Created: 4/1/1971

OCR scan of the original document, errors are possible

Intelligence Report

Indexes Of Soviet Industrial


Copy no






Description ol SPIOER 3

Evaluation of t

Coverage of the Product Sample 7

Reliability of thc Quantity cud Price- 9

Adequacy of thc Value-Added13

Companion of SPIOER with Other Indexes of Soviet Industrial Production 19

Comparison of SPIOER with Official CVO

Comparison of SPIOER with Soviet Input Output

Comparison of tbe SPIOEH Producer Durables Index with an Index of

Soviet Investment in24

Comparison of the SPIOEB Consumer Durables Index with the Official

Series on Production of Consumer

Comparison of SPIOER with Western Indexes of Soviei Industrial

Evaluation of Western

An Alternative Check on29

Conclusions 33



Appendix A. The SPIOER

. USSR: SPIOER Indexes of Industrial Production andof Industrial Materials. Machinery, and45

Appendix C. Derivation ol Final-Product Weights for Civilian and



USSH: Indexes of Industrial Production 4

Soviet Ruble Series Incorporated in 5

ol SPIOER Sample Compared with Soviet Cross Value of


Derivation0 Value-Added Weights at Factor Cost for

Branches ol Soviet Industry

l SPIOER Value-Added Weights, Alternative

Base Yean Moved0

8. Comparison of SPIOER Weights with Alternative Soviet Weights

Derivation0 Final-Product Weights lor Civilian and

Military Machinery

Comparison ol Average Annual Rates of Growth Derived from

SPIOER aod from Official Soviet CVO ludnei of Industrial

Average Annual Rates of

lu. USSR: Comparison of the SPIOER Index of Producer DurablesSoviet Index of Investment in Equipment

Comparison of Sectoral Components of the SPIOER Indexes

of Growth with Those of Western and Soviet Indexes of

Item* Included in SPIOER



CENTRAL INTELLIGENCE AGENCY Directorate of Intelligence1


Indexes of Soviet Industrial


The CIA index of Soviet industrial production (hereafter referred to by its acronym, SPIOER) serves as the basis for many CIA reports on the Soviot economy, including estimates of the growth of GNP and measures of the rate of growth of technical progress in thc USSR. This report describes SPIOER and explains why an independent estimate of Soviet industrial production Is necessary, lt also examines the possible biases and range of uncertainty in SPIOER and compares SPIOER with die official Soviet index of industrialThe report is onecries of reports designed to present the major components underlying CIA estimates of tho growth of Soviet GNP and to appraise their validity."


he industrial production indexes issued by tbe Central Statistical(CSA) of the USSR are indexes of gross value of output (GVO) in constant priceshey measure the change in tho value of outputiven branch or of industryhole, where output is defined as the total value of output of all die enterprises in the branch or industry.VO index represents something more than |ust the net contribution of each enterprise toe value that each enterprise addsroduct as it passes through various stages ofVO double counts many

, The Cnw National Product of the.ECRET, describe* ihe accounting framework uithin which the ertimMon of Soviet economic growth Uke* place. ERovfet CNP3,ECRET, dts-ana Ihe eonwrnon of CNP ettimate*08 price baits.

Note: Tha report tua prepered by Ihe Office of Economic Reteareh.

of the inputs ihat enter into production by including in the value of output of each enterprise tbe raw materials, power, and semifinished products purchased from other enterprises and already counted once (or more) as output at an earlier stage of production.

Because of this and other biases thc official Soviet index fails to meet the needs of intelligence estimates. The importance of industrial production in assessing Soviet capabilities in such areas as defease and foreign bade or in evaluating prospects for economic growthore exact measure of the growth of industrial production. An alternative, then, to the official Soviet index must be constructed.

Ideally, one would like to measure the change in the net value of output of industry or of industrialis, gross value of output net of all material inputs other than capital andigorous construction ofet output Index would require doubleis, for each enterprise the value of output deflated to constant prices minus the value of material inputs also deflated. These separate series for outputs and inputs and the deflators are not obtainable on an annual basis in most countries and for tho USSR not at all The compromise approach used by the Federal Reserve Board (FRB) for industrial production in thc United Statesample index ofproducts weighted by value addedase year.

SPIOER is CIA's alternative to the official index. It is intended to bein form and construction, as nearly as the data permit, to the FRBproduction index for tbe United States, la iu unclassified form, SPIOER is generally known as the Greeoslade-Wallace Index. This unclassified version of the index was prepared far the Joint Economic Committee (JEC) of tbe US Congress for inclusion in2 publication. Dimensions of Score*Power, and has served as the basis of contributions on Soviet industrial production appearing in subsequent JEC publications.

A preh'nunary calculation of industrial pioduction8 value-added weights is presented in this report and is being used in CIA estimates of Soviet GXP5 prices. NeverU*eLesi, this report focuses analytically on SPIOEB. in its more commonindex0 branch value-added weights5 price weights for tbe aggregation of pioduction within branches. The reasons lor usingbased version of SPIOER arc twofold:

So far,bascd SPIOER is incomplete because8 product prices necessary to replace5 prices used since the inception of SPIOER are unavailable.

Tbe discussions of the relative merits of SPIOER and other Western Indexes of Soviet industrial production have mostly concernedbased SPIOER presented in several JEC publications. In addition, these other Western indexes rely5 product price weights, as docsbased SPIOER.omparison of SI'lOER and other Western indexes can be carried out more conveniently ifbased SPIOER serves as thc standard.

At the same time the properties8 bated SPIOER are considered In this report,s clear that all of thc judgments applicablePIOER based0 apply also to SPIOER8 base.*


SPlOEItalue-weighted indexo measure the growthample of industrial products. At present the sampleino Item* representing Individual commodities ot commodity classes withinhysical production series are converted to value scrtea by the use5 prices. An annual indca of output for each branchew exceptions) directly from tbe total value of the sample items included in the branch. Each branch, in turn, in0 value-added weight compnled from the wage bill, depreciation charges, and imputed interest on fixed and workingPrices must be used for weighting items included in the branch samples been use of the absence of value-Added data for thc smaller subdivisions ofggregation of thcalue-weighted branch indexes yields the index of total Industrial production shown ia Table L

The sample of physical commodities whose production Is announced by thc Soviet Central Statistical Administration is deficient in two respects. First, most of the reported categories tend to be highly aggregative, thereby obscuring important changes in product mix. Second, MM important branches are omitted entirely for security or other reasons.

To reduce the Impact of the first problem, the series in tho SPIOKR sample have hern disaggregated wherever possible by estimating changes over time in the Soviet product mix. Production of motor vehicles, tractors, and rad-road locomotives, for example, is broken down in SPIOER by modelpermitting tlie addition of new models and the deletion of discontinued ones in the sample. The samples of such diverse commodiliei as ferrous metals, machine tools, cement, flour products, ami consumer electronicsele-

ii Italc lo ehnaie betweennd lUtiH-bond indexes from theof theirulntVuiion. where mmtU, of IWH price,0 prices hae onlyinor effect iter, pungrapheplacement0 value-added iceighliH value-added ueteht* hat even lea of an Impacl (ten rwnw-apn ffl.

"For analufral propose, eulmimed under three Industrialindustrialachinery (macho* hudd-ng endend cennmer nondurable goods The temhu of the Arte melonpredetermined beeights of the breathes assigned to them endexe na ^dependent tigndicemce. The eeemme err eOcxeted among, ihe three sector,he aeceenpanirtne. tabvletetn

Somber of Une Item,

meterleis Ueehmew*


the Index slnruld hate priori and weight, for tht same base near. Prices an taken tergela fromrice handbooks, uJuch are mailatA* oii/tfort of thev prion, however, remained In iffeclnd thus can be regarded0 prices

dlmkifidb UBifuuKjtf/xx ,




rpll| iwj



i'll tt

p*ol w




. niqajtipbda rahxb






















m n







w sb


16 1





i omi hsm *W


mil. :

doihbpaij mimmii>*b|



vision sets, tape recorders, and phonographs) are also significantly disaggregated to reflect changes in tbe product mix.

umber of recently published Soviet series that show the production of various commodities in rubles (allegedly in constant prioes) are bebeved to be constructedanner similar to that employed in SPIOER for motor vehicles, tractors, and railroad locomotives, as described above. Some of these series have been incorporated into the SPIOER sample, primarily to provide representation for groups of heterogeneous machinery products not otherwise included or inadequately represented in the reported sample of physical(agricultural equipment, electronic and nonelectronic instruments, chemical equipment, and thehcoviet ruble scries which have been incorporated either directly or indirectly into SPIOER are listed below in Tabic 2.

