By Li Shantong, Hou Yongzhi, Wu Yanrui, Liu Peilin

China has maintained a high-speed economic growth for two decades since 1978. With the start of the new century, the Chinese economy has also entered a new phase of development. Whether or not Chinese economy can maintain its rapid growth during this new period of history has aroused wide concern at home and abroad.

The answer to this question lies in one's understanding of the performance of Chinese economy, especially the understanding of the role of productivity in China's economic growth over the past 20 years. This paper will first present an analytical framework and then examine China's productivity growth in the past and its trend of development in the future. Finally, we will shed some light on the sustainable growth of the Chinese economy.

I. The Analytical Framework: An Extended Solow Approach

1. Documentary origins of the analytical approach used in this paper

The analytical approach adopted in this paper has two documentary origins: the analytical framework for calculating growth proposed by Solow in 1957, and the method for measuring technological efficiency put forward by Farrell et al.

According to Solow's analytical framework, economic growth is decomposed into input of key factors and contribution from improvement of efficiency of all factors, as shown in the following equation:

In this formula, Y represents the growth rate of output, F stands for the growth rate of key factors of production, αi, meanwhile, indicates the proportion of the remunerations of various key factors of production in output. (The sum total of the remunerations of various key factors in production under the condition of full competition and neutrality of scale remuneration is l, precisely. In other words, the remunerations of various key factors of production divide up total output completely.)stands for the growth rate of productivity of key factors of production. The economic meaning commonly attributed to it is technological progress.

As a matter of fact, however, whatmeasures is the result of joint action of multiple factors. According to recent documentary progress,can be further decomposed into two parts: (1). Movement of production function itself, or technological progress in a stricter sense. (2). Effect of changes in output resulting from changes in technological efficiency. To explain the second effect, it is necessary to define the concept of technological efficiency at first.

As is known to all, what production function stands for is the maximum output achieved by an industry under the condition of best software and management expertise. But the actual output level set by many economic entities under the condition of certain input is usually lower than that of their potential output. The difference between actual and potential output levels is technological efficiency. Diagram 1 shows the meaning of technological efficiency under the condition of input of a single key factor. The x axle in the diagram stands for the amount of input of the key factor, the y axle represents output, and TE1 and TE2 show the technological efficiency when the level of input stays at x1 and x2 respectively. Farrell (1957), Aigner, Lovell, Schmidt (1977), and Meeusen and van den Broech (1977) have constructed a method for measuring technological efficiency.

Diagram 1: Meaning of technological efficiency

If the method developed by Farrell et al. is used, thein Solow's framework can be divided into two parts -- technological progress and change of technological efficiency, thus deepening analyses of productivity and economic growth.

So far as China is concerned, economic reform influences productivity growth through two channels, theoretically. One is to improve the utilization efficiency of existing resources, that is, to improve the output level by way of revolution to let it approach the possible frontier of production (to improve technological efficiency), the another channel is to promote innovation and invention, that is, technological progress. It helps a lot to review the changes that have taken place in China''s productivity, especially the changes that have taken place in the two component parts of productivity, when we try to understand the root cause of China''s economic growth over the past two decades and forecast its prospect in the future. To this end, the paper extends the traditional Solow method by integrating it with the method mentioned above to construct a new analytical framework for studying productivity and economic growth.

2. The method adopted in this paper and its characteristics

The basic production function used in this paper is transcendental logarithmic production function. The key factors of production involved include input of capital and labour forces. The following assumptions have been included in the model designing: (1). The remunerations of production scale remain unchanged. (2). The output elasticity of capital and labour changes with time, and the structure of production function changes accordingly. (3). Technological progress moves at different speed during different period of time. (4). Technological efficiency changes with time. The different speed of these changes in different areas have also been taken into consideration.

According to the Solow growth accounting method, productivity growth comes mainly from technological progress. According to the method used in this paper, however, productivity can be decomposed into two parts: efficiency improvement and technological progress. The former is attributable to the move toward the possible frontier of production, while the latter is attributable to the move of the possible frontier of production itself. Even without technological progress, underdeveloped countries and regions can achieve positive productivity growth by narrowing the gap between their output and optimum output (improvement of efficiency).

