RUBLE ESTIMATES OF DEFENSE BURDEN

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP85-01156R000100010011-4
• Release Decision: RIPPUB
• Original Classification: K
• Document Page Count: 9
• Document Creation Date: December 21, 2016
• Document Release Date: June 20, 2008
• Sequence Number: 11
• Publication Date: February 19, 1982
• Content Type: MEMO

DATE
TRANSMITTAL SLIP ~ Mares 14RZ-
TO:
C/NIG
ROOM NO. I
BUILDING
REMARKS:
I called Andy to tell him
that misinterprets
the meaning of "hidden inflation"
and misuses the calculation.
Maurice Ernst
FROM:
NIO/Econ
ROOM NO.
I BUILDING
I EXTENSION
FORM Y0.
>< FES ,s 241
REPLACES FORM 36-5
WHICH MAY BE USED.
Approved For Release 2008/06/!20 :CIA-RDP85-011568000100010011-4
!.
Approved For Release 2008/06/20 :CIA-RDP85-011568000100010011-4
STAT
 Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
? ?
OFFICE OF THE SECRETARY OF DEFENSE
WASHINGTON, D.C. 20301
MEMORANDUM FOR HARRY ROWEN
SUBJECT: Ruble Estimates of Defense Burden
19 February 1982
T rPrPntly received the attached letter from Professor
raises an important prodiem
given that Defense and others ma a regu ar use of the CIA's ruble estimates
and particularly the dollar cost numbers.
argument about Agency underestimates of the ruble cost of Soviet
,
defense effort has some plausibility, and apart from his own version of the
argument about why the numbers need to be revised, the people at CIA (he
says) are convinced that they may have to adjust the ruble estimates
sometime in the not too distant future. Jumping from about 15% to a burden
rate of around 20% will require an explanation if the whole enterprise is
not be be called into question.
After you have read this, let us talk about how best to get the whole set
of problems addressed in a constructive way. Defense has a great stake in
the continued production and repute of these estimates.
STAT
STAT
STAT
Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
 Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
February 9, 1882
STAT
Andrew Marshall
OSD/NET Assessment
3A930 Pentagon
Washington, D.C. 20301
Dear Andy,
I would like to alert you to a problem that has arisen which may
seriously complicate national security policymaking in the near future.
The CIA is on the verge of discovering that its estimates of Soviet
defense spending are off the mark in 1980 by tEarly as much as they
were in 1970. The agency's own data indicate that the Soviet defense
burden was in the vicinity of 20 percent in 1980, a fact which when
made public is likely to cause a political malestrom. Critics of U.S.
national security policy will almost certainly challenge the veracity
of this figure by noting that CIA estimates of the Soviet defense burden
have implausibly jumped from 6.5 percent in 1975 to 11-13 percent in
1976, to 20 percent in 1980. I am concerned that criticisms of this
sort will not only be used to discredYt the agency's official ruble and
dollar estimates, but will be employed by extension to disparage the
wide range of corroborative evidence confirmin4 the existence of a
rapid Soviet arms buildup. -
My 3udgment that the agency will soon be compelled to revise its
burden estimate upward is; based on information provided at the CIA
Conference on "Price Change in the Soviet Defense Sector," December 7-8,
1981, held in N rn.Vir inia. Durin the course of these sessions,
remarks made by revealed that CIA
estimating procedures yield three wi e y ivergent measures of the Soviet
defense burden in 1980. Tabled summarizes these measures computed. -
variously at established prices and at factor cost. The official direct
cost estimates for 1980 are 12.4 percent at established prices and 13.9
.percent at factor cost. The agency's corresponding best current value
estimates, assuming a 4 percent rate of hidden inflation are 18.8
percent and 21.2 percent, and its residual method estimates are 20.3
percent and 22.7 percent.
STAT
Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
 Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
._ ~. ? ?
Andrew Marshall
Page 2
February 9, 1982
As ?~: readily apparent, the CIA believes that the Soviet defense ~ ~.
burden measured at current prices is a~ least 50 percent greater than
its estimate in constant 1970 prices, using the direct cost method.
This means, other things eoual, that the agency should scrap its direct
cost estimates predicated as they are on outmoded 1970 prices, and
endorse the higher burden figures which reflect prevailing 1980
opportunity costs.
