China's industrial performance : a review of recent findings

RESIEARCH PAPIER SERIES ALA6 INDUSTRIAL REFORM AND PRODUCTIVITY IN CHINESE ENTERPRISES ENTERPRISE BEHAVIOR AND ECONoMIC REFORMS: A COMPARATIVE STUDY IN CENTRAL AND EASTERN EUROPE RESEARCH PROJECTS OF THE WORLD BANK CIUA SERIES CH-lPS 25 August 1993 (revised) CHINA'S INDUSTRIAL PERFORMANCE: A REvIEw OF RECENT FINDINGS by Gary Jefferson Department of Economics Brandeis University Inderjit Singh Transition Economics Division Policy Research Department The World Bank TRANSITION ECONOMICS DIVISION POLICY RESEARCH DEPARTMENT THE WORLD BANK 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 Economics Division (PRDTE) 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: for the project in China, The Institute of Economics of the Chinese Academy of Social Sciences (IE of CASS), The Research Center for Rural Development of the State Council (RCRD), and The Economic Systems Reform Institute (ESRI), all in Beijing; and for the projects in Central and Eastern Europe 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 Cat6lica Portuguesa (UCP) in Lisbon; The Czech Management Center (CMC) at Cel6kovice, Czech Republic; The Research Institute of Industrial Economics of the Janus Pannonius University, Peds (RIIE) in Budapest, Hungary; and the Department of Economics at the University of L6di, in Poland; and the National Center for Development Studies, Australian National University, Canberra, Australia. The research projects are supported with funds generously provided by: The World Bank Research Committee; The Japanese Grant Facility; The Portuguese Ministry of Industry and Energy; The Ministry of Research and Space; The Ministry of Industry 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 author(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 affiliated agencies. 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. ***** For additional copies, please send your written request to: Transition Economics Division The World Bank 1818 "H" Street, N.W. Room N 11-029x Washington, D.C. 20433 Attention: Mr. Christopher Rollison or FAX your request to (202) 522-1152 CONTENTS A cknow ledgm ent .....................................................................................................i 1. Introduction ..................................................................................................... 2. P erform ance ..................................................................................................... Studies of Total Factor Productivity..................................................................... 3. Ownership, Structure, and Behavior of Enterprises.................................................... 10 State-O w ned Industry ........................................................................................ 11 4. Persistent Problem s..................................................................................................15 5. C onclusions ................................................................................................... 16 R eferences ................................................................................................... 29 LIST OF TABLES TABLE 1: Estimates of Bias in Measures of Output and Productivity Growth............... 17 TABLE 2: Correlations between Gross and Net Profits................................................. 17 TABLE 3: Losses in State-Owned Industrial Enterprises............................................... 18 TABLE 4: Indicators of Lost Work Time at Industrial Units.......................................... 18 TABLE 5: Product Innovation....................................................................................... 19 TABLE 6: Sources of Productivity Gain........................................................................ 20 TABLE 7: Sources of Growth in Subsectors of Manufacturing...................................... 21 TABLE 8: Level of Total Factory Productivity in State and Collective Industry............. 22 TABLE 9: Adjusted Rates of Capital Productivity Growth............................................ 22 TABLE 10: Nominal Marginal Revenue Products: Labor, Capital, Materials................. 23 TABLE 11: Coefficients of Variation for Factor Returns............................................... 24 TABLE 12: Patterns of Convergence within Subsamples............................................... 25 TABLE 13: New Products as a Share of GVIO............................................................. 25 TABLE 14a: Incentives for New Product Innovation, 1989 Sample Data....................... 26 TABLE 14b: Regressions Wage Bills with Profits and Retained Earnings for State Enterprises.............................................................................................. 26 TABLE 15: Key Measures of Enterprise Conduct: A Panel of 900+ SOEs..................... 27 TABLE 16a: Measures of Enterprise Autonomy............................................................ 28 TABLE 16b: Obstacles to Product Innovation............................................................... 28 1. INTRODUCTION This essay evaluates the overall performance of China's state industry by reviewing the recent literature. It examines changes in state industry from several perspectives. In the 1980s, major changes occurred in the structure of state industry that is, the incentives within state-owned enterprises, the autonomy granted to their managers, and the market and regulatory environment in which they operate. These changes in structure should give rise to changes in the behavior and performance of state industry. We examine various studies that have attempted to measure such changes in state industry. Throughout, we also provide a comparative perspective for our analysis by showing how China's state industry compares with nonstate industry. 2. PERFORMANCE A large body of literature now exists on the change in total factor productivity (TFP) in China's industry. Most pertain to the state sector, but a growing number of studies also investigate the nonstate and TVE sectors. Since TFP analysis entails comparing the growth of inputs relative to output measured in physical or constant price units, the accuracy of TFP studies depends directly on the quality of the relevant deflators for inputs and output. Given the acceleration of inflation in China in the 1980s, TFP analyses cannot be considered rigorous unless they focus on the problems associated with deflating outputs and inputs-namely, capital and, if included in the analysis, intermediate inputs. Several pitfalls are discussed in various contexts in this section: using gross output value (GVIO) deflators to deflate net output, failing to deflate the capital stock, and relying on GVIO deflators to measure the rate of inflation, particularly within certain industrial branches. Most studies of TFP include only capital and labor-that is, productivity gains in the production of net output or value added. In practice, however, just as measures of single factor productivity may give yield impressions of improvements in overall efficiency because capital is substituted for labor or labor for capital, value-added TFP measures may also yield misleading results if intermediate inputs are substituted for value added. Below, we review the results of both net and gross TFP analyses. Studies of Total Factor Productivity State Industry Studies based on Aggregate Data to Derive Multifactor Productivity Growth. Chen et aL (1988). By deflating the capital stock in state industry and eliminating nonindustrial inputs of capital and labor, this study attempts to create a consistent set of inputs and outputs. Using 1980 as the price base for output and investment goods, the authors found that total factor productivity grew at an average annual rate of 5.2% to 5.9% during 1978-85. During the preceding two decades, 1957-77, their reported estimate of TFP growth was much lower, ranging from 0.8% to 1.4%. 1 2 Research Paper Series: China Beck and Bohnet (1988). Using the data provided by Chen et at, the authors adopt a more sophisticated method for estimating TFP change. They estimate a nonparametric frontier production function technique to measure overall technical efficiency and its components. Beck and Bohnet found virtually no gain in productivity during the 1957-71 period, followed by a decline during 1971-77. Consistent with Chen et at, the authors find that TFP in the state sector accelerated to 3.8% during 1978-85. While this estimate is somewhat lower than the estimate of Chen et al., the incremental increase over the TFP performance of the prereform period is comparable. Both studies properly deflate capital and omit nonproduction capital and labor from their estimates. However, they also use the GVIO (gross value of industrial output) deflator (constructed as the ratio of the GVIO) in the current year to GVIO measured in 1980 prices) to deflate the net value of industrial output (NVIO). But using the GVIO deflator to deflate NVIO may introduce bias, because the intermediate input deflator should have risen more rapidly than the GVIO deflator. The reason is that the GVIO deflator is based on ex-factory prices, while the intermediate input deflator is based on purchase prices. As the administered distribution system is replaced by market distribution arrangements, the spread between the ex-factory and the purchase price of intermediate inputs is likely to increase. The introduction of this market mark-up will tend to make intermediate input prices rise more quickly than ex-factory output prices, biasing TFP analyses in two ways: " Since the intermediate input deflator is likely to rise more rapidly than the GVIO deflator during the transition, the GVIO deflator (a weighted average of the net output deflator and the intermediate input deflator) will rise more quickly than the true net output deflator. Using the GVIO deflator the NVIO deflator will thus tend to overdeflate net output growth and create a downward bias in estimates of TFP. " Alternatively, using the GVIO deflator to deflate intermediate inputs will tend to adjust intermediate inputs insufficiently for inflation, since the GVIO deflator is less than the true intermediate input deflator. Again, because a key input is not deflated sufficiently, the TFP estimates will tend to have a downward bias. State Industry Studies of TFP Growth. Jefferson, Rawski, and Zheng (JRZ, 1991). This study compares