How fast has Chinese industry grown?

Policy As"esch WC*IKING -PAPERS Transition and Msero-Adjuatment Policy Research Department The World Bank Septeniber 1993 WPS 1194 How Fast Has Chinese Industry Grown? Tom Rawski An upward bias in measures of China's real industrial output in the past decade may substantially alter our perception of the rate and pattern of Chinese industrial growth. The extent of such bias should be investigated and analyzed for possible links with other economic patterns that may be more readily measurable. Policy Rescjh WotkingPaper. dimninatethe finp of wotk in pm and nc=gethe exchanucoid mwig%Dank mff And alothmineadin dopinaiTues. llepapen, diLibuted by theReRarb Adviamy Staff,cany ditename of de audio,aa ny diirvic and shouldbetuad and citedaccardingly.lhe fndings,imapxuati,nd canclusios a Cathe amhox 'own. Theyibould not be auinbuted to dte Wodd Bank. its Bsrd of Dijto. its magaenat, or any of ita manber comta.  WPS 1194 This paper- aproduct of theTransition and Macro-Adjustmnent Division, Policy Research Deparmnenit- is parn of the division's research initiative, Indusurial Reforms and Productivity in Clinese Enterprises. The study was funded by the Bank's Research Support BudgeL under research project "Reforms and Productivity in Chinese Enterprises" (RPO 675-M8). Copies of this paper are available free from the World Bank, 1818 H Street NW, Washington. DC 20433. Please contact Emily Khine, room NI 1-065. extension 37471 (September 1993,44 pages). Data for recent years indicate an acceleration of The specific consequences of decentralized Chinese industtirl growth, from the annual rates decision~making, growing price flexibility, of about 10 percent recorded in the quarter inflation, dual pricing systems, the emergence of century before economic reform to figures enterprises with few or no ties to the system of approaching 15 percent in the mid- and late state planning, and other emerging features of 1980s. the industrial system may be unique to China but the broader issues raised are relevant in many Evaluating the statistics underlying these countiies. reports requires an appraisal of how economic reform has affected the ability of China's stztisti- Rawski finds considerable evidence of an cal system to measure economic performiance. upward bias In measures of China's real indus- Erroneous informnation about the rate and pattern trial output in the past decade. The issue is not of industrial growth could distort measures of whether such bias exists but whether its presnce productivity change considered to be central substantially alters our perception of the rate and indicators of the effectiveness of Chinese pattern of Chinese industrial growth. industrial reformn. To clajify this issue requires investigating Rawski describes the statistical materials and the extent of possible upward bias. This in tumn procedures used to provide information on the calls for an analysis of possible links between growth of industrial output. He investigates upward bias - which is itself difficult to ob- sources of bias in the official statistics to indi- serve - and other economic patterns that may cate, whenever possible, how these biases be more readily measurable. affected reported output totals, and to appraise the impact of adjustments to reported output growth on measures of industrial productivity. Pmoducsd by dw Pobicy R.earsch Disminazia Cmtar  How Fast Has Chinese Industry Grown? Tom Rawakl University of Pittsburgh  CONTES ACKNWLEDGEMENT ....................................... i I. INTRODUCTION ..........................................1 II. CONCEPTS AND CATEGORIES FOR CHNESE INDUSTRIAL STATISTICS . ......... 1 III. CHINESE INDUSTRIAL PEILORMANCE SUMM-ARY OF RECENT OFFICIAL DATA ..... 3 IV. SYMPTOMS OF UPWAR BIAS IN CHINESE INDUSTRIAL STATISTICS ... ........ 6 A. MISMATCH BETWEEN OUTPUT AND VALUE DATA ...... .................. 6 B. Do RURAL INDUSTRIES ADHERE TO STANDARD AccOUNTING REGULATIONS ...7 C. HAS THE RATE OF DOUBLE COUNTING INCREASED? ................ 10 D. Do LOCAL AUTHORITIES FALSIFY INDUSTRIAL OUTPUT DATA? ... ...... 12 E. Do INDUSTRIAL OUTPUT DATA ACCURATELY REFLECT RECENT INFLATION EXPERIENCE ........ 12 F. ARE REPORTED GAiNS IN ENERGY PRODUCTIVITY UNREALISTICALLY LARGE? 15 V. CONCL.USIONS ............. 20 REFERENCES ..43  TABLES TABLE : ALL INDUSTRY . ............................................ 21 TABLE 2: BREAKDoWN STRUCrURE OF GVIO AT 1989 PRICEs ..... .......... 23 TABLE 3: BRANCH STRUCrURE oF GVIO AT 1989 PRIcES ..... .............. 24 TABLE 4: NET OurPuT AT CURRErr PRICES ....... ...................... 26 TABLE S: CHMCAL INDUSTRY ........................................ 27 TABLu 6: ALTERNATIVE DATA TO MEASURE INrLATION ..... ............... 28 TABLE 7: PRICE INDmxs EXTRACTED FROM INDUSTRTAL OUTPTr VALuz DATA . 29 TABLE 8: ENERcY PRODUCTnvTY IN CHNESE INDUSTRY, OFFICIAL DATA, 19781987 ....................................... 30 TABLE 9: OFFICL.L ENERGY DATA FOR CHNESE INDUSTRY BY BRANCH, 1980-'98S .................................. 31 TABLE 10: GVIO DATA FOR 1S BRANCHES ............................... 32 TABLE 11: REvISED CALCULATION: ANNUAL PERCENT INCREASE IN ENERGY PRODUCTIVrrY BASED ON PHYSICAL ENROY CONSUMPTION AND PUBLISHD GVIO DATA ............................ 34 TABLE 12: ENERGY DATA FOR STATE SECTOR INDEPENDENT UNrrS, 40 BRANCHES, 1980 AD 1985.. 35 TABLE 13: ROUGH CONVERSION OF SOE ENERGY DATA TO 15 BRANCH FoRmAT . 37 TABLE 14: TRIAL CALCULATION OF ENEROY CONSumpON AND PRODUCTIVrY FOR COLLECTIVE INDUSTRY . . 38 TABLE 15: PERCENT CHANGE IN INDUSTRIAL ENERGY PRoDucTIVrrY FOR 15 BRANCHES, 1980-1985 ... .......... .................. 40 TABLE 16: DATA ON "ENERGY SAvwNGS" (1980-1985). ... .. 41  ACKNOWLEDGEMENT The research projects on "Enterprise Behavior and Economic Reforms: A Comparative Study in Central and Eastern Europe", and "Industrial Reforms and Productivity in Chinese Enterprises are research initiatives of the Transition and Macro Adjustment Division (PRDTM) of the World Bank's Policy Research Department and managed by I.J. Singh, Lead Economist. These projects are being undertaken in collaboration with the following institutions: The London Business School (LBS); Reforme et Ouvertures des Systemes Economiques (post) Socialistes (ROSES) at the University of Paris; Centro de Estudos Aplicados da Universidade Catolica Portuguesa (UCP) in Lisbon; The Czech Management Center (CMC) at Cellkovice, Czech Republic; The Research Institute of Industrial Economics of the Janus Pannonius University, Pecs (RIE) in Budapest, Hungary; and the Department of Economics at the University of LddI, in Poland. The research projects are supported with funds generously provided by: The World Bank Research Committee; The lapanese Grant Facility; The Portuguese Ministry of Industry and Energy; The Ministry of Research and Space; The Ministry of T'dustry and Foreign Trade, and General Office of Planning in France; and the United States Agency for International Development. The Research Paper Series disseminates preliminary findings of work in progress and promotes the exchange of ideas among researchers and others interested in the area. The papers contain the views, conclusions, and interpretations of the attho:(s) and should not be attributed to the World Bank, its Board of Directors, its management or any of its member countries, or the sponsoring institutions or their affliated ag ncies. Due to the informality of this series and to make the publication available with the least possible delay, the papers have not been fully edited, and the World Bank accepts no responsibility for errors. The authors welcome any comments and suggestions. Request for permission to quote their contents should be addressed directly to the author(s). For additional copies, please contact the Transition and Macro Adjustment Division, room N-1 1065, World Bank, 1818 H Street, N.W., Wasbington, D.C. 20043, telephone (202) 473-1442, fax (202) 676-0083 or 676-0439. The series is also possible thanks to the contributions of Donna Schaller, Vesna Petrovic, Cecilia Guido-Spano and the leadership of Alan Gelb. -1i-  I. INMODUCnON Data for recent years indicate an acceleration of Chinese iwdustrial growth from the annual rates of approximately ten percent recorded during the quarter-century prior to the introductiun of economic reform policies to figures approaching 15 percent annual growth during the mid- and late 1980s. Evaluation of the statistical materials underlying these reports requires an appraisal of how economic reform has affected the capacity of China's statistical system to measure economic performance. Since industrial output dominates China's national income totals, possible inaccuracies in statistics of industrial growth have the potential to affect overall measures of the size and structure of China's economy u well as perceptions about the size, structure and growth rate of the industial sector itself. Of particular concern is the possibility that erroneous information about the rate and pattern of industrial growth will distort measures of productivity change, that we regard as central indicators of the effectiveness of industrial reform efforts in China's economy. The objective of this paper is to describe the statistical materials and procedures that stand behind published information on the growth of industrial output, to investigate sources of bias in the official statistics, to indicate, whenever possible, the quantitative impact of these biases on repcrted output totals, and to appraise the impact of adjustments to reported output growth on measures of industral productivity.. The problems explored in this paper arise primarily from thea interaction between the growing complexity of industrial organization and market structure and a statistical network designed to collect information ior the pre-reform system of industrial olanning and administration. Although the specific consequences of decentalized decision-making, growing price flexibility, inflation, dual pricing systems, emergence of enterprises with few or no ties to the system of state planning, and other emergent features of the industrial system may be unique to China, the broader issues raised by these developments would seem applicable to other socialist economies at various stages of maiket-oriented reform programs. II. CONCErr A CATEORE FOR CNSE INDUSRIAL STsTIMMCS Manipulation and interpretation of data portaining to Chinese industry (which includes mining, manufacturing and utilities) requires an appreciation of important concepts and data categories used by Chinese statisticians. State enterprises (puanmin suoyouzhif are those in which the legal ownership of post-tax profits resides in the hands of some level of the government. Nearly all of China's largest enterprises belong to this category, which accounted for 83.2 percent cif indL -ial output in 1978 and 59.7 percent in 1987 [Industry 1949-84, p. 98 ;TJNJ 1988, p. 311]. L.ollecti.