COMPARING PLANNED AND ACTUAL GROWTH OF INDUSTRIAL OUTPUT IN CENTRALLY PLANNED ECONOMIES

• Document Type: CREST
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): CIA-RDP08S01350R000200340001-8
• Release Decision: RIFPUB
• Original Classification: K
• Document Page Count: 16
• Document Creation Date: December 22, 2016
• Document Release Date: March 7, 2012
• Sequence Number: 1
• Publication Date: August 1, 1980
• Content Type: REPORT

Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
/r Y'~'Tf,Y 1\QI~V~~4~
'c
Foreign
' ~. ~ ~'' Assessment
e?~'` Center
Comparing Planned and Actual
Growth of Industrial Output
in Centrally Planned Economies
A Research Paper
ER 80-10461
August 1980
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
This publication is prepared for the use of US Government
officials, and the format, coverage, and content are designed to
meet their specific requirements. US Government officials may
obtain additional copies of this document directly or through
liaison channels from the Central Intelligence Agency.
Requesters outside the US Government may obtain subscriptions to
CIA publications similar to this one by addressing inquiries to:
Document Expediting (DOCEX) Project
Exchange and Gift Division
Library of Congress
Washington, D.C. 20540
or: National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
Requesters outside the US Government not interested in subscription
service may purchase specific publications either in paper copy or
microform from:
Photoduplication Service
Library of Congress
Washington, D.C. 20540
or: National Technical Information Service
5285 Port Royal Road
Springfield, VA 22161
(To expedite service call the
NTIS Order Desk [phone])
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
~.E~~ ?.~ rv~r~nnn~
Foreign
~~ ~ ~ ~ Assessment
~~ Center
Comparing Planned and Actual
Growth of Industrial Output
in Centrally Planned Economies
Comments and queries on this report
are welcome and may be directed to:
Director of Public Affairs
Central Intelligence Agency
Washington, D.C. 20505
[phone]
For information on obtaining additional copies,
see the inside of the front cover.
ER 80-10461
August 1980
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Comparing Planned and Actual
Growth of Industrial Output
in Centrally Planned Economies
Understanding the planned growth of industrial production is an important
ingredient in the study of centrally planned economies. Comparison of
the planned rate of growth of total industrial production with what the
statistical authorities eventually claim creates the mistaken impression
that plans are generally fulfilled. Yet, when the achieved output levels
of individual products are compared with plan figures in physical terms,
it becomes apparent that the overall plans could not possibly have been
achieved.
The official rate of growth reported by the statistical authorities is biased
upward for two reasons. With the gross output indexes used, an increase
in ou a counting causes an overstatement of reported achieved growth.
More importantly, the official data allow disguised inflation to enter the
indexes under the guise of new-product pricing, so this inflation is counted
as real growth. This is a particularly severe problem in the machinery
industry, where products are complex and heterogeneous.
One way to circumvent the problems in the official data is to construct
synthetic indexes based on a sample of commodities in physical terms
In this manner the impact of the biases resulting from double-counting
and disguised inflation can be controlled. These synthetic indexes show
a growth rate substantially lower than the official measures and suggest
that overall plan targets are seldom achieved.
and then to aggregate them with fixed prices and value-added weights.
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Comparing Planned and Actual
Growth of Industrial Output
in Centrally Planned Economies
Introduction
Periodically the Soviet Union and other centrally
planned economies publish annual and five-year plans.
An important feature of these plans is the expected
growth of industrial production. One method of
assessing the performance of these economies is to
compare the actual growth of industrial production
with the planned growth. This paper demonstrates that
it is inappropriate to use the officially published data
on aggregate industrial production for such a compari-
son because they systematically overstate the real
growth of industrial production, create the erroneous
impression that the plans are generally fulfilled, and
cloud the picture of the relative performance of the
branches of industry.
The fact that the officially reported growth rate of
industrial production overstates the true growth rate
can be demonstrated easily. Figure 1 plots the planned
growth rate of Soviet industrial production versus the
officially reported rate for 1960-79. A point above the
45-degree line means that the plan has been
overfulfilled and a point below, underfulfilled. In only
four of the 20 years did industry fail to meet its plan.
Yet each year, especially recently, Soviet leaders
carped about a long list of key commodities for which
output fell below plan. To show the reason for their
complaints, planned and actual production data for a
sample of nine key industrial products were collected
for the same 1960-79 period. The plan was
underachieved more than 46 percent of the time.'
Figure 1
USSR Industrial Production: Plan and
Performance, 1960-79
Percent
3 4 5
Planned Growth
level.Z The true picture may have been even worse,
because the Soviet Union often omits the publication of
unfavorable production data. Thus the aggregate data
create the impression that 1977 was a successful year,
while the detailed commodity data give the opposite
impression.