Table 2

Share of Output

Soviet Buhl*SPIOER Sample0

of forest product* sample

Agricultural machinery (excluding spareof producer durable* sample

CliMtiicai equipment (intruding spareof producer durables sample

Equipment for canpimcr-otlcntcd Industriesof producer durablesspare partsotherwise noted)

Equipment for trade and public dining of pirjaueer durable* sample

Food processing equipment nf producer durables sample

Milling, granary, and grain elevatorof producer durable* sample

Papermakmg equipment (possiblyof producer durablesparte)

Printing trade equipment .of producer durables sample

Textileof producer durables sample

lectronic and nonclectroakof producer durablesspire parts)

of consumer durabks sample

Sewnf soft goods sample

lthough these Soviet ruble series account for only aboutf the total value of Ihe SPIOER sampleheyignificant weight in their respective components of the sample: forestnd civilian. Within civilian rriachinery, thc Soviet rubles series accountf the producer durables samplef the consumer durables

he SPIOER sample is limited insofar as possible to tlie final products of each branch in order to minimize th* incidence of double counting that characterizes thc Soviet CVOumber of SPIOER scries arc adjusted to net out production that is included as inputs into other series in the same branch. Products of the extractive industries such as ores, rough-hewn timber, and crude oil are omitted entirely from SPIOER or are included only to die extent that they arc consumed directly outside their respective branches in unprocessed form. Net production of crude oil, for example, is computed for the indexesidual, by subtracting from gross production of crude oil that portion which goes on to be refined. 'Ihe scries that are adjusted to eliminate

Soviet Ruble Scries incorporated in SPIOER

or minimize double counting within branches are identified in thetabulation;

to Exclude


Crude oil



Military machinery

Industrially processed meat Fish and fish products (usable part of fish catch)

Cramrlated sugar

Vegetable oil

power consumed by power stations in the generation of electricity.

Crude oil used as stock in tlie production of refined petroleum products.

Cement used in precast concrete articles, asbestos cement shingles and pipe, and concrete blocks.

Rock products used in precast concrete articles, cement, lime, concrete blocks, and bricks.

Cypsum used lo cement and dry gypsum plaster board.

Common-use durables, such as trucks, tractors,equipment, and instruments which arc pio-cuied by the military but already counted as part of civilian machinery production.

Meat consumed in the production of canned meat.

Fish included in the production of canned fish.

Granulated sugar undergoing further refining. Vegetable oil consumed In the production of margarine.

forest products, chemicals, soft goods, and processed foods branches of the SPIOER sample contain significant amounts of internal double counting, but it was impossible to make appropriate adjustments to all line items in these branches. Therefore, internal value-added weights were introduced within each of the four branches to dampen the effect of such double counting.

Although most of the production scries in the SPIOER sample are taken directly, or are derived indirectly, from Soviet open sources, publication of production data for uniquely military products and some strategic materials is expressly prohibited in the USSR. These components of industrialare too important to be ignored, so certain production estimates must be made for SPIOER. Data in the following series in the SPIOER sample are estimatesenotes that thc estimate relics on material classified SECRET)':

ntimony (SS Cadmium (S) Copper Lead

Magnesium Mercury (S) Molybdenum (S) Nickel (S) Tin (S)

<S) Tungsten (S) Zinc


Chlorineammonia (S)



Maciuhekv Civilian aircraft (Si-Civilian shipbuilding (S) Military machinery (S)

addition to (fees*eHet that ore esfrmoW elmott inll won. th* breakdown by model of motor vehicles, intclon, ond railroad toeemoliuejlw estimated for ell yam, uUh reported total umt production servingontrol fifiirre. Simitofy. most of the net production series doKrtbed above are oho rttimaied fo< all veers butthe hosts of adjutimenti in reported gross prOdurtiO'iumber of Other terie, in the SPIOER nrrapfe ate estimated or intefpolaltA for iWicioW yeari that have been omitted from StnUet Hatirtieal reporting.



mwt critical of these estimates for SPIOER is the series on tbeof rnihtary machinery (including; spacen the absenceofficial Soviet information on this significant category of production,scries on military procurements of machinery andubles and estimated by thc Office of Strategic Research, isOER for us* in the index. The "military machinery" Index is thenthe civilian machinery index0 weights based on the value ofof the two categories of machinery.


SPIOER Is such an important element of many CIA reportsSoviet economy, it is important to know Its strengths and weaknesses.of these can be gained by considering the adequacyhe price, quantity, and value data used to construct thegrowth for individual products,he weights that arc used tocomponent indexes into more aggregated indexes.

Coverage of the Product Sample

The SPIOER sampleine items is not the resultandom selection of products but ratheronscious effort to assemble as complete andasket of final products asbe sample appears to be reasonably representative of the majoi branches of industry, although the total number of products is not ui great as might be desired. The FRB Index of US industrial production, for example, it based on an averageeries (but includes goods at all stages ofmaterials, intermediate goods, and final products).

imple count of the number of line items in the sample isomplete indication of the size or relevance of the sample. In manyarge number of differentiated products are subsumed under one line item in the SPIOERore valid comparison of coverage is the share of total value-added in industry that is represented by tbe products in the SPIOER sample. On this basis, the SPIOER sample coversi the FRB sample (instead ofn terms of the number of lineimilarly, in titrms of value, thc SPIOER sample of final products amountedf Soviet official industrial gross production0 (seelthoughraction of all thc items in the Soviet industrial classification code. Furthermore, the coverage of SPIOER is substantially largerince the official CVO includes proportionately mon- double counting than does the total value of output of the SPIOER sample.

he comparison of the value of final products in thc SPIOER sample with thc CVO values indicates that thu SPIOER sample equals half or more of the gross output In all branches except nonferrous metals and machinery. Thc very low percentage in these branches reflects heavy double counting in the Soviet CVO, and perhaps inadequate coverage of the SPIOER sample, or aof these factors. 'Ihe SPIOER coverage appears to be bent in the case of electric power, fuels (coal products and petroleumorest products, paper and paperboard. and processed foods. The very Urge percentage shown

a lot of lht> commadHlet indudtd In the SPIOKHppendlt A,

Coverage of SPIOER Sample Compared with Soviet Gross Valuefor

row Output*

Sample as a


of Soviet CVO













and papeiboorii


. -





industrial productr)n


For Ihe furls induitrry (coal, petrobum, peer, ond other fuels comhrned).

ajiolualc price* have been approximated by deducting estimated lurnowr fawt and dlstribwion "uarglnr from the value of the tample to null prices.

4 Global figure, includingcitfion rubier of gross output produced hu other fcrflncAea of industry not covered fry (he SPIOER sample.

for electric power results from SPIOER's coverage of electric power inattached to enterprises in sectors outside tic electric (and thermal) power branch of industry.*

inally, the types of products not represented in SPIOER are believed to be either relatively unimportant or fairly typical in terms of growth so that their absence docs not greatly affeel the index: minor fuels (peat andome glass and porcelain products, abrasives, pots and pans, spare parts for some types ofearings, jewelry, sporting goods, musical instruments, some metal products, civilian corrimumcation equipment, toys, drugs andpublishing, and office supplies.

There may be other factor, than rmeraee uihfch account for ihe differences between the tekri in Tableovtble dentations of SPIOER prices from true "axeroe.e' prices, conceptual* differences in the method of computing and atX-eaUtoe output al the branch level and possible errors in the estimates of CVO branch catties. In the mo-shinery branch, for eiampU, the SPIOER value is low* than the CVO oofu* because it does not include the large amount of secondary production iproducU produced bv machinery plant* vhkh should be classified in the output of another branch) turned out by the machinery sector.

"Several of thee-uofue series wed to SPIOER (see fablenclude spore patlt production, but other output of spareot included in SPIOER. Soviet ruble scries on .pore port* production are available for tome products but are no* wed in SPIOER. because evidence indicates that these include only tenlralited production of spare parts, for most type* of spore ports, homeoer, the bulk of ihe output it apparently produced by the uteri themselves, the to-colled noneentrohied production.

Reliability of the Quantity andData

The majority of line items Id SPIOER ire based directly on official Soviet production series expressed in physical units. In most cases these time series are now complete for all years0f course, thc USSR does not report all categories of industrial production, nor are all of those it does report expressed in physical units. As indicatedandful of categories arc reported in rubles.

The commonly held view of Western scholars is that Soviet production series expressed in physical units are reliable in the sense that they reflect quite accurately the actual number of given Items produced throughout the Soviet economyiventate control over statistical reporting is highly centralized, the reUtrvely simple process of counting and recordingunits of production precludes some of the problems that arise whenthc value of production, and the physical production data are susceptible to verification through invoices and records pertaining to the final disposition of industrial goods.

Similarly, the reported prices obtained for the various commodities in the SPIOER sample seem adequate Most of these prices were taken from two multivolume price lists issued by tbe Soviet government for the use offinancial, and control organs throughout thchese are Ests of so-called "enterprise wholesalehich are set by the government with the aim of making them uniform throughout the economy and operative for long periods of*

Although the available statistics on physical production series and Ihe reported wlmlesale prices probably aro reliable, they may be inadequate in Several respects for the purpose of constructing an index of industrial production.ifficulties are discussed in the following sections.

New Products and Changing Product Mix

low example, Crtgoru Unman, Soviet StatitUc* of Phynal Output of IndoHrtal


Ministry of Finance, Spnvorhnflitraitefnyye ma tnHandbook of Price, for Construction Material, ami Equipment)nd USSK Ministry of the Coal Industry.lxiruiinvaiilyveol'noyspravochnlk (Handbook of. Material* and Equipment used to ihe Coal

"'The enterprise wreAcsale prices tstablHhed5 remained in effect with only minor adluitount,etpHa much controversy about the rale prices should playthc Soviet economy, they ham Imdrdan accounting rather than an oltoeatwnal function.