-e enterprises (iiti suoyouzhi are those in which this residuai ownership right resides with the ente-"ise itself. Residual profit of private or individual (gc) enterprise accrues to thcz owners. -1-  2 Now Fae Mm Ch*F.e hu*oy row? The concept of 'independent accounting units, (dulihesuan.giye) refers to industrial enterprises that function as separate accounting iltities. Statistics for industrial output or input (fixed assets, worldng capital, labor) often refer exclusively to independent accounting units, which contributed 85.8 percent of national industrial output (including 96.1 percent in the state and 86.4 percent in the collective sector) to the 1988 gross value total [Jefferson, Rawski, and Zheng, Table 1; these data exclude village-level industry]. Industrial activity may occur within non-industrial units, as when transport companies repair their own equipment or when universities or other non-industrial entitia operaie factories whose accounts are subsumed within their owri. The output value of these 'non-independent accounting units' (feiduliheau giy is incorporated into output totals that include all industrial production (rather than only tha' of independent accounting units). The coverage of industrial production was expanded in 1984 to include industrial enterprises managed at and below the village level (formerly descr.bed as nanaged by the production brigades of rural communes). These enterprises were previously claisified as part of agricultural rather than industrial production [Field 1988, pp. 584-585]. Recent: vubl cations have begun to retroactively incorporate this category into the industrial totals for previous years, leading to apparent inconsistency with previously published statistics (e.g. newly published labor figures for collective industry inclusive of "village-managed' enterprises are miuch larger than identically-labelled employment totals published in earlier sources not because of any change izi underlying statistics, but merely because the "village-managed" enterprises were formerly included in the farm -ector). Industrial production statistics are valued in terms of 'current' or "constant' prices. The current price value of industrial output indicates the value of each year's output according to the market prices of that year. For th. vast majority of products that are sold in the year of productioil, output value at current prices is identical with sales revenue. From this perspective, the existence of multiple prices for individual products poses no conceptual or practical difficulty for Chinese accountants and statisticians. Following Soviet example, China's wconomic statistics have long been cs'culated according to "constant" as well as current prices. Constant prices of 1952 were used for the period 1952-57; 1957 prices were in force from 1957-71; national statistics for the years 1971-81 are based on 1970 prices; and a 1980 price base has been in force since 1981. Calculations based on 'constant' or fixed prices are made by multiplying quantities of output by the relevant "constant p.ices," the latter being supplied to enterprise accountants by planning and administrau've agencies of the Chinese govemment. It is important to note that, as economic reform leads enterprise managers to focus more closely on financial results based exclusively on current prices, the calculat<on of output value at "constant" prices becomes increasingly peripheral to enterprise objectives'. In the absence of 1. One official of Chia's State Stisca Bureau commeAed ht thi hnge enhace te veracity of rpotd data.  Flow Fast HOa Chinese Indwry Gro*"? 3 price indexes for industrial products, output value calculated at "constantu prices becomes the chief indicator of industrial growth and structure. Intertemporal comparisons within a time period spanned by a single set of fixed prices pose no difficulty. When the period of analysis crosses from one to another fixed price base, as in como';ting growth between 1975 and 1985, ratios of output totals in the bridging years 1957, 1971 o ,981 are used to link figures across time periods served by different sets of fixed prices. If a calculation crosses more than one such gap, e.g. when comparing the levels of industrial output in 1952 and 1987, a series of chain-linked calculations is used (see Field, JEC 1986, p. 509] and the resulting time series is described as being based on "comparable* (kmi) rather than "constant' (bbian) prices. III CHN S INDusiAL PEmFQFMANCE-SUMMARz OF RECENr OmaAL DATA Official statistics of China's industrial performance during the past decade indicate a rapid expansion of output coupled with considerable structural change. This is the picture that emerges from the summary figures presented in rables 1-4. Part A of Table 1 reproduces data on the growth of overall industrial output in constant and in current prices for a comprehensive industrial aggregate (designated 'Industry+') that includes enterprises managed at and below the village level. Output totals for village-level industry are shown separately; subtraction of these figures from the global total generates an output series (designated as "Industry") restricted to enterprises ope.ated at and above the township (xiang) level. Indexes of output growth from a 1978 base derived from these figures appear in Part B of Table 1; annual output changes for each category are derived in Part C. These data indicate a continuation of the rapid and sustained growth that has characLerized Chinese industry throughout the history of the People's Republic of China. Using the d-aa at constant prices (labelled GVIO), the ate of output growth clusters around the 10 percent annual level observed over the long term in China. If village industry is excluded, the average annual growth rate of OVIO for 1978-87 is 10.4 percent. Even though village industry reported annual growth averaging 24.1 percent in real terms during 1978-R7, the village component is so small that adding its inclusion in the total produces only a marginal increase in average growth to 11.3 percent for 1978-87. This is important because, as will be seen below, there can be little dovbt that the figures shown in Table 1 considerably exaggerate the growth of real output in the village segment of industry. The figures reported in Table 1 indicate a distinct acceleration of industrial growth during the mid-1980s, with reported real output growth approa.hing the 15 percent mark in 1983/84 1. One official of China's Sate Statitca Bumau oomniud tht Shim chnp enhanc Uth vemcity of repored daa.  4 How Fan Has ChMe Indrj&y ro*wi? and 1986/87 and surpassing 15 percent in 1984/85 even without the inclusion of village-level industries. Data showing the level and growth of industrial output at both current and constant prices can also provide information about changes in the pdice level for industrial goods. This is particularly important because Chinese sources give no systematic information on price trends fer industrial goods until 1984/85; branch indexes for prices of industrial goods exist only ftwm 1985/86 [Chirna Price 2 (1989): 59]. Implicit price indexets fer industrial output appear in Panel A of Table 1. These data reflect the gradual emergence of inflationary pressures beginning in the mid 1980s following yeaws of near-stability in industrial prices. Whether these data acciiat'ly reflect trends in prices paid for industrial goods and/or received by industrial produce." will be discussed below. Table 2 indicates the brealkdown of output into several ownership ca.egories: state-owned, collective, private, and other. These figures show that output from the c )llective sector, particularly its sub-categories of township and vilage enterprises (jointly described by the tem xiangzhen Qa , has grown much faster than producton in state firms, leading to a rapid decline in the formerly dominant share of state firms in total output value. The output totals in Table 2 are divided into two major sub-categories, state and coUective. Additional detail is given for two segments of collective industry, firms managw at the township (xi) and viLage (M) levels within the coUective sector. Since coUective enterprises also operate in urban areas, these segments do not exhaust the entire output of China's collective industries. Table 2 also displays output data for a separate category, xian3gzbhCnqy, which is often translIzted as *township and village enterprises' (abbreviated lVE), even though the term's literal meaning is 'township and market-town enterprises.' These data are of interest for two reasons: first, reported output has grown with extreme rapidity in recent years, so that the total amounts in 1987 to nearly one-third of national gross output. Furthermore, the scope of the TVE data remairs uncertain for the period beginning in 1984. Prior to 1984, the TVE output data are almost precisely equal to the sum of output value at fixed prices for township and village enterprises. Beginning in .984. however, we find a large and growing gap between output of township and village enterprises and the much higher TVE total. In addition, the price basis of the TVE data reproduced in Table 2 is not specified. We know that the data for 1978-83 represent the sum of output from township and village enterprise at constant prices. The enormnous growth of the TVE category starting in 1983/84 suggests that the. figures shown in Column E of Table 2 may represent outpu. valued at current rather than constant prices. Table 3 presents information on the branch structure of gross industrial output value at 1980 prices. These data, which exclude the minor category of non-independent accounting units and also exclude village enterprise, indicate very little change in the importance of major branches of Chinese