TFP growth in the state and collective industrial sectors. Unlike previous studies, it includes intermediate inputs in addition to capital and labor. A major advantage of their methodology is that, by -using gross output reported in constant 1980 prices, they need not deflate net output with the gross output deflator. Conversely, computing TFP with intermediate inputs requires constructing an appropriate set of deflators for the intermediate inputs. Thus, a major thrust of the JRZ study, is formulating deflators for capital stock and intermediate inputs. The intermediate input deflators, a key innovation of the study, are constructed with the following calculations: a share of materials obtained through plan procurement and market purchase, an increase in the ex-factory prices for material inputs (the analysis uses input-output tables for 1981 and 1983), and an extra markup paid for inputs purchased through markets. As explained earlier, the intermediate inputs deflator used in the study generally rises more quickly than the GVIO deflator, due largely to a rising markup associated with the expansion of the distribution system for industrial goods. China's Industrial Performance: A Review ofRecent Findin2s 3 The 1980 price-based deflators used in the JRZ study are shown in Table 1. The ratio of the material deflator to the GVIO deflator for 1988 was 1.26 for state industry and 1.47 for collective industry. These figures imply that the appropriate NVIO deflator would be somewhat less that the GVIO deflator. As suggested by the analysis above, estimates of net output TFP growth in studies that use the GVIO deflator to deflate net output are biased downward According to the authors' findings, capital, labor and materials in state industry showed productivity growth rates of 2.14%, 5.21%, and 2.06% respectively during 1980-88. Thus, TFP growth formed by any linear combination of these rates would yield a composite rate of productivity growth somewhere in the range of the highest and lowest of these figures. The authors' report an annual average rate of TFP growth of 2.40% during the 1980-88 period. They estimate that TFP growth averaged just 1.80% during 1980-84 and accelerated to 3.01% during 1984-88. TFP Studies based on Sample Enterprise Data Studies. Virtually all samples of data on China's enterprises suffer in some degree from similar problems--they are neither random nor representative of the population of state-owned enterprises. Specifically, selection procedures for enterprise samples often tend to be biased toward larger and more successful enterprises. It is difficult to say with certainty how this bias affects TFP analyses. The choice of large and successful firms at the beginning of the sample period may bias estimates of TFP growth downward as performance averages out over time. Conversely, at the end of the sample period to select enterprises that have become relatively successful over the period will yield upwardly biased estimates of TFP growth. For this reason, we tend to discount estimates of TFP growth based on samples of enterprise data. These data are valuable for analyzing the comparative performance and behavior of enterprises in cross-sections and over time, but the overall measure of TFP growth generated by them is no substitute for studies based on the population of enterprises. Naughton and McMillan (1992). Using a sample of 769 enterprises in four provinces, the authors find that TFP (capital and labor) rose by 36% during 1980-89, an average annual rate of 3.4%. They obtain their results by estimating a semiparametric production function. Although they make certain adjustments, the manuscript does not describe how the data may have been adjusted to correct for inflation or the inclusion of nonindustrial inputs. The sample of enterprises was selected at the end of the sample period, imparting an upward bias to the estimates. Yet, they use the GVIO deflator, which, as explained earlier, imparts a downward bias to estimates of net output TFP growth. Hay, Liu and Yao (1992). Using a sample of state-owned enterprises, the authors found no evidence of a steady improvement in total factor productivity or a reduction in costs over time, such as might be attributed to technical progress. They found that TFP improved slightly between 1980 and 1985, but then experienced a set back after 1985. This relatively low estimate of TFP growth is likely due to three factors: using GVIO deflators to deflate NVIO (the issue discussed earlier), failing to deflate the capital stock, and drawing estimates which show that the elasticity of capital increased over time, which, since capital productivity grows more slowly than labor productivity, would lead to a decline in TFP growth. Other studies have not been able to reject the constant elasticity of substitution hypothesis. 4 Research Paper Series: China Gordon and Li (1989). Using a sample of 400 state enterprises, the authors found that total factor productivity (capital and labor) rose by 4.6% annually during 1983-87.' Woo et at (1993). This study is the direct refutation of the results of Chen et at (1988) and Jefferson et at (1992) which most often reappear in the literature. Using a sample of 300 large and medium-size enterprises, Woo et al. found no productivity growth during 1984-88. They are able to obtain positive and statistically significant rates of TFP growth in the state sector only by using the deflators for intermediate inputs that JRZ used in their study. Woo et al. claim that JRZ misconstruct the intermediate input deflator. Their deflator allows a markup for producer goods that are purchased through the market, rather than directly from the upstream factory under the plan. The deflator should reflect this market. But Woo et al. contend that JRZ's estimate of the market share in producer good purchases is too large. Using their sample of 300 large and medium-size enterprises, Woo et al. estimate that the share of outside plan sales grew from 12% in 1984 to 30% in 1988. Using a sample of small, medium, and large enterprise, JRZ compute larger shares of outside-of-plan sales. Had Woo et al. taken these market shares into account in constructing their intermediate inputs deflator rather than using the GVIO deflator, which is based on ex-factory prices, they would have obtained higher estimates of TFP growth (although perhaps not as high as the JRZ estimates). TFP Studies based on Industrial Branch Data. Notwithstanding the problems identified with the data and methodology used in the Hay, Liu and Yao (1992) study, interindustry comparisons can be viewed with some confidence if measures of the TFP performance of enterprises in each industrial branch are distorted in a similar direction and magnitude. The authors report these interindustry comparisons. Using intercepts, they found that these extractive industries have the lowest productivity levels; that textiles, energy, chemicals, and medicine have the highest levels; and that building materials, metals, food and drink, and electronic and engineering are in between. The authors also found that location is very important: firms located in the coastal cities exhibit much greater productivity and lower costs. These differences persist when the functions are adjusted for labor quality, capital intensity, and scale. Jefferson (1990) measures TFP growth within the iron and steel industry. Adjusting inputs and outputs for inflation and nonproductive capital and labor, he found that, TFP in China's steel industry grew at an average annual rate of 2.5% during 1980-85. Prior to the reforms, the study estimates that TFP growth was 9.7% during 1952-57, the reconstruction period, and -1.6% during 1957-80. This study demonstrates how sensitive the estimate of capital's output elasticity is to the deflation of investment-goods prices, the exclusion of nonindustrial capital, differences in product mix, and other factors. Li, Gong and Zheng (1992) use Jorgenson's methods to compute rates of TFP growth (capital, labor, and intermediate inputs) for 21 industrial branches during 1981-87. The simple average of the annual productivity growth rates for these 21 branches is 2.33%. High-performing industries include miscellaneous manufacturing (10.9%), motor vehicle (5.9%), and machinery, except electrical (5.1%). The poorest performers are crude petroleum and natural gas (-5.1%), tobacco (-4.4%), and nonmetallic mineral mining (-3.1%) (see Table 2). 1. We have not yet been able to retrieve the manuscript in order to evaluate the methods and data used in Gordon and Li analysis. China's Industrial Performance: A Review ofRecent Findings 5 Zheng (1992) reports rates of productivity growth for the 1980-90 period; the last year of the period was the one in which the greatest slack in the industrial system was associated with low aggregate demand. Also based on three inputs, Zheng's estimate for average annual manufacturing productivity growth during 1980-90 is 1.93%. In descending order, industrial branches with the highest rates of TFP growth are electronics and communication (6.58%), home electronic appliances (5.69%), pharmaceuticals (4.52%), furniture (4.31%), chemical fibers (4.12%), machinery, except for daily use (3.78%), and transportation equipment (3.72%). At the other end of the spectrum, the low TFP growth industries include tobacco (-3.66%), electric power (-2.11%), coal (-2.10 percent), and petroleum products (-1.93) (see Table 3). Although neither study distinguishes between state and nonstate production, several of the high-growth industries, including electronics and communication, pharmaceuticals, chemical fibers, machinery, and transportation equipment, are dominated by state-owned producers. Industries that have unusually low rates of TFP growth are frequently energy-related industries operating under the close supervision of state or local authorities, including plan-price setting for a substantial share of sales. Studies of the Comparative TFP of State and Nonstate Industry. Very few studies have compared the TFP of state and nonstate industry directly. Comparing the productivity performance of the state and nonstate sectors requires distinguishing between rates of productivity growth and levels of productivity. We examine first the literature on rates of productivity growth. Jefferson, Rawski and Zheng (1992). Using the same procedures for constructing deflators for output, capital, and materials in their analysis of state