ITA* problem, however, is not unique loen ihe PUB index is atteged to understate the growth of US industrial production as aof failing to measure Incnese* to Ihe average quality of products.

jii T

*ul*lrl j

all of the physical series can be disaggregated sufficiently toof the effect of new products and changing product mix. This ishandicap because the entire outputiven commodity orhas to beingle "average" or "representative" unit product mix within aggregative seriesiven branchto raise the average quality of the commodity group, thenrelying oa (he aggregate total alone will understate the actualproduction of final products by the brancb-l

hc SPIOER treatment ol ferrous and nonferrous metals indicates the nature of the product mil problem. The ferrous metals sample is broken down intoategories of finished steel products, allowing SPIOER to reflect changes in the product mix. The nonferrous metals sample,ased on the output ofifferent nonferrous metab in their crude form, pre-doding recognition of changes that may be occurring in thc mix of finished products of each of tho nonferrom metals. Other series in SPIOER which may not adequately reflect changes in the product mix include tho series for synthetic fibers and plastics in the chemical industry sample. These aro areas where significant quality ehange may have been occurring, but where the SPIOER samples are not disaggregated enough to reflect these quality changes. Similarly, the SPIOEH series on such consumer durables as refrigerators, washing machines, und sewing machines are not broken down by model and thus do not reflect changes in quality or complexity over time-

The ftepresenfonVentss of Soviet AnncHirtced Value Series

a number ofinvolving great heterogeneity inUSSR publishes ruble series of output in place erf physicalAlthough complete time series for all years0 have notthese ruble series have been used in SPIOER for thcSince the ruble-vmhi* scries presumably cover all products in athey have thc merit of including new products and reflectingtime in the product mix. In addition, in some cases there is no otheravailable for constructing series covering the product lines inprincipal caveats in using the ruble-value series are that thethe products is not known (there may be some intermediate orincluded with finalnd the prices,ay include some temporary prices that Introducebias."

s noted earlier, die SPIOEH series for tractors, motor vehicles, andarc broken down by model and thus should also reflect changes in product mix or average quality. These sencs, however, may aUo lietrice bias. Soviet list prices are used to value tbe new models, and it is often uncertain whetho the difference in price between new and old mods-Isquality improvementidden price increase.

he bulk of the ruble-value scries as well as the series disaggregated by modd are found in the SPIOER sample of civilian rnachinery. Furthermore. Ihe share of ihcse quality adjusted series in the SPIOER producer durables sample increased0ith physical quantity series representingemainder. Because such quahty-adjustcd scriesthose of Soviet origin) overstate growth due to the "new product" pricing problem while machinery series not adjusted for quality change tend to understate growth, combining these two types of scries into one sample results in an index which, if not precisely calibrated, probably falls within an acceptable lange.

Possible Divorgonc* Botwoen Growth of Final Output ond Growth of Nil Output

tttaOMafc* off.

if SPIOEH were to measure accurately the growth of finala branch or sector, it docs not necessarily follow Ihat il would reflect the

growth In net output of that branch or sector. Algebraically, final output and net output arc differeonated as follows:

Finalross output minus sales of sector to itself.

Net output "final output minus material purchases from other sectors


Net output =gross output minus total material purchases.

In other words, purchases from other sectors are included in final output, but not in net output The relation between growth of final output and net output, then, will depend on bow tbe ratio of final output to purchases from other sectors changes.

It could be argued that over time the production of increasingly complex products require! more processing and hence greater value added or net output per unit of final output. Even when the nature of tbe product remains the same, the value of net output per unit of production may rise if production methods become more efficient in thc use of raw materials. The SPIOER branch indexes, many of which rely on physical production series unadjusted for changes in product complexity and all of which assume fixed proportions of inputs to output, might well be challenged on this count.

N>vertheless, the available evidence suggests that for industryhole there is not an appreciable divergence between the growth of final output and the growth of net output According to the CSA. total gross output ingrew at virtually thr same talr a> net output8f course, even if valid, the CSA findings could have been the result of net output rising faster than gross output in some branches and more slowly in others. Preliminary resultsomparison of98 Soviet input-output) tables (which are presentedatereem to indicate that this is indeed the case. Differences between the growth of final output and the growth of net output among the branches tended to be offsetting.

Possible Bios Duo to the Price Weight*

questions arise with regard to the price weights employed inthr SPIOER indexes of branch production: (I) are the indexesaifectcd by the use of some estimated pricesowrices for the intended purpose? Ideally, all commodities in tbeshould Isn represented by enterprise wholesale pricesuniformity in the valuation of all products in the sample ntwl anrelative weighting ofumber of the commoditiesSPIOEH, however, were not represented in the available price lists, sowholesale prices had to Ih- fosimlor somehow devised.


he absence5 enterprise wholesale prices rxfmarily affects tbe consumer goods indexes. Enterprise wholesale prices of connimcr durables were estimated by amply7 retailo eliminate turnover taxes and distribution charges. Because retail prices ofgoodsarge percentage ol turnover tax which varies over time and from product to product, it is highly unlikely' that this uniformfactor (which preserved the relative- prices of thc retail price list) resulted

in correct relative weights (or consumern the case of consumer non-durables (soft goods and processed foods) the prices adopted for use in SPIOER were retail sales prices adjusted for trade charges {but inclusiveere too. this procedure probablyet of relative price weights for tbe consumer goods sample different from the price weights which would result from using wholesale prices. Unfortunately, so little is known of individual turnover tax rates that the direction of the index bias cannot be determined. To minimize possible distortions arising from the price weights, soft goods and processed foods have been aggregated into subhranches and assigned value-added weights.

Assuming that most of the SPIOER prices represent the actual enterprise wholesale prices5ore important question is whether such prices reflect the relative economic cost of producing different productsranch.5 prices have been criticized for not covering the full cost of the capital used as well as for allowing wide variations in branch profit rates. Afterrice reform for several years, the Soviet government carriedhoroughgoing revision of industrial prices inapital costs were taken into account in thc pricing formulas, which were designed to insure roughly equal profit rates for all types of production. After the price reform, thc dispersion of profit rates narrowed, mostdisappeared, and thus the new prices would seem toetter index of relative costs than5 prices. But so far, the basic price documents necessary ro reprice the SPIOER commodity sample7 prices have not appeared.

For some branches, however, enough information about the new prices has been published toest of how much the substitution7 prices for5 prices would change the SPIOER branch production indexes. The comparison for production of ferrous metals, nonferrous metals, petroleum products, construction materials, aud the soft goods branches is shown in the accompanying tabulation.

Average Annual Ratei of Growth (Percent.)


5 prices



5 prices








Theor only IS of thetems In the construction materials sample7 prices could not be identified fo, the full sample.

at least one case, subsequent information hoe thown the price to have been in great error. The price forand Hocks derived uriag the above-describeduble, pe, unit) rciultcdotal oafue of production0 ofillion rubles,ecently published Soviet figure computedhe same methodology at SPIOER (except that each unit icas weoihted by Its own true price) shoved that thc total production of aB "timepieces" amounted to onbjillion rubier in trUcrpriee wholesale prices.

n these branches, atonversion of price weights57 base is hardly worth the trouble. There is reason to believe that thc

same is true for most of thc other branches. Where the sample is essentially limited tosingle homogeneousin coal and electricnew internal price weights will not affect tbe branch production index. The forest products branch index has internal weights based on shares of branch employment for its four separate production series, ao new prices would not change the index. Not enough is known about relative price changes in the food and chemical industries to judge whether thc SPIOER production index would be greatly affected, but here again any significant price changes would be muted by the use of internal value-added weights for the major subbranches.

Substitution7 prices5 prices would probably have as its maloreduction in thc growth of the machinery index. Prices of heavy machine building products were raised inhileof the instrument and radio industry were cutnd prices onproducts were reduced. Enterprise wholesale prices wereiii rvamplc. for television sets, radios, watches, and refrigerators. Tbe percentage changes do not represent the difference57 prices in those cases where prices changed betweenndut their not effect would surely be to lower the weights of thc faster growing machlnrry products relative to those of slower growing machinery products.

A partial test of the effect of7 price weights into the producer durables sample can be made.8 the CSA published thc ratios7 prices5 prices for all of the ruble-value% of the value of the SPIOER producer durables sampley restating the rublo-value series7 prices, tbe growth of the ruble-value series is reducednd the growth of producer durablesear. As the physical productionarc predominantly heavy machinery items, presumably their weight would rise if they could be valued7 prices. If so. ihn rate of increase of civilian machinery would be pared still fmllver.

Adequacy of the Value-Added Weights

formula'ion ofeights

weights are of two kinds: the value-added weights for theof industry and thc price weights for the individual products within

tbe branches. As explafixtd above, the sample of products in moil branches is simply aggregated into branch indexes by the use of wholesale prices Value-added weights would beut there is no way of calculating value

added in the production of individual commodities.

weights employed in aggregating the branch indexes into sectoral

indexes and thc index of total production, however, are true value-added weights in that they attempt to measure the earnings of the productive factors thatfor the value added by each branchase year. To conform with thc

ase-year weights employed until now in the estimation of Soviet CNP.

as selected as the Iwse year. The steps in the derivation of tin! SPIOEB

'The ratio,7 prices5c. imrmnurnW.hemtailood proefsting; textile industryquipment for Irade and public diningaper industry equipment, lOS; and printing trade.

ranch weights are set out in Tabic 4.