industry during the past decade. This is confirmed by the rank correlation coefficient linking the size structure of industrial branches for combined state and collective industry (Panel  How FA M Chha induy Orow? 5 C of Table 2) in 1978 and 1987.2 The stabi!tv of br-rch stricture in both the state and collective segments of industry is a significant factor in analyzing trends in reported energy productivity. A stable branch structure eliminates the ossibility of raising energy productivity by raising the share of industil output produced in branches with 'ow energy-intensity. The figures in Tables 1-3 arc based on the 'gross value of industrial output," which represents the combined total of enterprise output val6s inclusive of materal costs. Table 4 contains information on the growth of net industrial output, which is calcu1bted by subtracting the vaiue of material inpuu (minerals, semi-fabricates. energy etc.) from gross output. The net value figures are usually rendered only in current prices; when nret output is presented in constant price terms, the figures are apparently obtained by multiplying net output in current prices by the ratio of gross output totals Li constant and current prices, which is equivalent to "single deflation rather than tU conceptually preferable "double deflation" used id contemporary industrial econormies, in which separate price indexes are used to remove the impact of price change from gross output and from the intermediate goods purchased hy industrial enterprises. The general picture emerging from shese data - rapid growth, acceleration after 1983, with differentially rapid cxpansion of the collective sector - parallels the results reported in Tble 1. The ratio of net to gross output is nearly identical in the state and collective sectors. Following decades of near-constancv in this ratio rsee Industry 1949-84, p. 41J, we now see a gradual but steady decline in the net output ratio for both state and collective industry. There are several possible reasons for this change: - changing product mix within individual brmnches of industry (note that stability of branch structure precludes change in this area as a source of decline in net outpmt ratios). - changing technology and efficiency in some branches relative to others. - differential inflation of raw matcrials prices relative to prices of finished products. - changes in the rate of double counting arising fr..a response to new marketing opportunities, enlarged ialization and inter-enterprise division of labor, expansion of subcontracting, and growth of joint production and transprovincial cooperation Of particular relevance here is the possibility that reform-induced increases in inter-enterprise specialization may have raised the, owth rate of gross output value above the growth rate of real industrial product. This outcome is not certain: we lack systematic data on the ratio of interentcrprise purchases to total out(ut (both level and time-path of this ratio) for 2. Braches 13 nd 14, nd also branch 11 ad 12 au combined for te oalwlaton becaus they amw not diMpiobshd in th 1978 fiSue; with 13 bmancr, he rmak ording of 7 (including the six lu brarchu) . identia in 1978 and 1957. The am of squa. of differen0e, nak orde nga hrd th tmainhng 6 branchs 136.  6 Now Fwt J.a.x Chwi-e lndvWy Growns? various types of erterprises. In addition, some units, isicluding the large and widely publicized Capital Iron and Steel Corporation, have taken advantage of reform policies to increase rather than reduce the degree of vertical integration, which has the opposite effect of causing gross value to lag behind the growth of real output. IV. SYMPTOMS OF UJPWAp1 BLAS iN CmNS INDUSTRIAL STATImCS Careful inspection of Chinese statistical publications raises the possibility that recent output totals based on wconstant prices* may exaggerate real outpit growth in the industriai sector. Indications of upward bias in the constant-price industrial output totals for the past decade appear from the following types of materials: - apparunt mismatch between growth of physical output and output value in certain branches of industry - indicaLions that township and village enterprise may confuse current and constant vices - possible increases in the rate of double counting, which would have the effect of artificially increasing the reported growth of gross output - indications that local govemments may falsify industrial output statistics to gain administrative benefits attached to achievement of lai ? output totals. - evidence that industrial output data do not adequately reflect price increases that have become pervasive in recent years. - indications that reported gains in energy productivity are unrealistically large, especially in the fast-growing machinery branch. A. Mlsmatch between Output and Value Data In some cases, the reported growth of real output value appears to outrun the expansion of physical output for major products. The most notable example of this occurs in the chemical industry, for which relevant data appear in Table 5. These data show that the arithmetic average, of annual physical output growth for nine major commodities lags behind the reported growth of output value at constant prices for every year beginning with 1979/80. The positive difference between reported growth of real output value and production volume for major commodities ranges from 2.56 percentage points in 1986/87 to 12.66 percentage points in 1984/85. It is, of course, entirely possible for real output to outgrow physical production of major commodities in a large and complex industry that turns out a wide range of products as well as  How Fau Has ChintE Indistry Gro"? 7 a variety of items within broad categories such as 'plastics." Furthermore, the calculations reported in Table 5 give identical weight to each of the nine products. On the other hand, the size of some of the annual differentials is troubling. Is it reasonable to anticipate that chemical output could rise in real terms by 11 or 12 percent during years in which average output growth of major commodities amounted to only 3.3 percent (1983/84) or even declined slightly (1984/85)? At the -ery least, these results suggest the possibility of upward bias in the value data for one of China's larger industrial branches. Figures for the machinery industry raise similar, though less serious, issues. B. Do Rural Industries Adhere to Standard Accounting Regulations? The extraordinary growth of township, village, and 'TVE' enterprises noted in connection with Table 1 raises questions about the veracity of the underlying output reports. In principle each enterprise is expected to compile output values based on both current and constant prices As reform causes managers to focus increasing attention on enterprise financial performance enterprise leaders (and presumably accountants and statisticians as well) pay growing attention to current cash flows. Despite the complexity arising from sales of similar products at multiple prices, output valued at current prices is closely related to enterpnise sales revenue, and therefore appears to pose little conceptual or practical difficulty even for the inexperienced and unsophisticated accountants available to small rural enterprises. To calculate gross output at fixed prices requires information about the fixed (1980) price o each item produced. Industrial ministries publish large compendia containing relevant price lists. Administrative units at all levels a,e responsible for passing on appropriate price information to enterprises under their jurisdiction. While this system has functioned smoothly for many years among large-scale enterprises in the staLe sector, the recent explosive growth of collective enterprise, especially in rural areas, raises the possibility that enterprises, their administrative superiors in township or county industrial bureaus, local statistical personnel, or all three group may have failed to implement the system of calculation in constant prices that is of crucial importance for industrial output statistics even though it is of little or no interest to enterprise personnel and perhaps to local government officials as well. China's statistical rystem operates according to a vertical hierarchy in which each administrative level compiles and processes statistical reports received from its immediate subordinate in the bureaucratic structure. Thus national agencies receive material prepared by provincial agencies, which in turn rely on data compiled by municipal or county authorities, who base their reports on data from local enterprises. This means that statisticians at higher administrative levels cannot easily evaluate the quality of the raw data underlying the reports that arrive on their desks, particularly when, as in the case of rural industry, these reports come from literally thousands of widely dispersed units. Under these conditions, it is entirely possible that data supposedly based on fixed 1990 prices could contain a substantial component based on (much higher) current prices. With inflation, substitution of current for constant prices imparts an upward bias to the resulting data. The view  8 How Fas: Has Chiae IndurJg Grow,? that failure to implement accounting conventions artificially inflates reported output growth, especia.Uy in the TVE sector, is widely shared within the Chinese economics community. An experienced accountant now working with a TVE machinery producer insists that even in the capital, failure to adhere to statistical regulations is not uncommon in the TVE sector; in rual areas, neglect of these systems is said to be widespread (personal communication, May 1989]. Officials of the State Statistical Bureau [SSB] agree that output data from TVE enterprises are problematic, that current prices are often used to calculate output values identified as based on fixed prices, and that the output totals tend toward bias in the upward direction (personal communication, May 1989; May 1990]. A position paper issued by the SSB in response to claims that its figures exaggerate the rate of industrial growth in recent years points specifically to rural enterprise as the chief source of what the SSB sees as a modest upward bias in its estimates of real industrial growth [SSB 1988; World Herald 1988]. An extenal researcher reports that visits to Jiangsu enterprises in the