industry, the authors estimate an average 4.63% rate of TFP growth for the collective sector (urban collectives and TVEs established at or above the township level) during the 1980-88 period. Prime (1992) investigates the relative TFP performance (capital and labor) of state and collective industry in a single province-Jiangsu province. She found that an overall industrial TFP growth rate of 4.4% during 1979-88. Distinguishing between state and collective industry, she estimates an average annual rate of TFP growth of 4.1% for state industry and 6.2% for collective industry. Prime adjusts the capital stock for housing, and, by using the investment-goods deflators constructed by JRZ, she deflates the capital stock. Prime actually presents two sets of results-one based on GVIO and the other based on NVIO. Both are deflated with the GVIO deflator. The results differ significantly. In contrast to the calculations based on GVIO in the preceding paragraph, Prime's estimates for TFP growth in state and collective industry based on NVIO are 2.3% and 1.7% respectively. That is, based on net output measures, TFP growth in state industry outpaces TFP growth in the collective sector. While the reversal of the relative performance of state industry and the collective sector has no apparent explanation, the lower estimates for both sectors can be explained by the tendency of GVIO deflators to overdeflate NVIO. Hence, using the GVIO deflator to deflate NVIO, Prime probably generates an underestimate of real net output and TFP growth within both state and collective industry. Xiao (1990) uses sample data on enterprises to estimate rates of growth for each sector. He found that TFP growth among the sample of state-owned enterprises was nearly 4% during 1980-85, compared with 8.8% among the collective sample. 6 Research Paper Series: China Svejnar (1990) uses a sample of 122 TVEs to investigate the extent to which productive efficiency is affected by type of ownership. Within the sample, 64% of the usable observations are TVEs; 17% are village and production teams; 11% are private firms (partnerships and households, and family and individually owned enterprises) and 7% are joint ventures of all types. Whether the data are analyzed for 1970-86, 1981-86, or 1983-86, the time dummies indicate that TVEs have achieved a significant rate of technical progress over time, with the most rapid gains occurring in 1985. Technical change becomes significant only after 1980, proceeding at an average annual rate of 13.3%. Unfortunately, this study appears not to have used deflated measures of output or capital stock input. While it is well established that in the TFP growth in the collective and TVE sectors 1980s, outpaced TFP growth in the state sector, it is not apparent what the relative levels of productivity are in the two sectors. The following studies shed light on this issue. Jefferson and Rawski (1992) use measures of gross output TFP to compare the efficiency of the state and collective sectors during 1980-88. The results are shown in Table 4. Their point estimates indicate that, while productivity levels were comparable in 1980, the average level of TFP in the collective sector exceeded average TFP in the state sector by nearly 16% by the end of the decade. Using a sample of 903 state enterprises and 55 collectives, Xiao (1990) found no statistical difference in productivity levels between the state and collective sectors for 1985. Jefferson (1993) uses a sample of 204 state enterprises and 132 TVEs to compare productivity levels within seven three-digit industries. The pooled data show that, in 1990, productivity among the sample of TVEs was significantly higher than among the sample of state enterprises. At the branch level, however, the data reveal considerable variation in productivity performance. In two industries-coal and home appliances-productivity among the sample of TVEs exceeds productivity among state enterprises. In the other five industries, the data show no significant difference in productivity between the two sectors. Distinguishing the "Pure Ownership" Effect. The aggregate data indicate that, by the end of the 1980s collectives and their constituent TVEs exhibited higher levels of TFP than did their state industry counterparts. This advantage reflects a higher average rate of productivity growth among collectives and TVEs during the 1980s. In light of this aggregate picture, this section examines whether this performance difference is driven by pure ownership effects or whether, independent of ownership, certain policy and institutional differences (for example, the degree of marketization, the type of management system, and so forth) can explain a significant part by the superior performance of the collective/TVE sector. Three studies address this issue. Jefferson (1993) also uses his samples of state enterprises and TVEs to distinguish the effect of various institutional and policy effects from the pure ownership effect. Pooling the data, he found that controlling for scale differences between TVEs and state enterprises tends to strengthen the productivity advantage of TVEs. Controlling for other factors-including the provision of housing and social services, the level of enterprise subordination, the degree of market exposure, and various aspects of the management system-tends to weaken the productivity advantage of TVEs due to the pure ownership effect. The inclusion of all these control variables does not negate the finding that relative efficiency varies considerably across different branches of industry. The conclusion of the study is that, even after policy and institutional factors are controlled for, TVEs China's Industrial Performance: A Review ofRecent Findings 7 have a significant efficiency advantage over state enterprises. At the branch level, however, considerable variation exists. Zou (1992) focuses specifically on the respective contributions of ownership and marketization to explaining differences in the TFP performance of state enterprises and TVEs. Using a sample of 300 enterprises (200 state enterprises and 100 TVEs), Zou has somewhat more complete price data (the share of the market sales of outputs, market purchases of inputs, and spreads among plan-market prices) than available for the Jefferson study. Zou's critical conclusion is that the underlying performance of China's industrial enterprises has been "market-driven rather than ownership-driven" (p. 55). He found that market exposure explains a substantial part of the T`FP growth among enterprises, more than can be explained by the dichotomy between state enterprises and TVEs. Finally, Svejnar (1990) uses his sample of 122 rural enterprises described previously to explain differences in the performance of various types of enterprises. Svejnar's results show that, when differences in inputs and other variables are controlled for, productive efficiency is not related to ownership. In summary, the aggregate data show that in terms of both the levels and rates of TFP growth are greater among TVEs than among their state industry counterparts. The disaggregated data show a more mixed picture. Moreover, there is evidence to suggest that when certain policy and institutional controls are introduced the measured productivity advantage of TVEs may not be as large or consistent as suggested by the raw data. This is an important area that requires more research. Estimating the Industrial Growth Rate. The broad consensus is that the deflators used to deflate state industrial output are biased downward, thus leading to excessively high reports of industrial output growth. The primary source of this bias is new product innovation. When a new product is introduced, enterprises are expected to report industrial output in current prices and in constant 1980 prices but perhaps because there are no comparable products with known 1980 prices, enterprises often use the price set at the time they introduced the product, rather than the 1980 price that the statistical authorities expect the enterprise to use in computing constant-price GVIO. To the extent that this practice occurs, it introduces systematic bias into measures of GVIO in 1980 prices, causing measures of real output (in 1980 prices) to be overstated. This bias is particularly serious in industries that have a higher incidence of new product innovation. Jefferson (1992) attempts a preliminary investigation of the magnitude of the product innovation bias on reported measures of GVIO in constant prices. The study responds to reports of rates of output and single- factor productivity growth that seem implausible. The most extreme case pertains to large and medium-size enterprises within the electronics and communications equipment industry. In this industry, capital and labor productivity reportedly grew at rates of 14.3% and 21.7%, respectively, during 1980-85. Within the machine- building industry, comparable rates were 5.92% and 8.82%. According to Jefferson's computations, both output and single-factor productivity growth should be adjusted downward by 7.6% for the electronics and communications industry, and by 1.74% for the machine-building industry. These adjustments are shown in Table 5. Rawski (1991) discusses the tendency of enterprises, particularly in the collective sector, to overreport their rates of real output growth valued in constant 1980 prices. This tendency occurs either because enterprises 8 Research Paper Series: China that were created after 1980 often do not know about or care about the 1980 prices of their products. Hence, rather than using 1980 prices, they use prices prevailing in the year in which they were established as the price base for computing and reporting gross output in constant prices. The upshot of these tendencies to overstate real growth in constant 1980 prices is that deflators used to measure real growth may be understated, and industry-primarily the newer nonstate sector, but also the state sector-may not be growing as rapidly as reported. For the type of bias identified by Rawski, where the newer firms tend to be downstream (so that measures of output are more biased than measures of the inputs they use), the result would be an upward bias in estimates of TFP growth. Conclusions about the Performance of Enterprises. This overview of analytical studies of productivity growth within China's industrial sector during the first decade of reform (excluding the 1988-91 stabilization period) yields