In 1977, a year for which unusually detailed plan and
performance data are available, the reported value of
industrial output grew 5.7 percent, slightly in excess of
the planned rate of 5.6 percent. Moreover, every
industrial ministry except the Ministry of Ferrous
Metallurgy and the Ministry of the Meat and Dairy
Industry fulfilled their plans. Yet_the production of
more than two-thirds of the individual commodities
whose production was reported fell short of the planned
' Ray Converse and Robert E. Ramsson, "The Linkage from Plan to
Performance in Soviet Industry: A Statistical Analysis," paper
presented at the annual meeting of the American Association for the
Advancement of Slavic Studies, New Haven, October 1979.
First, this paper describes the institutional and statisti-
cal sources of the bias in the official measures. Second,
it presents several statistical tests that were conducted
to verify the existence of the bias. Finally, it discusses
some reasons why the bias is allowed to continue.'
' [bid.
'Although most of the evidence presented here relates specifically to
the Soviet Union, Thad Alton has shown this is also the case for
several East European countries. For example, see Thad P. Alton,
Elizabeth M. Bass, Laszlo Czirjak, Gregor Lazarcik, Statistics on
East European Economic Structure and Growth, OP-48, (New
York, L. W. International Financial Research Inc., 1975).
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Sources of Bias
We believe that the inconsistency between plan and
actual growth rates arises because they are constructed
according to different approaches. The Council of
Ministers and the State Planning Committee
(Gosplan) formulate the plans with some feedback
from the ministries and enterprises. The plan (ex ante)
data are_created mainly in physical units andare
believed to rovide relatively unbiased growth rates.
The output (ex post) data are compiled by t e t e
Central Statistical Administration (CSA) based on the
accounting reports of the enterprises and ministries.
The managerial incentive system and the methods of
reporting production data both bias the results. The
principal cause is the inclusion of disguised_inflation. A
secondary cause is the use of gross output rather than
value-added as a measure of growth. Because the ex
post Baia are exaggerated, unlike the plan data,
comparing planned and actual growth rates is, there-
fore, akin to judging the size of grapefruits by the
normal size of apples and concluding the grapefruit is a
good apple.
This section summarizes the method of plan construc-
tion used by the Soviet Union and why we think it
results in a relatively unbiased estimate of planned
growth. Then it summarizes the method used to
compute the measure of industrial growth and why it
overstates the real growth.
Plan construction. To understand why the plan meas-
ures are relatively unbiased, we must examine the key
stages in the planning process.' The party leadership
makes the first step in formulating the annual plans by
establishing general guidelines for the upcoming plan.
These guidelines may include output targets for a few
key commodities. Gosplan uses these guidelines to
develop control figures for output in physical terms for
a set of commodities which constitute about 80 percent
` For a more detailed exposition of how the annual plans are drafted,
see Paul R. Gregory and Robert C. Stuart, Soviet Economic
Structure and Performance (New York: Harper & Row, 1974), ch.
5; Howard J. Sherman, The Soviet Economy; Igor Birman, "From
the Achieved Level," Soviet Studies 30 (April 1978) pp. 153-172;
David Dyker, The Soviet Economy (New York: St. Martfin's Press,
1976), ch. 2; Alec Nove, The Soviet Economic System (London:
George Allen & Unwin, 1977), chs. 2-3; and Gertrude E. Schroeder,
"The Soviet Economy on a Treadmill of 'Reforms,' " in US,
Congress, Joint Economic Committee, Soviet Economy in a Time of
Change (Washington, D. C.: Government Printing Office, 1979),
vol. 1, pp. 312-340.
of the value of industrial production. Then Gosplan
splits these targets among the various ministries and
republics, and they in turn allocate the targets to
individual enterprises. At this point there ensues a
protracted period of negotiations over the production
expected from each organization and the inputs needed
for that production. Once this phase is completed,
Gosplan checks the plan for consistency by using
material balances. This involves keeping a tally for
each commodity, usually in physical terms, of the
amount planned to be produced and that quantity
required by other segments of the economy to achieve
their targets. When planned demand exceeds the
planned supply, the plan must be rectified by boosting
production, cutting demand, expanding imports, or
increasing the efficiency of material use. When a
change is made in one material balance, it causes
imbalances in other material balances, which cause
further imbalances. In the end, changes are required in
all of the balances to restore equilibrium. In reality,
probably only the initial effects are considered by
Gosplan.
Once the commodity targets are established, the
planned growth in total industrial production is com-
puted, probably as a weighted average of the planned
growth rates of all of the commodities for which
material balances are kept. Each ministry's planned
growth is probably computed in the same way. The key
point is that, until the commodity targets are aggre-
gated, the planning occurs primarily in physical units.
