallen by enough in the US to regain control of the epidemic. We are very much in an exponential phase in the US: daily new COVID-19 hospitalizations are already at 59% of the spring peak and the number of new hospitalizations is doubling every two to three weeks.