xiangzenQive category do indicate extensive mixing of data in current and constant prices, with enterprise leaders finding it difficult to explain which data are based on which prices (personal communication, July 1989]. A paper prepared by personnel at China's State Statistics Bureau [SSB 1988] reports that industrial units at the township (xiang level and above are required to submit monthly reports of GVIO at constant prices. The small, dispersed and numerous village-level, jointly operated (at or below villane level) and private (gi) enterprises submit only annual figures, and these are in current rather than constant prices. Provincial and local statisticai bureaux then adjust these submissions on the basis of coefficients derived from sample surveys or surveys of key enterprises. The SSB's position paper asserts that these procedures remove most, if not all of the shuifei (literally "water content") or upward bias from the output data associated with village-level industry. The SSB also points out that figures for these units are not included in monthly reports of industrial output [See World Herald 1988 for a summary of debate on the "shuifen"issue]. Despite this explanation from the SSB, examination of available data confirms the impression that output data for TVE enterprises, especially figures said to be based on 1980 fixed prices, probably overstate the actual expansion of real output, especially during recent years of extremely rapid enterprise formation. Prior to 1985, compilations of TVE data carefully specify which data are based on 1980 prices; more recent publications conspicuously omit any mention of the price base for output data pertaining to 1985 and subsequent years r[VP, 1978-85; TVP 1987; TVP 1988; TJNJ 1988, p. 294. Note that Agriculture 1988 gives TVE GVIO for 1987 at 1980 prices (p. 314) using ;he same data that appear with no price attrbution in TVP 1988, p. 261.  How Fan Hu Chia, Indtusy Gro'? 9 Several specific examples can iUustrate what appears to be considerable inconsistency in these data: Data for Beijing industrial enterprises for (million yuan): 1986 1987 Index 1986-100 1. Gross output, 1980 prices (GVIO) a. Including village and sub-village units 34858.07 39512.33113.3 b. Excluding village and sub-village units 32177.14 35723.28111.04 c. Difference: GVIO for vWillage enterprise 2680.93 3789.05141.3 2. Village & sub-village output, current 2480.06 3382.27136.4 prices (CVIO) 3. Ratio for village output: CVIO/GVIO 0.925 0.893 Source: Beijing 1988, pp. 255 (la-b), 364 (2). These figures imply that village level industrial output is higher in constant than in current prices. But a decline in average prices for industrial output betwean 1980 and 1986 is most improbable. These data also indicate that industrial prices continued to decline during 1986/87, which is definitely incorrect. Shanghai industrial output figures for 1987 (million yuan) raise further questions: A B Ratio 1980 B/A prices 1. Township enterprises 5428 8122 1.496 2. Village enterprises 4361 5029 1.153 Sources: Shanghai 1988, p. 127 (A); TVP 1988, p. 26 (B)  10 How Fat Has Chine. lndwy Grow? The data marked 'A," identified as based on 1980 prices, are much smaller than the figures marked "b," which appear initially without price attribution, but are also used to compile a total described as based on 1980 prices in a separate source (Agriculture 1988, p. 314 - this is a table of 1987 gross output for township and village industry by province "at 1980 prices"; although the line for Shanghai is blank, the national totals and data for other provinces are virtually identical with figures in TVP 1988, p. 26, which makes no mention of a price base]. Scattered data for several provinces imply improbable increases in labor productivity. GVIO for Tianjin's village enterprises reportedly increased by 81 percent between 1985 and 1987 even though employment rose by only eight percent [Tianjin 1988, p. 111]. In Liaoning, reported GVIO from township enterprises rose by 21 percent during 1986/87 despite a two percent decline in employment [Liaoning 1988, p. 508]. In Heiongjiang, township enterprises reported real output growth of 15 percent during 1986/87 while employment fell by one percent (Heilongjiang 1988, p. 291]. The impression of widespread substitution of current for constant price output figures is confirmed by Robert M. Field, who finds that 'a large and growing number of provinces have not distinguished output [of village-level industry] in current and constant prices. . . . in 1985 the output of village and below-village industry in current and constant prices were identi:al for all provinces [Field 1988, pp. 586-87, with emphasis added]. C. Has the Rate of Double Counting Increased? Measuring industrial output growth using information about changes in the gross value of industrial output can produce misleading results if changes in industrial organization alter the frequency with which materials, components, and services are exchanged among enterprises. Assuming no change in real output, a trend in the direction of vertical integration (e.g. merger of iron mines with steel plants) will cause measured output to decline (the sale of iron ore to the steel mill disappears from reported GVIO). A trend toward interenterprise division of labor, on the contrary, will cause measured output to increase. There are two reasons why one might expect the use of GVIO data to artificially inflate measures of industrial output growth during the 1980s. First, industrial reform has created new opportunities for inter-enterprise specialization and division of labor. Chinese economists and planners have long criticized the excessive vertical integration typical of Chinese industrial operations. Despite ample evidence that integration raises production costs, managers have persisted in building daerguan (large and complete) or i (smaUl and complete) manufacturing establishments in order to limit their dependence on unreliable external suppliers. Economic reform has increased the availability and reliability of external suppliers for a wide variety of commodities and services. There are many reports of new sub-contracting an-angements, inter-provincial joint ventures and other institutional changes that point in the direction of an increase in the overall ratio of inter-enterprise transactions to real output within the industrial sector. [Note, however, the apparent counterexample of the Capital Iron and Steel Corporation, which has used new opportunities for independent decision-making to reduce its  How Fast Ha Chsuu Indwitry Grow? 11 reliance on external suppliers through such measures as building its ov n power plant and acquiring a fleet of ocean freighters.] Even if there has been no trend toward inter-firm specialization at existing enterprises, the growing weight of rural and collective enterprises in the industrial output total has almost certainly brought a decline in average scale of industrial operations. This probably implies an increase in specialization simply because small enterprises lack the capacity to achieve the high degree of vertical integration typical of China's larger industrial units. We thus have two reasons for anticipating an increase in the ratio of inter-enterprise exchange of materials and semi-fabricates independent of any shift in product mix or branch structure of industry. Such a change would build an upward bias into the output totals reported in Table 1. Such a change would also systematically reduce the ratio of net to gross output value, which is exactly what we see in Table 4. Since several other factors also influence the ratio of net to gross output, the downtrend in the ratic of net to gross output observed in Table 4, although suggestive, is not sufficient to demonstrate either the presence of vertical disintegration or its possible impact on measures of real output growth. A more promising approach would be to look at levels and changes in the ratio of interenterprise purchase of energy, materials and semi-fabricates to gross output value in various subdivisions of the industrial sector. For example, the industry-wide average ratio of interenterprise purchases to gross output at current prices (CVIO) [call this ratio IEP/CVIO0, can be expressed as a weighted average of distinct IEP/CVIO ratios for large and small industry. Thus: [IEP/CVIO] = a[IEPl/CVI01] + (1-a)UEP2/CVI02] where 1 and 2 indicate large and small-scale industry and a is the share of the former in national industrial output. I hypothesize that: (1) the ratio (IEP/CVIO] is considerably larger for large than for small firms (2) the weight attached to large firms, a, has declined irn recent years (3) economic reform tends to raise the ratio (IEP/CVIO] for both large and smaii firms It should be possible to use panel data from a enterprise surveys to investigate both the level and time path of [IEPl/CVIOlj and IEP2/CVI02]. Together with data on the parameter,it should be possible to obtain some rough quantitative idea of possible bias in the output figures arising from changes in industrial organization.? 