Support for the proposition that Soviet plans are
constructed primarily in physical units and, therefore,
provide an accurate expected rate of growth can be
obtained from plan data. We can use enterprise
wholesale prices to aggregate the commodities for
which plan growth is expressed in physical units in
order to estimate the planned growth rate of total
industrial production. If this estimate approximates
the officially published planned growth rate, then we
may claim that the planned growth rate is relatively
unbiased. This assumes that the growth rates of the
commodities for which plans are not published are well
correlated with those that are published. Considering
that the published commodity plans usually include the
main products on which most industrial activity relies,
this is fairly realistic.
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Noren and Whitehouse performed this exercise for
several key branches of industry for the Ninth Five-
Year Plan (1971-75), a plan for which unusually
detailed commodity data were published, and ob-
tained growth rates which were very close to the
published values.s The following tabulation compares
the average annual rates of growth estimated by
Noren and Whitehouse with the published rates (in
percent):
Industrial Branch
Estimated
Growth
Rate Based
on Com-
modity Plans
Officially
Published
Planned
Growth Rate
Electricity
7.5
7.9
Coal ____
2.2
3.0
Oil extraction
7.3
7.5
Gas extraction
_ _ _
10.1 _
_
_13.5 _
Ferrous metals
4.9 __
___
5.1 __
Paper and paperboard
7.9 _ _
_ 8.5__ __
Construction materials
6.5
7.1 _
Machinery
12.1
11.4
Light industry
5.7
_
_ __
6.6
Moreover, if these branch estimates are aggregated
with gross output weights t e gross output for each
branch in 1970, as reporte y the CS_A) to derive
a total in ustry esttmate ase on commodity plans
where available, the resulting fieure of 7 9 percent
per year is quite close to the official growth of
8.0 percent.b The same task has been done on a
m red scale for the 10th Five-Year Plan (1976-
80). In this case the estimated average annual
planned growth of industry works out to 6.5 percent
compared with the officially published 6.4 percent.
Figure 2 uses commodity data from the 1977 plan
to illustrate the validity of the planned growth rate
James H. Noren and F. Douglas Whitehouse, "Soviet Industry in
the 1971-75 Plan," in US, Congress, Joint Economic Committee,
Soviet Economic Prospectsjor the Seventies (Washington, D.C.:
Government Printing Office, 1973), pp. 206-245.
For completeness of coverage, it is necessary to use officially
published planned growth instead of a rate based on commodity
plans for some branches such as processed food and chemicals.
Figure 2
USSR: Plan and Performance for aCross-
Section of Products, 1977
annual plan. The planned rate of growth of each
product is plotted against its realized rate. The line
segment DBE represents the planned growth rate of
total industrial production for 1977 and ABC denotes
the realized rate as computed by the Central Statisti-
cal Administration. By comparing the distribution of
points along the horizontal axis, one gains a notion of
how the planned growth of total industry compares
with that for individual commodities. It is readily
apparent that the planned growth for total industrial
production falls near the midpoint of the sample.
Although a majority of products were targeted to grow
more slowly than the plan for total industry, they are
clustered near the total industrial rate. The growth
rates of those products that were slated to grow more
rapidly than total industry are somewhat more dis-
persed, which probably compensates for the slightly
more numerous cases of planned slow growth. The
1977 planned growth rate for total industry, therefore,
was probably an unbiased average of the growth rates
of the separate commodities.
25
'
E
I
20
I
I
15
I,
~.
1o
i
.
l3
.
A
-
- --
----
--C
5
~?
,
D
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
In light of the preceding discussion, the plan goals do
not appear subject to any large bias. Noren and
Whitehouse concluded, "the plan goals ...appear to
be based firmly on a broad range of commodity goals
and can thus be used in further analysis."'
Accounting for Actual Production. The ex post value
of industrial production computed by the Central
Statistical Administration overstates real growth
mainly because the overriding goal of each enterprise
manager is to meet his production goal. The method-
ology use to compute ex post levels of industrial
production perm[ s e manlpu atlons n ,~SSctry to
ensure ffiaf'tTe en erprtse an nzt,nn~l ntan;~ fulfilled
i value terms., if not in physical terms.8 This section
describes tie various measures of output and shows
how the Soviet use of one of them distorts a sector's
contribution to industrial output, both in a given period
and over time.
The correct method of calculating the real contribution
of any enterprise to industrial output and total GNP by
sector of origin is to count only the net contribution or
value-added of each enterprise in constant prices. To
do this one should deduct the cost of materials and
services purchased from other enterprises because
these intermediate items already are included in the
output of the enterprise producing them. What re-
mains after this deduction is the value-added by the
producing sector-more specifically the total of prof-
its, wages, depreciation, and payments to the other
factors of production.