five conclusions that appear to be quite robust, and are confirmed with a broad range of different data sets: a Rates of productivity growth within state industry are in the vicinity of 2% to 5%. Including intermediate inputs, TFP growth may have been somewhat slower due to material deepening. We believe that the best point estimate of TFP growth for gross output during 1980-88 is 2.5%. . TFP growth within the nonstate sector exceeds TFP growth within the state sector. Comparative estimates suggest that rates of growth within the nonstate sector are 1.5 to 2 times higher than within the state sector. * A degree of uncertainty surrounds these estimates, given problems with the accuracy of deflators. It is likely that official estimates of real output growth in the industrial sector overreport actual growth rates. Within the state sector, these biases vary significantly by branch; in the aggregate, does not 1% annually. It may be larger within the nonstate sector. " Considerable differences in productivity growth exist among various industrial branches. Productivity growth among the extractive industries is typically low, while the highest rates tend to be in light industry, particularly in the electronics industry. " It is useful to distinguish the role of pure ownership from the policy and institutional factors that are associated with ownership when assessing differences in productivity performance. Preliminary and scattered evidence suggests that this distinction may be important; some of the stronger productivity growth in the TVE sector may be explained by market exposure and other factors that can be distinguished from ownership. Potential Sources for Evidence of Gains in TFP. This review of various statistical above shows that estimates of TFP are sensitive to the choice of deflators, the coverage of inputs, the nature of samples and other factors. If in fact TFP in state industry rose during the 1980s, we would expect to find changes in the microeconomic conditions that give rise to measured productivity gain. These consist of gains in static efficiency and technological change. Gains in static efficiency stem from the tendency of factor intensities to equalize among firms with homogeneous technologies. Several studies China's Industrial Performance: A Review ofRecent Findings 9 examine the tendency of factor returns to converge within state industry. Static efficiency also stems from reductions in X-inefficiency--that is, in the presence of technical efficiency enterprises operate below the frontier isoquant. Jefferson, Rawski and Xu (1992). Comparing the marginal revenue products of capital, labor, and materials across the state and collective sectors, the study found evidence that factor returns converged modestly between these two sectors (see Table 6). Among a sample of 20 Wuhan enterprises, Jefferson and Xu (1991) found gains in allocative efficiency. Using a more extensive sample, Jefferson and Xu (1992) evaluate gains in allocative efficiency among 226 large and medium-size state enterprises at the core of the state system. During 1980-89, average productivities for labor, capital, and, to a lesser degree, materials showed patterns of convergence. If within a sample of enterprises the technologies are identical, then the convergence of factor returns (marginal revenue products) can be proxied by the convergence of average revenue products. However, in an aggregation of enterprises across different industries with diverse technologies, relative average products may not be suitable proxies for relative marginal products. Consequently, examining patterns within specific industrial branches the authors found evidence of convergence. The pattern of convergence is most notable among enterprises that operate under similar price regimes. Convergence is most rapid and most complete among enterprises that operate fully outside the plan. In addition, both their full sample and 8 of 10 industries show a convergence of TFP (measured in current prices) from 1980 to 1985 and from 1985 to 1989. By controlling for the price regime, the authors show that the firms operating 100 percent within the plan experience the lowest convergence of TFP (25 percent), while enterprises operating only partially within the plan demonstrate greater convergence (42 percent); those operating in the absence of plan sales show the greatest convergence (44 percent). These patterns of convergence are provided in Tables 10.7 and 10.8. Naughton (1992) measures changes in profit rates [(profit+tax)/total capital] across 37 industrial branches. These rates show an impressive tendency to converge; he reports that the coefficient of variation declines from 0.78 in 1980 to 0.44 in 1989. Figure 10.1 demonstrates this rather dramatic convergence of branch profit rates in state industry. Cao (1992) found an improvement in allocative efficiency among his sample of 99 state-owned steel plants. His estimate of allocative efficiency was higher in 1980 (-0.77) than in 1985 (-0.60) suggesting that workers were overpaid and capital costs were undervalued. During 1985-88, allocative efficiency then declines slightly to -0.65, due primarily, Cao concludes, to an increase in allocative inefficiency among small-scale plants. He also estimates a frontier production function which indicates that average technical efficiency rose from 0.46 in 1980 to 0.53 in 1988. At the same time, the spread between the average TFP of the more efficient (large-scale) plants and the less productive (small-scale) plants declined from a factor of 2.2 to 1.9. Among his sample of 903 state owned enterprises, Xiao (1990) found that the TFP gap between his subsets of so-called good firms and bad firms fell during 1980-85; average TFP in the subset of good firms declined from 162% of the bad firms to 110% by 1985. 10 Research Paper Series: China Jefferson, Rawski, and Zheng (1992) estimate 1987 scale parameters of 1.030 for the state sector and 1.020 for the collective sector. These estimates are somewhat smaller than 1984 estimates. Since the average scale of state enterprises grew (in constant prices) from 4.5 million yuan of gross output in 1980 to 7.5 million yuan in 1988, JRZ estimate that increasing scale efficiencies in state industry contributed 0.36% annually to TFP growth during this period. Within their sample of state enterprises, Hay, Lia, and Yao (1992) found elasticities of scale of 1.29 for light industry and 0.93 for heavy industry. Because the latter estimate is not statistically significant, they cannot reject the hypothesis that returns to scale are constant in China's heavy industry. Gains in static efficiency can increase productivity only so much. In the long-run, productivity increases require continuous innovation, in both products and processes. This section investigates evidence about the innovative capacity of state enterprises. An enterprise survey conducted by Jefferson, Rawski, and Zheng (1992) investigated the incidence of innovation and the extent to which resources for innovation are distributed efficiently between state enterprises and TVEs. Their survey consists of a sample of 254 enterprises drawn from the three types of ownership from three industrial branches--cotton textiles, electronics components, and machine building. They found evidence of an increase in both rates of innovation and the commitment of resources for innovation. From 1980 to 1989, the incidence of new products as a share of total production rose in each of the three types of ownership. Among state enterprises, the share rose from 12.7% in 1980 to 21.6% in 1989. The study also investigates the allocative efficiency of state industry and the TVE sector. The data, confirm the impression that state-owned enterprises continue to be lead innovators, while TVEs, although adopting new products vigorously, nonetheless tend to be followers (as shown in Table 9). In -summary, these studies provide surprisingly consistent evidence of advances all areas of productivity gain. These findings provide strong indirect evidence that the overall total factor productivity of state industry has grown. But evidence of growing static efficiency within state industry in the 1980s also suggests that further gains in productivity will require that state enterprises increase innovative capacity in the 1990s. 3. OWNERSHIP, STRUCTURE, AND BEHAVIOR OF ENTERPRISES This section examines studies that link the gains in productivity, now well-established, to changes in structure, especially incentives, autonomy, and the role of markets. Is there evidence that enterprises are seeking profits, exercising greater autonomy with a view toward increasing efficiency and profits, and responding to market pressures? Evidence of clear changes in structure and behavior can indicate further that the performance of state industry has improved. China's Industrial Performance: A Review ofRecent Findings 11 State-Owned Industry Incentives. A managers' questionnaire that accompanies the JRZ (1992) survey of innovation within Chinese industry, asks enterprises to identify the relative importance of various objectives for undertaking innovative activities. On a scale of 0 = "not important," I = "somewhat important," and 2 = "important," managers are asked to rank the following objectives: to increase profits, to fulfill plan requirements, and to increase the incomes of workers. The results, indicate that profit motivation dominates the other two motives by a large margin (Table 10). Hay and Liu (1992) investigate the relationship among wages, bonuses, productivity, and profitability. Analyzing a balanced data set of 208 firms for the 1983-87 period, they conclude that wages tend to be set largely outside the firm according to national and local norms, and allowing only partially for conditions within the firm. In contrast, bonuses are strongly linked to the productivity and profitability of the firm. Bonus incentives have had a dramatic effect on productivity, costs and output growth, particularly when they were introduced in the 1980-85 period . The authors conclude that bonus payment reform probably had the single most important effect on the performance of the enterprises. They do acknowledge a key reservation-that causation could run from efficiency to profits to bonus payments, rather than the other way around. Naughton and McMillan (1992) found that a shift in the decision-making responsibility for output levels from the state to the firm, and an