42 million dead?

Discussions on all aspects of the USSR, from the Russian Civil War till the end of the Great Patriotic War and the war against Japan. Hosted by Art.
Art
Forum Staff
Posts: 6196
Joined: 04 Jun 2004 19:49
Location: Moscow, Russia

42 million dead?

Post by Art » 29 May 2017 15:47

This topic is a continuation of the discussion which started here:
viewtopic.php?f=79&t=227772&start=15

The discussion is about an article published online by I. Ivlev in March of this year:
http://www.soldat.ru/news/1068.html
The article generated much reaction both in Russian and foreign media, examples:
https://www.novayagazeta.ru/articles/20 ... yaet-schet
https://www.bloomberg.com/view/articles ... llion-dead
and was even discussed in the Russian Parliament.
The principal point made by the article is that Soviet population casualties in 1941-45 were about 42 million men, including 20 million military casualties which is considerably more than it was believed thus far. Below we will examine this statement in detail.

Art
Forum Staff
Posts: 6196
Joined: 04 Jun 2004 19:49
Location: Moscow, Russia

Re: 42 million dead?

Post by Art » 29 May 2017 16:25

Total population losses.

The official figure of Soviet population losses (26.6 million) was made public first in 1990. It was estimated based on demographic calculations described in "Soviet casualties and combat losses...." ed. by G. Krivosheev and in much more details by Andreev, Darskiy and Kharkova (or ADK) in a Russian-language book "Population of the Soviet Union in 1922-1991" (1993) which is somehow a standard source on Soviet demographics. This method uses the following figures:
Soviet population on 22.06.41 = 196.7 million (estimated from the 1939 census and data of births and deaths in 1939-1941).
Soviet population on 1.01.1946 = 170.5 million (estimated from the 1959 census and data on births, deaths and migration in 1946-58)
Of them born before 22.06.41 = 159.5 million (estimated from the 1959 census and data on births, deaths and migration in 1946-58)
Population deficit for ages born before 22.06.41 196.7-159.5 = 37.2 million
"Normal" number of deaths expected in 4.5 years (22.06.41-1.01.46) = 11.9 million (estimated from pre-war stats on mortality)
Excess deaths for ages born before 22.06.41 37.2-11.9 = 25.3 million
Excess deaths for children born 22.06.41-31.12.45 = 1.3 million (estimated from war-time data on births rate and pre-war children mortality rate)
Total population losses 25.3+1.3 = 26.6 million

It must be stressed that this method utilizes a very inclusive definition of "population losses" which is defined as "population deficit" (total number of deaths + net foreign migration during the war) - "normal" number of deaths during the war (expected from a pre-war level of mortality). This definition is different from a simple decrease of population, albeit the latter is almost the same when expressed in numbers (196.7-170.5 = 26.2 million people).

Let's see what is different in the Ivlev's article. Ivlev has the following calculation:

Soviet population on 1.07.41 - 205 million (estimated by the author using docs of the Soviet statistical office)
Soviet population on 1.07.45 - 169.8 million (estimated by the author using docs of the Soviet statistical office)
Number of births in four years (1.07.41-1.07.45) = 17.6 million (estimated by the author from the records of births in the year 1941)
Normal expected number of deaths in 4 years = 10.8 million (estimated by the author from the records of deaths in the year 1941)
Total number of deaths in 4 years = 205+17.6-169.8 = 52.8 million
Population loss (excess deaths) = 52.8-10.8 = 42 million

It is obvious that Ivlev uses the same definition of population loss and almost the same method of calculation as ADK but simply pastes different numbers. So to understand why there is a difference we must examine how he had arrived to them.

Stiltzkin
Member
Posts: 1083
Joined: 11 Apr 2016 12:29
Location: Germany

Re: 42 million dead?

