Because of how final and unmistakable death is, death reports for the very same region should be pretty close to one another. Death, because of how obvious it is to investigators, is one of the easiest statistics to report on.
There is never a time when investigators are in “disagreement” about whether a dead body represents a dead person or not. Much of the reporting of death statistics is not that much more complex than simply counting body bags.
When wide discrepancies are found in death reports, it can help to see if one party had been making unrealistic assumptions, while the other party was staying true to the actual situation on the ground (ie., the actual count of the dead).
According to the World Bank, the year-to-year increase in crude death rates for 2019 to 2020 was higher than 18.2% in 8 regions of the world. According to the Human Mortality Database, the number of regions with that much excess death was 22 regions.
It could easily be the case that some of the recorded change in death counts was “expected” — so that even if the deaths increase year-to-year, it would not be considered excess death unless it was unexpected.
Alternatively, any death rate increase at all may get recorded as being unexpected — as when no increase was expected when examining recent trends.
Trends matter.
But trends don’t explain everything (eg, Armenia)
But the magnitude of disparity between the Human Mortality Database and the World Bank seems insurmountable by such adjustments. An example is Armenia.
According to the World Bank, the death rate in Armenia fell from 9.835 per 1000 in 2019 to 9.801 per 1000 in 2020 — yet the Human Mortality Database reports 40% excess death in 2020.
How can crude deaths be falling while excess deaths are 40% more than expected?
The answer, of course, is the senseless statement that crude deaths were expected to fall by more than 40% in a single year alone (in 2020) and — because they didn’t — it’s counted as an increase in mortality.
A small decrease in death is counted as an exceptionally large increase in excess death?
Another example is Bolivia
Bolivia reported the exact same death rate to the World Bank in 2019 and 2020 — yet the Human Mortality Database says that excess death in Bolivia in 2020 was 43%.
What must that mean?
That must mean that deaths in Bolivia were expected to fall, in one single year, by 43%!
Another example is Peru
According to the World Bank, the crude death rate in Peru in 2020 as 1.5% higher than it had been in 2019, but the Human Mortality Database says that excess death in Peru in 2020 was 80%.
That’s a 78-percentage-point swing.
The discovery of inconsistency is not necessarily a discovery of guilt or even of incompetence, but when disparaties this wide exist it is fruitful to discover how they came about.
The notion of crude death rates being expected to change by over 40% in a single year is not very credible. The standard deviation of annual crude death rate changes worldwide is something around 7% or so.
An expectation of a 42% change in death is then equivalent to an expected “6-standard-deviation” swing in death rates — something known as a “Six-sigma” event. It just doesn’t make sense to be expecting that much change in death in a single year.
Reference
[Cumulative P-scores using projected baseline] — OurWorldInData [Human Mortality Database data]. https://ourworldindata.org/excess-mortality-covid
[Crude death rates by nation and year] — World Bank. https://data.worldbank.org/indicator/SP.DYN.CDRT.IN