3. Wdiliam Byrd observes that the foregoing discusion overloolk the pouible impat of changes in the price of materils and intemediate goods relative to the price of final producs; such changer might be systeuailcaUy different for large and s1A1 finr.  12 How FM Ha Cbae ldutiy ww? D. Do Local Authoritles Falsify Industril Output Data? Well-informed statistical personnel report the existence of incentives for exaggeration of industrial output growth by local govenments in some regions of China. In Jiangsu, for example, municipalities that surpass threshold levels of industrial output (e.g. 1 billion yuan) are granted special privileges in the form of exemption from certain types of regulation (e.g. direct access to provincial funds vs. application through county offices). Falsification, if extant, presumably taces the form of inflating gross rather than net output [note that the former can be easily inflated by anranging exchange of materials or semi-fabricates anong producers of similar commodities]. If so, falling ratios of net to gross output will be typical of entities with inflated gross output totals. Unfortunately, there are many other factors influencing this ratio, so that falsification might easily escape detection, especially if the amounts are small relative to local and national totals. At the same time, other enterprises or localities may conceal some portion of their actual production in the hope of avoiding taxes. E. Industrial Output Data Accurately Reflect Recent Inflation Experience? After decades of considerable price stability, China has experienced growing inflationary pressures during the 1980s. Since Chinese price index compilation has previously focused almost exclusively on consumer prices, the impact of recent inflation on industrial prices is not easily discerned. The only systematic effort to monitor trends in industrial prices appears to come from the State Price Bureau, which now surveys price conditions in several thousand industrial enterprises and uses the resulting data to compile indexes of ex-factory prices for industrial products as well as purchase prices for major raw materials, fuels, and power [personal communication, May 1989]. Summary figures for 1984/85 and more comprehensive results, including price indexes for 15 industrial branches and purchase price indexes for eight classes of materials appear in the Bureau's journal rChia Pnricc 2 (1989): pp. 59-60]. These figures confirm that industry has experienced the inflationary trend reported for urban consumer goods, Annual increases for ex-factory prices averaged 8.7 percent in 1984/85, 3.8 percent in 1985/86, 7.9 percent in 1986/87 and well over 10 percent during the first eight months of 1988 [China Price 2 (1989): 59]. Price increases for materials, fuel, and power were substantially larger in each year [ibid., 60]. The influence of commercial intermediaries (including government agencies), in the inflationary process is visible if we compare trends in ex-factory prices of mining products and raw materals" (these categories appear not to overlap) with trends in purchase prices for wall raw materials" (percent increases over the previous year):  How Fan Hu Ouaw hmaaty 1r Ex-Factory Prices Purchase Prices Mining Raw All Raw Materials Products Materials 1985 108.8 110.9 118.0 1986 100.6 107.5 109.5 1987 114.1 106.9 111.0 Aug.1988 108.8 117.8 123.9 Source: China Price 2 (1989): 59-60. Profitable intermediation, whether by state agencies or by legitimate or illegal private enterprise (sometimes involving persons with official connections) is an important componen of the "official speculation' (guandao) widely reported in the Chinese press. Beijing's large Yanshan Petrochemical Company reportedly "sold almost all their products to the State at official low prices" only to discover that 'most of their products' users paid much higher market prices" for the same materials [China Daily 6-3-1989, p. 1]. Although inflation of industrial prices seems to have acclerated in recent years, substantial price increases certainly occurred prior to 1984/85. Data in Table 6 show a rising trend for coal prices paid by electric power plants beginning as early as 1978/79. Data for the construction industry show a consistent pattern of rising building costs from the stat of annual time seIes data in 1978 [TJNJ 1988, p. 590]. Since these figures exclude land costs, it would appear that rising costs of construction mateials, which prompted complaints in the Chinese press during the late 1970s, also date back to this period. Interview data collected by foreign reserchers give a strong impression of rising machinery prices during the late 1970s and early 1980s [personal communication]. Information on prices paid by power plants for coal illustrates the possible inconsistency between information about inflation and the price changes implied by industrial output statistics. The electric power industry is dominated by large, stateowned enterprises. It seems reasonable to assume that thermal power plants obtain the bulk of their coal requirements through planned allocations at low official prices, and that they enjoy considerable protection from the "official profiteers' attacked in the Chinese press. This would imply that, relative to otber consumers of coal, power plants are somewhat insulatid from inflationary pressures, and that trends in their coal costs should unmad the average rise in coal prices for the entire economy. Data reproduced in Table 6 show that averagc coal costs in China's power industry havi risen steadily since 1978. The index of coal costs, which should represent an underestimate o the aveage rise in coal prices, shows an increase of 88.9 peracnt between 1978 and 1987. The implicit price indicator calculated from gross output at cunrent and constant prices for the coal industry, however, shows much smaller increases of 66.7 percent for the state sector and 48.9 percent for the (much smauller) collective mining sector. These data lead to the conclusion that the annual price changes implicit in statistics of gross output value at current (CVIO) and  14 How Fast Has Chine Inditay Oroww? constant (GVIO) prices probably understate changes in the sales price of coal received by the producers. Specifically, we anticipate that, Jf t indicates time, the ratio of annual output data CVIO(t)/GVIO(t) is too low. Inconsistency among several data series raises the question of which is most likely to be in error. The conclusion that the ratio CVIO(t)/GVIO(t) is probably too small is based on the judgment that figures for average coal cost, which come dirctly from records of financial and material transactions maintained by large, well-establishri units, are less subject to distortion than ynthetic calculations of output value. If the distortion is contained in the output figures, the previous discussion suggests the constant-price figures (GVIO) as the probable locus of difficulty.4 This reasoning is not foolproof, but it appears that the most likely explanation of inconsistency between coal costs to the power industry and the 1VTO and CVIO data relating to coal production is that the series of output value at fixed w faster than the real value of industrial output in the coal industry. Inconsistency between information about inflationary patterns and the price indexes implicit in China's industrial output statistics is not limited to the coal industry. More general difficulties become evident when one compares the implicit price indexes derived from output data for the state and collective segments of industry. Despite the progress of economic reform, markets for Chinese indistrial output remain heavily regulated. Many firms are obliged to sell substantial portions of their output at low controlled prices. Even when firms are allowed to sell output at 'negotiated' or "imarket' rather than plan prices, they encounter numerous controls [Ishihara, 1989]. Government agencies set maximum prices, as when the State Price Bureau issues 'upper price limits for means of production outside the plan," including petroleum products, aluminum ingots and steel products [frice Thcy 3 (1988): 51-53). Extra-plan sales of steel and non-ferrous metal products are also restricted to designated commodity exchanges [rinc TheoX 5 (1988): p. 57]. These measures are clearly designed to contain and restrict the rise of industrial prices. "law-boning," or personal official advice intended to limit price increases, has the same effect. Each of these measures is directed primarily toward, and applied most forcefully to the activities of large, stateowned enterprises. Under conditions of excess demand in which govemment struggles to prevent prices from rising to market-clearing levels, it is difficult to doubt that small, widely dispersed collective enterprises encounter less restriction on product pricing than large, highly visible firms in the state sector. For this reason, there is a strong presumption that the rate of price increase for commodities produced in the collective sector will tend to outpace comparable inflation rates for similar products in the state sector. 4. Itis also pouible that riing makups by cooomeil intennsdiarim have widenod Xw p berwes tbe primes oseved by ooal produm and the price paid by oal user for the powing porton of astha oocur oute X pia frmwork. However, Jcfkon. RAwWi nd Zheng find no evidence of A genral ris in markups for industrial intemediat goods.  How Fail Hai Chuis Indu4wry Grow,? IS Unfortunately, the price indexes derived from statistics of industrial output value point in the opposite direution. Table 7 shows that implicit price increases reported for the collective sector fall short of comparable state sector data in every year since 1978. Typically, reported inflation is the coUective sector is half or less of ieported intlation in the state sector. Comparison of annual inflation rates for state and collective industrial ou.put witiin the same branch produces the same result. Between 1978 and 1987, nearly three fourths of the instances where comparable data exist (91 of 126 cases; 9 cases are excluded because of data incomparability), the implicit inflation is higher in the state sector. In recent years, this result is even clearer: during 1984/87, we find 35 instances of higher implicit branch price increase; in the state sector compared with only 8 instances of higher price increases in the collective sector (two items are discarded as incomparable)S These conclusions are not acceptable, particularly since data from the coal sector suggest that price indicators extracted from the reported growth of industrial output in constant and current prices already understate industrial inflation for the state sector. The figures shown in Table 7 suggest that output statistics for the entire collective sector, which now accounts for nearly one-third of industral production, systematically understate the impact of inflation. If this is true, the most likely mechanism is that rorted output yalue at fixed prices forChir1. collective industries Systematically overstates the growth of real out= during the 1980s, particularly in recent years of strong inflationary pressures. F. Are Reported Gains in Energy Productivity Unrealistically Large? Data reproduced in Table 8 indicate that China's industries achieved very substantial gains in energy productivity during the early 1980s. These figures, which exclude village industries (see below) indicate annual gains averging 6.7 percent in real output per ton of standard coal equivalent. This compares well with figures for energy productivity in mining and manufacturing for major industrial nations showing average annual productivity growth of 2.6, 3.4 and 7.0 percent for West Germany, the United States and Japan respectively during the period 1973-86.' If correct, these figures indicate a highly effective response to energy shortages within Chinese industry, which accounts for more than two-thirds of China's energy consumption. Many observers have noted that China's energy prices are much lower, in relative terms, than comparable prices in cther nations and in the world market, and also that domestic energy prices have not risen in paralle with global market trends. Despite the inflexibility of official prices, the data in Table 6, showing that avemge coal prices paid by thermal power plants nearly S. ..ece commenu am based on a workshea DEFLATE2.wkl. 6. Based on a scpare workshee NENOCOMP.wkl, which is not included in this paper.  