Post by Stiltzkin » 29 May 2017 16:30

To this can be added,
Counting Soviet Deaths in the Great Patriotic War
https://www.jstor.org/stable/3594591?se ... b_contents
http://sovietinfo.tripod.com/ELM-War_Deaths.pdf

death probability (excess war risk) = k*(1-d)
combined mortality risk = d+k*(1-d)
Total wartime mortality risk m= d+k-d*k
normal death conditions = d*P
thus the lower bound (method by Goskomstat) k*(1-d)*P, 24 million
normal risks absent = k*P
d*(1-k)*P = upper bound of war excess deaths. 25.8-26.6 million
War mortality risk in the absence
of normal causes, k=(m-d)/(1-d)

If you include Net Migration (2.4-2.7 million) from the total value of 196.7 million (with 16.4 million wartime births), 194 as the prewar population you will land at approx. 159 million post war.

Sokolov argued that the numbers were inaccurate but his figures are mostly concerning excess and total deaths (42-46).
The only problem I have is still the number of military deaths in relation to civillian deaths.

Art
Forum Staff
Posts: 6196
Joined: 04 Jun 2004 19:49
Location: Moscow, Russia

Re: 42 million dead?

Post by Art » 29 May 2017 18:20

Everything in due time.
Ok, about pre-war population. Ivlev strongly criticizes ADK for their estimates of the pre-war Soviet population. In particular he makes a points that there are "absolutely accurate" (as he says) stats on population from the files of the Soviet office of statistics which make any indirect estimates redundant. He refers a single document from the statistics office's files in the Russian State Archive of Economy which gives a Soviet Union's population as of 1 July 1941 equal to 199.9 million. Then follows an important step. Ivlev says that this document counts civilians only. It must be said that it's his own assumption since there are no references to "civilians" in the original document. He advances two points to support this assumption:
1. Civilian statisticians could have no access to top secret data on military personnel and therefore only counted civilians
2. During the census of 1939 the armed forces and NKVD made a count of their personnel which was separate from a "normal" civilian census. Ivlev assumes that the practice in 1941 must be the same.
As a result about 5 million military personnel (a number known from secondary sources) is added to the totals from the document, and the sum is almost exactly 205 million.
As one can see the crucial issue here is a nature of statistical documents. Does it counts only civilians or the entire population. First let's examine Ivlev's point

Point 2. Indeed there was a separate count of military personnel in the 1939 census. However the results were included in the official totals (170.5 million people in January 1939). In other words the census result included both civilians and military. So in principle the figure from 1941 could include both as well. Then contrary to what Ivlev says during the census NKVD counted not only personnel of the NKVD troops but also various NKVD administrative and police personnel, associated civilians (family members, as far as I understand), prisoners in camps and prisons etc for a total number of 3.7 million people. Following Ivlev's logic all those categories must be added to his 205 million number. One can't understand why he stops halfway here.
Point 1 is more interesting. Apparently Ivlev believes that statisticians regularly counted civilian population in the same manner it was made in the military - company sergeant mustering his company every day and calling soldiers against his roll and then passing the report to superiors. It was not how statistics of population worked. In fact civilian statisticians had the 1939 census as a reference point and then projected a current population from this point. So if they wanted to estimate the population in the year 1940 they took the population in 1939 added the number of birth and subtracted the number of deaths during the year. And births and deaths were covered by civil registration. You can see that on military strength was simply nod required to that kind of math. So this point also not conclusive.

An important consequence relates to accuracy of statistical count of population. Ivlev says that it was more accurate than indirect method (projection from the 1939 census). In fact this count was based exactly on this kind of projection as described above. Moreover if we talk about the Soviet situation in 1941 this count also included a population of territories annexed in 1939-40 which were not covered by the January 1939 census. This population was an unknown value and was approximated with a more or less realistic guesstimate. So in fact what population data the Soviet statistics office had in 1941 were not a direct and accurate count but more a value estimated using certain assumptions.
to be continued

Art
Forum Staff
Posts: 6196
Joined: 04 Jun 2004 19:49
Location: Moscow, Russia

Re: 42 million dead?