Table 4

USSR; Derivation0 Value-AddedFactor Cost lor Branches of SovietNew Rublesrice.)

and Other InetMM'

+ Interest Charge*

Added - al

Factor OkI

Distribution ot Value-Added Weights





metals .



and paper board .







Including social insurance.

' Calculated as anharge on fixed and working capital.

'The amortiratson rates Introduced3 werereference to thorn actually tn effect0 because they are believed to reflect mote accurately the actual service licet of fixed assets. Only that part of the amortisation rate that relates to replacement (in contrast to capital repair) was used to calculate depredation.

"The imputation of anather than some other choree oa capital Is largely of academic importance insofar as the behavior of Ae index It concerned. To test this proposition, charge* ofer* assumed as the extremesange within which the true rale would fall. Ai shown below, tlie difference in growth rate, resulting from these various onumpriiHu about the rate of return on capital Is very aught.

Average Annual nates of Growth (Percent)

Imputed Intereit



he data inrc based largely on Soviet statistics that arereliable. Value added to output in the form of wages and other income was computed using fairly detailed information on annual average wages and employment, social insurance rates, and the like, reported by branch of industry. Depreciation allowances were calculated using amonOzation rates reported by branch of industry and reflect the cost of capitalFinally, interest charges were estimated by imputing anharge on the fixed and working capital in each branch of industry. No separate provision for profits was made in computing the value-added weights inasmuch as the imputed return on capital conceptually covers profits. Tbeharge is somewhat arbitrary, but preferable to accepting profitseasure of the return tooviet pricing0 permitted excessive variation in branch profit rates.

Branch WoeaiSfs for DmVonf Years

Value-added weights computed for the5 were used In SPIOER at first. These weights consisted only of estimated wages and depredation.ew value-added weights wore calculated foi then concept,0 weights were quite different from5 wdghrs, not only because of the sdcctiooifferent base year but also because of more complete and reliable statistics on emptoyment and wages and tbe indusion of an imputed interest charge on fixed capital.90 base-year weights were further revised to indude an imputed interest charge on working capital. Finally,aluo-added weights were calculated for base8 as part of aestimate of Soviet CNP8 prices. Despite these changes in the methodology used in calculating the weights, thc actual differences in the weights are small, as can be seen innder the revised procedures, the weights of most of the materials branches and tbe processed foods industry are somewhat higher while the weight of mac htaory Is lower.

Changing base-year value-added weights does not affect overall industrial growth significantly. Substitution of Ibe earlier and somewhat5

wcights does not yield very different results from0 weights, as shown in the accompanying tabulation.

Annual Hates of Growth (Percent)

Base-Year Weights



'Bated on ooUte-^dded weights in ute. moved forwardValue-added

' Value-added weights8 bated on industrial value-added at factor cott.

Moreover, indexes aggregated using value-added weights87 price reform) show rales of growth of industrial production that are almost identical with those calculated0 weights.*

Product vs. Establishment Weights

n SPIOER. the industrial activity represented by the branch weights and the branch product samples do not match precisely. The weights are computed using wage, employment, and capital statistics based on an establishmentof the branches of industry, whereas tbe indexes arc constructed using final products wherever produced in the

Under an establishment classification, the outputiven branch of industry represents the activities of the producing units (establishments) that formally "belong" to the industry (usually on the basis of ministerialStatistics on employment, wages, fixed and working capital,and gross value of outputiven branch of industry are computed simply as the sum of the corresponding statistics reported by the individual producing units within the industry. For some branches of industry, particularly machine building, metallurgy, and chemicals, many producing units have "captive" plants and sliops engaged in tbe production of secondary goods tbat are produced mainly in some other branch of industry. Therefore, the statistical data published on an establishment basis do not precisely match the statistical data on final products as far as classification is concerned.

To evaluate the effect of thcet of alternative value-added branch weights basedroduct classification were constructed from9able. This set of alternative value-added weights is onlydifferent from thc SPIOER set (seeo the substitution of

rales of growth of individual branch indexes are unaffected bu the choice of base welshts. The rales of growth of SPIOER sector indexei dcrieed8 weight! are sh In the accorrivoniito* lobulation.


i ihe accompanying tabulation.

Machinery .


Total iotluitrial production .

Thc indexes from which these rates of gtowth art derived are given in Appendix B.

Weight, derived from9 Soviet his-.toutput table and moved forward0 by SPIOER branch mdeve, to facilitate comparison.

trtimatrd from Soviet data on the percentage of fou output claimed by the rne-vr branehe, of industry classified on anboats. Tha Sooirt data have been ad-Med so lhat the there, of Ae branches add

the IO weights for the SPIOER weights has an insignificant effect on thegrowth of total Industrial production (see the accompanyingn fact, even the substitution of Soviet GVOare at much greater variance with the SPIOER weights (see Tablenot produce substantially different growth rates. Tbe lower rate of industrial growth obtained using GVO weights is largely explained by the fact that the CVO weight for the slow-growing consumer nondurable* is inflated not only by tbe many stages of processing that occur in tbe soft goods nnd processed foods industries (most of which are counted cumulatively) but alao by the inclusionizable amount of turnover tax. These distortions arc absent in the SPIOERalue-added weights-

Annual Bain of


SPIOER (value-added)


CVO (turns output)

Baaed0 weight.

be lack of sensitivity of the index to alternative value-added weights demonstrates that the accuracy of SPIOER does not depend much on tbeset of branch weights used, although the0 weights adopted9 are conceptually superior to the weights used in the earlier years.

I *^

n at least one instance, however, the choice of weightsifference. The growth of output of military hardware has varied markedly over lime from the trend in the production of civilian machinery, so thc behavior of the overall machinery index depends on the share of output assigned to military products. In the absence of value-added weights for the various branches of machine building, final product weights0 were constructed for the civilian and military components of machinery production on the assumption that the final product in each component is roughly proportional to the valuehe same procedure is used to separate civilian machinery into producer durables and consumer durables."

Table 7

USSR; Derivation0 Final-Product Weights for Civib'an and Military Machinery

of Output (Billion Now Rubles)'



Producer durables


parts lor repair






o( end items



and maintenance




For the deriietton o/ these valuer, te

Appendix C.

Enterprise wholesale price,

The present method of0 final product weights relies on es-timates of the value of final products manufactured in the machinery sectorhese estimatedthe weights derived fromshown in Tablet should be noted thats assumed tourely industrial activity and is assigned wholly to the machinery sector, while only half ofs assumed to originate in machinery.

Thc weights allocated to producer durables and consumer durables within the SPIOER machinery sector are also based on tbe final-product weights shown in Tablef it is assumed that the outlays for spare parts andre divided proportionally between producer durables and consumer durables, then weights may be calculated directly from thc values shown for producer durables and consumerillion nihles5 billion rubles, respectively).

'These is not sufficient information acatlable on the pricing of military machinery in the USSR to prove or ifiiprot* the validity of (his arsumpUon There it riason to beliete that0 the profit margin frt the price* of military machinery icas lest then in the prices of ciWfan machinery. However, the present method of valuing the final product of mUilory machinery for usePIOER largely dreurrnxntt thu problem.

"Separate weight! for producer durables and contutner durable* were frnt introduced into SPIOERhe two rets of machinery weights are rhoicn in TahiV 7.


inny be compared with other Indexits of Soviet industrialThe most important of the Soviet indexes Is. of course, tlie officialof industrial CVO. but other Sovietthe Soviettables9oviet series on investment inan official series on the production of consumerwith SPIOER branch and total indexes. Second, two Western indexeeand MoorsteenPowdlSPIOER in concept andcompared with it. In addition, another Western index recently devisedThornton deserves mention, as it restsuite different methodFinally, SPIOER is compared with an index of Sovietderived from the statistical relation (ti other countries betweenproduction and the output of certain basic industrial commodities.

Comparison of SPIOER with Official GVO Indexes

A priori there is reason to expect SPIOER and CVO ii>dexes. both branch and aggregate, to be at sixes and sevens. Indeed, for several reasons tbe CVO indexes should gross' faster than the corresponding SPIOER indexes. First, the GVOiven sectorood deal of double counting. Should double counting increase over time, as would occur with increased specialization inVO index svould be uiflatod. SPIOER attempts to uvold this problem by measuring the output of final products and excluding Intermediate stages of production. Second, the introduction of new products at inflatedprices could inflate thc CVO index. But measures of CVO may diverge from SPIOER for other reasons. CVO measures all production whereas SPIOER is basedample, and this could cause tbe growth of the two indexes to differ. Since the CVO indexiven branch includes production of secoridary products by that branch, whereas SPIOER measures only primary products, indexes will generally differ on this account. For thc industrial outputhole, the CVO weights used to aggregate the brunch indexes differ markedly from those used in SPIOER (seentroducing another discrepancy.

ompares average annual growthor all the branches of industry included in the SPIOER sample with corresponding Soviet GVO branch indexes. The growth registered by CVO indexes in all cases exceeds tbe corresponding SPIOER indexes, but tn no case is the discrepancy as great as in the respective machinery indexes (civilian and rnihtary machineryhe differenceercentage points between the SPIOER ma chincry index and tlie GVO index of machineryredibility gup. For thc materials and consumer goods sectors combined (that is. total industrial production minus thc machineryhe dllfeieno between the average annual growth rates ol SPIOER and CVOs onlyercentage point Although partly due to the particularly large weight given to processed foods and soft goods in tbe CVO index (seehis CTHnparlsoe. nevertheless shows that the (sulk ofercentage points difference between the two indexes of total ittdustrial production is due to the differences in the machinery indexes.

pattern of divergence is significant. Machinery and chemicalsonly two branches wliere the difference between growth in GVO and growth


in thc SPIOEH index is consistently greaterercentage points. In all of the other branches except ferrous metals and soft goods the gap ranges from Vtercentage points. Ferrous metals and soft goods are the best behaved matches by far, the difference amounting toercentage point. Thus the difference between GVO and SPIOER growth is greatest in the branches where the assortment of products is the largest, the problem of new products is most important, and the opportunities for increased specialization in production have been most obvious.

