COVID-19 is spreading around the world and has reached the level of a global pandemic. First estimates suggest that together with the large numbers of deaths, there will also be large economic consequences of orders of magnitude seen with the global financial crisis or even higher (Baldwin and Weder di Mauro 2020a, 2020b). Scientists from all disciplines are putting their expertise to use to combat the consequences of the pandemic. Our research (Bayer and Kuhn 2020) aims at highlighting the role of social networks in spreading the virus to most vulnerable strata of the population, which is responsible for a large number of deaths and the risk of overwhelming local health systems.
Our starting point are the large differences in case fatality rates (CFRs) across countries. As of 12 March, Italy had one of 6% while countries like Norway, Denmark, Sweden, and Germany have rates still close to zero (top panel of Figure 1, which shows countries with at least 100 cases as of 12 March). These differences persisted as of 15 March (bottom panel of Figure 1, which shows countries with at least 200 cases as of 15 March).
Figure 1 Case fatality rates across countries
a) Countries with at least 100 cases as of 12 March
b) Countries with at least 200 cases as of 15 March
The rates may converge over time as the virus spreads, but they may also diverge as heath systems get overwhelmed by the escalating crisis. What can we learn from these data to help organise our lives in a way to minimise the risks to us and the risks we pose to others?
It has been established now that, unlike the Spanish Flu (Taubenberger et al. 2001), COVID-19 is particularly deadly for the elderly (Dowd et al. 2020) and creates a disproportionate need for intensive care in this group. Medical research and treatment are front and centre in combatting this crisis, and as scientists we believe in the power of medical research and science. But can we, as social scientists, also contribute to combatting the crisis?
The question we are interested in is: why are more elderly are infected in some countries than in others? Existing research has already emphasised the importance of social contact for the spread of infectious diseases (Mossong et al. 2008 and, on the current crisis, Liu et al. 2020, Bi et al. 2020). The two latter studies explicitly look at within-household transmission and suggest that very close contact, for example among household members, is of particular importance. Mossong et al. (2008), in contrast, shows (for a limited sample of countries) large differences in contact between the non-working age population with working-age persons, while within working-age contact is rather similar across countries. Our hypothesis is that differences in social interactions and social networks play a key role in explaining the cross-country differences in mortality during the early phase of the coronavirus outbreak.
The reasons for differences in social structures are complex. They can stem from cultural or institutional differences, such as the prevalence and affordability of childcare facilities, laws on not leaving children unaccompanied, the labour market and economic situation for young workers across countries, or the scarcity of housing. In the light of current policy measures – in particular, school closures – taken around the world, social structures might quickly reshuffle if grandparents move in with or visit their grandchildren to accommodate families’ need for childcare. Understanding better how such intergenerational interaction relates to CFRs is therefore a key and pressing concern for policymaking.
Our idea for the relationship between social interaction and the spreading of the virus is simple. Suppose that in country A, almost all interaction is within one group of people – i.e. working age people interact mostly among themselves and a second group of people, the elderly, do the same (top panel of Figure 2). At the same time, in country B, interaction is often across generations. The young and the old live together and interact – for example, for childcare – or young workers still live with their parents as they cannot afford to live on their own (bottom panel of Figure 2). If COVID-19 has been imported to Europe through work-related travel, then a country of type A should see an initially much more contained outbreak, with much less need for intensive care and many fewer fatalities relative to the size of the outbreak. This correlation is what we explore in our analysis.
Figure 2 Stylised social structures
Given this hypothesis, how can we operationalise this idea? We turned to the World Value Survey data (http://www.worldvaluessurvey.org/wvs.jsp) and calculated from this source the share of people between age 30-49 who live with their parents. Figure 3 shows that this share varies dramatically across countries. From shares below 5% in countries like France, Switzerland, and the Netherlands, to cases like Japan, China, South Korea, and Italy with shares above 20%.
Figure 3 Share of 30 to 49 year-olds living with their parents across countries
If we take the data from Figure 3 as measure of intergenerational interaction (how many red arrows there are), then already a simple figure highlights our key idea. Figure 4 (top panel) shows the CFR for all industrialised economies with more than 100 cases (as of 12 March). For the bottom panel of Figure 4, we updated our data to 15 March. There we also take into account that public health systems started to become overloaded and that this likely increased CFRs further. We therefore restricted our sample to the period when the total number of diagnosed cases was still below 5,000 in the respective country. Now, only a few days later, further countries had large case numbers and therefore entered our sample – yet the picture remained the same. We also refined the specification and allowed for a different relationship for the East Asian countries, because these countries differ from the rest not only in the cultural dimension but also in their preparedness for the outbreak (for example, through the presence of fever clinics). In this case, we find the same correlation between social interaction and CFRs within the two country groups (green and light blue line).
Figure 4 CFRs and intergenerational interaction
a) 12 March
b) 15 March
Both figures support our key hypothesis that in countries with more intergenerational interactions, CFRs are initially higher. These differences do not necessarily persist, as further developments will depend on the policy measures taken and the capacities of the public health systems across countries. We present detailed regression tables in our paper.
What we conclude from this analysis is that the structure of social interactions matters for the case fatality rates for this outbreak, and that social distancing needs to focus particularly the elderly.
This effect will go away as soon as the virus finds its way into the elderly populations and can spread within this group because the elderly are not a entirely socially disconnected set of people. Those countries with low fatality rates, such as Germany, should take this as a warning sign. Unfortunately, it is likely the low initial fatality rates are not here to stay once the virus spreads.
At the same time, we hope this little piece of data analysis helps us to better understand how pivotal it is to keep the elderly uninfected and what role social networks and links play in this. It may also provide a warning sign for those countries where the elderly and the young live close together on how important it is to contain the virus there early on. Figure 5 shows that countries within Europe that are at risk include Serbia, Poland, Bulgaria, Croatia, and Slovenia.
Figure 5 Share of 30 to 49 year-olds living with their parents
Baldwin, R and B Weder di Mauro (2020b), Economics in the Time of COVID-19, a VoxEU.org eBook, CEPR Press.
Baldwin, R and B Weder di Mauro (2020b), Mitigating the COVID Economic Crisis: Act Fast and Do Whatever It Takes, a VoxEU.org eBook, CEPR Press.
Bayer, C and M Kuhn (2020), “Intergenerational ties and case fatality rates: A cross-country analysis”.
Bi, Q, Y Wu, S Mei et al. (2020), “Epidemiology and transmission of covid-19 in Shenzhen China: Analysis of 391 cases and 1,286 of their close contacts”, medRxiv.
Dowd, J B, V Rotondi, L Andriano et al. (2020), “Demographic science aids in understanding the spread and fatality rates of COVID-19”, mimeo, Leverhulme Trust.
Liu, Y, R M Eggo and A J Kucharski (2020), “Secondary attack rate and superspreading events for sars-cov-2”, The Lancet.
Mossong, J, N Hens, M Jit et al. (2008), “Social contacts and mixing patterns relevant to the spread of infectious diseases”, PLoS Medicine 5(3).
Taubenberger, J K, A H Reid, T A Janczewski and T G and Fanning (2001), “Integrating historical, clinical and molecular genetic data in order to explain the origin and virulence of the 1918 Spanish influenza virus”, Philosophical Transactions of the Royal Society of London. Series B: Biological Sciences 356(1416): 1829–1839.