This method of decomposition distinguishes, from a quantitative angle, the effect produced by economic reforms on the level and growth of long-term economic growth (Lucas, 1988). On the one hand, the level effect of economic reforms leads to growth of actual output (move toward the possible frontier of production), and on the other hand, the growth effect means that economic reform leads to sustainability of economic growth by raising the level of production in the short run and stimulating technological progress. A fundamental difference between these two effects is that level effect may disappear with the elapse of time, while growth effect will stay or even expand.

II. Source of Data and Their Processing

The model specified above is used to analyze the panel data of 27 Chinese provinces and autonomous regions between 1981 and 1997. The data from Tibet are incomplete. There are no data of the time sequence from Hainan because it was established as a province only in 1988. Chongqing was established as a municipality in 1996, so it was included in Sichuan Province for calculation in this paper. Preliminary graphic analysis shows that the data of the time sequence from Qinghai Province are highly questionable. For this reason, these four regions are not included in the data samples.

The GDP data for the provinces, autonomous regions and municipalities are from the State Bureau of Statistics. The 1981-1984 statistics of labour forces are taken from the China Yearbook of Statistics, with the statistics for the period 1981-1984 obtained by the extrapolated method. In order to turn the nominal GDP data into GDP of invariant price, a price reduction index is worked out with nominal GDP and actual growth rate of GDP. The total number of the employed is used in this paper due to lack of data on the "man/hours".

The net capital in stock for each year is estimated with the perpetual inventory method. Annual data on investment between 1952 and 1997 are available for all regions. Apart from data on annual flow of investment, two other parameters are necessary for estimating capital in stock: capital in stock at the initial time and depreciation rate of capital. Based on the growth rate (20.83%) of capital in stock in 1953 and the result of calculation of capital increases in 1953 provided by Li et al. (1995), it is estimated that China's capital in stock stood at RMB 95.2 billion in 1952. Choice of the rate of depreciation has been rather willful in existing documents. In order to make a right choice of the depreciation rate, three possibilities - 4, 7 and 10% -- have been tried. On the basis of tests with an appropriate method of measurement, the 7% rate of depreciation is decided upon in this paper.

The final sample has 459 observations. A summary of the sample is presented in Diagram 2. During the 16 years from 1982 to 1997, China saw its economy grow at an average annual rate of 10.4 per cent. Growth of employment, however, slowed down with the progress of time. Capital, meanwhile, maintained a steady growth of 11.6 per cent during this period of time. It can be seen clearly from Diagram 2 that growth of GDP and that of capital in stock moved close to each other gradually during this period of time. This shows, perhaps, that capital input has played an important role in promoting economic growth in China.

Diagram 2: Growth of GDP, labour force and capital in stock (1982-1997)

III. Estimated Results and Their Interpretation

The results of calculation indicate technological progress since China's initiation of reform and an acceleration of its speed along with the deepening of the refine. See Table 1 for the changes in technological efficiency, rates of technological progress, and productivity of all key factors. The relation between the three columns of figures in Table 1 is that improvement of productivity equals the total sum of changes in technological efficiency and technological progress. The bigger the algebraic values in these three columns, the greater the improvement of technological efficiency, the faster the technological progress, and the quicker the pace of improvement of productivity.

Table 1: Estimated rates of technological progress, efficiency improvement, and productivity growth in China

Year	Change in technological efficiency	Technological progress	Improvement of productivity
1982	0.39	1.23	1.62
1983	0.79	1.24	2.03
1984	2.17	1.25	3.42
1985	1.07	1.25	2.33
1986	-1.13	1.26	0.13
1987	-0.35	1.26	0.91
1988	0.79	1.26	2.06
1989	-1.25	1.28	0.03
1990	-0.40	1.29	0.89
1991	-2.74	1.3	-1.44
1992	1.9	1.3	3.2
1993	1	1.3	2.3
1994	0.42	1.3	1.72
1995	0.64	1.3	1.94
1996	-1.53	1.3	-0.22
1997	0.27	1.31	1.58
Mean rates in selected periods
1982-85	1.11	1.24	2.35
1986-91	-0.84	1.27	0.43
1992-97	0.45	1.3	1.75
1982-97	0.13	1.28	1.41