It is revealing that one of the smallest differencesVOPIOER index is found in ferrous metals, where the coverage is fairly complete and the information on thc assortment of steel products Ls reasonably detailed, the sample consistingist of finished steel products. Thus the SPIOER ferrous metals index has been able to reflect the widespread shift in steel output from hot-rolled to cold-rolled products which causes the SPIOER index to growear, while steel ingots increasedear. On the other hand, the production data for nonferrous metals are not detailed enough to reflect the trendore finished product that undoubtedly has takenparticularly in view of the rapidly changing requirements of the space, rnihtary, and electronics sectors.

In some instances, the differences between thc GVO and SPIOER indexes are plaiuly the result of differences in coverage. Two examples are electric power and construction materials, branches where the product list is narrow and quality change has not been significant, Tbe SPIOER electric power index covers all electric power wherever produced while tbe GVOxcludes electricity generated in powerplants attached to enterprises classified in other branches of the economyncludes the generation of centrally supplied heat. Since large, independent installations have supplied an increasing share of total electricity and thermal power has also grown rapidly, thc SPIOEH index Is better than thc CVO index in describing trends in the total output of electricity alone. One difference in the product list accounts for virtually all of the margin separating the CVO and SPIOER indexes for constructionThc SPIOER sample includes glass and refractory materials, but the CSA classifies them in the glass-porcelain and ferrous metals industries,Removal of these products from Ihe SPIOER construction materials index raises tlw growth rateear.f apointhan Ihe growth rate registered by tbe CVO index.

CVO-SPIOER comparison for fuelspecial case in that theof Ihe different fuels in the GVO index is uncertain. Some(he part of SPIOER is possible as thc average value of the refined oilincreased over time. As for the remaining branches ofproducts, soft goods, paper and paperboard, and processedsamples seem to be fairly representaHve. While they may nolenough to Identify some changes in the average quality (ofor processed foods, forhe CVO indexes are probably biasedby increased double counting. Soviet consumers have been buyingclothing instead of buying fabrics to sew their own garments,of thc basic products of the food industry (like flour and sugar)processing before being consumed.


rum, ihe difference* Itcrwoer) SPIOER growth rales and Sovietrates might be explained in part by difference* in coverage, butbiases in the CVO indexes and changes in tbe extent of double countingin the CVO indexes The question of coverage has been exploredlength above. The amount of bias caused by the new productcannot be measured but is judged to be important in tbe case ofbuilding and chemicals. Willi regard to changes In double countingtlie Soviet input-output tables provide some Interesting evidence.

Comparison of SPrOER with Soviet Input-Output Indexes

The input-output) tables98elatively recent development In Soviet statistical reporting. These tables, reconstructedan be compared with the SPIOEH and official CVO indexes, but three major caveats must be observed in their use. First, because the data inables is in current prices while both official GVO ami SPIOER indexes arc based on data in constant prices,ata must be converted to constant prices with the help of official price indexes which are themselves open to question. Furthermore,alues are given in purchasers prices which differ from the enterprise wholesale prices used in SPIOEH and the official GVO index by the amount of turnover tax and trade and marketing costs, elements which may change over time and hence influence the growth shownata. Finally, errors undoubtedly have occurred in the reconstruction of th*ables.

xxmpares the average annual rates of growth of SPIOEH and CVO branch Usdcxesith corresponding branch indexes derivedata Because of certain intom parabiotics in the96 tables, machinery must be aggregated with "other industry- inalculations.

First, those changes in double counting large enough to deserve attention are indicated by the compaiisonross outputinal output. These comparisons show that gross output grew faster than final output in fuels, forest products, machinery, and soft goods. This was the result of amore rapid growth of mtrabranch deliveries, which are included in GVO but not in final output. In the metals, electric power, and processed foods sectors, the giowth of gross output and final output was roughly the same, while in chemicals and construction materials final output grew faster than gross output, suggesting that growth of gross output in these branches is not inflated because of increased double counting. The comparison of course covers only seven of tbe years9 and makes usereliminary recoostnKtvon ofable in0 prices. Nevertheless, the comparison Ltenough to support the view that increased double counting accountsarge part of the difference botwecn growth of Soviet GVO and tbe growth of SPIOER in the fuels, fount products, and soft goods industries and apart of lite difference in machinery.

C. TrrW. ReseawACoepceettem.9 Sovtrt Uter-teetani Flew Table, MeLemm.nd II, end Tteml et. ml. feaparcA AneketUThe Structure of theotnon ol6 Input-Oitput Table.


Comparison of tbe growthross oulputinal output with thc growth of official Soviet gross output and SPIOER, respectively, yields ambiguous results. In most cases the growth of official GVOross output diverge significantly. This, however, may be th* result of the prior andproblems mentioned above and tbe fact that official gross output in dudes production of secondary1 products while IO gross output indudes only primary product production On the other hand, theinal output indexes and the branch SPIOER indexes formatch very closely except for machinery and chemicals Although narrower than the gap between thc growth of official GVO and SPIOER. the difference between growthinal oulput and SPIOER final output in machinery and chemjeals Is considerable, suggestingajor portion of the deference cannot be bydouble counting.

The average annual rate* of growth of gross output and final output which are shown inor industryhole are incomparable because the branch weights used inables differ from those used by either the CSA in calculating official GVO or by OER in calculating SPIOERranch gross output and final oulput weights include turnover taxes and trade and distribution charges which are excluded from the official CSA branch weights. If branch rates of growth of gross output are aggregatedonirnon set ofhat is, the SPIOERrossoutput increasesearnd official GVO Increasesear. Thus much of the discrepancy in Ihe rates of growth of gross output shown inisappears.

Conversely, tbe apparent identity of growth of the final output ol industry as measured by SPIOEHmisleadinginal output weights are based on thc value of shipments destined for use outside


of industry (and Include any turnover tax or distribution charges on these shipments) while thc SPIOER weights are value-added weights- Ifranch rates of growth of final output are aggregated with SPIOER value-added weights,inal output of industry increasesearnsteadear.

in six of the nineet output grew faster thanoutput. The difference was particularly striking in metallurgy,construction materials, TJiis implies that in these sectors the rawother inputs required per unit of output declined. Thus tbe SPIOERon final output, may understate the actual growth in net outputsectors. On the otherinal output exceeds the increase inin machinery, fuels, and forest products and paper andwith high value-added weights. Aggregating branch ratesinal outputet outputommon set of (SPIOER)in an identical rate of growth of industrial' final output anda year. Therefore, at thc level of industryhole,docs not understate growth in net output because of its reliancein final outputeasure of change in net output in individual brandies.

Comparison of the SPIOER Producer Durables Index with on Index of Soviet Investment in Equipment

partial confrontation of SPIOER with official Soviet indexesby comparing the SPIOER producer durables index with anSoviet investment in equipment adjustedroductionhesecompared inhe similarity of tlie growth of thc two indexesAt first glance this is encouraging, since thclobal production series with each individual product valued atprice. Thus the SPIOER producer durables sample index wouldreflect accurately the growth of this part of the machinery sector.

investment series may inflate growth, however, because of thepricing" problem. Although the CSA claims that new orare given prices comparable to5 prices attached Ioproducts, Soviet authorities continue to express concern over thewith the formation and use of prices for serially produced newNew products are assigned "temporary" prices set high enough toto recover expenses of the enterprise connected with research,and thc tooling up for production- These prices are supposedin effect only fur the first year or two of production. Later, aprice is supposed to be established based on normal production cost plus

'The official Soviet data on investment In equipment Is lagged one year (on the assumption thai tquipmera "inoeaetT rliii year was produce'! last year} and reduced bv the amount oi net imports af equipment.

Khrushchev the regional economicvnartuWy) were authorized toprice, for new products, with the result lhat enforcement of various regulation, mt Out of hand. In recent war, there has been significant tightening of both regulations and control by the State Committee on Prices, but some abuses apparently still continue.