Since continuous time series of value-added are seldom
available, a technique known as double. deflation has
been developed to estimate value-added. This entails
deriving constant price indexes for the amount of gross
' Nolen and Whitehouse, p. 212.
'The writings of the late Rush V. Greenslade in particular are
prolific in discussing the problems with the Soviet industrial pro-
duction indexes. For a more detailed treatment of this topic, see
Rush V. Greenslade, "Industrial Production Statistics in the
USSR," in Vladimir G. Treml and John P. Hardt, eds., Soviet
Economic Statistics (Durham, Duke University Press, 1972), pp.
155-194; Rush V. Greenslade and Wade R. Robertson, "Industrial
Production in the USSR," Soviet Economic Prospects jor the Sev-
enties, pp. 270-282; and Rush V. Greenslade, "The Real Gross
National Product of the USSR, 1950-75," in US, Congress, Joint
Economic Committee, Soviet Economy in a New Perspective
(Washington, D.C., Government Printing Office, 1976), pp. 269-
300.
output and also for purchases of intermediate materi-
als by each producing sector. In many cases complete
data are unavailable so the indexes must be derived
based on samples of output and the consumption of
intermediate product. These indexes are combined into
a value-added estimate by deriving weights for a base
year when a plethora of data on industrial structure are
available; this could be the year of an industrial census
or of an input-output table. These value-added weights
of the base year are then applied to the indexes to
subtract material purchases from gross output and
arrive at value-added.
In contrast to this procedure most of the centrally
planned economies use a gross output method to
aggregate national output. In general this entails
totaling the value of output of individual enterprises in
a ministry, branch, or all industry to determine the
respective total output figure. This has the effect of
placing too high a weight on enterprises producing for
final consumption, because it includes the value of
intermediate products that are also counted elsewhere.
As a result the importance of enterprises producing
raw materials is understated, because their output
estimates include few intermediate products. Figure 3
is instructive, showing the different impressions of
industrial structure that are implied by the 1972 input-
output table depending on whether value-added or
gross output weights are used. Processed food, for
example, has twice as large a share when measured by
gross output instead of value-added, and the impor-
tance of light industry is also overstated. In contrast
the contribution of the basic branches of industry, such
as fuels, electricity, ferrous metals, and construction
materials, are understated by use of gross output
weights.
Enterprises usually use one of two methods for
reporting their production.9 The first is known as gross
turnover of output (valovoy oborat). It includes
intraplant consumption of a plant's own product, so
enterprise output is calculated by adding together the
output of individual workshops within the plant, even if
For a discussion of the two types of gross output measures, see
M. R. Eydel'man, Mezhotraslevoy Balans Obshchestvennogo
Produkta (Moscow: Statistika, 1966), pp. 200-03; A. I. Yezhov,
Statistika Promyshlennosti (Moscow: Statistika, 1977), pp. 57-61;
and Vladimir G. Treml et al., The Structure q/'the Soviet Econ-
omy: Analysis and Reconstruction gjthe 1966 /nput-Output Table
(New York: Praeger Publishers, 1972), pp. 45-46.
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Figure 3
USSR: Share of Industry Output by Different
Branches and Measures, 1972
Ferrous Metals
~ 6.0 Gross Output
7
1
N
f
M
t
l
.
Value-added
- 3.2
errous
on
e
a
s
~ 2.4
~ 5
7
Fuels
'
9
0
-
.
3
1
Electricity
.
5.0
Machinery
24.5
32
3
Chemicals and
~ 6.1
.
Petrochemicals
- 6.8
Wood, Pulp, and
4.9
~
Paper Products
6.6
Construction
? 4.2
Materials
~ 6.0
?
16
0
Light Industry
.
10
1
.
Proce
d F
d
23.4
sse
oo
11.0
n
Indust
e
c
2.s
.
ry,
.
.
3 ~
that output is consumed within the plant. Many of the
Soviet production statistics recorded on this basis
include but are not confined to consumer products. The
second method is known as gross product (valovaya
produktsiya). It excludes the intraplant turnover and,
therefore, represents shipments from the enterprise
adjusted by inventory changes. In general,.production
reported in value terms is computed on this basis.
The branch and total value of gross industrial output is
calculated as the sum of the gross output of all of the
industrial enterprises within a branch, ministry, or the
entire nation. The values of enterprise output are
allocated among branches according to their primary
product. No adjustment is made for secondary pro-
ducts that normally would be allocated to a different
branch. That the branch indexes are computed on an
establishment basis as opposed to a commodity basis.
means that individual branches receive either too high
or low a weight depending on the structure of
production.10 Ferrous metals and machinery are two
prime candidates. Many machinery plants fabricate
their own steel, and the value of the steel would be
counted as part of machinery output. Similarly, most
enterprises fabricate some of the metal parts they use;
this metalworking production would be counted in the
respective branch rather than machinery. While this
type of commodity-establishment problem could bias
the measures of branch output, it should not present
any inherent problem for the total industry index.