A review of the trend of technological progress shows that it tends to grow steadily along with the passage of time. During the whole period from 1982 to 1997, the average value of the three indicators is positive, with technological progress playing a dominant role. Based on this study, the conclusion can be made that "technological progress has contributed positively to China's economic growth in tile past two decades" . In addition, it is obvious from Table 1 that the movement of the three indicators can be examined in three separate periods, i.e., the early 1980s (1982-985), the second half of the 1980s (1986-91), and the 1990s (1992-97). For the convenience of presentation, a summary of the mean rates of the three indicators during the three periods is presented in the lower part of Table 1.

It is clear from Table 1 that the Chinese economy performed best during the early 1980s when China began to launch reforms. Both efficiency and technological progress experienced positive growth. In this sense, China's economic reform has indeed brought about significant improvement in efficiency, or evident "level effect", to put it in another way.

Table 1 also shows, however, China suffered a noticeable downturn in its productivity during the latter half of the 1980s, with efficiency declining annually during the whole period from 1986 to 1991, except in the year 1988. This resulted in a negative average of changes in technological efficiency during this period and turned the period into one with the poorest achievements during the whole sample period from 1981 to 1997. Efficiency decline may take place in all economic departments. Jefferson et al. (1992) observed a decline in efficiency in the industrial sector in the second half of the 1980s. Fortunately, efficiency performance recovered in the early 1990s, and all the three indicators showed a tendency of upturn.

Finally, according to Table 1, all factor productivity recorded an average growth rate of 1.41 per cent during 1982-1997. This growth is attributable mainly to technological progress. Thus, China's economic reform has produced both a "level effect" and a "growth effect". This finding may be used to support the argument that China's economic growth is sustainable. Table 1 also shows, however, that the evident downturn in efficiency has something to do with the austerity programmes implemented by the government in 1985, 1989, 1992 and 1995. Credit freeze under the austerity policies is partially responsible for the decline in efficiency during the two to three years. An only exception is the readjustment carried out in 1995, which curbed the downward mm of efficiency in 1996 for the time being. This situation is partially attributable to the soft landing (Oppers, 1997) achieved through a combination of multiple factors (such as good harvests in agriculture, improvement of economic structure, implementation of gradually tightening economic policies, and application of policy measures, such as interest and exchange rates with greater effectiveness). This observation has an important policy implication. It reflects the cost of macro economic management relying on administrative interventions rather than economic policy instruments.

IV. Percentage Share of Contributions of Different Sources to China's Economic Growth

Based on the estimates made above, it is possible to get the percentage shares of contributions of different sources to China''s GDP growth. Specifically speaking, China's GDP growth can be decomposed into four components, that is, contributions from labour input, contributions from capital input, contributions from efficiency changes, and contributions from technological progress. The decomposed results are shown in Table 2. According to this table, a considerable part of China's economic growth between 1982 and 1997 was attributable to capital accumulation, whose contribution stood above 50 per cent on the average. At the same time, contribution from all factor productivity was 13.5 per cent on the average. This contribution came mainly from technological progress. Contribution from changes in technological efficiency to China's economic growth was significant in the early 1980s but only modest in the 1990s. The overall contribution of technological efficiency to economic growth was positive during the sample period. In comparison with other studies (excluding the study by Woo in 1988), this paper has derived a much smaller share for the contribution of all factor production.

Table 2: Percentage share of contributions of different sources to China' s economic growth

Source	Period	Increase in labour input	Increase in capital input	Change in all factor productivity	Change in technological efficiency	Technological progress	Residual
This paper	1982-84	12.2	44.3	19.3	9.1	10.2	24.2
	1986-91	15.2	64.4	5.3	-10.4	15.7	15.2
	1992-97	5.9	62.3	15.2	3.9	11.3	16.6
	1982-97	10.4	57.7	13.5	1.2	12.3	18.4
Other studies	Borensztein and Ostry (1996)	1979-94			41.3
	Hu and Khan (1997)	1979-94	12.8	45.6	41.6
	World Bank (1997)	1978-95	17	37	46
	Maddison (1998)	1978-95			29.8
	Woo(1998)	1979-94	14	52.7	12.9

Two factors have resulted in the difference. The first factor is the choice of a different output elasticity coefficient (share coefficient) of all factor productivity, which leads to a different result. Under similar other conditions, for instance, application of a bigger share coefficient to a fast growing production factor will result in a comparatively smaller share of contribution from all factor productivity. Many studies have been extremely subjective in their choice of share coefficients. The share coefficient used in this paper has been obtained through careful calculation and estimation.