L i












3 Is

mi fill

j IU

ut according to the Soviet critics, the temporary prices are oftento run on for years, and even when they do not, Soviet statisticiansdo not go back and replace temporary prices in thc production series of previous years. The net effect of practices of this kinduilt-in price inflation embedded In tho official indexes of producrJon. Investment data could be similarly affected, but not necessarily to the same degree. If so, theof the SPIOER Index of producer durables with Soviet investment in equipment may indicate that SPIOER overstates the growth of producer durables.

artial check may be periormcd by viewing the growth of the SPIOER producer durables sample excluding Soviet ruble production series (that is, excluding those series most likely to be affected by price inflation) andfor thc deleted series, wherecries basedixed sample. The growth oferies is compared to tbe growth of SPIOER producer durables and Soviet investment in equipment In the accompanying tabulation.

Average Annual Bate* of Crowth (Percent)

SPIOER producer durables

including value,

Exdudlne value series

Soviet investment in

The growth of thc SPIOER index which excludes the producer durablesin rubles does diverge significandy from the other series. This maythat the indexes of Soviet investment and SPIOER producer durables are biased somewhat by price inflation. The exact amount of such bias, however, cannot be determined, since thc alternative SPIOER index (without the value series} understates growth by fading to record changes in quality and

Comparison of the SPIOER Consumer Durables Index with the Official Series on Production of Consumer Durables

The CSA publishes the ruble value of the production of "goods of cultural-living significance and domesticainly, the CSA series includes consumer durables of the land included in the SPIOER consumer durable series, but thc coverage of the CSA series Is broader. The CSA series includes, forumber of household articles (like kitchen utensils, sporting goods, some electrical appliances, and metal furniture) which arc produced in the machinery sector but which are not included in the SPIOER sample. The CSA series,also includes such goods as wooden furniture and musical irtsfrumeots which are not produced in thc mactiinery sector. Beyond this, the CSAof thc series are unclear; school supplies, nonmetal toys, and jewelry may lie encompassed in CSA* definition of "goods of cultural-living and domestic significance."

To improve comparability with the SPIOER index of output ofdurables, wooden furniture and musical instruments bave been removed

manyItems noprice is ever ret, uxomuch atrices ere limited to products that enter serial or large-batch production.


from CSA'* ruble series. The resulttsallenge to tbe SPIOER index, asxxxnpanying tabulation shows.

Avaeaar. Annual Rata of Crowth (Tarcaat)

SPIOER coBsumw durables

CSA cubit

"Condi of csiAuraJ-bolne lutriificunt* o'td domeilic uie" Jcii value of output of wooden furniture and musical tnrtrumtrUi

Surprisingly, the CSA series grows more slowly than the SPIOER consumer durables sample, particularly in.

he explanation for the divergence appears to Ite primarily in tbe absence from the SPIOER series of basic household articles such as pots and pans and tableware. These items probablya for part of consumer durables output before TV sets, refrigerators, and washing machines were produced in more than token quantities. During, the production of housewares increased but not nearly as fast as the products included in the SPIOER sample Moreover, the CSA ruble series probablyumber of otherwhose production lagged or even declined in. Presumably such items would be covered under the ruble series, although the CSA tends to regard sluggish production series as unworthy of being reported individually. Because of this spotty reporting of individual sories and because of therecording of thc ruble series, the ruble scries could not be substituted readily for the SPIOER sample, even Ifubstitution seemed warranted.

nother possible reason for the dlffeienoo between the SPIOER and the official series is the use of relative retail prices in tho SPIOER sample of consumer durables- The official series, onther hand, is given in wholesale prices and the gap between retail and wholesale price (mainly turnover tax)ood deal from product to product. It may be that the faster growing components of tbe SPIOER sample have higher than average turnover tax rates jnd are weighted too heavily in tlic SPIOEH index of consumer durables output

hus, the SPIOER consumer durablesin theprobably overstates actual growth Still, thc weight of consumer durables in thechinc buildmg index is relatively slight0 and. so the impact of an exaggerated consumer durables index is not great0 tbe electrical and electronic appliances and tbe other items in tbe SPIOER sampleominated the production of consumer durables, and tbe SPIOER index should be quite representative.

Comparison of SPIOER with Western Indoxes of Soviet Industrial Production

umber of Western scholar* have constructed cither partial or total Indexes of Soviet industrial production, three have been selected for comparisonNutter, Moorsteen Posv.rUnd Thornton indexes. Tlie Nutter and Moorsteen-Powcll indexes resemble SPIOER inand thc Moorsteen-Powell index luis tho further advantage of having been recently extended forwardhornton, on the other hand, first

reccmstructs value added in Soviet industry and then deflates it by thc Soviet official price index. For contrast, the rates of growth of SPIOER and these other indepcodendy constructed indexes are compared with the rates of growth of official industrial CVO in thc accompanying tabulation.

Average Annual Bates of Growth



Total industrial production





Soviet CVO

Civilian industrial production'



Wanen G. Nutter, The Growth of Industrial Production In the Soviet Union,


Kaplan. RANDhe Record of Soviet Economic. SantaS,nd Abraham Seeker. Richard Moor-steen, and Raymond PcnceM, Two Supplements to Richard Moorstocn and Raymond P. PowelL The Soviet Capital,

'Judith Thornton. -Value-Added and Factor Productivity In Soviethe American Economic Review,.


' Thornton does not estimate on index of civilian industrial production.


the one comparable period, the Nutter, Moorsteen-Powellindexes of total industrial production are remarkably similar."indexes of industrial production fall short of the average annual rateof the Soviet CVO indexo 2Vt percentage points.index, however, is closer to the official GVO index than to theindexes.

Evaluation of Western Indexes

similarity between the Thornton and the official GVO index isThornton uses official data on the distribution of CVO byto derive value added for industryhole, and she deflatesby an official price index- As Thornton recognizes, however, it isprecarious toalue-added series by an output priceall, the Soviet industrial price index is so biased that its usean overstatement of growth of industrial output in constant prices."

the case of the Nutter index, however, the overall resemblance is somewhat misleading in that it conceals the fact that the civilian index of Induttnet production gnrw, much -lower and his index of military production rites much faster than thend SPIOER indexes.urther diwumon of the Nutlet index see CIA/RR, Index rrf Civilian Industrialhc..f. UNCLASSIFIED.

relatesroup of economists at the Central Economic Mathematical Institute In Mr-wow criticized her results on Ihe grounds that "the official price index provides anmeasure of the rise in the average price of new and nonstandard product?."

Tableompares the behavior of the major components of the other mdustrial production indexes; the Thornton index has no branch or sectorBecause tbe MP index of military machinery has very little basis in reported Soviet data or in independent analysis, the examination centers on indexes of civilian output. Thc most mteresting features of thc comparisonhc relative closeness of the Moorsteen-Powell and SPIOER indexes ofof producer durables3 coupledtriking divergence thereafterhc general similarity of all the indexes (including GVO) with respect to production of materials and consumeris, total Industrial pioduction exclusive of machinery.

Since thc machinery indexes arc so critical to Ihe behavior of the overall industrial index, they deserve further examination. Because the Moorsteen-Powell index of consumer durables is classified in thc consurncr goods sector, whereas in SPIOER it is part of civilian machinery, only producer durables can be considered in comparing the SPIOER civilian machinery index withThese indexes are also compared with thc scries representing Soviet investment in equipment which washis comparison of average annual percentage rates of growth of producer durabless shown in thc accompanying tabulation

Source of the


Official investment (adjusted)

Moorsteen-Powell machinery scries is derived from the Sovietinvestment in equipment In its original version the Moorsteen-Powellindexample series, but this index was dropped and thescries usedurrogate for machinery production. Because themachinery series is based on thc official Soviet series lor investment(adjustedroductiont is subject to thc samethis latter series.*'

An Alternotlve Check on SPIOER

alternative method of measuring total Soviet industrial growthcheck onrancis Seton used regression analysisthe relationship between thc growth of consumption of certain(steel, fuel, and electricity) and the growth of manufacturing insection ofountries. He then applied the average relationship toto estimate an index of Soviet industrial production. The Setonspecifically designed to be "rough andsingmallproducts for which information was readily available.

SI. Although the Seton method rests on an oversimplified assumptionthe relationslup between the pioduction of certain major commodities

'Ideally, the consumer durable components of the Mwrrfeen-Posietf index rhould be added lo their retpectioe machinery components to mole them comparable with the civilian machinery component of SPIOER. Unfortunately, the necessary data on consumer durable, air notfrom Ihe Mooraeen-Powell Index for the period after IS5S. so at is nece&ery to adjust the SPIOER civilian machinery data to conform ,oith the Mocrfteen-Pouvtt classification.


Seton, "The Tempo of Soviet Industrial Exposition" Bulletin of the Instrtute of Statistics, Oxford.

it r

and industrial output, bis results suggest that there is much to recommend the method, particularly when thc aim is to express the industrial outputountry like the USSR in terms comparable with market-oriented industrialized nations. The Seton method sidestep* distortions due to problems ot valuation and likewise avoids the difficulties encountered in trying toufficient amount of data on quantity and price to construct adequately representative indexes of Soviet industrial production.

eton's original methodology has been modified to estimate the index presented in this report Instead ofross sectional approachingle time periodumber of countries, observations based on fourUnited States, tbe United Kingdom, Japan, and Westwere taken for each year. Another modification was the addition of paper to the list of independent variables (steel, fuel, and electricwo variants of the regression were run First, thc rate of giowlh ofin each of the four countries was regardedunction of rates of growth of the consumption of steel, fuel, electricity, andn thc second variant, thc rate of growth of industrial production (manufacturing plus rruriing and utilities) was regardedunction of the rates of growth of the same four independentn both cases, the resulting; regression coefficients from each country were applied to the oonesponding Soviet data lo estimate indexes for manufacturing and industrial production,lie indexes thus derived were averaged toingle index of Soviot manufacturing or industrial production.