The Soviet measures of gross output include items in
addition to what are normally considered to be finished
production." Increases in inventories of semiprocessed
goods are treated as output, although in recent years
this has been limited to items with production cycles
exceeding two months. In addition, some repair
expenditures are treated as industrial production.
Finally, many machinery enterprises are permitted to
count expenditures for the development of new pro-
ducts as gross output.
Regardless of how measured, gross value of output is a
misleading indicator because it includes intermediate
products already counted elsewhere. For example, the
value of coking coal used in manufacturing steel is
counted in the output of the coal industry and,
implicitly, in that of the steel industry. If the steel is
used elsewhere in industry, the coking coal will be
counted again.
If the amount of double-counting of production is
constant and the industrial structure is stable, then the ,
bias in the computed growth rate is minimal since the
growth rate of value-added and gross output of each
enterprise will be nearly identical. A small bias arises
from the fact that the output of the separate enter-
prises will be combined by gross output weights rather
than value-added weights.
The principal bias from double-counting arises, how-
ever, from increasing vertical specialization in the
production of a given commodity. This causes the gross
output of an enterprise to rise faster than value-added.
The bias caused by this type of double-counting is
particularly severe over time, where the economic
structure is rapidly changing. By any standard, Soviet
industry has grown rapidly over the last three decades
and the degree of specialization has increased
somewhat.
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Soviet industry is notorious for its autarkic nature.
Because of the vagaries of its distribution system,
enterprise and ministry managers want to control as
much as possible the production and distribution of the
material inputs needed by their enterprises. Enter-
prises and ministries frequently produce goods which
clearly lie outside their area of responsibility. Soviet
planners and academics have long realized that this
excessive vertical integration hampers efficiency and
there have been numerous campaigns over the years to
encourage specialization. To the extent these cam-
paigns have been efficacious, this would increase
double-counting and the upward bias of gross output as
a measure of the growth of industrial production.
The construction materials industry is a prime example
of how a changing industrial structure causes double-
counting to increase. At one time, most cement was
shipped directly to the construction industry. Soviet
leaders decided, however, to "industrialize construc-
tion" by fabricating as many construction elements,
such as walls, bathroom units, and railroad tracks, as
possible in a plant instead of on site by the construction
industry. This means that much of the cement is now
converted into precast concrete products by other
plants within the construction materials industry and a
larger share of cement output is now double-counted.
Finally, industries with a high ratio of material inputs
to gross output are more susceptible to biases imposed
by double-counting. This means we would expect a
greater danger of significant bias in the measurement
of machinery, chemicals, and construction materials
output and a smaller bias in the remaining branches.
But the most significant source of bias in the official
production data seems to be the disguised inflation that
creeps into the data under the guise of new product
pricing and the deterioration of quality. While most of
the value series published by the Soviets are ostensibly
reported inconstant rubles, we believe that these series
reflect prices that become inflated over time owing to
Soviet pricing procedures.
The task of the Central Statistical Administration is to
report on what occurred in the previous year. There-
fore, it is interested in the value of production rather
than an average of physical units. To compute the real
rate of growth, industrial production must be valued in
constant prices of some base year. This is easy when
the same product was produced in the current and base
years. When a new product is produced, it is difficult,
but necessary, to impute a price to it that corresponds
to the prices of the base year. The primary flaw of the
Soviet procedure is that excessively high prices are
imputed to these new products. In addition, it is
believed that existing products are frequently altered
in insignificant aspects so that they may be classified
as new products and be given higher prices.
Many enterprise managers find it advantageous to
boost prices to meet their gross output plan. Although
prices of existing products are strictly controlled by the
government, an enterprise or ministry may have
considerable influence over the price of a new prod-
uct.'Z In theory a new product is given a price
sufficiently high to recover research, development, and
introductory production costs. After these initial costs
are recovered, the new-product price is lowered and a
permanent price is established.
In practice two things commonly occur, both of which
overstate the growth of industrial production. First,
Soviet managers will slightly alter the specifications of
old products, and charge substantially higher prices for
the new products, which in reality are virtually
unchanged from the old ones. At the same time, the
enterprise frequently halts production of the cheaper
"old" item. Purchasers have little choice but to accept
the more expensive "new" item and inflation.
Second, when a genuinely new product is introduced,
the temporary price often becomes an unjustifiably
high.. permanent price. In effect, this practice intro-
ducescurrent prices of a later year into a constant price
series and biases the growth index upwards. As the
proportion of new products increases, the bias due to
using current prices increases.
