British physicist Lord Kelvin liked numerical measures: “ … when you cannot express it in numbers, your knowledge is of a meagre and unsatisfactory kind … ” To which one modern statistician responded: “and even when you do express it in numbers, your knowledge often remains meager and unsatisfactory.”
This is the case with a lot of COVID metrics, including “excess mortality,” a demographic measure that some hope eventually will settle lots of present disagreements about the real medical effects of the virus. It is a very useful tool, but one needs to understand its limitations.
Excess mortality is “the number of deaths from all causes during a crisis above and beyond what we would have expected to see under ‘normal’ conditions.” It is a measure, albeit always somewhat after the fact, of the total death toll from, for example, a war, epidemic, famine or other large-scale disaster. It is a measure through which some current COVID controversies may eventually gain clarity — such as whether the same criteria are being used across different states or nations to report infections and fatalities from the virus.
Understand, however, that taken very simply, excess mortality conclusions are examples of the “post hoc fallacy” — which philosophy and econ profs warn students about. The fallacy is to assume that because one thing happened before something else, the first caused the second: Lotterman went to work for the Minneapolis Fed and our nation had eight years of prosperity and high employment. He moved on and we were in recession less than two years later. Cause and effect? No. Ridiculous? Yes.
So deaths in general worldwide have gone up sharply during COVID. Many are directly due to people contracting the virus. However, there are other factors, maybe related, maybe not. Domestic violence and murders went up. Addiction overdose deaths rose. And some people died of cancer or heart problems — maybe because COVID-overcrowded emergency rooms, real or perceived, deterred people from seeking or receiving help for chest pains or possibly cancerous lumps; or maybe these people would have died anyway. However, during this same period, commuting and vacation travel deaths went down.
Trained researchers have ways of teasing this all out, but the layperson should be cautious about comparing global deaths in 2019 with 2020 and concluding that the entire difference was COVID related.
More generally, researching the social and economic effects of disastrous events like pandemics, wars, great floods and famines, often means separating deaths by gender and age.
The “population pyramid” is a useful graphical tool that separately shows the number of males and females in a series of age cohorts, typically 0 -5 years, 5-10, and so on up to “95 and older.” The pyramid is two bar graphs set on end with bases touching and bars going in opposite directions for the two genders.
The graph is a snapshot in time that usually looks like an evergreen tree — male and female populations on either side. A rich county with low birth rates and low death rates would have steep sides. A poor country with a high birth rate and still high death rates would look like a broad-set pine, very wide at the bottom but tapering in rapidly. World War I caused a big divot in the male side of cohorts in their late teens and 20s, with little effect on women. As the time snapshot advances through the years, the divot of male deaths shifts up as that cohort ages.
The U.S. “birth dearth” during the Depression first looked like the tree had been put on a narrow stand. As those people aged, subsequent graphs show this divot on both male and female sides moving upward. The “baby boom” for the 18 years starting in 1946 showed up as the opposite, a greatly broadening base that advances up in later snapshots.
The once-broad baby boom base is now near the top, but as the general population grew in the following five decades, the base remained quite broad even though the number of children born slid. Now, in terms of “natural increase,” we are near replacement levels. Except for immigration and the higher fertility of immigrant women and their daughters, the U.S. population will shrink.
Epidemiologists and demographers are now preparing to study COVID’s excess mortality, just as they had looked back at the 1918-20 Spanish Flu epidemic. There will be more clarity about many COVID issues, including whether different state or national governments took steps to diminish or inflate the number of officially tabulated infected cases and deaths, possibly for political purposes.
This works in countries, like ours, with reliable “vital statistics” — birth, death and illness numbers. But it also helps if there are no simultaneous events that muddy the waters. Consider the world a century ago. North America and Western Europe had kept good records. But there was political and social chaos further east, as shifting borders, ethnic cleansing, the dismemberment of the Austro-Hungarian empire and the Russian Revolution and civil war raged. Keeping track of who died, when and where was a low priority. Hundreds of thousands of intentional killings happened, millions died from malnutrition or other diseases like typhus among refugees. These all also occurred at the same time as the new, deadly influenza.
So accurately breaking down the enormous excess mortality for those years into component parts is difficult at best for many nations. Moreover, going to the Central Asian countries from Turkeys’ eastern border out to China, vital stats records often did not exist at all. Historians of the pandemic write things like “ … and 10 to 25 million deaths in Central Asia.” One hopes we do better than this with COVID, but there will be problems
There are many more examples of how historic events like pandemics or warfare drastically affected economic and social structures for many decades. These remain for another day.
St. Paul economist and writer Edward Lotterman can be reached at stpaul@edlotterman.com.