Ih. renffre-resnon,ommealem train the denvetwn ef the -Modified Seton" index af meemfeetuetng aeehe fotlou.tnghere U. F. S. thef fovIA of memujeKtvlmeL fuelteeielectric power proA^SBM, end paper production, rcspectraelu (J) Vnaed Stout






"The rrmJl, of theurn mnth the derivation of the -Modified Seton' mdci of tndustnalrei ra the foOoumg eqvatems,he rate of tfrouln ofproduction and the otherf defined a* in thebove.




i-sp. ir-oae



ofn obtainingconsumption data for the USSR and betause foreign trade UlUelatively minor nit In the Soviet economu. productionre used Instead of roruumption data for all the products except tteei

ne results [wesenlcd In ihe acxompanying tabulation vhow thai thr growth of indexes based on an avenge of tbe four regrenumparts favorably with the corresponding SPIOER indexes over thc whole. Tbe fit is noi as good for ther for indexes based on individual country regressions. The index of industrial production estimated from regression dataaried from an average annual rate% for the United Kingdomear for West Germany. These results therefore arc not conclusive. Nevertheless, SPiOER's rate of increase does fall within tbe range predicted by the correlation between industrial output and key industrial products in ma for industrial countries with quite differentstructures.

Annual Rate* of Growth (Percent)





The CIA index of Soviet industrial production (SPIOER) is the most comprehensive of such indexes constructed to date in the West. Thc product sampleivilian series and one consolidated military series based on classified estimates of Soviet procurements of military and space equipment-It resembles the Federal Reserve Board Index of US induslriol production in form.

The SPIOER branch indexes represent the total value, expressed5 wholesale prices, of those Items in the SPIOER sample which arc produced within thc givenhe branch indexes arc, in turn, aggregated0 value-added weights to estimate the index of total industrial production. SPIOER isybrid index rather than an "ideal" Index of net output. In an attempt to minimize the effects of this departure from tho ideal, thc SPIOER sampleroducts is limited to final products insofar as possible to curtail -double counting- of intermediate industrial products. SPIOER's explicitis that the trend in final outputranch corresponds to thc trend in net output in that branch.

The SPIOER sample of commodities and commodity classes isumerical count of line items would indicate The final products in thc SPIOER sample representf the coverage of all the value-added series (final and Intermediate products) in tha FRB index of USproduction andf Soviet CVO.

In order to cover the most important industrial commodities, production ofonferrous metals and chemical products for which the USSR docs not publish statistics has been estimated by CIA/OER from collateral information.andful of Soviet ruble-value scries, which appear to be constructed in the same manner as several SPIOER scries and which take account of changes in product assortment, arc included in the sample to provide coverage of important heterogeneous categories of machinery that would otherwise not be represented at all or would be only inadequately represented. Tlie SPIOER series which reflect changes in product mix (motor vehicles, tractors, and railroad locomotivos) and most of the Soviet ruble-value series (agricultural machinery, chemicalequipment for consumer industries, and electronic and nonelectronicconslsl largely of producer durables and are the dominant influence on the producer durables index within thc machinery sector.

lthough the production and price data used in SPIOER arc considered reliable, they are deficientumber of resr>ects for purposes of indexExcept for the machinery series cited above, physical production reported by tbe Soviet Central Statistical Administration (CSA) is very often

however. Is tarried outumhes of tubltrancher in the chesrttestL, forest products, toftnd practised foods industries.

too aggregative to indicate changes in average quality resulting from changes in the product mix. While thc CSA series areeasonableof the growth of many relatively homogeneouspower, coal, bricks, lumber, and thearc less suitable for more differentiated commodities. In the case of several SPIOER scries (ferrous metals, cement, flour products, and consumer electronics) adjustments have been introduced to allow for changes in average quab'ty arising from changes in product mix. Although it is impossible to assess with certainty tbe net effect of these quality adjustments, many of thc remaining biases are probably offsetting. The SPIOER indexes that are not adjusted for quality change mayownward bias, but the Soviet ruble-value series included in SPIOER probably are biased upwardesult of the procedures used in pricing new products in the USSR.

everal pieces of evidence suggest that in some branches thc actual growth in net output may differ from the growth in final outputercentage point or two per year. Nevertheless, the information is contradictory with respect to the trend in the ratio of material purchases to final output in individual branches. On balance, it appears that SPIOER's use of final producteasure of change in branch value added leads to overstatement of growth as often as it leads to Wilder statement*

Insofar as tbe SPIOER prices are concerned, the available Soviet sources listed only producer goods, Even for producer goodsarge number and wide range of prices was reported for some productsedian price had to be selected as the "representative" price. The "wholesale" prices fordurables are derived7 retail prices and, in thc case of consumernclude turnover tax.

Except for some categories of machinery,5 prices employed by SPIOEH are relatively reliable in terms of what they rjurport totransaction prices of that period.5 prices, however, did not represent thc relative costs of production of different products very well. Profit rates varied enormously. After some abortive attempts at price reform, the USSR carriedeform of industrial prices7 in which the relative prices of products went changed significantly, especially in the industrial materials branches. But the substitution of the new prices for5 prices as weights for thc branch samples of physical production does not affect thc SPIOER branch indexes appreciably.

Thc SPIOER branch value-added weights0 are sound. First, they are formally correct and have been derived quite rigorously. Second, the substitution of alternative weights alters the overall index of industrialvery little. The estimate of annual increases in Industrial production is almost unchanged58 base-year weights arc used instead0 weights, and thc SPIOER index is almost as insensitive to the substitution of Soviet CVOalue-added weights.

Tlie major weighting improvement that could be made in SPIOER would be to introduce value-added weights for subbranches of the civilian machinery sector. This sector is large and diverse, and the SPIOER sample represents some

- cccnur

subbranchcs more completely than others. Therefore, growth of some of the subbranches may influence unduly the aggregate index of civilian machinery output.

onfidence In SPIOEReasure of final output is bolstered byit with official Soviet indexes and with other independent reconstructions of an index of Soviet industrial production. Tbe growth of Soviet CVO exceeds thc growth in SPIOER for all branches of industry. The difference is the greatest, however, in those branches (chemicals and machinery) in which questionable procedures for valuing new products and/or increased double counting bias the Soviet index upward. Furthermore, confidence in SPIOER is buttressed by the fact that the SPIOER civilian machinery index agrees well with reported Soviet Investment in equipment, the dominant component of civilian machinery. Although thc SPIOER index of producer durablesslightly slower than reported investment ia equipment, the official statistics on mvestment in equipment overstate actual growth according to knowledgeable Soviet experts. Thus. SPIOER's approach to the problem of quality change in the output of producer durablesredible average of growth for civilian machinery ofear. Other Western indexes corroborate SPIOER in most industrial branches. Where differencesin thc machineryhas the advantagearger and more representative sample. In addition, the trend of SPIOER over thcs supportedegression index keyed to the production of four basic industrial commodities.

Our evaluation of SPIOER indicates that, despite drawbacks due largely to the limited availabdity of Soviet data, this indexore reliable indicator of Soviet industrial growth than other indexes of either Soviet or Western origin. Still, because no absolute standard exists against which to measure theand accuracy of SPIOER, this evaluation Is necessarily an interim report.

In thia connection, parts of SPIOER clearly require further improvement as the availability of Soviet data permits. When the SPIOEH index of chemical products, for example, is compared with thc Soviet GVO for the chemicalthe large discrepancy is not explained conclusively by bias in the GVO index. The SPIOER index of chemical products therefore may need to beand adjusted for quality changes, if that is possible. Other areas open to improvement in SPIOEH include enlargement of the product sample in some brandies, further disaggregation ol production series, verification olprices used, and thc derivation of price coefficientsefinitive conversion of the index from prices5 to prices

Finally, an industrial production index8 value-added weights is presented in this report. It does not differ significantly from the index0 weights. However, since it is more up to date, it will be used as the standard CIA index of Soviet industrial production.

re i


The SPIOER Sample

Tableists all line items included in the SPIOER sample of Sovietproduction. Unless otherwise noted, the line items incoTporated in SPIOER are based on official Soviet physical production series as reported by the CSA. When these series are unavailable, however, Soviet reported ruble-value series (designated in the luting asIA estimated physical production seriesr CIA estimated ruble-value series (ERS) are used.

A simple count of the series detailed in Tableeveals Ihat thereine items in the SPIOER sample. In this count, different models which are assigned thc same wholesale price weight arc regarded as being one line Item in their respective branch sample. For example, the.assenger cars allnit wholesale priceubles and hence are counted as one line item.

Tableine Items Included in SPIOER


Industrial materials vector Electric power Branch

Electric power generation

Coal products branch



Noncoking bard coal

Petroleum products branch

Crude oil

Refined products

Associated natural gas .

Natural errous metals branch

Rails and rail accessories

WW. rod

Plain wire

Seamreu pipe and tube ..

