Germany COVID-19 impact: alternative point of view
German Statistisches Bundesamt had announced ad-hoc evaluation of mortality figures for 2020/2021 which show that COVID-19 impact was huge. I’d like to challenge their point here.
Let me start from representing used data by Statistisches Bundesamt in table view with aggregation by year:
Let me draw it as three independent graphs from 0 to 50, 50 to 70, and 70+
As you can see only death statistics are quite wired. Age group 35–40 had its decrease and increase, 45–55 are decreasing, etc. Why? Because the population is aging. Special in developed countries.
Thus, if you present the population in Germany via a population pyramid it isn’t a Stationary pyramid. A lot of different things impacts it: World War II which was killed some peoples and forced a baby boom after it was over, huge depends on migration, declining fertility rates, rising life expectancy, etc.
The official population pyramid for 2019 without any projection is:
Fortunately, statistics about populations ages are available, it is actual and has good resolution. This allows me to create a table for the same age group which I’ve used in death statistics:
and draw it as three independent graphs from 0 to 50, 50 to 70, and 70+:
You may see that the next generation or up to 15 years and 30–40 stay constants. I guess by migrants and their kids. 15 to 30 and 40 to 55 is decreasing, but 80+ is increasing.
I guess it is clear that the population in Germany has quite a dynamic structure.
People die and older people die with a bigger probability. Nobody dies from old age, usually, the cause of death is illness or accidents.
My point? Using only death statistics without taking into account the size of the group isn’t correct. Why? Because for a bigger group more people die in the same timeframe. Nobody tries to make panic from the fact that a number of deaths in the year in Germany significantly higher than in Finland. Why? Because Germany is home to more than 10 times more peoples.
Same here. If some group is bigger when it is expected that more people die in next year or five.
So, let me join both tables to compute the percent of the population in Germany per age group who is died in the next year for 2016–2020. How? I do have population age groups at 31 December before of each year and it is quite easy to compute:
as usual I’ll draw it as three independent graphs:
Anyway, this data has a least a few sources of noise:
- when someone was at the upper bound of age group and died in the next year after his birthday, he is reported as dead at the next group;
- I haven’t taken into account migration;
- I haven’t taken into account medical tourism that may have some impact on death statistics.
Anyway, it allows me to conclude number of deaths in 2020 by comparing it with 2016–2019:
- it was slightly increased (~0.0025%) for group 35–45;
- it was increased (~0.05%) for group 70–75;
- it was greatly increased (~0.15%) for group 85+;
- it was decreased for all other groups.
The greatly increase (~0.15%) can be explained by a lot of things that happened in the last 85+ years. Which included not only Germany but the Soviet Union also. Why? Because a lot of peoples was ethnic German and immigrated to Germany only after 1991 as German citizens.
I guess I made clear that COVID-19 hasn’t got such a significant impact on death statistics in Germany. Or at least I can’t say that increased death for whole year for group for near 0.15% is huge. Why? I have no idea. The virus might not be so deadly as its narrative. Is it possible that Government can hold it? I can’t nor dismiss nor accept that with these numbers.
This method can be used to measure Government impact. But it required detailed population ages for a week, month, or quarter which aren’t available.
It also very interesting to see dynamic not for only 5 years, but since 2010 or 2000 but death by age is available only since 2016.
Yes, this is another example of Simpson’s Paradox and you can read also a good example of such paradox in taxes here.