Another factor is the difference of methodologies. Under the traditional method of accounting, all residuals remaining from deduction of contributions from key factor input to output growth are counted as contributions to productivity growth. Just as we have pointed earlier, this paper differs from other studies in two aspects: decomposition of all factor productivity into effect of changes in technological efficiency and effect of technological progress, and the exclusion of residuals from contributions of all factor productivity.

V. Productivity Growth at the Industrial and Regional Levels

Our discussions have focused on the aggregate level. In order to study China' s economic growth in greater depths, we will shift our discussion to the industrial and regional levels in this section, with the aim to analyze the difference of performance of different industries in different regions and find out the factors leading to such difference.

The empirical study has been based on data from China' s third national industrial census. Compared with studies based on data from China''s industrial census in 1995, the specialty of this paper lies in its attention to achievements in productivity at the industrial and regional levels. Four-digit data for different industries are used for observation. Altogether, this paper analyzes 60 industrial sectors classified into three-digit industrial categories and uses more than 5,000 values of four-digit industrial samples for observation. A two-stage approach is applied in this paper. At the first stage, standard frontier production functions are estimated to calculate regional and sectoral technological efficiency. At the second stage, the Tobin model is used to analyze the influences of regional and sectoral factors on technological efficiency.

At the first stage, total output is counted as output, and fixed capital in stock, working capital and labour force are counted as input of production factors. Transcendental logarithmic production functions are used to estimate the random frontier production functions of 60 three-digit sectors. When specific estimates are made, it is assumed that the same three-digit sector in all parts of the country has the same production technology frontier functions. This paper does not apply, however, any assumption of unchanged scale remunerations to production functions.

Apart from obtaining various estimated parameters of production functions, corresponding technological efficiency of each value of observation is also obtained at the first stage. See Table 3 for specific estimates. On the average, the technological efficiency of China's industrial sectors stood below 80 per cent of their potential level in 1995. The transportation equipment manufacturing sector (379) and the sugar-making sector (133) performed the best, and the electronic component manufacturing sector (417) and the daily electronic appliance manufacturing sector (418) faired the poorest. As a whole, industrial sectors in coastal regions performed better than those in the central and western regions of the country. The technological efficiency of the tram manufacturing sector (376), the radar manufacturing sector (413) and the electronic component manufacturing sector (417) differs much in the three regions mentioned above. As expected, provinces along the eastern coast, including Jiangsu, Zhejiang, Guangdong, Fujian and Shandong provinces, performed better than other provinces along the coasts.

Table 3: Comparison of technological efficiency between different industrial sectors

<" Marcro economy

Code	Sector	Observations	Coastal	Central	Eastern	National
131	Grain And Forage Processing	188	0.7690	0.7479	0.7393	0.7542
132	Vegetable Oil Processing	53	0.9443	0.9439	0.9437	0.9440
133	Sugar Refining	48	0.9906	0.9906	0.9905	0.9906
134	Slaughtering, Meat And Egg Products Processing	95	0.9862	0.9859	0.9860	0.9860
135	Aquatic Products Processing	76	0.9789	0.9789	0.9789	0.9789
136	Salt Processing	22	0.6617	0.6861	0.5288	0.6393
139	Other Foodstuffs Processing	27	0.9420	0.9428	0.9411	0.9420
141	Cakes and Candies Manufacturing	167	0.8510	0.8415	0.8363	0.8442
142	Dairy Products Manufacturing	27	0.9604	0.9607	0.9572	0.9595
143	Canned Food Manufacturing	114	0.8964	0.8966	0.8916	0.8955
144	Fermented Products Manufacturing	104	0.6491	0.6025	0.6363	0.6308
145	Condiment Manufacturing