12 The new-product pricing phenomenon has been widely discussed.
[n particular, see Abraham S. Becker, Ruble Price Levels and
Ratios 4jSoviet Machinery in the /960s, Report R-1063-DDRE
(Santa Monica, California, The RAND Corporation, 1973); Joseph
S. Berliner, The Innovation Decision in Soviet Industry (Cam-
bridge, The MIT Press, 1976); Padma Desai, "On Reconstructing
Price, Output and Value-Added Indexes in Postwar Soviet Industry
and its Branches," Oxford Bulletin ojEconomics and Statistics
(February 1978): pp. 55-77; and Central Intelligence Agency, An
Analysis ojthe Behavior ojSoviet Machinery Prices, 1960-73,
ER 79-10631 (December 1979).
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
[t is not difficult to identify two characteristics that
would greatly increase the likelihood that new-product
pricing would be a serious problem in a given branch.
First, this phenomenon would prevail in industries with
a heterogeneous assortment of products. Not only are
value measures the only meaningful quantification of
output, but the presence of unique products provides
many opportunities for price increases. Second, it
would occur in technically dynamic industries where
many opportunities for innovation exist and product
complexities make it difficult to distinguish between
real and cosmetic improvements. These features are
most characteristic of the machinery and chemicals
branches. It has been estimated that unique products
comprise one-half of the output of the machinery
branch. Moreover, the rapid rate of change in the mix
of machinery products is confirmed by the estimates of
Soviet scholars that from 10 to 17 percent of all
machinery products are renewed annually." This is
also the branch whose growth is believed to be the most
overstated.
Statistical Evidence of Bias
Soviet literature provides abundant evidence of the
existence of the biases in the officially published
growth rate discussed above. The use of Western
reconstructions of Soviet data provides additional
support for this proposition and gives some estimates of
the size of the bias. This section presents the statistical
evidence of a bias by comparing our index of Soviet
industrial production with the officially published
data. Some limited comparisons are made with East
European data.
There can be little doubt that the ex post measures of
performance in value terms are upwards biased.
Gregory Grossman found this the case in his analysis
of the Eighth Five-Year Plan (1966-70)." Although
the Central Statistical Administration claimed that
the overall plan had achieved its targeted 50-percent
increase in industrial production, Grossman found the
targets for only four of 37 key industrial commodities
were achieved, and two of those four were expressed in
rubles. The median level of plan achievement for the
"Abraham S. Becker, Ruble Price Levels and Dollar-Ruble Ratios
ojSoviet Machinery in the 1960s, pp. 8-9.
"Gregory Grossman, "From the Eighth to the Ninth Five-Year
Plan," Analysis oJ'the USSR's 24th Party Congress and Ninth
Five-Year Plan (Mechanicsville: Cremona Foundation, 1971), pp.
54-66.
products in his sample was only 83 percent of the
respective targets. Even allowing for the limited
sample, an enormous and unlikely overfulfillment by
commodities whose planned and actual production
levels were unpublished would have been required to
bring the overall growth rate up to the claimed level of
performance.
Interpretation of the Ninth Five-Year Plan results
leads to similar conclusions. The official measure of
industrial production increased only 43 percent, short
of the planned 47 percent. Analysis of the commodity
data shows an even larger shortfall. More than two-
thirds of the products in a sample of 58 products grew
less than 40 percent, and nearly half grew less than 30
percent. The performance of the machinery branch is
an apt example. That branch supposedly fulfilled its
goals for the five-year plan, yet only two of 20
machinery items in the sample achieved their plan.
Output of both items is measured in rubles rather than
physical units.
Another way of demonstrating the upward bias of the
official measure is to compare the official indexes with
our synthetic index of Soviet industrial production
(SPIDER). This synthetic index is based on the official
data on commodity output, usually in physical terms.
These data are aggregated into sectors with constant
price weights and into total industry with value-added
weights.15 Figure 4 summarizes this procedure. The
resulting indexes are believed to avoid most of the
problems of the Soviet official measures discussed
above.
Figure 5 plots the planned growth rate, the Soviet
officially published rate, and our synthetic SPIDER
estimate for the period 1960-79. It shows that, despite
Soviet claims, the plan was fulfilled only in 1967, 1970,
and 1973.
Table 1 compares the claimed growth and our esti-
mates by branch of industry. It shows that the official
measure is higher than the synthetic measure for total
industry by one to two percentage points per year.
Among the branches, the largest differences are in the
same branches where we would expect to encounter the
" A detailed exposition of the indexes and the methodology behind
them will be published later this year.