16 How Fwat Mar Chine Irdmrbv Grow? doubled between 1978 and 1987, imply still larger increases in the mumsinal prices paid even by high-priority energy consumers in China's industrial economy, Even where energy prices have not risen steeply, reports suggesting widespread energy rationing point to the binding nature of energy constraints across much of Chinese energy. Under these conditions, managers seeling higher financial returns will impute a high opportunity cost to inessential energy consumption even if direct costs are low. It thus seems reasonable to conclude that many Chinese managers feel intense pressure to economize on energy consumption. Changes in industrial energy productivity can be decomposed into three components: changes in the branch structure of industry that decrease the reladve weight of energy-intensive industries, reductions in energy requirements for producing specific commodities, and changes in the commodity structure of output within individual branches of industry. We have already seen that Chinese industry experienced no significant change in branch structure during the past decade (Table 3). Review of Chinese publications, which include numerous descriptions of physical input-output coefficients related to energy consumption, sugest only modest gains from reduced unit energy requirements for specific products. Electric power consumed in producing one ton of crude oil or raw coal increased every year during 1980/85 (Energy 1986, pp. 86, 504]. Coal consumption per idlowatt of power turned out by large power plants or per ton of cement produced by major plants declined, but by less than five percent [ibid., 508, 527]. Since the share of output coming from small plants, which are criticized for their excessive energy requirements, has risen in many branches, the potential for major growth of energy productivity from reduced unit energy requirements for specific products seems very limited. This leaves structural change within individual branches of industry as the main locus of improved energy productivity for Chinese industry during 1980/85. Tlhis conclusion is problematic because substantial intra-branch restructunng appears limited to a few branches - machinery, chemicals, and perhaps textiles. This outcome draws attention to the possibility that a portion of the productivity gains repot.ed in Table 8 may be attrbutable to measurement error. To explore the possibility of measurement error, and also to investigate the branch pattern of cnange ip. sarmgy productivity, we turn to an examination of energy data for 15 branches of Chinese industry during 1980/85. Table 9 pnsents energy consumption data for 15 branches during 1980/85. Panel B uses these figures to derive annual percentage changes in energy productivity (GVIO/E, where E represents energy consumption in terms of standard coal) for 15 branches. These data conta a number of improbable items: can we believe, for example, that energy productivity in food processing rose by 27.3 percent during 1984/85, or that energy productivity in machine-building rose by 21.1 percent in the same year?  How Fad Has CAsa Inr4ustry Oro,? 17 We can investigate the consistency of the data for energy consumption (not shown) and energy consumption per unit of GVIO by extracting the implicit branch figures for GVIO and comparing them with GVIO data from other sources. Tis is done in Table 10, which uncovers unacceptably large discre)ancies for branches 9-14 in the 15-branch clasification. Fortunately, data for the sectors that consume the largest quantities of energy, namely metallurgy, power, chemicals, building materials, and machine-building, are not involved in these inconsistencies. Table 11 presents a recalculation of EIGVIO and of annual percentage changes in GVIO/E for 1980/85 based on published branch data for E and on information from other sources giving branch time series of GVIO. These revised data show fewer improbable entries (readers should ignore the problems in branches 13 and 14, which should be merged in a future recalculation). These results suggest that, among the major energy users, machinery, chemicals and, to a lesser extent, metallurgy, have achieved considerable success in raising energy productivity, while electricity and building materials have recorded much smaller gains. The veracity of these data, however, depend substantially on the accuracy of GVIO data for chemicals and machinery - exactly the sectors for which comparison of physical production and value data indicate the possibility of upward bias in the value totals. Since energy productivity in these sectors, alone among the major using branches, rises much faster than the reported national avenge, this dependence is considerable. Removal of the chemical and machinery branches, which contribute over half of the overall energy savings attained during 1980/85 (Table 16) reduces the cumulative growth of energy productivity during 1980/85 from 28.6 percent to 17.3 percent (Table 16, Panel 1). If we were to assume that the growth of energy productivity in machinry and chemicals was limited to this lower amount, rather than the much larger figures shown in Tables 1lB and 16, the average annual growth rate of the entire industrial sector (excluding village enterprises) during 1980/85 would be reduced by two percentage points, from 10.8 to 8.8 percent annually.7 Note, however, that the World Bank anticipates large reductions in unit consumption of electricity in the chemical brarich because of slow relative growth of synthetic ammonia, "dramatic reductions' in unit power requirements for synthetic ammonia, and the international trend toward reduced power intensity in chemical manufacture (1985-A3, pp. 47-49]. Furthermore, Chinese specialists regard the materials contained in Energy (1986] as preliminary; this book was never released to the general public. However, it is my impression that data issued in subsequent publications (notably Energy [1989]), will support similar results. The importance of energy issues and the major differences in scale and technology separating state and collective industry makes it important to provide separate analysis of energy consumption in state and collective industry. As far as I can determine, China's statistical agencies have made no effort to do this. Industrial census data giving energy consumption for 7. Thi calculzion is based on first column of Table 2B, following a separae worksheg 'What if Energy Savings ate Trimmed?' dated 7-4-1989.  18 How Fa Ha CHasM . Indaw Gou? state sector independent accounting units in 1980 and 1985 allow a trial calculation of energy consumption and energy productivity trends for collective industry durinig the period 1980/85. NOTE: this can perhaps be extended to 1986 (using data in TJNJ 1988, pp. 424-436). This is done in several steps: (1) exanine energy data for state sector independent accounting units divided into 40 branches; (2) collapse data for 40 branches into 15 branches; (3) obtair. estimates of collective sector energy consumption in 15 branches for 1980 and 1985 by subtracting energy consumption by state-sector firms from the national totals underlying Table 9. (4) calculate changes in energy productivity for the collective sector using derived or published figures of branch OVIO. Energy data for 40 branches of industry in 1980 and 1985 are reproduced in Table 12, where I find no significant inconsistency between implied and publisned data for branch GVIO (readers should ignore the discrepancy for branch 7, which is trivial in size, and in the residual branch 40). Again, we see substantial increases in energy productivity over 5 years; again, these gains depend crucially on reported increases for a small number of branches that consume large amounts of energy and report above-average productivity gains: chemicals (branch 26); machine manufacture (branch 35). These data can also be used to compute the 'energy savings' arising from the presence of lower unit energy consumption coefficients in 1985 than existed in 1980. As before, two sectors, machinery and chemicals, dominate the calculated savings, accounting for 45 percent of the total amount (Table 16). If we were to assume that the path of energy productivity in these two branches palleled the (considerably slower) gains reported for other branches of industry, the average annual growth rate for state industry during 1980/85 would decline by one percentage point, from 8.2 to 7.2 percent.' Table 13 reports the result of calculations that collapse data for 40 state-sector branches into 15 branches and derive figures for changes in energy productivity for the 15 branches between 1980 and 1985. The transition from 40 to 15 braiiches is incomplete because it is not possible to make adjustments for 18 sub-branches (as is done in a separate worksheet, DATA87.wkl); the discrepancy, however, is not large. Energy productivity for 15 branches is calculated in two ways: first, using GVIO data that is derived from the 40-branch figures shown in Table 12; and second, using data from other sources that give GVIO for independent accounting units in the S. Thuis i bued on Table 2 and a s.paz workshea What if Energy Savings ar Trimmed?'.  