increase on the firm's marginal profit-retention rate, prompt managers of state- owned enterprises to strengthen the discipline imposed on workers. But they also increase the proportion of the workers' income paid in the form of bonuses, and increase the proportion of workers on fixed-term contracts; in principle, they give themselves greater flexibility to manage the workforce. The new incentives have been effective, according to the authors. As the share of bonus payments in total compensation and increases in contract workers as a share of total employment have risen, productivity has also increased. The study uses various econometric methods that seem to demonstrate that causality runs from incentives to performance. Lee (1990) uses a sample of 75 iron and steel enterprises from the 1985-88 period to examine the enterprise contract systenm-the contract management system (chengbao), the manager responsibility system, and the internal contract system, also called the CMI model. Lee found that the isolated adoption of any single reform measure has no significant effect on the output of enterprises while some combinations of reform measures do have a positive effect on output. The effect of the overall reform is small, less than 3% when the effects of initial superiority of a firm are accounted for. To examine the incentive effect of the contract system, we tested whether enterprises whose actual retention rates in the first period exceeded the rates implicit in their contracts also had higher remittance rates subsequently in the second period. In fact, we found that these enterprises were largely able to validate their success in the second period. That is, enterprises that exceeded their expectations in the first period could anticipate that their effective rate of profit retention would rise in the second period. This increasing marginal rate of profit retention provides powerful support to reports that the profit motive is central to the enterprises. It also suggests that managers and workers may maintain a longer profit horizon than is commonly assumed among agents in the state sector. 12 Research Paper Series: China Autonomy. In their survey of innovative capacity, JRZ (1992) ask managers to rank the distribution of decision-making authority between the enterprise and supervisory organ. Along each dimension of authority, the firm is asked to identify the decision-making arrangement as "supervisory organ decided," "joint decision," or "enterprise decided." The results are shown in Table 11. The average score for thereby is 2.62, indicating a relatively high level of autonomy. The degree of autonomy varies considerably across different areas of decision making- from complete autonomy in decisions to sell products outside the plan to shared decisions about importing equipment. With a sample of 20 enterprises from 1988, Jefferson andXu (1991) report three relevant findings on the impact of changes in the policy environment of enterprises. First, managers enhance productivity by using the optimal labor combination program to reallocate workers within the firm; in contrast, the contract labor program does not seem to enhance labor productivity. Second, firms that finance a greater share of investment from retained earnings have higher rates of capital productivity growth. Third, firms that purchase larger shares of their material inputs on the market appear to achieve higher rates of material productivity growth. These findings suggest that enterprises which are granted autonomy generally use it to increase efficiency. Market Structure. Naughton (1992) investigates the hypothesis that the state's monopoly over industrial production has been relaxed substantially; nonstate industry has expanded its share of industrial production from about 20% of total industrial output in 1978 to more that 50% currently. If growing competition has eroded the state's monopoly, we should observe a pattern of changing profitability. Three changes are found in this important study. First, consistent with the competition hypothesis, Naughton notes that the sum of remitted enterprise profits and indirect taxes, representing 24.7% in 1978, fell to just 10.7% of GNP in 1989. Second, profits have tended to equalize across state industrial branches. In 1980, profit rates (profit + tax/total capital) across China's 38 industrial branches varied considerably. In that year, the standard deviation of branch profit rates was 19.7, while the coefficient of variation was 0.78. By 1989, these figures had fallen to 7.4% and 0.44% respectively. Third, Naughton found that the superprofits earned in the nonstate sector after initial entry declined over time, from 40% in 1978 to about 13% in 1990. Naughton argues that the ability of the nonstate sector to bid away superprofits within the state sector has reduced profitability within industry in general, as the state's monopoly has increasingly been eroded. In addition, Chen, Jefferson and Singh (1992) point out that if productivity is growing at a rate of 2.4% in state industry and 4.6% in nonstate industry, then growing losses in the state sector may be explained not only by competition from the nonstate sector, but also by the rapid rates of growth in productivity and reductions in costs within the nonstate sector. Following up on this point, Singh, Ratha and Xiao (1993) found that profits in state industry have been falling most rapidly in those provinces in which the share of nonstate industrial output has been growing most rapidly. In addition, they found that, with a lag, the productivity of state industry has grown most rapidly in areas in which competition from the nonstate sector has grown most rapidly. The fact that growing competition from the nonstate sector explains the paradox of rising productivity and falling profitability in the state sector is a crucial finding. It should be examined not only with regional data, but also with industrial-branch data. It should be noted that both these studies took place before the new regulations on autonomy came into effect. China 's Industrial Performance: A Review ofRecent Findings 13 Wages and Investment. Hay and Liu (1992) investigate the interaction among employment, wage, and bonus behavior. They found that the wage bill is determined largely by rule of thumb, and only partially according to the conditions within the firm. But bonuses are linked strongly to the productivity (and thus the profitability) of the firm. An unpublished analysis by Rawski shows a consistently strong relationship between retained earnings per worker and bonus earnings per worker. Examining data on 18 different two-digit industrial branches during the 1986-89 period, he found t-statistics on the relevant coefficient of 3.0 or larger in 70 of the 72 cases (see Table 12). Morris and Liu (1992) investigate investment behavior within state industry. Their main conclusion is that despite the fact that the environment in which investment decisions are made differs widely from the one in industrialized market economies, a neoclassical approach, allowing for financial constraints, is "not inappropriate" for analyzing and explaining investment spending in China. Jefferson and Xu (1991) examine the significance of the relationship between average profitability during 1984-86 and growth in productive capacity during 1985-87. For both a sample of 20 enterprises in Wuhan and a sample of 110 large- and medium-size iron and steel factories at the core of the state system, they found statistically significant relationships between profitability and subsequent growth in capacity. In order to investigate this key behavior within state industry further, we tested four relationships with the World Bank data set: the relationship between retained and gross profits; the relationship between wages and labor productivity; the relationship between bonuses and labor productivity and between bonuses and bonuses and profitability, and the relationship between profitability and capacity growth (investment). We tested the relationships for three subperiods: 1980-83, 1984-87, and 1988-90 (Table 13). The results indicate that each of these relationships became increasingly stronger throughout the 1980s, or, if already strong in the earlier period, maintained its strength during the decade. In summary, the studies reviewed here strongly indicate that both managers and workers within state industry face powerful incentives to improve the performance of their enterprises. The evidence also indicates that considerably greater autonomy exists now than prior to the reforms, although the degree of autonomy is clearly uneven. Enterprise managers appear to use their autonomy to develop new products and retire old products, reallocate labor, and adjust factor input mix. Finally, the data show strong evidence of profit- maximizing, or at least profit-seeking, behavior among state-owned enterprises. The incentive to increase labor productivity and profits is substantial. China's state enterprises thus show strong evidence of neoclassical profit- maximizing behavior under the reforms. Comparisons with Nonstate Enterprises. As described earlier, the evidence indicates that, despite considerable variation across industries, TVEs often match or outperform their state enterprise counterparts. While some of the superior TFP growth performance may be attributed to greater market exposure and their relatively young age, which enables them to move along the steeper segment of their learning curves, many TVEs do appear to be performing at higher levels of TFP than state enterprises in the same industrial branch. Here, we compare the structural conditions governing TVE operations with those governing state enterprise operations'. Simply put, the question is: "If both state-owned enterprises and TVEs are publicly owned, what might account for differences in their performance?" In their survey of innovative capacity among 250 state enterprises, collectives and TVEs, JRZ (1992) ask TVE managers to rank the relative importance of their objectives in undertaking innovative activities (as 14 Research Paper Series: China reported earlier for state enterprises). Table 14 shows that TVE managers tend to assign the same degree of importance to the profit motive as do state enterprise managers. The only, surprising difference in the overall ranking is that TVEs appear to lend more importance to fulfilling plans. The survey also asks factory representatives to identify the degree of enterpnise autonomy along various dimensions. As shown in Table 15, differences across different ownership types are not significant, certainly less than anticipated. The only major difference between state enterprises and TVEs is that the latter appear to have greater autonomy in choosing their sources for raw material inputs. These two results on incentives and autonomy suggest that state enterprises and TVEs differ not by the intensity of interference by their supervisory authorities, but, rather, by the quality of the intervention. But several authors suggest that TVE managers, workers, and government supervisors may have a more common culture and a consistent set of motivations than the larger, more bureaucratic state enterprises. For example Weitzman and Xu (1992), address paradoxes and dilemmas presented by the successful "Chinese model" for standard theory. The suggestion that China's TVEs-exemplary of the Chinese model-are "vaguely defined cooperatives" that lack a well-defined ownership structure is consistent with traditional property rights theory. To explain the success of this institutional form, the authors demonstrate how the "cooperative culture" embodied by TVEs can resolve various types of free-riding problems internally, without the imposition of explicit legalistic rules of behavior. As Byrd and Gelb (1990) note, the limited mobility of local supplies of labor and capital, combined with a relatively free flow of goods, may make local populations depend particularly on the prosperity of their local community and its TVEs to generate the desired employment, savings and revenue. This immobility of factors of production within an otherwise competitive environment may help explain the phenomenon of the cooperative culture. Svejnar (1990) provides tangible evidence that is consistent with the Weitzman-Xu hypothesis. He found that enterprises which offer their workers a fixed-time wage with a bonus and those that use the internal work- point system tend to be more productive than those operating under piece rates, a combination of a fixed wage and piece rates, or a combination of a fixed wage and a year-end dividend. Other things equal, group incentive schemes seem to be associated with higher productivity than individual incentive schemes. Svejnar concludes that group coordination and team spirit may thus be important to efficiency. In summary, TVEs appear to have caught up to and to be surpassing the efficiency of state enterprises. This is not to suggest that TVEs will in all cases challenge and surpass them. TVE production remains concentrated in light industry. Few have yet demonstrated that they can mobilize the capital and technologies necessary to compete on a scale comparable to the large and medium-size state enterprises found in such heavy industries such as steel, industrial chemicals, and transportation equipment. As suggested from the discussion on new product innovation, it also appears that TVEs tend to be followers rather than leaders in the area of product development. In fact, a kind of symbiotic relationship between state enterprises and TVEs seem to exist in many areas, in which they operate along similar "quality ladders." In this context, state enterprises develop new product designs or improve quality, and TVEs imitate, reducing production costs and expanding competition. In turn, low-cost competition forces TVEs to move up the quality ladder. It is important not to lose sight of the fact that, while patterns of actual ownership vary widely among TVEs, both TVEs and state enterprises are forms of public ownership. China's Industrial Performance: A Review ofRecent Findings 15 4. PERSISTENT PROBLEMS The review of studies herein reports the largely positive results on gains in productivity and output. Nevertheless, the studies also reveal that difficulties persist. Enterprise losses and subsidies. Losses by state-owned enterprises are shown in Table 16. As a percentage of GDP, these losses ranged from 0.5% to 0.6% over the 1986-88 period. The macroeconomic stabilization program of 1988 aggravated the losses, which as a share of GDP peaked in 1990 at 1.98%. Subsequently, as the industrial sector began to recover during 1991, losses recovered 1.73% of GDP. In 1991, more than 30% of industrial state enterprises within the central budget incurred losses, an increase from only 13% in 1987 (Hwa, 1992). Unemployment. China's government reported that it had laid off 1.4 million state industry employees in the first half of 1992. In 1992, in the state-owned coal industry, which employs 3 million persons, 100,000 workers were reportedly laid off, and another 300,000 are scheduled to be laid off by the end of 1995 (New York Times, Dec. 29, 1992). To soften the disruption of layoffs and minimize the political impact, the government will make each laid-off coal worker eligible for a no-interest loan of up to $1,720, to be used to help start a business or find work elsewhere. In the textile industry in Beijing, state-owned enterprises are transferring workers to service firms that are being established as independent accounting units. These transfers are being made to curtail the syndrome in which China's state enterprises manage society (qiye ban shehui), and instead make them tend to their production and profit-making missions. Certain steel mills (most notably Wuhan Iron and Steel) are also implementing similar changes. Liquidations of state-owned enterprises are also becoming increasingly common in China. The China National Coal Corporation reported its intent to close 30 inefficient mines in 1993. In September 1992, the Chongqing city government declared the bankruptcy of a state-owned knitting mill, involving the layoff of 3,000 workers. At that time, it was reported to be the single largest bankruptcy allowed in China. Excess Labor. In their review of the literature on "on-the-job unemployment," Jefferson and Rawski (1992) found that actual hours worked cluster within 20% to 30% of statutory work hours. Estimates drawn from firms in Shanghai, Tianjin, Shenyang and Wuhan show that absenteeism was by far the largest source of lost time in November 1988 (Table 17). In December, the relative importance of lost time due to import shortages rose, given that shortages of energy and materials typically rise during this month of peak activity as enterprises rush to meet their annual plans. Investment Hunger. The investment hunger of China's state-owned enterprises given the availability of cheap capital and its complement, cheap energy, is suggested by interviews with factory managers. The managers show a preoccupation with issues associated with fixed capital and investment, including such predictable problems of obsolete equipment (shebei luohou) and insufficient capital (zi/in buzu). By displacing attention to business strategy, training the workforce, and improving management skills, this preoccupation with fixed assets inhibits promising opportunities for upgrading the performance enterprises. Soft Budget Constraint. A classic criticism of state-owned enterprises is that they tend to manipulate various state organs in order to "soften" their budget constraint in the face of various inefficiencies. In a World Bank survey of state industry, enterprises were asked whether they had adjusted their chengbao contracts during the period in which they were in force (generally 1987-90). Among state enterprise respondents, 26.4% reported having negotiated alterations. For the urban collective sector, the comparable figure was 24.7%; among TVEs it 16 Research Paper Series: China was 21.8%. As the dual pricing system is phased out and prices are set increasingly by markets, the instruments of choice used to soften budget constraints are tax concessions and subsidies and loans. These have a direct effect on the fiscal and financial health of the macroeconomy. 5. CONCLUSIONS This review of quantitative analyses of the performance of China's industrial sector during the reform period of enables us to provide provisional answers to our three main questions fully recognizing that performance must continue to be studied and monitored in greater depth. Have state-owned enterprises substantially improved their performance? Most analyses of total factor productivity growth in state industry show that it has accelerated in the reform period. An examination of specific sources of TFP growth-gains in allocative and technical efficiency and scale economies, and an acceleration of innovative activity--reveal patterns that are consistent with higher rates of TFP growth. Likewise, an examination of specific changes in the structure and behavior of state enterprises reveals important changes that tend to corroborate more rapid TFP growth. Due to various data issues, the extent of improvement remains an open question. During 1980-88, the annual rate of TFP growth appears to have been in the vicinity of 2% to 5% depending on coverage (net or gross output) and on product and factor input deflators. Is this performance sustainable? It is useful to think of China's state-owned enterprises as operating along an efficiency continuum ranging from 0 to 100 (where 100 is the international efficiency frontier defined by "best practice," holding technology fixed but allowing for changes in how institutional arrangements are governed). Prior to reforms, the institutional frontier of state enterprises may have stood at 40 (that is, the maximum attainable efficiency given institutional constraints). The industrial reform program of the 1980s created a set of conditions/opportunities that shifted the institutional efficiency frontier to 65. The accelerated growth of TFP in state industry reflects the movement of enterprises from 40 to 65 on the scale. This is the essence of the gradual reform strategy. The main report shows that the Chinese government and local governments are taking serious initiatives to reform this sector further. These reforms represent a further shift into the institutional efficiency frontier, perhaps to 80 or 85. In this sense, TFP growth should be sustainable under the existing gradual reform scenario, although perhaps at lower rates given the easy gains in static efficiency that existed in the 1980s. Why is the TVE sector outperforming the state.enterprise sector? Here it is important to distinguish between rates of TFP growth and levels of TFP. During the 1980s, productivity within the TVE sector clearly outpaced productivity within the state sector. Differences in levels of TFP were not so dramatic. Some of the uneven advantage of TVEs in their levels of productivity may be attributed to conditions that are unrelated to pure differences in type of ownership, including the fact that TVEs have greater market exposure. Moreover, as provider of various social welfare services, state enterprises are put at a disadvantage. The next several years will be telling. If TVEs are able to extend their productivity lead over state enterprises and continue to close the gap in product mix and quality, even as state industry sheds redundant labor, then there will be strong reason to believe that the performance advantage of TVEs is due to intrinsic differences in ownership. TABLE 1: Estimates of Bias in Measures of Output and Productivity Growth Output Labor prod. Cap. prod. Growth growth growth growth adjustment Al/ 7.70 4.66 -0.73 -1.43 Electronic and communications equipment 24.97 21.73 14.32 -7.77 Machine building 10.74 8.82 5.92 -2.34 Electric machinery 12.59 9.38 5.86 -2.02 Transport machinery 11.23 9.20 8.26 -2.83 /a Estimate based on a simple average of the adjusted growth rates of the 13 industrial branches included in the study. (In addition to those listed above, the remaining nine industries are chemical, nonferrous metals, ferrous metals, coal, power, petroleum refining, food processing, textile and oil and gas extraction.) TABLE 2: Correlations Between Gross and Net Profits 1980 0.88 (646) 1981 0.84 (670) 1982 0.93 (682) 1983 0.90 (812) 1984 0.89 (853) 1985 0.96 (868) 1986 0.93 (831) 1987 0.95 (812) 1988 0.95 (828) 1989 0.94 (751) 1990 0.97 (600) TABLE 3: Losses in State-Owned Industrial Enterprises (Y 100 million) Year 1986 1987 1988 1989 1990 1991 Loss 54.4 61.0 81.9 180.2 348.8 367.0 of which: Coal 21.8 25.2 37.0 46.6 66.4 75.1 Oil - 3.3 13.1 36.0 50.6 57.0 Iron & steel 0.6 0.7 0.6 1.5 12.8 12.1 Chemical (fertilizer) 8.4 2.5 2.4 5.1 12.4 13.1 Textile 1.7 1.6 1.3 4.2 25.0 33.7 Light industry 4.1 5.0 4.6 13.0 41.0 37.9 Tobacco 0.1 0.1 1.3 4.1 6.5 14.8 % of SOE losses 13.1 12.7 15.7 24.0 37.4 39.4 % of GDP 0.6 0.5 0.6 1.1 2.0 1.9 Source: Fax from Hwa to Harrold (9/19/92), pp. 6 and 7. TABLE 4: Indicators of Lost Work Time at Industrial Units (in four cities, Nov./Dec., 1988) ('000s of staff-hours) Work time lost due to lack of Number of firms Actual work time Inputs Demand Workers absent November 1988 662 124,062 2,399 903 8,839 December 1988 126,406 7,171 1,306 7,575 652 Materials, energy, and equipment. TABLE 5: Product Innovation DGV DNPF LAB'S DINT - DEFO DEFK DEFM (1) (2) (3) (4) (5) (6) (7) State 1980 379.14 205.05 31.78 222.84 1.000 1.000 1.000 1981 387.53 211.20 33.27 229.69 1.004 1.059 1.007 1982 414.58 221.73 34.55 246.89 1.005 1.066 1.017 1983 453.56 233.25 35.28 266.25 1.005 1.151 1.030 1984 497.01 242.36 35.73 278.28 1.025 1.233 1.083 1985 565.68 262.89 37.04 307.05 1.090 1.320 1.199 1986 616.69 289.26 38.85 331.21 1.126 1.421 1.281 1987 669.42 311.74 39.82 349.15 1.205 1.508 1.424 1988 748.03 340.95 41.33 372.78 1.340 1.575 1.691 Collective 1980 96.10 23.54 20.80 56.47 1.000 1.000 1.000 1981 105.44 26.56 21.97 62.64 1.000 1.048 1.003 1982 115.67 29.21 22.77 67.98 0.990 1.054 1.005 1983 131.34 31.54 23.48 76.33 0.986 1.126 1.015 1984 168.43 34.22 24.94 91.58 0.993 1.199 1.129 1985 217.53 39.28 26.99 113.76 1.036 1.284 1.236 1986 252.84 47.13 29.00 129.03 1.051 1.374 1.305 1987 299.22 55.24 29.98 146.36 1.090 1.451 1.447 1988 372.72 63.78 30.56 169.36 1.175 1.506 1.724 /a Gross industrial output value: billions of yuan, 1980 prices. /b Net value of productive assets (average of consecutive year-end figures): billions of yuan, 1980 prices. /c Total work force (average of consecutive year-end figures): millions of workers. /d Intermediate inputs in constant 1980 prices. /e Product price deflators (1980= 1.000). If Capital goods price deflator (1980 = 1.000). Lg Price deflator for intermediate inputs (1980 = 1.000). TABLE 6: Sources of Productivity Gain Contributions to growth in output Rate of output Intermediate Rate of pro- Industry growth input Capital input Labor input ductivity growth Rubber and plastic products 0.1705 0.1078 0.0303 0.0102 0.0222 Leather and leather products 0.1428 0.0851 0.0241 0.0091 0.0245 Stone, clay and glass products 0.1357 0.0947 0.0352 0.0270 -0.0212 Primary metals 0.1047 0.0789 0.0215 0.0054 -0.0012 Fabricated metal products 0. 1332 0.0909 0.0137 0.0080 0.0206 Machinery, except electrical 0.2006 0. 1356 0.0048 0.0090 0.0512 Electrical machinery 0.2118 0.1521 0.0251 0.0053 0.0302 Motor vehicle 0.1975 0.1231 0.0167 -0.0015 0.0593 Other transportation equipment 0.1034 0.0557 -0.0064 0.0186 0.0355 Miscellaneous manufacturing 0.1602 0.0552 0.0039 -0.0078 0. 1089 Other manufacturing 0.1529 0.0823 0.0210 0.0047 0.0449 Transportation 0.1146 0.0575 0.0515 0.0176 -0.0120 Communication 0. 1146 0.0426 0.0772 0.0202 -0.0254 Electric utilities 0.0941 0.0788 0.0569 0.0049 -0.0464 Trade 0.1084 0.0600 0.0071 0.0 196 0.02 18 Other services 0.2 105 0.0076 0.0879 0.0041 0.1108 Finance, insurance 0.1282 0.0497 - 0.0041 - Government 0.1800 0.0389 - 0.0334 TABLE 7: Sources of Growth in Subsectors of Manufacturing Average annual rate Average annual Contribution to output growth (output growth = 100) of productivity rate of output growth (%) growth (%) Intermediate input Capital Labor Productivity 1. Food 2.46 7.88 47.12 16.59 4.23 32.06 2. Drink (beverage) -0.89 12.99 59.89 41.12 6.48 -7.48 3. Tobacco -3.66 12.64 31.97 97.99 1.38 -31.34 4. Forage 2.66 42.40 82.18 7.94 3.55 6.33 5. Textile (except cotton textile) 0.46 12.80 67.23 22. 11 6.85 3.80 6. Cotton textile 0.22 6.88 57.19 31.54 7.95 3.32 7. Sewing 1.56 12.17 63.68 17.39 5.41 13.53 8. Leather and fur products 2.67 11.60 54.20 15.43 6.31 24.06 9. Lumber and wood product 1.32 9.14 60.13 13.57 11.30 15.00 10. Furniture 4.31 11.58 49.82 10.88 0.84 38.47 11. Papermaking 1.57 11.16 60.89 19.01 5.41 14.68 12. Painting 2.10 11.22 59.27 16.11 4.12 19.50 13. Culture, education sport products 1.27 13.06 61.58 20.42 7.74 10.26 14. Industrial art products 40 16.08 60.06 22.97 7.67 9.30 15. Supply of electricity, steam, hot water -2.11 7.59 67.45 55.92 5.57 29.14 16. Petroleum products -1.93 6.77 71.74 54.94 3.10 -29.78 17. Coke, coal products -2.10 8.91 100.22 12.83 11.85 -24.89 18. Clemical product (except 19) 0.85 9.90 69.07 18.01 3.98 8.93 19. Chemical products for daily use 1.08 11.77 61.89 26.34 2.13 9.64 20. Medicament 4.52 17.05 51.36 17.51 3.07 28.06 21. Chemical fiber 4.12 22.08 53.64 23.21 2.89 20.26 22. Rubber products 1.19 9.60 56.82 26.14 4.16 12.88 23. Plastic products 2.31 15.28 59.10 20.12 4.74 16.085 24. Construction materials 1.53 11.31 59.58 17.67 8.57 14.18 25. Ferrous metallurgicals -0.64 7.48 82.96 20.46 5.44 -8.86 26. Nonferrous metallurgicals -0. 18 7.63 83.55 14.25 4.59 -2.39 27. Metalproducts 2.82 12.75 61.33 12.06 3.46 23.16 28. Metal products for daily use 2.46 12.16 59.21 16.63 2.96 21.21 29. Machinery (except for daily use) 3.78 11.50 55.76 7.56 1.58 34. 10 30. Machinery for daily use 1.10 9.43 59.83 24.43 3.57 12.17 31. Transportation equipment 3.72 12.61 59.45 7.50 2.29 30.76 32. Electric machinery (except 33) 1.67 11.60 62.97 17.29 4.64 15.09 33. Home electrical appliance 5.36 1 31.24 62.57 15.40 2.82 19.20 34. Electronics and communication 6.58 20.03 451.55 10.93 2.63 34.89 35. Home electronic appliance 5.69 27.61 65.60 9.32 2.40 22.48 36. Instrument 3.92 9.61 41.70 13.68 2.65 41.96 37. Other 3.63 18.28 58.66 14.68 5.40 21.27 A// manufacturing total 1.93 11.11 59.66 17.58 4.61 18.15 TABLE 8: Level of Total Factory Productivity in State and Collective Industry State Industry Collective Industry 1980 2.18 2.28 1981 2.16 2.27 1982 2.16 2.31 1983 2.22 2.38 1984 2.34 2.64 1985 2.42 2.74 1986 2.46 2.72 1987 2.52 2.78 1988 2.63 3.04 Index for 1988 (1980 = 100) 120.6 133.3 Source: Authors' calculations. The data for total factor productivity measure the changing ratio of gross output value per unit of resource input (labor, fixed capital, materials) for independent accounting units at or above the township level. For details of these calculations, see Jefferson, Rawski, and Zheng (forthcoming). TABLE 9: Adjusted Rates of Capital Productivity Growth (percent) (A) (B) (C) (D) Capital prod. Share adjust Innov. adjust Adjusted prod. High Growth Electronics and communications equipment 14.32 0.1 -7.77 7.55 Machine building 5.92 -1.84 -2.34 1.74 Electric machinery 5.86 -1.48 -2.02 2.35 Transport machinery 8.26 -2.42 -2.83 3.01 Medium Growth Chemical -0.89 -0.61 -0.45 -1.95 Nonferrous 1.14 -2.08 -0.22 -1.15 Ferrous 0.11 -1.73 -0.19 -1.80 Low Growth Coal -4.41 -0.38 -0.07 -4.96 Power -3.21 0.04 -0.11 -3.27 Petroleum refining -5.65 0.31 -0.08 -5.42 Food processing -8.78 5.33 -1.48 -4.92 Textile -7.68 2.64 -0.57 -5.60 Oil and gas extraction -10.66 4.43 -0.51 -6.28 /a Also represents downward adjustment to measures of output growth reported in Table 1. TABLE 10: Nominal Marginal Revenue Products: Labor, Capital, Materials MRPL MRPK MRPM Level Index Level Index Level Index State 1980 1,520.75 100.00 .403 100.00 1.220 100.00 1981 1,490.44 98.01 .379 94.04 1.206 98.85 1982 1,536.99 101.07 .384 95.29 1.190 97.54 1983 1,647.53 108.34 .370 91.81 1.192 97.70 1984 1,817.42 119.51 .371 92.06 1.212 99.34 1985 2,093.43 137.66 .382 94.79 1.185 97.13 1986 2,219.35 145.94 .358 88.83 1.143 93.69 1987 2,480.96 163.14 .359 89.08 1.118 91.64 1988 2,974.24 195.58 .391 97.02 1.097 89.92 Growth rate (%) 8.38 -.37 -1.33 MRPL MRPK MRPM Level Index /a Level Index /a Level Index /a Collective 1980 737.29 48.48 .680 168.73 1.164 95.41 1981 766.16 51.40 .632 166.75 1.148 95.19 1982 802.34 52.20 .620 161.46 1.146 96.30 1983 880.30 53.43 .608 164.32 1.143 95.89 1984 1,070.48 58.90 .679 183.02 1.106 91.25 1985 1,205.97 57.61 .692 181.15 1.125 94.94 1986 1,188.05 53.53 .588 164.25 1.135 99.30 1987 1,251.82 50.46 .536 149.30 1.134 101.43 1988 1,648.60 55.43 .601 153.71 1.105 100.73 Growth rate (%) 10.06 -1.55 -.65 /a Relative magnitude (comparable state sector figure = 100.0). TABLE 11: Coefficients of Variation (CV) for Factor Returns L (Large and Medium-Sized SOEs) Group Year OIL O/W O/K G/M NTFP Full sample (226) 1980 0.91 0.86 1.04 0.29 0.32 1985 0.85 0.80 0.82 0.26 0.26 1989 0.76 0.73 0.64 0.25 0.21 Industrial equipment (35) 1980 0.64 0.68 0.80 0.21 0.16 1985 0.96 1.13 0.88 0.19 0.17 1989 1.05 0.93 0.79 0.29 0.17 Consumer durables (15) 1980 0.60 0.56 0.63 0.42 0.19 1985 0.67 0.60 0.62 0.31 0.18 1989 0.81 0.66 0.72 0.27 0.23 Steel (23) 1980 0.81 0.76 1.02 0.16 0.30 1985 0.72 0.72 0.89 0.22 0.19 1989 0.61 0.53 0.54 0.21 0.13 Nonferrous (4) 1980 0.68 0.53 1.23 0.27 0.22 1985 0.49 0.40 1.02 0.21 0.16 1989 0.43 0.39 0.74 0.17 0.09 Textiles (24) 1980 1.04 0.97 0.63 0.19 0. 