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Figure 4
USSR: A Capsule View of How the Index is Constructed
Products
Products aggregated
with 1 July 1967 prices
SeCtOfS
Sectors aggregated
with 1972 input-output
table value-added
production series from
1950 to the present)
Manganese ore J
(Each item is a
Figure 5
-~Ferrous ores
Ferrous metals
Coke products
Refractory materials
USSR: Growth Rates of Total Industry
According to the Plan, Official Indexes, and
SPIOER Indexes
1o v ~
z
1960 1965 1970 1975
measures for several East European countries leads to
similar results. The upward biases of the official
_ indexes are demonstrated in table 2. Except for
Bulgaria and Hungary in the 1961-65 period, the
official indexes are always growing faster than the
Br~
-Planned
--Official
-SPIOER
Branches aggregated
with 1970 value-added
weights
Industry
Group
r (Ferrous metals
Nonferrous metals
Fuel
Electricity
Chemicals
Wood, pulp, paper
Construction materials
Groups aggregated
with 1970 value-added
weights ~ TOtel
Industry
Industrial materials
Machinery
Consumer nondurables
most bias (machinery, chemicals, and construction
materials).
Because the bias in the officially published measure is
greater in some branches than others, one obtains a
distorted view of structural change in Soviet industry.
For example, machinery production grew twice as
rapidly as ferrous metals in 1966-70-11.7 percent per
year compared with 5.7 percent-according to the
official data. According to our synthetic measure,
machinery production grew more rapidly than ferrous
metals, but by the narrow margin-7.0 percent per
year to 5.0 percent.
Comparison of similar synthetic indexes derived by
~ Thad Alton's group with the respective gross output
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Table 1 Percent
Average Annual Growth Rates of the
Officially Published and SPIDER
Indexes for Total Industry and
Selected Branches of Industry
1951-
55
1956-
60
1961-
65
1966-
70
1971-
75
1976-
78
Total industry
SPIDER
10.9
9.6
6.8
6.4
6.0
4.0
Official
13.1
10.4
8.6
8.4
7.5
5.0
Ferrous metals
SPIDER
I1.1
7.6
7.2
5.0
4.0
2.2
Official
12.1
8.8
8.0
5.7
5.1
3.0
Fuels
SPIDER
9.4
8.9
6.3
5.0
5.0
3.7
Official
10.1
8.5
6.5
5.7
5.9
3.8
Electricity
SPIDER
13.1
11.4
11.5
7.9
7.0
5.0
Official 14.4 13.1
12.3
9.0
7.1
5.4
Chemicals and petrochemicals
SPIDER
11.6
10.5
12.0
8.9
8.6
4.5
Official
17.3
12.0
14.4
12.2
10.6
6.6
Wood, pulp, and paper
SPIDER
12.2
12.4
8.2
7.0
8.2
6.2
Official
16.7
14.2
12.4
11.7
11.6
9.0
Light industry
SPIDER
10.4
6.4
2.6
7.2
2.7
2.8
Official
12.3
6.9
2.6
8.6
4.6
3.8
Processed food
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Eastern Europe: Average Annual Growth
of Industrial Production
Adjusted
14._2
8.7
5.7
4.6 _
Czechoslovakia
Official
5.5
6.8
6.7
5.5
Adjusted
4.8
4.3
3.9
3.5
East Germany
Official
5.8
6.5
6.5
5.5
Adjusted
5.0
3.8
3.4
3.5
Hungary
Romania _ _ __ _
Official _ _ _ _ _ 11.2 14.4 13.0 11.1
Adjusted 10.6 11.5 9.2 8.7
synthetic ones by appreciable amounts. Typically, the
spread in growth rates is at least 2 percentage points
per year and at times approaches 4 points.
Unfortunately, the data are such that it is impossible to
decompose the differences between the growth rates
into that caused by changes in double-counting,
disguised inflation, or other reasons. T. P. Hill,
however, used OECD data to estimate the possible
impact of using a gross output index rather than value
added.'b He found that the use of gross output
measures increased the growth rate for several OECD
countries by an average of 0.1 percentage point per
year for total industry. While this suggests that the net
bias is not large, he found considerable variance in
individual sectors. In about one-fifth of the industries
surveyed, the gross output measure would increase
growth rates by 2 percentage points or more and in
about one-tenth, the gross output measure would
'? T. P. Hill, The Measurement ojReaJ Product (Paris: Organization
for Economic Co-operation and Development, 1971), pp. 97-106.
decrease growth rates by 2 percentage points or more.