HoaswFt fai Chi,e InduurJy Groy? 19 state sector according to the 15-branch classification in use before 1986. Although the differences between the two sets of calculations are not large (again, note that branches 13 and 14 should be merged), the latter figures are preferable. Here again we see the importance of machinery and chemicals, which are the only large branches reporting above-average growth of energy productivity. Table 14 presents trial estimates of energy consumption and productivity change for 15 branches of collective industry during 1980/85. In Panel 1 of Table 14, branch clergy consumption and branch GVIO are derived as residuals from the national totals and the state-sector figures presented above. Comparison of GVIO figures derived in this manner with published data showing branch GVIO for collective-sector independent accounting units at and above the &ing level (recall that the basic energy consumption data for 1980/85 appear to exclude village-level industry), reveals massive inconsistency. We therefore focus our attention on Panel 2 of Table 14, in which branch energy consumption for collective industries is derived as a residual, and then combined with published data on collective sector branch output to obtain productivity figures. Scrutiny of these results yields the following observations: 1. There is a major inconsistency in data for the electric power industry, in which data for the state sector alone indicate much larger energy consumption in both 1980 and 1985 than for the entire power branch! As a result, the calculation reported in Table 14 indicates large negative energy consumption in the collective power industry for both years. Less worrisome discrepancies appear in branches 4 (in which the collective sector minute) and 14 (probably reflecting need to merge with branch 13). 2. Output value per unit of energy consumed (partial energy productivity) appears much higher in the collective than in the state sector. 3. If valid, the foregoing observation appears to be the result of differences in output structure rather than superior coUective productivity on a branch-by-branch basis. Energy consumption per unit of real output by collective producers is markedly higher in branches 1, 6 and 9 (metallurgy, machinery, food processing) than in the state sector. Partial energy productivity seems to favor collective firms in branches 7, 8, 10, and 12 (building materials, forestry, textiles, leather), of which only 7 and 10 are major branches. In 7, the quality of small-plant output is far inferior to the state-sector norm [World Bank 1985-A3, p. 17]. Partial energy productivity is similar for state and collective firms in branches 5 (chemicals - here the comparison is blurred by major differences in product mix) and 11; the comparison is obscured by data problems for the remaining sectors. 4. Cumulative gains in energy productivity for the collective sector, summarized in Table 15, are far larger than comparable gains for the state sector in every significant branch except building materials (ignore the confused and minor branches 13-15). In most cases, the margin of difference is extremely large.  20 How Fast Has Chbwae Inidut Growe? This last result seems quite improbable, and once again calls attention to the possibility of upward bias in available measures of real output growth fnr the collective sector. A final point about e'nergy data. There is a variety of material suggesting that overestimates of output growth may not be confined to TVE enterprises and the collective sector. Some examples: Energy consumption per 10,000 yjan of output in Beijing's electric power industry dropped from 4.62 to 4.07 tons of standard coal during 1986/87, indicating a rise of 13.6 percent in energy productivity, even though coal consumption per kldowatt-hour of power produced did not change [Beijing 1988, pp. 291, 372]. Nationally, the electric power industry reports a 7.5 percent drop in unit energy requirements during 1980/85 (Table 11-B) even though coal consumption per kwh for large power plants (6000 kw and above), which produce a large share of total power output, declined by only 3.6 percent during the same period (Ene gy 1986, p. xxx]. The chemical industry reports thai energy productivity increased by 12.6 percent in 1984/85 (Table 11-B). During the same perio:'. !owever, 8 of 16 unit energy coeffici:nts for major plants actually increased! The remainiig eight coefficients declined by an average of 5.8 percent. Only 1 of 16 coefficients for major plants declined by more than the reported industry-wide average (Energy 1986, p. 87]. The building materials industry reported a 20 percent rise in energy productivity during 1980/85, but unit energy requirements at major plants decline by a maximum of 6.3 percent [Energy 1986, p. 88] Reported energy productivity in coal production improved in every year but 1984/85 (Table l 1-B), showing a cumulative gain of nearly five percent. Yet data for major enterprises show a rising treiid for unit energy requirements; power consumption per ton of raw coal rises in every year [Energy 1986, pp. 86, 493-500] V. CONCLUSIONS This survey reveals considerable evidence pointing to the existence of upward bias in measures of China's real industrial output during the past decade. The issue is not whether such bias exists, but whether or not its presence substantially alters our perception of the rate and pattern of Chinese industrial growth. To clarify this issue requ res an investigation of the possible extent of upward bias. This in turn will require an analysis of possible links between upward bias, which itself is difficult to observe, and other economic patterns that may be more readily measurable.  How Fait Has Ch;na Irdalwvy Grown? 21 TABLE 1: ALL INDUSTRY A. Level of Industrial Gross Output, 100 million yuan Year GVIO GVIO GVIO CvIO Price Annual % Iadusuy Viage Industzy + Indus + Iadex Inflazion Data in 1970 prices 1978 4231 161 4392 4237 96.47 1979 4591 184 4775 4681 98.03 1.62 1980 4992 222 5214 5155 98.87 0.85 1981 5199 241 5440 5400 99.26 0.40 Data in 1980 prices 1981 5178 246 5424 5400 99.56 N.A. 1982 5577 277 5854 5811 99.26 -0.29 1983 6164 325 6489 6461 99.57 0.30 1984 7030 460 7490 7617 101.70 2.14 1985 8295 661 8956 9717 108.50 6.69 1986 8979 841 9820 11194 113.99 5.06 1987 10307 1150 11457 13813 120.56 5.76 B. Index of Output Growth, 1978- 100 1979 108.51 114.28 108.72 110.48 1980 117.99 137.89 118.72 121.67 1981 122.88 149.69 123.86 127.45 1982 132.35 168.55 133.68 137.15 1983 146.28 197.76 148.18 152.49 1984 166.83 279.91 171.04 179.77 1985 196.85 402.21 204.52 229.34 1986 213.08 511.74 224.25 264.20 1987 244.59 699.77 261.63 326.01 (continued)  22 How Fast Has Chine Inudwy Grown? TABLE 1: ALL INDUsTRy (continuation) Year GVTO GVIO GVIO CVIO Price Annual % Industry Village Industry + Indus + Index Inflation C. Annual Output Increase Over Previous Year (percent) 1979 8.51 14.28 8.72 10.48 1980 8.73 20.65 9.19 10.13 1981 4.17 8.56 4.33 4.75 1982 7.70 12.60 7.93 7.61 1983 10.52 17.33 10.85 11.18 1984 14.05 41.54 15.43 17.89 1985 17.99 43.70 19.57 27.57 1986 8.24 27.23 9.65 15.20 1987 14.79 36.74 16.6 23.40 Source: Rawsli written files GVIO-tables.  How Fast Hai Chinw Indwray 7rown? 23 TABLE 2: BREAKDOwN OF GROSS INDUSRAL OUTPUT AT CONSTANT PRICES DATA EXCLUDE VILLAGE-LEYEL UNITS B C D E F G H Total State Collective Sector Categories Xiang Total Xiang-A Xiang-B CunlDui zhen 1978 4231 3416 814 212 224 161 385 1979 4591 3720 871 234 241 184 424 1980 4992 3928 1034 280 286 222 509 1981 5199 4028 1131 310 321 241 579 1981 5178 4054 1089 323 332 246 579 1982 5577 4340 1193 354 369 277 646 1983 6164 4748 1354 413 432 325 757 1984 7030 5171 1758 539 575 460 1245 1985 8295 5840 2301 742 799 661 1827 1986 8979 6201 2637 948 2413 1987 10307 6902 1217 3243 Check H-(F +G) 1978 0 1979 -1 1980 1 1981 17 1981 1 1982 0 1983 0 1984 210 1985 367 1986 2413 1987 3243 Sourvs: Rawsld written file.s GVIO-tablea  24 How Foat Has ChinueIndwr^y Grow.? TABLE 3: BRANCH STRUCTURE OF GVIO AT 1980 PRICES (PERCENT) A. Branch Structure for State Enterprisea (SOE) Br 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1 11.80 11.99 11.81 11.08 10.96 10.73 10.77 10.51 10.58 10.52 2 4.82 4.79 4.87 4.86 4.82 4.59 4.45 4.66 4.61 4.59 3 4.26 3.99 3.61 3.42 3.36 3.25 3.15 2.99 2 83 2.65 4 8.01 7.81 7.49 7.16 6.83 6.69 6.68 6.52 6.43 6.37 5 11.81 11.60 11.92 12.02 12.41 12.44 12.43 11.70 12.61 12.36 6 21.12 21.34 20.26 18.55 19.86 21.34 23.12 24.58 25.86 25.51 7 2.75 2.69 2.67 2.50 2.57 2.56 2.59 2.81 3.45 2.70 8 1.89 1.86 1.78 1.73 1.67 1.56 1.48 1.24 1.14 1.13 9 12.44 12.65 13.01 14.30 14.58 13.86 13.52 13.19 12.99 12.99 10 13.83 14.20 16.11 18.10 16.67 16.78 15.56 15.19 14.56 14.42 11 1.04 0.51 0.54 0.60 0.56 0.55 0.56 0.56 0.52 0.53 12 0.00 0.53 0.58 0.63 0.57 0.52 0.48 0.47 0.48 0.46 13 3.07 3.18 3.18 2.93 1.44 1.43 1.46 1.48 1.60 1.64 14 0.00 0.00 0.00 0.00 1.48 1.44 1.42 1.48 0.46 0.47 iS 3.14 2.85 2.17 2.12 2.24 2.26 2.32 2.62 1.89 3.67 Sum 100.00 00.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 B. Branch Structure for CoUective Enterprises (COE) Br 1978 1979 1980 1981 1982 1983 1984 1985 198b 1987 1 1.97 2.28 2.34 2.17 2.32 2.47 2.54 2.89 3.39 3.56 2 0.11 0.15 0.17 0.21 0.23 0.25 0.22 0.20 0.20 0.20 3 2.27 2.12 1.94 1.87 1.93 1.94 1.92 1.64 1.63 1.42 4 0.12 0.13 0.13 0.11 0.11 0.14 0.13 0.15 0.18 0.21 5 10.94 10.74 10.76 10.93 11.45 11.94 11.26 10.94 11.17 11.30 6 36.07 34.84 32.63 30.38 30.68 31.55 30.95 33.27 32.83 33.07 7 8.37 8.63 8.32 7.91 8.57 8.46 8.17 7.50 8.89 7.51 8 2.55 2.73 2.72 2.64 2.71 2.52 2.33 2.23 2.85 2.28 9 2.96 3.39 3.70 4.39 4.69 4.71 5.04 4.96 5.37 5.46 10 7.90 8.58 10.04 11.92 12.08 12.26 16.02 16.3 15.76 15.96 11 11.28 9.36 10.49 10.96 9.59 9.22 8.44 7.10 6.39 6.22 12 0.uO 2.35 2.94 3.05 2.58 2.34 2.09 2.08 2.17 2.20 13 5.48 5.06 5.35 5.53 5.26 4.61 1.02 1.01 2.45 2.61 14 0.00 1.02 1.02 1.06 1.08 1.08 4.18 5.13 3.62 3.94 15 9.97 8.61 7.43 6.88 6.74 6.50 5.71 4.59 3.11 4.07 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 (continued)  How FPa Hat ClaAmaal Industry Ora"? 25 TABLE 3: BRLNCH STIrUCT1z o GVIO AT 1980 PRiCzS (PEcCNT) (coaninuatin) C. Branch Structure for SOE and COE Combinod (peamt) 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1 9.99 10.21 9.90 9.18 9.07 8.84 8.69 8.39 8.49 8.37 2 3.9S 3.94 3.92 3.86 3.82 3.61 3.38 3.43 3.33 3.23 3 3.91 3.6S 3.27 3.08 3.0S 2.96 2.84 2.61 2.48 2.27 4 6.56 6.40 6.00 5.6S 5.36 5.22 5.02 4.7S 4.61 4.47 5 11.65 11.45 11.69 11./9 12.20 12.33 12.14 11.49 12.19 12.03 6 23.87 23.81 22.76 21.08 22.22 23.63 25.10 27.00 27.88 27.35 7 3.79 3.78 3.81 3.65. 