7. Industrial Reform and Efficiency Change in China: Methods, Data, Literatures and Existing Results

Solow (1957), Denison (1969), Jorgenson and Griliches (1967).

Jun and Xianwang (1996).

To be familiar with the research literatures on East Asian economic growth patterns and their interpretations, see the World Bank’s research reports. For details, see: World Bank (1995). In addition, Felipe (1999) provides a clear and beautiful overview of the research literature and its debates on total factor productivity (TFP) changes in East Asia’s economic growth. See Felipe (1999).

Here, I used part of Shi Shaohua’s earlier work.

The early literatures on production frontier approach mainly include R. Shephard (1953), Solow (1957), M. J. Farrel (1957) and D. J. Aigner & S. Chu (1968). Among them, Farrel’s thesis is recognized as a groundbreaking classic. In empirical research, there are promising frontier approaches, stochastic frontier methods and panel data methods. A review of these literatures can be found in Wu (1996).

This part is a further expansion of our earlier literature review.

World Bank (1982).

World Bank (1985).

Zhizhuang (1984).

Qingqi et al. (1986).

Shizhong and Sangtao (1986).

Chen Kuan et al. (1988).

Jefferson et al. (1992).

Dollar (1990), Groves et al. (1994), Huang and Meng (1995), Wan Guanghua (1995), Guo Kesha (1993).

Woo et al. (1994).

Sachs and Woo (1997).

A few years ago, Jefferson and Rawski mentioned that among the 13 empirical studies of total factor productivity of China’s state-owned enterprises they had observed, 9 indicated that the efficiency of state-owned enterprise has been significantly improved s after the reform, with an annual 2–4% growth rate of TFP on average. Of the remaining 4 studies, the results of 2 were higher than this growth rate, while the results of the other 2 were lower. See Jefferson and Rawski (1994).

Keijiro et al. (2000).

Wu (1995).

Bai et al. (1999).

Chen Shiyi provided me with the data I needed to prepare this section.

The small-case letters here mean growth rates.

Moreover, the selection of sample for fixed assets investment price survey adheres the following principles. First, the principle of choosing the survey site of construction installation projects: ① the sample units should have a certain degree of coverage; ② the investment activities should be excellent typical examples; ③ take into account different economic types; ④ the choice of key projects; ⑤ both the various kinds of economic sectors and different categories of projects should be considered. Second, the principle of choosing other expense survey sites: this principle is basically the same as the principle for construction installation projects, but special attention should be paid to those large investment projects. Since the other costs are not easy to get, key investigations should be made in the construction and construction units at the same time coupled with typical case surveys (to get information from the management department). Third, price survey method. The combination of corporate statements and visits by investigators should be adopted. Fourth, determination of weight. The weight calculation of price index of investment in fixed assets is the average proportion of construction and installation engineering projects, purchase of equipment and other expenses of past three years.

Jefferson and Xu (1991), Cao (1994).

Assuming the production function is \(Y=f (K, L)\), \(Y\), \(K\) and \(L\) are the output, capital and labor respectively. Since the scale returns are assumed to be constant, the two sides of the production function can be multiplied by \(1/Y\) changing it into the form of \(1=f(K/Y, L/Y)\).

Battese and Coeli (1992).

Jefferson and Xu (1994).

Naughton studied the coefficient of variation in the profitability of China’s state-owned enterprises, and also found the trend of average profit margins. For example, his calculations show that during the period of 1985–1989, the coefficient of variation in the profitability of state-owned enterprises in China’s 38 industries dropped from 0.77 to 0.44, which means improvement in allocative efficiency. See Naughton (1995).

Jinghai et al. (2002).

Jefferson (1990).

Zheng et al. (1998).

Lo (1999).

Murakami et al. (1996).

Feng Qu has recently provided a literature review of the theoretical and empirical research on China’s township enterprises. See Feng Qu: A Review of Research on Chinese Rural Industry. See Jun (2002).

Byrd and Lin (1990), Weitzman and Xu (1994).

Out of which, the data of 1979 come from Meng Lian and Wang Xiaolu (2000); the data of 1984–1992 from Jefferson, Rawski and Zheng (1995). In fact, the Chinese Statistical Yearbook released in 2001 presents all of these data since the 1980. See Lian and Xiaolu (2000).

The deflator method is from Jefferson, Rawski and Zheng (1995).

For example, according to calculation, the township industry produced 18,812 yuan of added value in 2000, accounting for 69.27% of the 27,156 yuan worth of the added value of township enterprises nationwide and rose to 47.3% in the proportion of industrial added value of the whole country. In 2000, the current price industrial output value of township enterprises was 82,456 yuan, accounting for 70.99% of the current price total output value of all township enterprises of 116,150 yuan.

This paper uses the fixed capital price index of 1979–1992 estimated by Jefferson and others (1996), the data from 1993 to 2000 is the price index of investment in fixed assets provided in volumes of the Chinese Statistical Yearbook released after 1993. Since the weighted average calculation of construction and installation deflator and equipment purchase deflator is also used, the data is identical with that of Jefferson.

The units of total output value, fixed capital and number of employment in the table are billion yuan, billion yuan and million person respectively, and all data were processed by obtaining the approximate values.

In the appendix, the growth rate of TFP calculated by Solow’s constrained residual value (formular 9) is slightly lower than the unconstrained residual value (formular 8). The gap between the two narrows gradually and is almost equal finally. We have only presented formular (9) here. Even so, the TFP growth rate of township enterprises is still higher than that of the Chinese economy as a whole, the same is true with transcendental function. See: Jun and Shaohua (2003).

But we still use the C-D function and transcendental function in our following regression later on to compare the tested data.

In the growth factor accounting, the figure obtained by trans-log function unconstrained evaluation was used for TFP growth rate of township enterprises in order to make sure that the addition of contributing shares of the growth factor will be 100, though not absolutely equal to 100, of course. Moreover, all the other analyses in this paper is based on the TFP obtained by the constrained estimation of transcendental function, and the analysis of the two is consistent.

The average growth rate is calculated by using compound interest formula.

Jun (2002).

When calculating the TFP index, the first phase of the sample time series is 1.000.