In analyzing his sample, Hill found the difference
between gross output and value-added growth rates
was apt to be the largest under two conditions. First,
the difference between the two measures tends to be
high in those industries with a high ratio of material
purchases to total output. Second, the difference is
more likely to be large for sectors with high growth
rates. The fact that Soviet industry is more material
intensive than its Western counterparts and has
sustained a rapid rate of growth in the postwar period
suggests that the extent of bias arising from using gross
output measures may be larger than the 0.1-
percentage-point difference computed by Hill.
The data used to construct the SPIDER indexes
indicate that disguised inflation exists in the official
indexes. Commodity data are either unavailable or
meaningless for some branches of our synthetic index,
however. To ensure a comprehensive sample for our
indexes it is necessary to rely on official data expressed
in value terms for parts of the wood, pulp, and paper;
chemicals; light industry; consumer durables; and
producer durables branches. The inclusion of these
value series may bias the synthetic indexes upwards
but not to the same extent as the official measures. By
dividing the samples for these sectors into physical and
value categories and computing the relative growth
rates, we can obtain a notion of the possible extent of
the pricing bias. Table 3 shows that in virtually every
case the value series grow more rapidly than the
quantity series. The most serious case seems to be
producer durables, where the value series are an
important share of the sample and the growth differen-
tial is largest-averaging 4 percentage points per year.
These results only suggest the possible extent of the
distortions that enter the Soviet value of output series.
The actual bias in growth rates may be somewhat
smaller for two reasons. First, there is some evidence
that the value series represent sectors of the economy
that probably are growing more rapidly than sectors
represented by quantity series. For example, transport
machinery is a quantity series and a sector that
traditionally grows slowly once a nation's transporta-
tion infrastructure is in place. On the other hand,
instruments and computers is a value series that would
be expected to grow rapidly at this stage of develop-
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Average Annual Output Growth of Selected Branches
by Product Sample Components
Wood, pulp, and paper
Quantity series
7.0
5.0
Official value series
18.1
17.8
Chemicals and petrochemicals
Quantity series
11.4
10.2 11.8 _
8.9
Official value series
_
20.8
19.4 16.1
8.7
Light industry
Quantity series
9.3
5.9 3.2
4.9
Official value series
13.4
7.6 1.2 1
2.1
Consumer durables
Quantity series
18.2
_ _
11.8 9.4
4.9
ies
Official value ser
17.5
10.0 1
10.0
3.3
_ ___
_
Producer durables
_______
Quantity series
1 l .l
12.7 7.3 _ _ _ _ _
3.6
Official value series
13.1
12.6 10.9
9.9
ment. Second, the quantity series may somewhat
understate growth by failing to catch quality improve-
ments-especially those for machinery items-that
may have occurred over the years.
The Bias Will Continue
Why have these biases in the official Soviet measures
been allowed to persist? The system works in a way
that profits everyone. Since plans are taut and difficult
to fulfill in real terms, enterprise managers find it
easier to meet their production targets and to receive
their bonuses by encouraging double-counting and
disguised inflation. Because labor productivity growth
is measured as the change in output divided by the
change in average employment, the goals for labor
productivity are also easier to achieve. Moreover, the
availability of incentive funds that can be used to
increase wages and fringe benefits to workers, and
thereby reduce labor turnover, and to increase invest-
ment are dependent on plan fulfillment. The produc-
tion ministries do not complain because their perform-
ance is evaluated on the same terms as the enterprises.
Finally the Council of Ministers and the Central
Statistical Administration may complain about spe-
cific shortfalls, but still the optimistic aggregate
statistics present impressive propaganda to demon-
strate the benefits of a centrally planned economy.
Which statistics should be used in analyzing Soviet
economic performance? The measures of the actual
gross value of output should be strictly avoided. As
Rush Greenslade wrote:
I cannot think ojany proper usejor aggregated
gross value ojestablishments unless all other
measures are missing. Gross value ojoutput of an
enterprise ...has utility for management analysis
and control. Conceivably gross value might be used
for questions ojoptimum scale of plant or optimum
location. However, more detailed breakdown for
products within plants would be even more useful
for careful analysis ojthese questions. The estab-
lishment classification, and the gross weighting, of
the GVO's would distort almost any economic
analysis or projection."
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
 Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8
Abetter picture of industrial performance can be
obtained by constructing a synthetic index that uses
constant prices and value-added weights.
Since the planned growth measures apparently avoid
the biases of disguised inflation and double-counting,
they may reasonably be compared with the synthetic
ex post growth measures. Also, the planned growth in
one year may be compared with planned growth of
another year to discern the shifting priorities of
economic growth. Finally, the planned growth for some
segment of industry in a given year may be compared
with the national average or some other component to
distinguish the relative priorities placed on them by
Gosplan. But never should the planned erowths.~_
gross output be compared with the official moss output_
measures.
Declassified and Approved For Release 2012/03/07 :CIA-RDP08S01350R000200340001-8