Post by Art » 29 May 2017 18:47

Continuing with the pre-war population.

Although it was not mentioned by Ivlev some information about estimates of Soviet Union's population made by statisticians before the war was available long ago. As far as I know the first publication of this kind was an article by V. Kozhurin in the Soviet Military History Journal (No. 2 from 1991). Kozhurin discoverer several documents submitted by the Soviet statistics office on 16.06.1941. These documents give the following figures for the population:
January 1939 (within old borders) - 170.5 million
January 1939 (with annexed territories) - 192.3 million
January 1940 - 196 million
January 1941 (preliminary) - 198.7 million
As Kozhurin says based on the rate of population growth one can expect the population to reach 200.1 million by July 1941.

What looks striking is that both figures 199.9 (Ivlev) and 200.1 (Kozhurin) are very close. They are practically the same given the accuracy of estimates. So we must conclude that the document published by Ivlev belongs to the same set of data as documents found by Kozhurin. Then the January 1939 figure from Kozhurin is the same as official census results. From here (an also from a common sense) we can make a conclusion that all those documents pertain to the total population. Hence the document found by Ivlev gives the total population not civilians only as he claims. Hence his addition of 5 million military personnel is redundant. Or another way to look at it: if Ivlev is right and the real population on 1.7.41 was 205 million then there must be a growth by about 13 million people in 2.5 years (January 1939 - June 1941). Such a rapid growth would be in a complete disagreement with registered numbers of births and deaths from which the annual growth was about 3 million (or some 7-8 million in 2.5 years).
So it can be stated quite conclusively that Ivlev's estimate of the pre-war USSR population is erroneous.

Art
Forum Staff
Posts: 6196
Joined: 04 Jun 2004 19:49
Location: Moscow, Russia

Re: 42 million dead?

Post by Art » 29 May 2017 19:12

Finally there is a natural question. Ok, according to the statistics office Soviet population in June 1941 was 199.9 or 200.1 million. Why Andreev and coauthors take a much smaller figure (196.7 million)? The answer consists of several parts:
1. ADK believe that the official results of the 1939 census were exaggerated for political reasons. In fact that is a popular opinion shared by other researchers (I'm not going to discuss arguments in support of it here). ADK put a Soviet population in January 1939 about 2 million smaller than census results - 168.5 million vs. 170.5. This estimate at least fits well with results of the abortive 1937 census.
2. Population of annexed territories was an unknown value as it was not covered by the census. Apparently pre-war estimates were 21.8 million (as of January 1939). ADK say they were exaggerated and put it at 20.3 million or 1.5 million smaller.
3. A related factor is a border change. ADK calculate population within post-war borders. Between 1941 and 1945 a Białystok and Przemyśl regions were given back to Poland (estimated population 1.4 million), Tuva republic joined the Soviet Union (population about 0.1 million) as well as Transcarpathia (0.7-0.8 million). Also Konisgberg region and Sakhalin were annexed in 1945, the original population almost completely migrated to Germany and Japan and was not apparently included in the population stats. So the difference between 1941 and 1945 borders is some 0.5-0.6 million men. Ivlev in his article ignores the effect of border changes altogether. It is not the major factor but it still counts.
All that combined gives us some 3 million difference.

Art
Forum Staff
Posts: 6196
Joined: 04 Jun 2004 19:49
Location: Moscow, Russia

Re: 42 million dead?

Post by Art » 29 May 2017 19:48

And a final note: there is a good summary of various estimates of Soviet population here (in Russian):
http://rybakovsky.ru/demografia4a6.html
For example for the Soviet population as of 1.1.1941 different scholars provide the following figures:
198.7, 197.1, 197.5, 196.7, 190.5-195.1, 196.7, 196.6, 195.4 million.
As you can see there is quite a variation. In any case all numbers are much smaller the one used by Ivlev. ADK estimate belongs to a lower ebb but still doesn't go beyond a reasonable range error. A simple conclusion is that you can't really know what was the real Soviet population with accuracy better than say 1 million.

User avatar
stg 44
Member
Posts: 3222
Joined: 03 Dec 2002 01:42
Location: illinois

Re: 42 million dead?

Post by stg 44 » 30 May 2017 13:32

So Ivlev's estimates are completely bunk?

mars
Member
Posts: 1164
Joined: 03 Oct 2002 19:50
Location: Shanghai

Re: 42 million dead?

Post by mars » 30 May 2017 15:59

Thanks, art, it is very interesting

Art
Forum Staff
Posts: 6196
Joined: 04 Jun 2004 19:49
Location: Moscow, Russia

Re: 42 million dead?

Post by Art » 30 May 2017 20:55

Post-war population. Ivlev gives 169.8 million as of 1.07.45. ADK - 170.5 million as of 1.01.46. The difference is small and is easily explained by natural growth in 6 months. I believe Ivlev has some methodical mistakes in his estimates but I don't want to dwell on them since there isn't much difference from the mainstream version. Again there are different estimates from different author but all they fall within 167-170.5 million range.

Art
Forum Staff
Posts: 6196
Joined: 04 Jun 2004 19:49
Location: Moscow, Russia

Re: 42 million dead?

Post by Art » 30 May 2017 21:39

Wartime births. That's another curios part. Ivlev takes another document from the files of statistics offices which gives the number of birth in 1941 for the Soviet Union without occupied territories. The raw number is 2.9 children. This raw number is used for the extrapolation to the entire Soviet territory. In essence we have the entire population 199.9 million. Population without Baltics, Ukraine, Moldavia and Belorussia = 131.9 million. Estimated total number of births =2.9*(199.9/131.9) = 4.4 million. Then the number of births in four years (July 41 - June 45) is 4.4*4=17.6 million.

My remarks:
1. Ivlev assumes that births stats for the year 1941 cover the entire territory of the Russian SFSR. A natural question is how on Earth it was possible given that, for example, Smolensk and Pskov were occupied already in July 1941. Even larger territories were occupied by the end of the year. A commons sense suggests that no registration data was available from them in the annual report. So Ivlev's procedure is at least dubious and most probably underestimates the real number of births.
2. Comparison with the previous year. In 1940 the raw registered number of births was 6.1 million and after adjustment for incomplete registration - 7 million (from Andreev, Darskiy and Kharkova). It is reasonable to expect about the same birth rate in 1941. That confirms the conclusion from 1.
3. The most crucial part: Ivlev assumes that the number of births in 1942-45 was the same as in 1941. That is in fact a very dubious assumption since it is known how strongly the birth rate is affected by large wars. Also the census of 1959 registered much less people born in 1943 or 1944 than in 1939 or 1940 (e.g. 2.1 million born in 1940 and 0.9 million born in 1943). The author comments on this problem saying that birth rate actually became even larger during the war yet huge children mortality nullified it. That is nothing but a leap of faith which is not based on any facts. In reality both contemporary registration on non-occupied region and post-war sample studies showed a dramatic decrease of birth rate during the war. In 1944 it was about 40% of the 1941 level. Hence extrapolation of the 1941 stats on births to other years is completely wrong.

The bottom line is that this method so crude that it cannot lead to any meaningful value. ADK based on a more sound method arrive to 16.5 million children born in 4.5 years (1.7.41 - 31.12.45). Ivlev has 16.6 million born in 4 years so he has an overestimation (about 2.5 million), albeit a moderate one because by an accident his different mistakes compensated each other.

AJFFM
Member
Posts: 607
Joined: 22 Mar 2013 20:37

Re: 42 million dead?

Post by AJFFM » 31 May 2017 01:10

Good Morning

Interesting information.

One thing I do not see here is the losses from forced deportations of ethnic minorities. Could it be that ethnic minorities were undercounted in the census and that their losses during the deportation were higher than the current estimates?

Art
Forum Staff
Posts: 6196
Joined: 04 Jun 2004 19:49
Location: Moscow, Russia

Re: 42 million dead?

Post by Art » 31 May 2017 10:53

Estimation of the "normal" death rate. The story is almost the same as for the estimation of births. There is an archival document published by Ivlev which gives the number of 1.8 million deaths for the unoccupied territories in 1941. Extrapolation for the entire territories gives 2.7 million deaths in 1941. For the 4 years (7.41 - 6.45) the number if 2.7*4=10.8 million.

Again comments:

1. It's not clear why the year 1941 must represent the normal peace-time mortality. One have was not peaceful at all. It would be more logical to take 1940.
2. The raw number of deaths registered in 1940 was 3.5 million, after corrections for under-registration - 4.2 million (ADK). There are no reasons for the deaths rate in 1941 to be smaller than a year before. Hence Ivlev's estimation is too low. First huge territories were occupied and didn't provide any stats. Second, there was a large scale evacuation and relocation of millions of men, one can expect that usual registration didn't work well in such conditions. Hence, Ivlev's estimate is by no means adequate.

From ADK the normal mortality in four years (assuming a 1940's level) can be estimated as 4.2*4=16.8 million. However, one should take into account the wartime decrease of births rate. Less babies born = less babies dying. Some more complicated calculations are needed. If we take results from ADK the normal expected number of deaths in 4.5 years for people born before 1.7.41 was 11.9 million. In addition 16.5 million children were born, of them 11 million were alive on 1.1.46. That means 5.5 million deaths. Excluding excess deaths attributed to the war (1.3 million) we have 4.2 million as a "normal" mortality. The sum is 11.9+4.2 = 16.1 million. For the 4 years instead of 4.5 the crude estimate is 16.1*4/4.5 = 14.4 million or some 3.5-4 million more than calculated by Ivlev.

Sid Guttridge
Member
Posts: 9480
Joined: 12 Jun 2008 11:19

Re: 42 million dead?

Post by Sid Guttridge » 01 Jun 2017 09:35

Hi Guys,

From memory, only 110 million of the Soviet population were Russians.

In 1941 the German Reich contained some 90 million, of whom 80 million were Germans.

If one strips away all the intervening countries, SSRs and peoples between the two, all of whose loyalties were more or less negotiable (as the Germans proved by half heartedly enlisting a million of them) one is left with a relatively rather larger Germany than might be assumed and a rather smaller Russia than its usual burial in a collective Soviet Union reveals.

Cheers,

Sid.

Art
Forum Staff
Posts: 6196
Joined: 04 Jun 2004 19:49
Location: Moscow, Russia

Re: 42 million dead?

Post by Art » 01 Jun 2017 14:15

A conclusion from all said above. Ivlev puts Soviet population losses 15 million above "official" number (41.9 vs 26.6 million). Those 15 million consists of:
a) 5 million military erroneously added to the pre-war population
b) 3.2 million - difference between ADK and contemporary estimates of Soviet population as of 1.7.41 by Gosplan's statistics office (part of it is an effect of border changes)
c) about 2.5 million - overestimate of wartime births
d) 3.5-4 million - underestimate of normal peacetime mortality.
The sum is about 14.5 million and almost perfectly explains the difference.
It must be said that a),c), and d) are the result of grave methodical errors. As explained above the situation with (b) pre-war population is less definitive, since there is a great deal of uncertainty regarding the exact number. Still it can be expected that contemporary estimates of population by Gosplan were exaggerated, yet I'm not sure that by those 3 million.

The bottom line is that Ivlev relies on 4 numbers: pre-war Soviet population, post-war population, wartime births and normal peacetime mortality. 3 of 4 were calculated either with mistakes or with wrong assumptions, hence the result.

Return to “The Soviet Union at War 1917-1945”