Welded pipe and tube

Light sections



Hot-rolled sheet

Cold-rolled sheet

Electrical alttvt

Tinplate sheet

of Production


Metric tons

rrrnn -tJLLrtLt-

Line.of Proooctlou

Cafvanixed sheet Metric tons

Products for reprocesriog In other

Foundry pi; iron

Iron ore riporla

Pif iron

NonlprriKis metal* branch

Copper <EPS)


Zinc (EPS)

Alkimlnuni (LPS)


Marmesium (EPS)

Antimony (EPS)


Nickel (EPS)

Titanium (EPS)

Cadmium (EPS)

Tungsten (EPS)

Molybdenum (EPS)

Forest products branch

Industrial lopsCubic meters




Paper and pu per board branch

Newsprint torn

Wrapping and packing


Writing paper


Offset printing



Deep printing



Cable msulatsan

Capadtor paper

Waiing paper


Pjper board

Construction material? branch




Ceramic tiles for facingsquare meters



Asbftstoi cementMillion standard units

Asbestos cementStandardam pipe

Metric Km*

Dinas (wfractory brick)

Magnetite, chrome toaenesfte

Magnetite, powder

Window glass Square rooter*

Polished glass

Precast concrete .Cubic mctrrs


Maieeal wool insulation



sewer pipe Construction brick Dimension and field Large concrete and silicate wall block

Other waD materials

Bock products

Chemicals branch

Nitrogen lertihreri

Phosphorite remitters

Phosphorous fertibxen

Potassium rcrtiliirri


Soda ash

Caustic soda

Ethyl alcohol

OsoorusUi. baros, and rosin

Raw and refined turpentine

Turpentine oil

Synthetic dyes

Artificial and synthetic libers

Synthetic rubber



Enamels and primers

Litharge and red lead

Natural drying oil

Oksol drying oil

Oil varnishes and. Nitrocellulose varnishes and solvents

Synthetic ammonia (EPS)

Benzol (EPS)

Chlorine (EPS)

Nitric acid (EPS)

Phenol (EPS)

Toluol (EPS)

Motor vehicle tires Machinery sector Civilian machinery Producer durable* Boilers branch

Steamcapacity .


Steam boilers-Jaw capacity lectric power equipment branch

Steam and gas turbines

Generators for steam and gasfor hydraulic turbinei .motontransformers

Electric bulbs

Metal cunine machine toot branch


Turret lathes

Automatic and semiautomatic lithcs


Goarmaking machines

Jig-boring machine*


Standard bricks it

Cubic meters

Metric tont,


Metric t



Metric tons of itcam capacity


aw A

KVA Units


Line Item



Hnrlronlal breaching machines .

Internal grinding machines

Drill grinders

Vertical dnlb

Radial drills

Special, specialized and unit type

Grinders, poliiben. bolt threader)

R.dway machine building branch

Refrigerator cars, mainline freight

Boa cars, mamline freight

Plat cars, mainline freight

Gondola can, mainline freight

Tank cars, mainline freight

Cement con, mainline freight

Steam locomotives


Sum (EPS)



Diesel locomotives




0 diesel electrics (EPS)


G-l, Car turbine (EPS)

iesel hydraulic*


Klocnic locomotive,

, dlwcl current (EPS)

, direct current (EPS)

fcetric (EPS)

kcttk (EPS)

lectric (EPS)

lectric (EPS)

VI. alternating cutrcnt .

Ratltoad pasenger

Trolley cars

Subway com

Trolley buses

Motor vehicles branch

of Pr Units

iedan (EPS)

Mrakvieh cabriolet model (EPS)

Moskvtcti station wagon (EPS)

edan (EPS)



ZJM wdan (EPS)

eep (EPS)


Pobeda (EPS)

Volga) (EPS)


Una Item















EPS) ..












nd miacvlbneuii- other. (EPS) ..
















Belarus (EPS)






Aufculturil machine bulMuifr branch

All agrtculfuril Burku buildm* equipeaen*


Contraction and read equipment branch EnmUi (wednbocket)


0uckrt eteavHon


ot Produdim


h h

Ruble* Unib

lineof Production

ingle-bucket ncivMn


r more single-bucket


Motor graders

Hoist-transport equipment branch


Truck cranes


Pneumatic tiro


Mining and other equipment branch

Metallurgical (eicept rolling mills) .tons

RoDtng mlD

Coal combine*

Coal cutting

Rock loading niactunes

Electric mine

Petroleum equipment (refinery) torn

Deep well pumps


Chemical equipment <

Equipment for consumer industries branch

For the paper Industry

For trade and public dining enterprise*

For the printing industry

For mitli, elevators, and granaries (RS)

thc fowlrnAistry (RS) .

For the textile industry

Civilian shipbuilding branch

CMUan shipbuildingS)

Civilian jirtTift abf&ftcb

Qvaian aircralt (ERS)

Sanitary technical equipment branch


Square meter*

pipe and fittings

Instruments branch

Power optical instrument* and apparatus (RS)

Electrical mc-Murine imtroments (RS)

Radio measuring instruments (RS)

Computer eogiDeering equipment (RS)

Instruments for the mechanisation nndof engineering and control

lnstruiwnU for the control and regulationpiocesses (RS)

Instrument* for physical research

Instruments for the mcasuicment ofmagnitudes (RS)

Instrument* foi niedKlnc. physiology, and


Consumer durable*

Consumer electronics branch

Consumer electronicsS)




Soft goods branch


Square meters




garments <RS)





foods branch



and meat products .

Metric tons

and fish .

_ _

granulated (net)




oil (net)

products .



ram cans




and vodka products


and vodka products



Derivation of Final-Product Weights for Civilian and Military Machinery

To help explain the derivation of final-product weights for civilian and military machinery.s reproduced below. Comments are keyed to the row headings.

New Rubles






parts for repair



machinery .


of end Neros





two majormachinery and militarywell as total machinery are simply derived as the sum of their respective components.

The value of producer durables is derived from their value as investment goods. The equipment component of investmentillion rubles in mvestjncnt pricess taken to represent industrial output of producer goods0 on the assumption that there isne-year lag between the production of goods and the acquisition of them forpurposes. This figure is reducedillion rubles) to convert the valuation from investment prices, which include the cost ofhandling, and transportation, back to enterprise wholesale prices. To this value is added an increaseillion rubles in inventories of uninstallod equipment (producer durables produced0 but not entering intone further step required to complete the adjustment is the elimination of net imports of equipment amounting0illion rubles (derived by adjusting the value of net imports of producer durables AS reported in Soviet foreign trade handbooks to domestic pricesatiohe resulting figure6 billion rubles represents the final-product value of Soviet producer durables0 expressed in enterprise wholesale prices

The value of consumer durables is taken directly from tbe SPIOER sample and excludes passenger cars- (Conceptually, some passenger cars belong in the consumer durables rather than the producer durables category, but only about one-fourth of all tbe passenger cars produced in tho USSR are. in fact, sold as consumer goods. The value of passenger can sold to the publicdid notillion rubles) The total value shown in this table Is thc sum of thc following two categories:

Billion New Rubles (Enterprise Wholesale Prk-soi

tknturnei durable* -

etactrs*lM 1

Tfd cemmmmm du-abUs

This total is thought to represent very nearly the universe of consumer durable production0 (allowing for the exclusion of passenger cars) on the basis of the flows of machinery to consumption shown in the Soviet statisticalUSSH. Central Statistical Aclministralion. Narodnoue ktuja/austvo SSSR ogoviu (The National Economy of thc USSR,hc consumption figures in the handbook are on an input-output basis and tlie purchaser prices in current rubles shown there must, of course, be significantly adjusted whenomparison with enterprise wholesale prices

A sizable part of the output of Soviet machine building consists of spare parts which are used in tbe repair of producer and consumer durables but axe not ineluoed in the value of final product going to mvestrnes* arsd con-sumption- Conseo.iscntry an estimate of cavilian spare parts production originat. ing in the machine building sector is an Important element in determining thc SPIOER weights for civilian and military machinery. Thc production of spare potts originating in the machine building sector is predominantlyproduction (in distinction to "noncentralixed" production of spare parts, which is generally carried out in the usingecause thc value-addedn SPIOER are based on an establishment classification, the value of -centralized- production of spare parts It most appropriate for use in txxostructing the weights in the rnacninery sector. Centralized production of spare parts forides, tractors, and agricultural cqiuprncnt0 was planned3 billion rabies (in enterprise wholesale prices. It is estimated that production of these spare parts alonef the total centralized production (civilian and military) of spare parts, based on the ratio of the deliveries of the various machinery branches to tbe machinery repair sector as shown in9able. The resulting total6 billion rubles is reduced by one-thitd to reflect thc division between civilian final products and spare parts.

The RM> figure for civilian machiitery Is estimated as one-half of totalor theillion rubles.

Procurements of military and spaceequivalent toestimated by OSR

The category of operations and maintenance, also estimated by OSIt, conAists largely of spare parts but also includes some rnaterials used in theand repair of military rnacbinery.

Tbe figure fors estimated by OSR from announced Soviet expenditures on science. It excludes outlays for investment in science inasmuch as such outlays are covered conceptually in the producer durables category of civilian machinery.

Original document.

Comment about this article or add new information about this topic: