Journal Issue
Share
Article

Why is China Growing so Fast?

Author(s):
Mohsin Khan, and Zuliu Hu
Published Date:
July 1996
Share
  • ShareShare
Show Summary Details

I. Introduction

Capital accumulation played a dominant role in China’s economic growth during the central planning era of 1952-78 (Perkins (1988, 1989); and Chow (1993)), because of the ability of the government to mobilize saving and undertake large-scale investments in physical and human capital to support its policy of rapid industrialization. Growth rates, while impressive, were substantiality below those that have been witnessed in more recent years. Since market-oriented economic reforms were launched in 1978, China has achieved phenomenal rates of growth, with real GNP growing on average by nearly 10 percent a year. What are the principal sources of China’s post-reform economic growth? In particular, what has the role of improvements in productive efficiency that are expected to typically accompany structural reforms been in China’s recent growth performance? These are the questions addressed in this paper.

It is useful to consider some stylized facts to obtain an overview of the issue under consideration. Chart 1 depicts the behavior of the investment-output (I-O) ratio, employment growth, and output growth in the 1952-1994 period. Real national income expanded by an average of 7.2 percent annually in the past four decades. The average 1-0 ratio was above 27 percent throughout the period, and in fact rose to over 40 percent towards the final years of the period. By contrast, employment growth was fairly steady, averaging 2.6 percent, though educational attainment of the labor force, and thus its quality, increased over this period.

Chart 1.China: Factor Supply and Output Growth, 1952-94

Sources: Data provided by the Chinese authorities and authors’ estimates.

This study examines the contributions of capital and labor inputs, and particularly productivity, to economic growth in China, within the standard neoclassical growth framework. To undertake this exercise a comprehensive data set based largely on official Chinese language sources was constructed specially. This data set contains consistent time series on the relevant variables for a longer period than hitherto available. The results of the decomposition of growth highlight the relative importance of each factor in explaining the growth of output and thus provide important insights about the underlying causes of economic growth in China.

The central finding that emerges from this study is that while physical investment continued to play a dominant role in China’s economic growth, its importance diminished over time, particularly during the reform period, 1979-94. Since economic reforms were initiated, productivity growth has become the more significant force in driving the Chinese economy. Chart 2 shows the relative importance of capital, labor and productivity in China’s economic growth before and after the reforms. Capital formation contributed over 65 percent of the output growth in the central planning era between 1952-78, while productivity growth contributed only 18 percent. However, a new pattern emerged following China’s economic reforms. In the 1979-94 period, the relative importance of productivity rose substantially and accounted for about 42 percent of China’s aggregate growth.

Chart 2.China: Sources of Economic Growth, 1953-94

Sources: Authors’ Estimates

This study adds to a growing literature examining China’s recent growth experience, which includes McMillan, et al. (1989), Li, et al. (1993), Woo (1995), Sachs and Woo (1996), and Borensztein and Ostry (1996). Since these studies are all motivated by a set of closely related issues concerning China’s growth, some of the differences between the findings here and results obtained in these other studies will be briefly outlined. Comparisons of the results for China in the paper are also made with available estimates for industrialized countries, some Latin American countries, and four East Asian newly industrialized economies.

The remainder of this paper proceeds as follows: Section II briefly outlines the methodology used in this study. Section III describes how the data set was constructed. Section IV presents the main findings of the analysis. Section V extends the results in a number of directions and compares the findings for China with those obtained in other studies. The final section contains some concluding remarks.

II. Methodology

The methodology utilized in the analysis of this paper is the familiar one based on the notion of an aggregate production function for the economy. As is well-known from a large body of growth accounting literature, this aggregate production function approach is an analytical simplification that makes it possible to summarize detailed information about the complex process of economic growth within a simple unified framework.1

The neoclassical growth framework takes as it starting point an aggregate production function, F, which relates a national output measure to the primary inputs and time:

where Y is output, K and L are capital and labor inputs, and T is time. The specific form for the production function selected here is the translog production function.2

The translog production function represents constant returns to scale technology if and only if the parameters satisfy the following conditions

The shares of primary inputs in output can be defined as:

where Pk, Pl, and Qy are prices of capital, labor and output, respectively. Assuming perfect competition, necessary equilibrium conditions are given by equalities between each value share and the elasticity of output with respect to the corresponding input, as follows:

Where εk and εl, are the elasticities of output with respect to capital and labor, respectively. Moreover, under the assumption of constant returns to scale, the following condition holds:

That is, the value shares and the elasticities sum to unity. Finally, the rate of growth in total factor productivity (TFP), vt, is defined as the growth rate of output with respect to time, holding the capital and labor inputs constant:

A Divisia index for vt, called the translog index of total factor productivity, can be derived by first differencing the logarithm of the production function given in equation (2):

where vt-1, t is the rate of growth of total factor productivity from year t-1 to t, and ∈k and ∈l, are the average elasticities of output with respect to capital input and labor input in year t-1 and in year t:

Later in the paper, the directly observed factor shares vk and vl, will be used as approximate measures of these elasticities to estimate total factor productivity growth, assuming that the restrictions underlying the methodology described above hold. The estimates of productivity growth for China may be biased in either direction if there are deviations from the assumptions imposed by the adopted methodology. However, since this methodology is widely used in the studies of sources of economic growth for OECD countries, for the newly industrialized economies of East Asian (NIEs), and for many developing countries with divergent income levels and economic structures, it is of interest, certainly as a first step, to apply the same analysis to the Chinese economy to obtain what could be viewed as “benchmark” estimates.

III. Data Construction

Decomposing China’s economic growth into its sources requires an enormous amount of detailed data for both prices and quantities of national output and inputs, and major efforts have been made here in data collection, compilation, and reconstruction. All original data were obtained from the State Statistical Bureau and various government ministries. Because of many unusual features of the Chinese statistical system, extra care was taken to evaluate the compatibility and consistency of each series from the Chinese national accounts data with the corresponding series defined by the System of National Accounts (SNA). Some of the major series used in this study are shown in the Appendix Table.3

1. Measuring aggregate output

Chinese central planners emphasized “gross social output” (GSO) as the principal indicator of national output. Contrary to what the name might suggest, in China this indicator is limited to the so-called material production sectors, comprising agriculture, industry, construction, transportation, and commerce.4 Output from all other services, such as banking, education, health care, real estate, social services, and so forth, have been excluded from GSO. Moreover, GSO does not net out the value of intermediate inputs such as energy and material consumption, rendering it difficult to use in international comparisons of output.

Since China opened up to the outside world in the 1970s and embarked on market-oriented reforms, considerable effort has been made to bring the country’s statistical system, particularly its national product and income accounts, more into the line with those typically adopted in developed market economies. The State Statistical Bureau of China has been compiling a new series for gross national product (GNP) since 1978, the beginning year of economic reforms. Unfortunately, however, estimates for the GNP series are not available for the period prior to 1978, making comparisons with the pre-reform period impossible.

An intermediate series called national income (NI), which lies somewhere between gross social output and GNP, is however available beginning in 1952 and ending in 1993.5 The Chinese NI series shares a common feature with GNP in that it is a value-added measure. Like the traditional GSO, but unlike GNP, the NI measure excludes the value of a variety of services. While the NI series is net of depreciation of fixed capital assets and other allowances for capital consumption, it is gross of indirect business taxes and business transfer payments. Therefore, the NI measure as reported by the Chinese authorities could be more appropriately dubbed a variant of net national product (NNP) originating in the five material production sectors of the economy. Since the existence of indirect taxes introduces a discrepancy between market prices and prices received by producers, all indirect taxes net of enterprise subsidies were deducted from the reported NI to derive the value of output at factor costs rather than at market prices. It is this derived series—adjusted NI—that is chosen here as the preferred measure of aggregate output for the Chinese economy over the period 1952-94.

In the rest of this paper, the terms output, value-added, and national income, all essentially refer to the adjusted NI series. The average rates of growth in real output for various subperiods are given in Table 1. It is noteworthy that output growth in the reform period (1979-94) exceeded output growth in the central planning period (1953-78) by 3.5 percentage points. Also note that, for all periods, the average growth rate of output per capita exceeds that of output per employee, owing to the faster expansion in employment relative to population growth, as can be seen from Table 2 below on capital and labor inputs.

Table 1.China: Real Output Growth in Selected Periods(Average annual rate of growth in percent)
1953-941953-781979-94
Output7.25.89.3
Output per capita6.04.58.4
Output per employee5.03.87.0
Sources: State Bureau of Statistics of China, and various Ministries; and authors’ estimates.
Sources: State Bureau of Statistics of China, and various Ministries; and authors’ estimates.
Table 2.China: Capital Formation and Labor Force Growth in Selected Periods during 1953-94(In percent)
1953-941953-781979-94
Real capital stock growth1
Aggregate economy6.86.27.7
Excluding residential capital6.36.06.8
Real capital stock growth2
Official estimates11.411.411.4
Investment-output ratio27.122.334.9
Capital-output ratio2.582.592.56
Employment2.62.52.7
Population1.82.01.4
Sources: State Bureau of Statistics of China, and various Ministries; and author’s estimates.

In 1978 base prices.

Changes in the net value of fixed assets in the state-owned sector.

Sources: State Bureau of Statistics of China, and various Ministries; and author’s estimates.

In 1978 base prices.

Changes in the net value of fixed assets in the state-owned sector.

2. Measuring capital input

The Chinese authorities regularly undertake fixed asset surveys for the state-owned sector, obtaining information on: (i) the gross stock of fixed assets valued at the original acquisition prices of the respective assets; and (ii) the stock of fixed assets valued at current prices in the survey years, net of depreciation. Comparing the net stock value series, as reported by the official asset surveys, and the capital stock estimated using cumulated investment flows and the official depreciation table for the state-owned sector, large discrepancies emerge. One possible explanation is that the state-owned enterprises and other state entities fail to use consistent price deflators for those asset surveys. Another possible reason is that official surveys suffer from serious reporting errors and omissions. In any event, such asset surveys are not conducted for urban collective and rural agricultural sectors, and thus do not cover the economy as a whole. Therefore, for the exercise here the capital stock values were constructed based on gross investment flows data drawn from the Chinese national accounts.

Measuring capital inputs for China involves an important decision concerning the treatment of “circulating funds” in computing the capital stock.6 The literature usually includes accumulation of liquid assets as part of investment (see Chow (1993) and Li, et al. (1993)). The precise definition and coverage of “circulating funds” are difficult to grasp. The official statistics alternately refer to this series as “working capital”, and more recently as “changes in inventory”, adding even more confusion. The term dates back to the central planning era when each state-owned enterprise was allocated a certain amount of “quota circulating funds” for purchasing fuel, raw materials, intermediate inputs, etc., according to the plans drawn up by the line ministries and local governments for output, material, and credit allocations. The main function of “circulating funds” was thus to meet enterprises’ demand for working capital in the production process, and there was little role for “inventories” in a command economy characterized by chronic shortages and unsatisfied investment needs. Economic reforms disrupted the working of the planning system, and the buildup of inventories became significant in China’s economy, with information on inventories lumped together with the conventional series on “circulating funds”. A puzzle that arises is that the ratio of “circulating funds/inventories” to fixed investment is extraordinarily high—typically above 20 percent—for much of the period. While the SNA treats changes in inventories as investment, it is difficult to justify also including the accumulation of “circulating funds”, as they are used in China, as investment. This is just one of the areas where the Chinese statistical system fails to deliver reliable information. Since it is inappropriate to include “circulating funds” in the measurement of capital for China, in this study only fixed investment data is used in estimating the aggregate capital stock for China’s economy.

To estimate the capital stock for the aggregate economy the standard perpetual inventory approach was used. This involved the following steps: first, investment data for the state-owned sector, urban collectives, rural collectives, the individual business sector, and the other private business sector, were collected and then combined into a single consolidated investment series at current prices. Second, an investment price index was constructed to deflate the aggregate investment at current prices to obtain aggregate investment at constant prices with 1978 as the base year. While for much of the central planning period, producer goods prices were fixed by the government, capital goods prices displayed a rising and variable trend since the early 1980s, as the planning authorities gradually loosened their grip on prices, and producer price inflation became significant, particularly for building materials such as steel, cement and glass, and so forth. Specifically, the investment price index was based on the implicit deflator for accumulation estimated by Chow (1993) for the period prior to 1978, the price series for building materials between 1978 to 1990, and the price series for fixed assets thereafter, which should mimic the path of asset inflation reasonably well.7 Third, the initial value was set for the aggregate capital stock in 1952 at 175 billion Chinese yuan (at 1952 prices), the estimate used by Chow (1993).8 Finally, official depreciation rates were used for constructing net investment data. It should be acknowledged that Chinese planners have long used artificially low rates of depreciation, averaging 3.6 percent a year, as compared to 13.3 percent of the unweighted average of the geometric depreciation rates for a broad class of U.S. equipment assets (Hulten and Wykoff (1981)).

Chinese official development policies emphasized new capital investment projects with an objective to expand China’s overall scale and capacity of industrial production, while neglecting the continual updating and replacement of the existing machinery, equipment, and infrastructure. It is commonplace for technologically obsolete machinery and equipment to remain in service in Chinese firms (World Bank (1983)). The official depreciation rates bear little relation with the true vintage price functions that characterize the relative efficiency profile of capital assets in China, but in the absence of independent estimates of depreciation, the official low rate of depreciation had to be used. Low rates of depreciation could lead to overestimation of the capital stock growth, and thus bias upward the contributions of capital inputs to economic growth, and bias downward the role of other factors. This caveat should be kept in mind when analyzing the results.

To summarize, the procedure for estimating the aggregate capital stock is based on the following formula:

where Kt is the stock of capital in period t, It is the total real investment (in constant Chinese yuan) in period t, and ξt is the rate of depreciation. The average growth rates of calculated capital stocks for the aggregate economy and for the nonagricultural sector are shown in Table 2. For comparison purposes, also shown are the average growth rates of the official real capital stock (for the state-owned sector).

3. Measuring labor input

The sources of the Chinese labor statistics are from annual employment surveys and the infrequently-conducted population censuses.9 These surveys and censuses produce fairly accurate estimates of total employment in terms of head counts, cross-classified by gender, age, industry, and forms of economic ownership. However, almost no official attempts have been made to measure the number of actual working hours so far. With the possible exception of the state-owned sector, where all employers tend to follow standard 48 weekly hours as mandated by the Chinese State Council,10 little information is available on the hours worked by the urban self-employed and the vast rural labor force. Thus, the labor input for the Chinese aggregate economy is simply assumed to be equal to the total number of employed persons. The average growth rates of the labor force (employment) for the aggregate economy and for its nonagricultural component are shown in Table 2.

4. Estimating factor shares

For labor compensation, total wage payments in each industry for state-owned sector and for the urban collective sector are summed to obtain the total wage bill for China’s formal employment economy. To this figure is then added total labor insurance and welfare payments to obtain total labor compensation for all formal sector employees, i.e., wage earners. For rural laborers, several surveys on rural household income and expenditure are used to obtain scaled-up total labor income of peasants for the period 1990-94. The average ratio of peasant labor income to labor compensation for formally salaried urban employees is then used to derive labor income of rural laborers for all other years. Considering the relatively rapid growth in rural income in recent years, this approach might lead to overestimation of rural labor income and thus of the aggregate labor income share. On the other hand, in-kind income, such as subsidized housing in urban areas and self-consumed agricultural products in rural areas, might be important enough to offset any upward bias of the labor share estimate.

Recall that under the maintained hypothesis of perfect competition and constant returns to scale (equation (8)), the share of capital income in total value added can be derived simply by taking it to be one minus the share of labor income.

As can be seen from Table 3, the measured labor income share averaged 0.411 in the 1952-94 period for the aggregate economy. If agriculture is excluded, the estimated labor share becomes 0.399. The estimate for the labor share is consistent with the assigned value of 0.40 in Chow (1993). It also appears a reasonable figure in comparison with the few available estimates of labor income shares for developing countries. The study by Li, et al. (1993), however, produced somewhat higher estimates for the labor income share in China, with an average value of 0.484 for the period 1952-90. That study, however, included implicit housing subsidies (state-controlled rent) in labor compensation. Such implicit housing subsidies were not included here for labor compensation on the grounds that the imputed value of housing services is not counted symmetrically as part of the aggregate value-added in Chinese official statistics.

Table 3.China: Estimated Factor Shares in Selected Periods, 1952-941(In percent)
1952-9421952-781979-942
Aggregate economy0.4110.3860.453
Nonagricultural0.3990.4000.397
Li et al (1993)30.4840.4620.536
Sources: State Bureau of Statistics of China, and various Ministries; and authors’ estimates.

Labor income as a share of total value added (NI); period average.

The period for Li, et al. (1993) ends in 1990.

Estimates by Li, et al. (1993) include implicit housing subsidies (state-controlled rent) in labor compensation.

Sources: State Bureau of Statistics of China, and various Ministries; and authors’ estimates.

Labor income as a share of total value added (NI); period average.

The period for Li, et al. (1993) ends in 1990.

Estimates by Li, et al. (1993) include implicit housing subsidies (state-controlled rent) in labor compensation.

IV. Sources of Economic Growth in China

1. Aggregate productivity growth

As shown in Table 1, China’s real national income (the derived net output measure) grew by 7.2 percent per year on average over the four decades from 1953 to 1994. At the same time, there was extraordinary input growth. With investment averaging 27 percent of national income, China’s real capital stock expanded by 6.8 percent per year for the period (Table 2). Surprisingly, the capital-output ratio had remained roughly constant at about 2.6. The absence of pronounced “capital deepening” in the Chinese economy suggests that capital accumulation is only a part of the story behind China’s economic growth.

Output growth in China for 1953-94 and the various sub-periods was decomposed along the lines indicated by equation (9) in Section II. From this decomposition it is possible to obtain the estimates of the rate of growth in total factor productivity, i.e., the growth of output with respect time, holding capital and labor inputs constant. The results are shown in Table 4.

Table 4.China: Aggregate Productivity Growth in Selected Periods, 1953-94(Average annual rate of growth in percent)
1953-941953-781979-94
Output7.25.89.3
Capital stock6.86.27.7
Employment2.62.52.7
Total factor productivity (TFP)2.11.13.9
Sources: State Bureau of Statistics of China, and various Ministries; and authors’ estimates.
Sources: State Bureau of Statistics of China, and various Ministries; and authors’ estimates.

As can be seen from Table 4, for the 1953-94 period as a whole real capital grew at 6.8 percent and labor grew at 2.6 percent. By comparison, total factor productivity (TFP) growth was 2.1 percent. However, productivity performance after economic reforms improved dramatically compared to the pre-reform period. While the average TFP growth rate amounted to about 1.1 percent during 1953-78, it rose sharply to 3.9 percent during 1979-94 when China was making progress in market-oriented economic reforms.11 It is interesting to note that Chow (1993) finds that there was no technological progress prior to 1978 in the Chinese economy. The results presented here indicate that productivity performance differed sharply before and after 1978, although it was positive in the pre-reform era. The opening up to foreign trade and investment, the gradual price decontrol, the rise of rural township and village enterprises, and the expansion of new private businesses, brought competitive forces into the Chinese economy, and helped make the economy more efficient during the reform period.

2. Sources of economic growth

Table 5 shows the relative contributions of capital, labor, and productivity to aggregate output growth in 1953-94 and various sub-periods. The percentage contributions of capital and labor are obtained by weighting the corresponding growth rates by their respective income shares. For the period 1953-94 as a whole, it is evident that capital is the most important source of growth in China. During the central planning period 1953-78, capital input alone accounted for 65 percent of the output growth. The relative importance of capital has diminished over time, however. For the reform period 1979-94, capital contributed less than half of the output growth. For the last 5 years of the sample period, capital accounted for only about 30 percent of the output growth. Labor is the least important factor in terms of its contribution to growth. For the period 1953-94 as a whole, labor accounted for 15 percent of the output growth. Labor was responsible for 17 percent of the growth for the central planning period 1953-78, but its contribution dropped to less than 13 percent during the reform period.

Table 5.China: Sources of Economic Growth, 1953-94(In percent)
Variables1953-941953-781979-94
Output growth7.25.89.3
Capital input growth6.86.27.7
Labor input growth2.62.52.7
Total factor productivity (TFP)2.11.13.9
Contribution of capital155.665.245.6
Contribution of labor114.916.812.8
Contribution of productivity growth229.518.041.6
Sources: State Bureau of Statistics of China, and various Ministries; and authors’ estimates.

Ratio of input growth, weighted by the corresponding factor income share, to output growth.

Ratio of TFP growth to output growth.

Sources: State Bureau of Statistics of China, and various Ministries; and authors’ estimates.

Ratio of input growth, weighted by the corresponding factor income share, to output growth.

Ratio of TFP growth to output growth.

While the Chinese economy experienced low productivity growth in the central planning era, productivity has clearly become a significant source of growth since economic reforms. Productivity growth contributed 42 percent of the output growth in the reform period 1979-94, rivaling the percentage contribution of capital.

China’s rate of output growth increased by more than 3 percentage points in the reform period compared with the central planning period. Although growth rates in both capital and labor inputs rose significantly in 1979-94, it is the productivity differential that appears to explain the bulk of the output growth difference between pre-reform and reform periods.

3. Causes of productivity growth

A variety of market-oriented reforms and increased integration into the global economy are the most likely explanations for the rapid productivity growth observed in China. Specifically, there are several factors that are directly related to productivity growth in the reform period.

First, there has been significant reallocation of labor from traditional agriculture to the industry and service sectors. Agricultural employment as a share of labor force fell from more than 70 percent in 1978 to 54 percent by 1994. Clearly, the process of labor reallocation into higher value-added activities such as manufacturing has been an important source of aggregate productivity growth for the economy as a whole, as noted by Woo (1995), Borensztein and Ostry (1996), and Sachs and Woo (1996), all of whom attribute a large part of aggregate TFP growth to the effects of labor reallocation.

Second, as can be seen from Chart 3, there had been a dramatic rise of the nonstate sector in the Chinese economy over the past one and a half decades. Even if agriculture, which had already been liberalized from the collective commune system by early 1980s, is excluded, the nonstate sector in China’s nonagricultural economy, which was once dominated by the state sector, rose from 25 percent in the late 1970s to 56 percent by 1994. Unlike their counterparts in the state sector, nonstate enterprises faced hard budget constraints, had strong profit incentives, and were relatively free of the interventions from the central and local bureaucracy. The rapid growth in the rural township and village enterprises played an especially important role by drawing tens of millions of workers from traditional agriculture into higher value-added manufacturing industry. The expansion of the nonstate sector relative to the state sector clearly helped raise overall production efficiency in the Chinese economy.

Chart 3.China: FDI, Foreign Trade, and the Expansion of the Nonstate Sector: 1978-94

Sources: Data provided by the Chinese authorities and authors’ estimates.

1/ The share of the nonstate sector output in total nonagricultural output.

2/ The share of the total foreign trade in national income.

3/ The share of FDI in total fixed investment.

Third, China’s open-door policy and “special economic zones” had helped attract massive foreign direct investment (FDI). FDI into China was negligible prior to 1979. By 1994, however, cumulative FDI had reached US$95.6 billion. Annual FDI inflows increased from about 0.5 percent of total fixed investment in 1979 to 18 percent in 1994. There was likely significant spillover in technology and managerial know-how through numerous joint ventures and wholly-owned foreign enterprises that have sprung up along China’s coastal provinces in the last decade.

Finally, China had emerged as an export powerhouse over the last decade. Total exports as a share of national income rose from less than 6 percent in 1978 to over 30 percent in 1994. Manufacturing exports in particular had been expanding in U.S. dollar terms at 19 percent a year in the 1981-94 period. The ratio of trade to national output increased from a little over 10 percent in 1978 to over 60 percent by 1994. The evidence from a large number of cross-country studies seems to suggest that export growth has strong positive correlations with productivity growth in domestic industries.

4. International comparisons

A recent study by Young (1995) documenting the East Asian growth experience has stirred a heated debate. Noting Young’s findings that East Asian economies had rather unspectacular TFP growth rates, Krugman (1994) dismissed the East Asian economic miracle as a myth and went so far as to put Singapore and the former Soviet Union in the same category of economic inefficiency! As it is widely suggested that China is well on its way, following the examples of her East Asian neighbors, to achieving an economic miracle of a far bigger proportion (e.g., Perkins 1989), it is clearly interesting to compare China’s growth experience with those of the four “little dragons” and other countries.

Table 6 tabulates estimates from previous studies for the Group of Seven (G-7) industrialized countries (Dougherty (1991)), several Latin American developing countries (Elias (1990)), and four newly industrialized East Asian economies (Young (1995)). While China’s TFP growth in the pre-reform period was lackluster in comparison with other economies, its TFP rate in the post-reform years was significantly higher than the rest. No other country was even close, and in most the TFP rate was less than half. Currently, Chinese TFP growth appears to be the most rapid of those countries for which estimates are available.12

Table 6.China: Productivity Performance in Selected Countries(Average annual percentage changes)
CountryPeriodTFP Growth
G-7 Industrial Countries1
Canada1960-890.5
France1960-891.5
Germany1960-891.6
Italy1960-892.0
Japan1960-892.0
United Kingdom1960-891.3
United States1960-890.4
Latin America2
Brazil1950-851.6
Chile1940-850.8
Mexico1940-851.2
Venezuela (manufacturing)1950-702.6
East Asia3
Hong Kong1966-912.3
Singapore1966-900.2
South Korea41966-901.7
Taiwan Province of China41966-902.1
China
1953-942.1
1953-781.1
1979-943.9
Sources: Authors’ estimates and references cited in the footnotes.

See Dougherty (1991).

See Elias (1990).

See Young (1995).

Excluding agriculture.

Sources: Authors’ estimates and references cited in the footnotes.

See Dougherty (1991).

See Elias (1990).

See Young (1995).

Excluding agriculture.

V. Further Explorations and Sensitivity Analysis

Though extremely useful, growth accounting is nevertheless a very delicate exercise. The question is whether the results reported in this paper constitute firm evidence that the Chinese economy has since the reforms truly moved into a sustainable long-term path characterized by high productivity growth. To check the robustness of the findings the analysis was redone using alternative data and methodologies suggested by other studies.

1. Impact of political disruptions

Comparisons of China’s efficiency and growth performance between the central planning and economic reform periods may be affected by major political disruptions that occurred before the country’s opening up to the outside world in 1979. During the central planning period 1953-78, China experienced periodic political upheavals, culminating in the Cultural Revolution between 1966 and 1969. Chart 1 shows the large swings in output growth associated with the disastrous Great Leap Forward beginning in 1958, and the subsequent tumultuous years of the Cultural Revolution. As noted by a number of studies, including Chow (1993) and Borensztein and Ostry (1996), it is important to separate out the influence of political disruptions on China’s growth performance. Borensztein and Ostry (1996) argue that, once the political turbulence of the pre-reform China is taken into consideration, the differences in productivity growth between the pre-reform and post-reform periods become less dramatic.

It is well known that in those politically chaotic years, China subordinated the national priorities of economic development to political objectives, and the ideological struggles caused huge disruptions of economic activities. To evaluate whether China’s traumatic political developments are mainly responsible for the low productivity growth rate in the pre-reform period, the 1958-70 years were excluded from the sub-period 1953-78. Table 7 shows that, when these politically disruptive years were excluded, the TFP growth rate was 1.6 percent, only somewhat higher than the TFP growth for the whole period up to 1978. As such, it is possible to conclude, in contrast to Borensztein and Ostry (1996), that political factors do not completely explain the low productive efficiency in China’s central planning period, and the large differential in estimated TFP growth between the central planning and reform periods remains.

Table 7.China: The Impact of Political Disruptions on Efficiency and Growth(Average annual percentage changes)
1953-781953-781
Output5.86.6
Output per capita4.54.7
Output per employee3.84.5
Real capital stock6.26.5
Employment2.52.2
TFP1.11.6
Sources: State Bureau of Statistics of China, various Ministries; and author’s estimates.

Excluding the politically disruptive years from 1958 to 1970.

Sources: State Bureau of Statistics of China, various Ministries; and author’s estimates.

Excluding the politically disruptive years from 1958 to 1970.

2. Efficiency of China’s nonagricultural economy

In the 1953-78 period, Chinese central planners favored the urban industrial sector over agriculture, allocating the predominant share of national savings as capital investment to industry, while forcing rural population to stay in agriculture by strictly controlling migration to the cities. In the reform period 1979-94, even though Chinese industry continued to receive a dominant share of capital input, the government also adopted a more flexible labor market policy that permitted rural surplus labor to move out of agriculture into rural industry or the urban service sector, At the same time, agriculture shrank from 40 percent of the national income in the early 1980s to about 25 percent by 1994.13

With the continuing reduction in the shares of agricultural output and agricultural labor force, and with more capital and labor resources drawn into the nonagricultural economy than ever before, an important question arises as to how such sectoral shifts in output and reallocation of labor affected the efficiency of China’s nonagricultural economy.

As can be seen from Table 4 and Table 8, growth in nonagricultural output outpaced output growth for the economy as a whole, but growth in inputs for the nonagricultural economy was considerably higher than growth in inputs for the aggregate economy. The capital stock expanded 10 percent a year for the nonagricultural economy, compared to 6.8 percent for the aggregate economy, while growth in labor inputs averaged 5 percent in China’s nonagricultural economy, as opposed to 2.6 percent for the economy as a whole. Excluding agriculture, TFP growth turns out to be negative (-0.8 percent) in 1953-78, and 3.1 percent in 1979-94.14 While TFP growth for the nonagricultural economy was a little lower than the 3.9 percent for the overall economy including agriculture in 1979-94 period, the contrast in productivity performances of the nonagricultural economy between the central planning era and the reform period is remarkable.

Table 8.China: Efficiency and Growth in the Nonagricultural Economy(Average annual percentage changes)
1953-941953-7811953-94
Output8.59.310.0
Capital input10.112.19.3
Labor input5.06.55.4
TFP0.9-0.83.1
Sources: State Bureau of Statistics of China and various Ministries; and authors’ estimates.

Excluding the politically disruptive years from 1958 to 1970.

Sources: State Bureau of Statistics of China and various Ministries; and authors’ estimates.

Excluding the politically disruptive years from 1958 to 1970.

3. Can China’s TFP growth be sustained?

Looking at the reform period alone, the question arises as to how sustainable is China’s productivity growth.15 If the dramatic productivity improvement in the 1979-94 period mainly reflects the one-time gains induced by the earlier reforms such as the dismantling of collective agriculture—the people’s communes, and the restoration of family farming in the rural areas—then one might observe a subsequent slowdown in aggregate productivity growth as the potential for efficiency gains was exhausted over time. In the first six years until 1984, Chinese reforms were concentrated in the rural sector and were largely successful. As shown by McMillan, Whalley and Zhu (1989), and by Khan and Khan (1995), productivity growth in Chinese agriculture during this period was impressive, partly due to increases in agricultural prices, and more importantly, due to positive incentives to farmers induced by institutional changes, such as the introduction of the household responsibility system, and the increased labor market flexibility permitting rural surplus labor moving into industrial and urban tertiary sectors. As shown in Table 9, the TFP estimate for the aggregate economy over the initial period of reforms (1979-84) is 2.5 percent, a full percentage point higher than the pre-reform period. Beginning in 1984, China shifted its reform priorities to urban, state-owned industrial sector, which proved far more complex and far more challenging than the earlier rural reforms (Perkins (1988), Bell, Khor and Kochar (1993)). During the ensuing period between 1985 and 1989, the TFP growth for the aggregate economy was 2.7 percent. In the last several years of the sample period (i.e., 1990-94), TFP growth reached an astonishing 5.8 percent, and productivity changes had for the first time overtaken capital as the predominant source of China’s economic growth. Instead of slowing down as one might expect, productivity growth reached stunning new highs as China moved forward on the reform path, albeit at an uneven pace.

Therefore, the evidence from this study points to a somewhat different conclusion from that reached by Sachs and Woo (1996). Even though the efficiency gains brought about by earlier agricultural reforms may have dissipated, the sharp growth in the rural industry, the surge in foreign direct investment, the export boom, the further dismantling of the central planning system and the increasing market-orientation in the state-owned sector, have combined to boost aggregate productivity growth in the 1985-94 period, and even more so during 1990-94.

Table 9.China: Productivity Growth in the Reform period(Average annual percentage changes)
Extension of Reform to
Rural ReformUrban State SectorAccelerated Reform
1979-841985-891990-94
Output8.08.811.5
Capital input7.18.67.5
Labor input3.02.82.1
TFP2.52.75.8
Sources: State Bureau of Statistics of China and various Ministries; and authors’ estimates.
Sources: State Bureau of Statistics of China and various Ministries; and authors’ estimates.

4. Adjusting capital stock growth

During the central planning period, much of the fixed investment was allocated to new machinery and equipment in order to increase the country’s industrial capacity, while infrastructure and residential investment were neglected. In the reform period, outlays in new housing by state and nonstate entities increased sharply, taking a larger share of gross fixed investment. Since the Chinese NI series excludes imputed value of residential housing services, a corresponding adjustment for the capital stock is made by deducting residential investment from gross fixed investment. As can be seen from Table 10, such an adjustment has no impact on the TFP estimates for the central planning period 1953-78, but raises the TFP estimates somewhat for the reform period. The rate of productivity growth for the period 1979-94 was 4.4 percent, for instance, as compared to 3.9 percent for the same period when residential investment is included in the measure of capital input.

Table 10.China: Sensitivity Analysis of TFP Estimates(Average annual percentage changes)
Central Planning Period 1953-78Economic Reform Period 1979-94
a. Alternative TFP estimates
Li, et al. (1993)1-0.82.5
Sachs and Woo (1996)21.7
Borensztein and Ostry (1996)-0.73.8
b. Alternative measures of capital input3
Excluding residential investment1.14.4
Reducing initial capital stock4-0.93.7
c. Assumed values of labor shares5
α = 0.30.73.2
α = 0.51.13.7
α = 0.61.84.6
Sources: Li, et al. (1993), Sachs and Woo (1996), Borensztein and Ostry (1996), and author’s estimates.

The economic reform period covered is 1979-90.

The economic reform period covered is 1978-93. Sachs and Woo (1996) also report TFP estimates of 3.3 percent and 0.8 percent for sub-periods 1978-84 and 1985-93, respectively.

Using authors’ own estimates of output growth, input growth and factor shares.

Setting initial capital stock at 58.9 billion in 1952 yuan, which reduces the initial capital/output ratio from 2.5 to 1 in 1952.

Using authors’ own estimates of output growth and input growth.

Sources: Li, et al. (1993), Sachs and Woo (1996), Borensztein and Ostry (1996), and author’s estimates.

The economic reform period covered is 1979-90.

The economic reform period covered is 1978-93. Sachs and Woo (1996) also report TFP estimates of 3.3 percent and 0.8 percent for sub-periods 1978-84 and 1985-93, respectively.

Using authors’ own estimates of output growth, input growth and factor shares.

Setting initial capital stock at 58.9 billion in 1952 yuan, which reduces the initial capital/output ratio from 2.5 to 1 in 1952.

Using authors’ own estimates of output growth and input growth.

TFP estimates can also be sensitive to the initial values of capital stock. To assess this possibility, the value of initial capital stock was set at 58.9 billion of Chinese yuan (1952 prices),16 which reduces the starting capital/output ratio from 2.5 when Chow’s (1993) estimate was used to 1. The effect of this adjustment is to increase real capital stock growth rates, particularly in the earlier central planning period. As a result, estimated TFP growth turns negative (-0.9 percent) for 1953-78, but it becomes only slightly lower for the reform period 1979-94, (3.7 percent), than when Chow’s (1993) estimate of the initial capital stock was used.

5. Measurement errors in output and input data

One hardly need point out the hazard of indiscriminately using China’s official statistics. The potential measurement errors in the official data warrant considerable caution in interpreting the results presented here and elsewhere. Woo (1995), Sachs and Woo (1996), and Borensztein and Ostry (1996) assert that the Chinese official output data overstated output growth by as much as 0.5 to 1.0 percentage points in the post-reform period. If their output adjustment were correct, it would reduce the TFP estimates in this paper by 0.5 to 1.0 percentage points, holding inputs and factor shares at the same levels.

According to these various recent studies, the errors in the official output data stem mainly from two sources: over-reporting of nominal industrial output and under-deflating of nominal industrial output by Chinese collective enterprises, especially township and village enterprises (TVEs). It is certainly puzzling that China’s nonstate enterprises in the reform period should over-report their nominal output to the authorities. These enterprises have strong profit incentives and would like to avoid taxes if possible. One possibility is that the over-zealous local officials overstated the total output in their jurisdictions to demonstrate their job performances to their superiors. However, such bureaucratic incentives are far weaker in an era of decentralization and increased local autonomy in post-reform China, than presumably they were in the central planning period. In fact, there is plenty of evidence to suggest that since economic reforms China’s informal sector expanded substantially, which had a growth effect as well as a level effect on output. As such, it is doubtful the bias from overstating output could reduce the sharp differences in China’s output growth before and after reforms.

The under-deflating of nominal output could be potentially more important source of bias. Woo (1995) and Sachs and Woo (1996) attributed much of the upward bias in official output growth estimates to the improper deflating of industrial output, especially in the collective sector. However, they failed to make the symmetric adjustment to the official investment series, which could be similarly under-deflated as industrial prices determine to a large degree investment goods prices. Thus the net impact of the Sachs-Woo adjustment on the true TFP estimates is not clear. One should also bear in mind that the official output statistics in the central planning period are not free of the “improper deflating” problem. Selecting an appropriate deflator is a delicate matter for that period because of the severe repressed inflation. The chronic shortages, queues, and excess demand for capital investment indicate that open inflation would be much higher in the absence of price control. Thus, while under deflating the output series poses a serious problem for the estimated values of TFP growth, it does not change the basic conclusion about the substantial improvements in efficiency in the reform period.

Essentially the biggest problem with the official statistics lies with the capital stock data. As noted in the discussion of data construction in Section III, the official asset surveys do not produce stock estimates consistent with the investment flow data from the national accounts. The differences between the official estimate of the capital stock and the estimate obtained by applying the perpetual inventory method to the disaggregated investment flow data, with proper deflating, are substantial. Woo (1995) and Sachs and Woo (1996), and Borensztein and Ostry (1996) all use a series of capital stock estimated earlier by Li et al (1993). As discussed in great details in Hu (1996), this capital stock series, although representing a significant improvement over the official series, seriously overestimates China’s real capital stock growth, in certain time periods.

Table 10 shows how TFP estimates vary if different measures of output, capital input and factor shares are used. Li, et al. (1993) estimate TFP growth rates to be -0.8 percent for 1953-78, and 2.5 percent for 1979-90. Woo (1995) and Sachs and Woo (1996) report TFP estimate for China’s aggregate economy to be 1.7 percent for 1978-93.17 Borensztein and Ostry (1996) estimate TFP growth to be 3.8 percent for the reform period 1979-94. These estimates do not seem substantially different from the TFP estimate of 3.9 percent for 1979-94 reported in this paper, considering their use of alternative output and input data. Li, et al’s (1993) results imply a net increase of 3.3 percentage points in TFP growth, for instance, as compared to 2.8 percentage points between central planning and reform periods estimated here. Using arbitrary values for factor shares, as for example done by Borensztein and Ostry (1996), also does not alter the results significantly. In particular the differential in TFP growth remains large between the pre-reform and reform periods irrespective for the value chosen or the labor share (α). By and large, the available evidence from these studies confirms that productivity improvements is a significant source of economic growth in China during the reform period, and the basic conclusion remains unchanged when alternative measures of capital input and values of labor income share are used.

VI. Conclusions

The central finding of this paper is that while capital accumulation was a dominant factor in China’s economic growth, productivity improvements have assumed an increasingly important role. While the importance of factor inputs in China’s economic growth is well understood, because of China’s high savings rates and abundance of labor, it is nevertheless somewhat of a surprise to find that China also achieved such an impressive record of productivity growth in recent years.

The evidence is clear that China has reaped considerable gains from its market-oriented economic reforms in the past two decades. The total factor productivity rate in the Chinese economy was fairly low for the period 1953-78 because its previous system of central planning stifled economic incentives, prevented competition, and distorted resource allocation. By contrast, China achieved spectacular productivity growth in the reform period, with TFP rate approaching 4 percent a year. Productivity growth contributed more than 40 percent of China’s aggregate economic growth during the reform period, and rapidly replaced capital input as the predominant source of economic growth in China. In the absence of an identified surge in research and development efforts, the observed productivity gains can be mainly attributed to improved allocative efficiency under market-oriented reforms, such as the emergence of family farming and the rise of rural industry, as well as to China’s decade-long “open door” policy that brought about an expansion in foreign trade and foreign direct investment. While at this stage it is not possible to pinpoint the exact sources of productivity growth in China, certainly there are good theoretical and empirical reasons to believe that the reform process that was initiated in 1978 has had, and apparently continues to have, an important effect on the rapid productivity growth observed over the past decade.

APPENDIX
China: Selected Time Series of Output, Factor Supplies and Factor Shares(In 100 millions of yuan, unless otherwise indicated)
NationalNationalNationalGross FixedCapitalLaborTotal
IncomeIncomeIncomeInvestmentStockIncomeEmployment
Year(Current(1978Deflator(Current(1978Share(10,000
Prices)Prices)(1978=1)Prices)Prices)Persons)
19525896640.88725716590.3020729
19537097570.936810617060.3121364
19547488010.934312517710.3021832
19557888520.925212318360.3122328
19568829720.907520219850.3523018
195790810160.893916521040.4123771
1958111812390.902030023520.3726600
1959122213410.911240226720.4326173
1960122013220.923043230640.4725880
19619969291.071616931790.5325590
19629248691.063311032800.5225910
196310009621.039615633640.4926640
1964116611211.040522034860.4427736
1965138713101.058437936070.3928670
1966159615341.040732937880.3629805
1967146714231.031225438980.4030814
1968141513291.064122039700.4231915
1969161715861.019131541600.3833225
1970192619550.984845644930.3334432
1971207720930.992351248560.3435620
1972213621530.992152652340.3735854
1973231823320.994256556220.3536652
1974234823580.995761760530.3637369
1975250325540.980172065890.3538168
1976242724860.976270471130.3838834
1977264426800.986574776430.3739377
1978301030101.000088082390.3740152
1979335032211.040294788500.3841024
1980368834281.0757102694890.4142361
1981394135951.096396199930.4143725
1982425838881.09491230106990.4145295
1983473642771.10731430115250.4046436
1984565248581.16341833126290.4048197
1985702055141.27302543139840.4049873
1986785959391.32333020153210.4351282
1987931365441.42323641168470.4252783
19881173872841.61144497185020.4154334
19891317675491.74544138194230.5455329
19901438479371.81224449204450.5656740
19911655785491.93695509217180.5358360
19922022398642.05027855233110.5159432
199324882113542.191512458255320.5060220
199433312134112.483616370282970.5361470
References

The authors are grateful to Eduardo Borensztein, Tamin Bayoumi, and participants at seminars at Oxford University, Tsinghua University, and the Chinese Academy of Social Sciences for helpful comments.

See Solow (1957), Denison (1962), Jorgenson, et al. (1987), and more recently, Young (1995), among numerous others.

The complete data set is available from the authors upon request.

The agricultural sector broadly covers farming, forestry, fishery and animal husbandry; the transport sector also includes postal and telecommunication services.

The NI figure for 1994 can be estimated from the reported GNP figure by making adjustments for intermediate purchases, depreciation of fixed assets, and the values of all services except transport, and post and telecommunications.

The term “circulating funds” is adopted from the official translation. The original term in Chinese can also be interpreted as “liquid assets”.

Prices of fixed asset investment are determined by the weighted average of prices of machinery and equipment, prices of construction and installation, and other prices.

Chow’s (1993) estimate of initial capital stock value includes the value of land. The estimate implies a capital/output ratio of 2.5 in 1978 prices.

China conducted its first population census in 1954, the second one in 1970, the third one in 1982, and the most recent one in 1990.

In 1995, the official working hours for the state sector were first reduced to 44 hours per week, and finally fixed at 40 hours per week.

For a more detailed description of the Chinese economic reforms, see Bell, Khor, and Kochar (1993), and Perkins (1988), (1989).

It should be noted, of course, that a fairer comparison would involve more up-to-date estimation of TFP for these other countries, and particularly for the East Asian countries.

In the 1950s the Chinese economy was predominantly agricultural, with the share of agriculture output in national income at 58 percent in 1952. Indeed, if the forced concentration of population in the rural areas is not taken into account, one could argue that China’s official development policies favoring industrialization over agriculture through much of the period had been largely successful.

Focusing on the nonagricultural economy is likely to produce more accurate TFP estimates because land is not included as a separate productive factor for the aggregate economy.

The sustainability of productivity growth in China has been recently questioned by Woo (1995) and Borensztein and Ostry (1996).

Recall that Chow (1993) put a value of 175 billion Chinese yuan (at 1952 prices).

Their TFP estimates are 3.3 percent, and 0.8 percent for sub-periods 1978-84 and 1984-93 respectively.

Other Resources Citing This Publication