As the Covid-19 spreads around the northern hemisphere, one wonders what effects this pandemic will have in lower and middle income countries in the southern hemisphere, which are currently experiencing the end of the summer.  Researchers previously believed that warmer weather retarded the progress of viruses, yet the emergence of the Middle East Respiratory Syndrome (MERS) in Saudi Arabia in the early summer of 2014 demonstrated that temperature and viral transmission need not be inversely proportional.  Adopting a more long-range view may yield greater insights and researchers have been looking for evidence from the so-called Spanish Flu pandemic of 1918 for guidance. We have familiarity with the repercussions of the 1918 Spanish Flu on demographic and human capital outcomes in the State of Sao Paulo, Brazil, where the State Government collected reliable health statistics every year at district or municipal level and the lessons then are still relevant today (see our NBER working paper on the topic here).

The experience of Sao Paulo in 1918 provides three relevant insights for today. First, geography and climate are ineffective barriers in an interconnected world.  Even if government and health officials in Sao Paulo in October 1918 had the advantage of enjoying milder weather conditions than in the United States and even if they had learned that best practices to prevent diffusion of the flu included quarantining people, shutting public spaces such as movie theatres and suspending large-scale events to limit transmission, still there was little to stop the flu when it arrived in the end of that month. Carried by soldiers returning from the First World War and arriving by ship in the port city of Santos, Brazil, the Spanish Flu quickly spread inland and wreaked havoc in large areas of Sao Paulo state affecting a large percentage of the population between October 1918 and the early weeks of 1919.  Estimates predict that it killed tens of thousands of people (approximately 36, 400 in 1918 alone). In the city of Sao Paulo (the state capital), the disease caused 5,331 deaths between mid-October and mid-December 1918, infecting up to 350,000 people, that is, two-thirds of the population of the city (i.e., there was no bending of the diffusion curve despite the warnings) (see Figure 1, Panel A).

Second, without proper training and methods to diagnose the flu, authorities then (just like today) were prone to make significant errors, under-estimate the morbidity and mortality rates, minimize the impact of the pandemic, and facilitate its diffusion. In Sao Paulo state, in 1918, the precariousness of the health centers led local medical authorities to misdiagnose cases and causes of death, thus thwarting measures to reduce the spread of the disease. In the health reports of the time, we noticed that the number of deaths from respiratory diseases spiked in 1918, but so did the number of deaths from “unknown causes,” most likely because of the difficulty of diagnosing people with expediency and/or accuracy. In fact, after adding the mortality rates from respiratory diseases and unknown causes, we estimate an overall mortality rate of 0.099 per thousand people in the districts outside the two major metropolitan areas at that time—a rate comparable of that of the worse-hit cities in the United States in 1918 (see Figure 1 Panel B). Because this is an average overall mortality rate, it is possible that this rate was higher in localities with lower socio-economic status.

Third, the long term consequences of the disease were worse for females than males. Census data from the period 1920-1940 in Brazil show that the places relatively more affected by the flu pandemic in 1918 ended up with a significant increase in short and long-run literacy rates for males (aged 15 and above), but not for women. That means that, in a way, women bore the immediate and future brunt of the pandemic. There are at least three explanations for this. Either, prime-age men died in greater numbers from the flu or male fetuses were relatively more susceptible to the health shock in the aftermath of the 1918 pandemic, i.e., leaving relatively stronger male children as survivors.  Finally, and more worrisome, it could also be that families chose to invest in males/boys at the expense of females/girls in these contexts in which labor markets favor men. This is consistent with the findings of John Parman in his study of the United States and in a recent paper on the effects of the tsunami in Indonesia.

In summary, we believe the lessons are three-fold. First, a flu pandemic can have dire consequences for lower and middle-income countries even if they have geographical and weather-related conditions that in theory should act as a bulwark. In contexts where basic public health infrastructure is either absent or of poor quality, the repercussions of Covid-19 are likely to develop rapidly.  Second, given the lack of training and accurate tests to diagnose the new flu strain, absence of sanitation and hygiene which is further amplified by lack of knowledge and awareness, cities and public transport facilities that are over-crowded, populations that are under-nourished and thus more susceptible to disease, and lack of quality health statistics and health management systems that afford accurate monitoring of health emergencies as they develop, the spectrum of challenges in the developing world may range from misreporting of cases to widespread collapse of the limited existing public health infrastructure as hospitals and clinics are overwhelmed with a rising tide of those succumbing to Covid-19.  Quarantining large sections of the population as China successfully managed to do seems harder to implement in environments where if nothing else, freedom is slightly more widespread. Finally, health shocks of this nature may have short and long-run detrimental effects that are gender-disaggregated.