17 1985 0.62 0.62 0.35 0.17 0.10 1989 0.69 0.68 0.36 0.16 0.12 Apparel (3) 1980 0.19 0.11 0.64 0.03 0.12 1985 0.11 0.14 0.33 0.06 0.06 1989 0.30 0.15 0.15 0.09 0.10 Chemicals (62) 1980 0.73 0.67 0.91 0.15 0.27 1985 0.60 0.51 0.56 0.15 0.22 1989 0.40 0.35 0.45 0.15 0.14 Food products (20) 1980 0.76 0.76 0.91 0.34 0.41 1985 0.84 0.80 1.04 0.31 0.43 1989 0.75 0.76 0.73 0.33 0.35 Building materials (13) 1980 0.44 0.42 0.35 0. 16 0. 17 1985 0.41 0.31 0.50 0.13 0.16 1989 0.45 0.39 0.56 0.18 0.15 Other (27) 1980 1.00 0.98 0.58 0.35 0.29 1985 1.10 0.96 0.62 0.32 0.27 1989 0.97 1.00 0.61 0.19 0.21 /a CV = SD (Zi)/Z, i = O/L, O/W, Q/K, Q/M and NTFP. TABLE 12: Patterns of Convergence within Subsamples (coefficients of variation) Q/W 0/K O/M NTFP A. Sales 100 percent within plan (74): 1980 0.883 0.998 0.314 0.363 1985 0.832 0.904 0.307 0.332 1989 0.749 0.701 0.294 0.271 8. Sales partially within plan (105): 1980 0.800 1.044 0.206 0.265 1985 0.737 0.693 0.201 0.194 1989 0.662 0.574 0.217 0.153 C. No within plan sales (44): 1980 0.834 0.800 0.272 0.221 1985 0.506 0.614 0.206 0.144 1989 - 0.452 0.490 0.220 0.123 TABLE 13 New Products as a Share of GVIO State-owned Total A// LIM Small UCOE TVE 1. New products as a share of GVIO - 1980 12.7% 13.5% 14.7% 6.9% 10.1% 17.3% - 1985 17.9 18.7 18.0 23.1 16.9 14.9 (19.3)- (17.4) (17.2) (18.5) (19.7) (27.7) - 1989 21.6 24.2 22.1 36.2 17.6 17.7 (26.7) (22.0) (24.2) (39.3) (23.4) (37.5) No. of enterprises 76 46 39 7 24 6 (101) (56) (46) (10) (35) (10) 2. Autonorny ---------------------- unchanged ----------------------- 3. Resources (a) Technicians (1989 only) 3.13 - 3.86 5.19 2.36 2.20 No. of enterprises 187 - 68 17 67 34 (b) Tech. innov. expend.! NVIO (avg. 1985-89) 8.22 - 8.66 6.07 7.48 9.72 No. of enterprises 559 - 243 51 188 75 /a The data in parentheses cover 1985 and 1989 only and consist of a balanced set of enterprises. TABLE 14a: Incentives for New Product innovation, 1989 Sample Data All firms grouped by: Equipment Ownership All firms branch SOE UCOE TVE Raise profits 1.79 1.80 1.77 1.84 1.75 Fulfill plan 0.96 1.20 0.87 0.92 1.12 Raise worker incomes 0.59 0.60 0.60 0.57 0.61 Note: Figures are averages based on 0 = "not important," 1 = "somewhat important," and 2 "important" reasons for pursuing innovations. TABLE 14b: Regressions Wage Bills with Profits and Retained Earnings for State Enterprises X = Profit X = Retained earnings Year b R2 b El Sample 1980 .08 .37 2.55 .41 1983 .16 .56 .79 .67 1984 .19 .56 .0 .06 1985 .26 .59 .89 .82 1986 .23 .43 .69 .76 TGS Sample 1986 .18 .60 .00 .00k 1987 .18 .59 .43 .60 1988 .24 .74 .43 .26 1989 .26 .64 .54 .62 /a T-ratio = 4.26; t-ratios for other b coefficients in this panel > 9.0. /b T-ratio = 0.52; t-ratios for other b coefficients in this panel > 15.0. Regression format: Wage Bill = C + b*X, where X = profit or retained earnings (RE); El data from E1308; Wage Bill = D8; profit = D39; RE = Dbl 51; TGS data from TGS309; Wage Bill = V6; profit = V31; RE = V33. TABLE 15: Key Measures of Enterprise Conduct: A Panel of 900 + SOEs 1980-83 1984-87 1988-90 A 1 % increase in gross 0.76/a 0.87 0.92 profit yields an X% (71.84)Lb (110.15) (137.25) increase in retained profit 0.67Lc 0.79 0.76 A 1% increase in labor 0.04 0.13 0.18 productivity yields an X% (8.76) (19.11) (25.69) increase in the average wage 0.03 0.11 0.22 A 1 % increase in labor 0.22 0.37 0.51 productivity yields an X% (17.02) (25.38) (27.75) increase in per capita bonus 0.09 0.18 0.26 A 1% increase in gross profit 0.97 1.01 0.99 yields an X% increase in the (58.02) (68.36) (43.69) per capita bonus 0.56 0.62 0.50 A 1 % increase in the profit 0.01 0.06 0.12 rate yields an X% increase in (1.57) (8.73) (5.87) the rate of capacity growth Ld 0.00 0.10 0.05 /a OLS estimate of X, includes an unreported constant. /b T-statistic. /c R2 . ./d Two-year average rate of capacity growth regressed on average profit rate over two previous years. TABLE 16a: Measures of Enterprise Autonomy SOE COE TVE Begin making new products 2.72 2.76 2.67 Stop making old products 2.72 2.72 2.60 Sell products outside the plan 2.91 2.93 2.96 Determine outside plan price 2.69 2.72 2.74 Sell products outside existing sales area 2.91 2.86 2.99 Export products 2.49 2.60 2.43 Determine total wage bill 2.87 2.91 2.92 Determine total bonuses 2.37 2.33 2.31 Determine total employment 2.46 2.46 2.41 Distribute and appoint workers within the enterprise 2.87 2.91 2.92 Choose enterprises supplying raw materials 2.59 2.67 2.99 Import raw materials 2.37 2.33 2.31 Import equipment 2.04 2.15 2.18 Note: Figures show the arithmetic average for all firms in each category; responses were coded as follows: 1 = supervisory organ decides; 2 = joint decision; 3 = enterprise decides. TABLE 16b: Obstacles to Product Innovation SOE COE TVE- Not enough capital 1.48 1.63 1.61 Old equipment 1.24 1.17 1.25 Not enough technical capability 0.97 0.96 1. 18 Not enough technicians 0.86 1.10 1.06 Raw materials are not suitable 0.57 0.45 0.64 Insufficient information from abroad 0.55 0.44 0.48 Reduction in short-term profits 0.48 0.49 0.48 Can not get necessary approvals 0.22 0.28 0.26 Do not have authority to set prices 0. 14 0.24 0.28 China 's Industrial Performance: A Review ofRecent Findings 29 REFERENCES Bain, Joe S., Industrial Organization. New York: Wiley, 1959. * Beck, Martin and Armin Bohnet, "Productivity Change in Chinese Industry: 1953-85: Some Further Estimations," Justus-Liebig-University Glessen (undated manuscript). Cao, Yong, "Chinese Iron and Steel Industry in Transition: Toward Market Mechanisms and Economic Efficiency" (dissertation) Australia National University. Chen, Kuan, G.H. Jefferson, T.G. Rawski, H.C. Wang, and Y.X. Zheng, "Productivity Change in Chinese Industry," Journal of Comparative Economics 12: pp. 570-91, December 1988. Dollar, David, "Economic Reform and Allocative Efficiency in China's State-Owned Industry," Economic Development and Cultural Change 39: pp. 89-105, October 1990. Gordon, Roger and Wei Li, "The Change in Productivity of Chinese State Enterprises, 1983-87: Initial Results." mimeo. University of Michigan, December 1989. Grossman, Gene M. and Elhanan Helpman, "Quality Ladders and Product Cycles," Quarterly Journal of Economics, vol. 106: pp. 557-86, 1991. Hay, Donald and Liu Shaojia, "Chapter Five: Incentives for Workers" and Shaojie Yao, "Chapter Seven: Productive Efficiency and Costs" Jefferson, Gary H., "China's Iron and Steel Industry: Sources of Enterprise Efficiency and the Impact of Reform," Journal of Development Economics 33: pp. 329-55, 1990. , "Growth and Productivity Change in Chinese Industry: Problems of Measurement" in Asian Economic Regimes: An Adaptive Innovation Paradigm. Research in Asian Economic Studies, Vol. 4B: pp. 427-42, eds. M.J. Dutta (1992). Hwa, Erh-Cheng, "Enterprise Reform in China," A Background Paper prepared for the World Bank, September 21, 1992. Jefferson and Rawski "Constraints on China's Labor Market Reform: Employment Policy and Factor Price Distortions" Jefferson, Gary H., Thomas G. Rawski, and Zheng Yuxin, "Growth, Efficiency, and Convergence in China's State and Collective Industry," Economic Development and Cultural Change, 40:2, pp. 239- 66, January 1992. "Innovation and Reform in Chinese Industry: A Preliminary Analysis of Survey Data," 30 Research Paper Series: China and W.Y. Xu, "The Impact of Reform on Socialist Enterprises in Transition: Structure, Conduct and Performance in Chinese Industry," Journal of Comparative Economics 15: pp. 45-64, March 1991. "Assessing Gains in Efficient Production Among China's Industrial Enterprises," Policy Research Working Paper #877, Socialist Economies Reform Unit, World Bank, March 1992. Jefferson, Gary H., "Are China's Rural Enterprises Outperforming State-Owned Enterprises?," prepared for the Symposium on Economic Transition in China, July 1-3, Haikou, China, 1993. Lee, Keun, "The Chinese Model of the Socialized Enterprise: An Assessment of Its Organization and Performance," Journal of Comparative Economics 14: pp. 384-4000, 1990. Li, Jingwen, Gong Feihong and Zheng Yisheng, "Productivity and China's Economic Growth, 1953-1990," (manuscript) presented at the Conference on Productivity, Efficiency and Reform in China, Chinese University of Hong Kong, August 3-6, 1992. Naughton, Barry, "Implications of the State Monopoly over Industry and its Relaxation," Modern China 18, January 1992. and John McMillan, "Autonomy and Incentives in Chinese State Enterprise," (manuscript) March 1992. Perkins, Dwight H., "Markets vs. Plans: The Key Role of Enterprise Manager Behavior," pp. 160-166 in U.S. Joint Economic Committee, China's Economic Dilemma in the 1990s: Problems of Reforms, Modernization and Interdependence, 1, Washington, D.C. Prime, Penelope, "Industry's Response to Market Liberalization in China: Evidence from Jiangsu Province," 41:1, pp. 27-50, 1992. Rawski, Thomas G., "How Fast Has Chinese Industry Grown?," Research Paper Series #7, World Bank, Country Economics Department, Socialist Economies Reform Unit, Washington, D.C., 1991. Scherer, F.M., Industrial Market Structure and Economic Performance, Chicago, Rand McNally, 1980. Stepanek, James B., "China's Enduring State Factories: Why Ten Years of Reform Has Left China's Big State Factories Unchanged," in China's Economic Dilemmas in the 1990s: The Problem of Reforms, Modernization and Interdependence. Study papers Submitted to the Joint Economic Committee, Congress of the United States, Washington, D.C., U.S. Government Printing Office, 1991. Svejnar, Jan, "Productive Efficiency and Employment," in China's Rural Industry, edited by William A. Byrd and Lin Qingsong, World Bank Research Publication, Oxford University Press, 1990. Walder, Andrew G., "Wage Reform and the Web of Factory Interests," China Quarterly 109: pp. 23-41, March 1987. China's Industrial Performance: A Review of Recent Findings 31 Xiao, Geng, "The Impact of Property Rights Structure on Productivity: Capital Allocation and Labor Income in Chinese State and Collective Enterprises," Research Paper Series, #24, Socialist Economies Reform Unit, World Bank, October 1990. Zou, Gang, "Enterprise Behavior Under the Two-Tier Plan/Market System" (unpublished manuscript) September 1, 1992.