3.87 3.88 4.00 4.11 S.03 4.19 8 2.01 2.02 1.97 1.92 1.89 1.77 1.70 1.51 1.64 1.48 9 10.70 10.96 11.13 12.18 12.42 11.81 11.38 10.91 10.77 In).66 10 12.74 13.17 14.89 16.78 1S.67 15.77 1S.68 1S.50 14.91 14.19 1 1 2.92 2.13 2.55 2.82 2.53 2.49 2.55 2.38 2.23 2.29 12 0.00 0.86 1.06 1.1S 1.01 0.93 0.89 0.92 0.97 1.00 13 3.52 3.53 3.62 3.49 2.28 2.1S 1.34 1.35 1.85 1.94 14 0.00 0.19 0.21 0.23 1.39 1.36 2.12 2.49 1.38 1.54 15 4.39 3.91 3.23 3.14 3.22 3.21 3.17 3.16 2.24 3.79 100.00 100.00 100.00 100.00 100.00 99.99 100.00 100.00 100.00 100.00 Note: these data exclude vilba-level ;mepris; they ar oonfingd to indepmida accounting units. Key to branches: i metallurgy 2 power 3 coal 4 petroleum S chemicals 6 machiney 7 building materials 8 forestry ad wood processng 9 food procesing 10 textile 11I 12 leuher prcesing 13 paper 14 cultnual and at products 15 other Source: Wodrhot STRUCTURB.wkl  26 How Fast Has Cine Inrdwtyr Grown? TABLE 4: NET OUTrT AT CURRENT PRICES (100 MILON YUAN) INDEPENDENT ACCOUNTiNG UNITS, EXCLUDING VILAGE ENTERRLSEs Total State Colective Xiang 1978 1358 1979 1486 1980 1648 1319 322 114 1981 1690 1317? 342 1982 1774 1373 369 1983 1930 1501 415 1984 2246 1721 506 183 198S 2767 2058 679 247 1986 2979 2178 763 296 1987 3488 2530 894 354 Source: Rawali wntten files GVIO-tables. Net output ratio (baed on CVIO data for independent accounting unit - not shown) Total State Colective 1978 1979 1980 0.35 0.351 0.348 1981 0.344 0.341 0.337 1982 0.336 0.332 0.334 1983 0.332 0.332 0.332 1984 0.333 0.340 0.314 1985 0.329 0.336 0.313 1986 0.316 0.322 0.301 1987  How Fast Has ChIat Indua Gretw? 27 TABLE 5: CEHsUcAL INDurrRy Averag % Change Differmtial Perew Change GVIO,P80 Growth of physical output OVIO, percentage Nine Products 1979 13.51 7.01 -6.60 1980 7.85 10.77 2.92 1981 -4.06 4.66 8.72 1982 7.30 11.43 4.13 1983 8.33 12.46 4.13 1984 3.30 12.04 8.74 1985 1.05 11.61 12.66 1986 5.77 12.17 6.40 1987 14.S2 17.C8 2.56 Source: 1988 TJNJ, pp. 34546 for commodity dat; OVIO dat aro from Indusry 1988, p. S4. Thea figtuem ezcluds villge-level onprises. Figures for 1978-80 woro converted from 1970 to 1980 prime usaug th ntio of .1981 gro output for chemicu_ at 1980 and 1970 prioe  28 How Fast Has Chks Indawmty rotm? TAILE 6: ALTmENATrVE DATA To MEASURe finLATION IN THE COAL INDUSTRY CVIO/OVIO Cumulatve A B C Annual Price Price Chang Costiton % change Price Change From Since 1978 Sid Coal from pan Indx Pat Year Percet Y/ton yur 1978-100 SOB COE Power Plats 1971 43.77 NA. 100.00 1972 38.99 -10.92 89.08 1973 39.86 2.23 91.07 1974 43.10 8.13 98.47 1975 44.17 2.48 100.91 1976 44.70 1.20 102.12 1977 44.80 0.22 102.35 1978 42.06 -6.12 96.09 1979 44.93 6.82 102.65 14.32 4.44 6.82 1980 46.18 2.78 105.51 7.25 1.38 9.80 1981 47.71 3.31 109.00 2.56 5.05 13.43 1982 55.40 16.12 126.57 1.66 4.99 31.72 1983 61.32 10.69 140.10 1.23 5.41 45.79 1984 64.01 4.39 146.24 2.71 3.82 52.19 1985 69.21 8.12 1S8.12 13.68 9.83 64.5S 1986 7S.21 8.67 171.83 5.13 2.01 78.82 1987 79.46 5.65 181.54 4.94 3.97 88.92 Source: Coal costs from Xu et Al (1989); implicit price indbexas tAken fom workshoet DEFLATE2.wkl.  How rar Hu ChLRue I1duvY Grow? 219 TABLZ 7: PRICt INDIn EXTRACTD MOM INDUSAL Ouwnr VALUz DATA SOB Year PR1CE INDEX ANNUAL % CHANGE DIFFERENTLAL SOE COE IN PRICE LEVEL PRCE CHANGE SOE COE percentage poin Combined Data for All 15 Banch 1978 0.96 0.97 N.A. N.A. N.A. 1979 0.98 0.97 1.84 -0.12 1.97 1980 0.99 0.96 1.09 -0.60 1.69 1981 1.00 0.96 0.38 0.04 0.35 1982 1.00 0.95 0.10 .1.06 1.17 1983 1.00 0.95 .0.05 4.36 0.41 1984 1.02 0.96 1.95 0.72 1.23 1985 1.08 1.00 6.30 4.28 2.02 1986 1.12 1.01 3.33 1.50 1.83 1987 1.20 1.05 7.07 3.71 3.36 Soure: Worksheet DEFLATE.wkil.  30 How F Has ChIldmsy Grows? TAJLT 6S ZNUGY PRODUCtMITY PI CN INDUMY OmciAL DATA, 1918-7 You Eneg UN Annual Pecmu PA 10 Milioa Chanp in Yuan of OVIO Ewa (ton std. coal) Productivity 1980 78411 ns. 1981 72380 8.33 1932 70374 2.85 1983 6704S 4.97 1984 62592 7.11 1985 56872 10.06 1986 1987  How FuM HMm Chlmue Ir4Wi Gru? 31 TAILE 9: OmciAL ENERGY DATA FOR CHNE INDUSTRY BY BRANCH, 1980-198S A. Energy consumption in tons of standard coal per 100 million yuas of GVIO at 1980 pricu branch 1980 1981 1982 1983 1984 1985 1 164549 156603 150471 147084 142632 132 . 2 99124 99492 96161 93911 93926 92062 3 162468 157580 155472 152577 151079 156652 4 83429 76069 73443 69541 65960 60040 5 148546 136517 127091 118445 110716 98176 6 32538 31160 29269 25823 22886 18894 7 217821 21552S 215280 20959S 193233 180283 8 36959 37293 33711 38041 33546 34263 9 131594 127529 128454 135213 128038 100577 10 32537 31258 29761 30040 28645 28135 11 701 666 712 662 633 717 12 6839 6371 6721 5813 5054 5224 13 173163 144298 161236 168497 165408 154314 14 11482 12205 12169 12370 11549 11476 1S 77476 73383 70686 66480 61444 58634 total 78411 72380 70374 67045 62592 56872 B. Annual percentage rise in GVIO per ton of standard coal consumed brmnch 1981 1982 1983 1984 1985 1 5.07 4.08 2.30 3.12 7.56 2 -0.37 3.46 2.40 -0.02 2.02 3 3.10 1.36 1.90 0.99 -3.56 4 9.68 3.58 5.61 5.43 9.86 5 8.81 7.42 7.30 6.98 12.77 6 4.42 6.46 13.34 12.83 21.13 7 1.07 0.11 2.71 8.47 7.18 8 -0.90 10.63 -11.38 13.40 -2.09 9 3.19 -0.72 -5.00 5.60 27.30 10 4.09 5.03 -0.93 4.87 1.81 11 5.26 -6.46 7.55 4.58 -11.72 12 7.35 -5.21 15.62 15.02 -3.2S 13 20.00 -10.51 -4.31 1.87 7.19 14 -5.92 0.30 -1.62 7.11 0.64 15 5.58 3.82 6.33 8.20 4.79 total 8.33 2.85 4.97 7.11 10.06 Source: Data taken or calculated from Energy 1986, p. 16.  32 How F Hmm ChL.e Iadaiy otrP TAMZ 10: GVIO DATA FOR 15 BRANCHS (100 MLION YUAN, 1980 PrICES) A. Derived from Statistics of Energy Consumption and Efficiency, 1980-1985 brunch 1980 1981 1982 1983 1984 1985 1 473.05 456.70 485.21 523.71 579.39 664.04 2 189.16 194.89 207.05 220.21 235.61 272.75 3 159.79 157.25 166.33 173.34 194.73 208.42 4 290.07 282.11 287.98 310.03 334.14 416.39 5 565.08 591.43 658.98 741.10 830.32 926.70 6 1!21.77 1079.91 1225.19 1440.58 1756.97 2235.10 7 196.80 195.10 222.59 24S.43 287.27 350.62 8 105.52 104.85 112.13 116.19 126.69 133.09 9 112.32 121.78 129.33 134.08 151.91 213.17 10 612.23 690.06 755.69 794.27 865.77 951.84 11 727.53 855.86 870.79 951.66 1090.05 850.77 12 128.67 147.54 141.35 153.11 178.08 199.08 13 51 63 58.56 55.69 57.03 62.21 76.47 14 26.13 69.64 73.96 81.65 92.65 108.05 15 168.18 172.25 188.72 212.70 250.80 309.55 total 4745.51 4946.81 5330.66 5900.37 6721.95 7962.79 B. Published GVIO daa: SOE+ COE Independent Units, 1980 pricas 1 470.42 452.32 481.13 519.19 578.22 657.4 2 186.35 190.35 202.29 211.32 224.72 268.29 3 15S.48 152.04 161.48 173.01 188.83 204.69 4 285.22 278.63 284.43 305.29 334.37 371.82 5 555.34 S81.23 647.04 721.12 807.57 899.91 6 1081.82 1039.19 1178.18 1382.13 1670.42 2114.46 7 181.02 180.18 205.45 227.13 266.29 322.17 8 93.75 94.75 100.47 103.71 112.93 118.36 9 523.82 600.53 6S8.76 690.6' 756.97 854.11 10 707.42 827.08 830.69 922.22 1043.18 1213.85 11 121.31 138.96 134.07 145.89 169.94 186.21 12 50.14 56,60 S3.S8 54.21 59.11 72.06 13 171.82 171.87 120.66 125.65 89.49 105.68 14 9.84 11.13 73.65 79.43 141.1 19S.18 15 153.69 154.89 170.61 187.9 211.27 247.82 total 4752.45 4929.75 5302.49 S848.93 6654.41 7832.01 (continued)  HOw Fuc Mu CAba. 1Ay Oeu t is TABLE 10: aoIO DATA FOR 15 BLgCNz (100 MWoN YUAN, 1930 PRC=) (On*imAsoa) branch 1980 1981 1982 1983 1984 1935 C. GVIO discrepancy: derived - publishd u X of derived figurn 1 0.56 0.96 0.84 0.86 0.20 1.00 2 1.48 2.33 2.30 4.04 4.62 1.64 3 2.69 3.32 2.92 2.99 3.03 1.79 4 1.67 1.23 1.23 1.53 -0.07 10.70 5 1.72 1.72 1.81 2.70 2.74 2.89 6 3.56 3.77 3.84 4.06 4.93 5.40 7 7.55 7.65 7.70 7.45 7.30 8.11 8 11.16 9.63 10.40 10.74 10.86 11.07 9 -370.84 -393.14 -409.15 -41S.11 -398.31 -300.67 10 -IS.SS -19.86 -9.93 -16.11 -20.49 -27.53 11 83.33 83.76 84.60 84.67 84.41 78.11 12 61.03 61.64 62.09 64.59 66.81 63.80 13 -232.80 -193.49 -116.6S -120.31 -43.85 -38.20 14 62.34 84.01 0.42 2.66 -52.30 -80.64 1S 8.62 10.08 9.60 11.66 15.76 19.94 tota -0.15 0.34 0.53 0.87 1.00 1.64 A7. Revised calcula*io: os sad. coal per Yi yua CIVIO,P80 1 16.55 15.81 1S.17 14.84 14.29 13.40 2 10.06 10.19 9.84 9.79 9.85 9.36 3 16.70 16.30 16.01 15.73 1S.58 15.95 4 8.48 7.70 7.44 7.06 6.59 6.72 5 15.12 13.89 12.94 12.17 11.38 10.11 6 3.37 3.24 3.04 2.69 2.41 2.00 7 23.56 2..4 23.32 22.65 20.85 19.62 8 4.16 4.13 3.76 4.26 3.76 3.85 9 2.79 2.59 2.52 2.62 2.57 2.51 10 2.82 2.61 2.71 2.S9 2.38 2.21 11 0.42 0.41 0.46 0.43 0.41 0.33 12 1.75 1.66 1.77 1.64 1.52 1.44 13 S.20 4.92 7.44 7.65 11.50 11.17 14 3.05 7.63 1.22 1.27 0.76 0.64 15 8.48 8.16 7.82 7.53 7.29 7.32 Toal 7.83 7,26 7.07 6.76 6.32 5.78  34 Now Past Mu C%kwe Indi&y Gro? TABLE 11: RriD CALCULATON: ANNUAL EczNT iNcRrAC IN ENEGY PRODUCTVTY BASED ON PCAL ENERGY CONSUmpON AND PUBnL GVIO DATA Cumuluivi Total branch 1981 1982 1983 1984 1985 1980/1985 1 4.65 4.20 2.28 3.81 6.70 23.53 2 -1.23 3.50 0.57 .0.63 5.22 7.50 3 2.44 1.77 1.82 0.95 .32 4.67 4 10.16 3.58 5.29 7.14 -1.97 26.19 5 8.81 7.32 6.33 6.93 12.60 49.S1 6 4.19 6.39 13.08 11.81 20.53 68.93 7 0.96 0.06 2.99 8.65 6.25 20.09 8 0.81 9.68 -11.72 13.25 -2.32 7.98 9 8.08 2.50 .3.89 2.16 2.36 11.34 10 7.97 -3.67 4.64 8.83 7.76 27.63 11 2.49 -11.30 7.09 6.36 23.94 28.34 12 5.67 -6.33 8.00 7.83 5.50 21.60 13 S.83 -33.94 -2.69 -33.48 2.98 -53.40 14 -60.06 524.68 -3.84 67.57 19.36 379.90 1S 3.89 4.37 3.90 3.17 -0.41 15.76 total 7.80 2.66 4.60 6.97 9.35 35.41  How Fast Hai Chinea Indawsy Growu? 35 TABLE 12: ENERGY DATA FOR STATE SECTOR INDEPENDENT UNiTS, 40 BRANCES, 1980 AND 1985 A. Raw Data from Indusuial Cesus Energy Consunption Tone of *td. coal per 10,000 tons std. coal 10,000 yuan of GVIO Branch 1980 1985 1980 1985 1 2283 2543 18.24 16.22 2 1084 1147 8.27 7.15 3 46 62 5.61 5.68 4 106 116 5.3 4.18 5 69 76 6.61 6.11 6 124 132 7.65 8.39 7 0.0001 0.0001 3.2 3.87 a 156 183 4.06 4.98 9 44 65 4.04 3.91 10 740 1120 2.21 2.31 11 311 447 5.64 4.53 12 36 68 0.41 0.43 13 1 5 0.68 0.2 14 952 1234 1.72 1.63 iS 10 13 0.44 0.42 16 42 48 1.87 1.78 17 91 132 4.S 5.46 i8 9 10 1.89 1.5 19 594 746 9.21 8.15 20 27 36 0.79 0.67 21 11 13 0.86 0.58 22 6 9 1.27 0.87 23 7431 9127 38.89 34.44 24 1560 1613 9.43 7.76 25 300 330 23.08 20.37 26 5829 6211 18.08 13.78 27 237 322 3.97 2.68 28 321 481 9.71 5.4 29 160 195 2.19 1.84 30 39 59 1.58 1.37 31 2631 3594 21.44 19.34 32 5687 6397 19.01 16.36 33 620 754 5.2 4.64 34 135 IS4 2.69 2.19 3S 1170 1343 3.17 2.1 3 370 43S A.51 1.66 37 194 224 1.8S 1.23 38 80 99 1.32 0.5 39 38 43 1.28 0.89 40 4 7 1.17 1.31 sum 33S48.00 39593.00 8.91 6.99 Source: Indistia ceamns materias, 3: 346-355 (