This revision however has been resisted to date because it will
embroil the CIA in further difficulties. Although the agency believes
that the 50 percent increase in the Soviet defense burden is attributable
to hidden inflation, it has virtually no evidence to support this
coniecture. At the conference on "Price Change in the Soviet Defense
.Sector it was acknowledged, for example, that preliminary statistics
issuedt,by the Soviets as part of their 1982 price reform indicate that
machinebuilding Writes will be unchanged from 1970, and even more
telling it was asserted that no ruble weapons prices have been collected
since the early ~venties.
The absence of corroborating evidence places the CIA in an awkward
position. On one hand the agency wishes to argue that hidden inflation
has caused the 1980 Soviet defense burden to increase to 20 percent,
but cannot convincingly demonstrate the existence of hidden inflation,
making its belief vulnerable to preemptory challenge. On the other hand,
the agency is reluctant to put forward its residual method. estimates as
proof of the 20 percent burden level because it would have to seriously
entertain the possibility that the 50?percent disparity between its
1970 price and current price estimates is explained by real increases
in Soviet procurement undetected by the direct Cost method.
These irreconcilable contradictions suggest that the time has come
for the CIA to cut the Gordian Knot. If the agency sticks with its
direct cost estimates, it~.~will necessarily suppress the mass of evidence
showing that the burden is 50 percent higher than is presently acknowledged,
thereby impairing objective national security policy assessment., If it
.endorses the 20 percent burden level, but tries to explain the disparity
.away by appeal to hidden inflation, its critics, will make the agency _
appear ridiculous. And if it takes the`bull by the horns, accepting
its own residual estimates, it will have to come to terms with the fact
that its direct costing method is faulty.
'Under the circumstances, the only way out is a thorough review to
:determine once and for all why the agency's estimates of Soviet defense
spending behave so aberrantly. My own work on intervintage parametric
cost estimating bias and learning curve distortion may be of some
assistance in this regard. Whether it is or not however is less
Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4 -~-~""`~?a
 Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
Andrew Marshall
Page 3
February 9, 1982
important than the need to achieve a 'clear disposition of the orobi~ems
obscuring the true dimensions of the size, rate and momentum of the
Soviet arms buildup.
Best regards,
Sincerely,
STAT
Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4 -------~--A
 Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
Alternative CIA Estimates
of the 198D Soviet Defense Burden
(Mean Values)
Table 1
Established Prices
Factor Cost
1.
Official [~irect Cost
12.4
13.9
2.
Official Direct Cost
(Adjusted for Hidden Inflation)
18.8
21.2
3.
CIA Residual Method (1970 prices)
21.9'
24.5
4.
CIA Residual Method (1980 prices)
20.3 ~
22.7
Sources: Table A1, and CIA, The Soviet Economy in 1978-79 and Prospects
for 1980, ER80-10328, June 1980.
Method: The GNP estimates reported in Table Al are converted to mean
values consistent with published CIA burden statistics by
adjusting them for the CIA's factor cost estimate of Soviet
GNP growth. According to agency calculations real Soviet
GNP grew 3.5 percent per annum 1970-79, compared with the
official Soviet NMP rate of 5.1 percent per annum. As a
consequence, the CIA's estimate of Soviet GNP in 1981
(extrapolated 1979-80) is~15 percent lower than the official
estimate, and its burden estimate is correspondingly higher
(see The Soviet Economy in 1978-79, Table 2, p. 25).
Notes:. The burden estimates reported above will be 15 percent
lower if the official Soviet NMP growth r is used to
compute GNP. See?,Table A1. Also some of late .year
estimates, which were confirmed by the agency, are based on
planning data and maybe too high. This bias is offset to a
?L limited extent because the agency's 1970 GNP estimates are too
low. Given the "new information" discovered in 1975, GNP at .
?. established prices in 1970 (including the underestimate of MBMW)
should be 391.6 billion instead of 380.7 billion rubles, and
350.2 instead of 340.2 billion rubles at factor cost. (See
Rosefielde, False Science, appendix 5).
. Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
 Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
?
Raw Data Underlying the Burden
Estimates Reported in Table 1
Table Al
Established Prices
Factor ~:st
1.
Official Rirect Cost
Burden Estimated Valued
9.4-11.8
10.5-13.2
in Constant 1970 Prices
(10.6)'
(11.9)
2.
Adjusted Official Direct
Cost Burden Valued in 1980
Prices Assuming Hidden
Inflation (4 percent per
14.3-17.9
16.1-20
'
annum)
(16:1) ~
(18
.1)
3.
CIA Residual Method Burden
17.2-20.1 ~
19.2-22.5
Valued in 1970 Prices
(18.7)
~ (20.9)
4.
CIA Residual Method Burden
15.9-18.6
17.8-20.9
Valued in 1980 Prices
(17)
(19.4)
Sources: CIA, Estimated Soviet Defense S endin in Rubles, 1970-1975,
SR75- 0 21U, May, 976; CIA, Estimated Soviet Defense Spending.:
Trends and Prospects, SR78-10121, June 1975; Rosefielde, False
Science: Underestimatin the Soviet Arms Buildu Transaction,
982; W~ lam Lee, Testimony in CI Estimates of Soviet
Defense Spending, House Committee on Intelligence, Washington,
D.C., September 3, 1980, p..22; Narodnoe Khoziaistvo SSSR,
1979, p. 405.
Method: A. GNP Estimates: Soviet GNP in 1980 can be calculated either
in esta fished prices or at factor cost by extrapolating the
CIA's 1970 Soviet GNP estimates to 1980 either with the
official real or nominal Soviet NMP growth index. The former
produces estimates in constant 1970 rubles, the latter in
1980 rubles. Soviet GNP for 1980 computed in established 1970
prices is 628.4 bill"ion rubles, at established 1980 vrices,
~~ 602.6 billion rubles. Soviet GNP for 1980 calculated at
factor cost for 1980 is 561.2 ,billion rubles valued in 1970.
?. prices and 538.2 billion rubles in 1980 prices. Nominal GNP
estimates (in 1980 prices} have been extrapolated from 1979
to 1980 at the compound nominal NMP growth rate 1970-79,
4.7 percent per annum. Real GNP estimates (in 1970 prices}
have been extrapolated from 1979 to 1980 at the compound real
:.NMP growth rate 1970-79, 5.1 per annum. ,
Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4 ~-------~---,~
 Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
.~ ?
B. Burden: 1. The CIA estimates that Soviet defense expenditure
in 1970 were between 40 and 50 billion rubles valued at 1970
prices. During the seventies the agency contends defei:~e
outlays rose 4 percent per annum. 1980 Soviet defense
expenditures therefore can"be estimated at
40(1.04)10 = 59.2 billion rubles usi~e the lower bound
U.S. defense definition and 50(1.04) = 74 billion rubles
using the upper bound Soviet definition of defense. These
estimates divided alternatively by GNP measured at 1970
established prices and ]970 factor cost produce the burden
estimates reported above.
2. At the CIA conference on "Price Change in the. Soviet
Oefense Sector" the agency reveated its often stated position
that Soviet military machinebuilding prices have been rising
4 aercent per annum due to hidden inflation. The effect of
hidden inflation can be taken into account by multiplying
the direct cost estimates valued in 1970 prices by (1.04)10
which yields an established 1980 price estimate of 86.4-
billion rubles, and a 1980 factor cost estimate of 197.9
billion rubles. The burden is then computed by dividing these
-estimates respectively with 1980 GNP valued at 1980
-established prices and at 1980 factor cost.
3. At the CIA conference on "Price Change in the Soviet Defense I
Sector" revealed that the agency's residual STAT
.-estimates o oviet defense spending corresaonded closely with - -
~~Lee's. According to Lee Soviet defense outlays in 1980 were ..
between 108 and 126 billion rubles valued at 197 prices. -
The associated burdens for 1980 are computed by dividing Lee's
estimates with 1980 GNP valued at 1970 established prices and
1970 factor cost.
4. The CIA's residual estimates can also be expressed in
1979 prices by, adjusting procurement with the official .
Soviet machinebuilding price index (including turnover tax),
.$(6'6) - .8(74) = 52.8 - 59.2 billion rubles, and nonprocurement?-
~vith the aggregate retail price index 1.02(42) - 1.02(52) _
-42.9 - 53.1 billion rubies. ,The corresponding aggregates.
-95.7 and 112.3, divided by T~980 GNP valued at 1980 established
-prices and 1980 factor cost, produce the desired residual
burden estimates at current prices.
Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4
 Alternative CIA Estimates of the 1980
Soviet Defense Burden
(Mean Values: Officiri .soviet GNP)
Table 2
Established Prices
Factor Cost
1.
Official Direct Cost
10.4
11.6
2.
Official Direct Cost (Adjusted
for Nidden Inflation)
15.8 _ _
17.8
3.
CIA Residual Method
(1970 Prices)
18.3
20.5
4.
CIA Residual Method
(1980 Prices)
17.0
19.0
Approved For Release 2008/06/20: CIA-RDP85-011568000100010011-4 ._ ~,~ _