decade of reforms (see chapter 3). The data are taken from Quanguo Gongye Pucha Zilao
(Nation-wide Industrial Investigation Documents) (Nation-Wide Industrial Investigation
Office, State Economic Commission of China 1987), Gangtie Gongye Nianjian (The
Year-book of Chinese Iron and Steel Industry) (The Ministry of Metallurgical Industry of
China 1986 and 1989). The estimation from data for 1980 was used to represent the situation in the early stage of the economic reforms. As few reform measures were implemented by 1980, it could be expected that the behaviour of enterprises and performance were stni mainly influenced by the traditional planned system.
The year 1985 was chosen because, in 1984, China's State Council issued provisional regulations expanding the autonomy of enterprises. The regulations authorized decision making of enterprises on pricing, production levels and sales, investment planning, internal organization, personnel management and the disposition of wages and loans. It could be expected that these regulations had an impact on enterprises' performance one year later, that is, by the end of 1985. The year 1988 was chosen because substantial reform measures have not been implemented since then. The estimations from the data in 1985 and 1988 are compared with those from 1980s. Hence, it is possible to investigate whether there was any systematic change in enterprises' efficiency during the reform period.
Ninety-nine different scale enterprises were chosen as the sample population. Statistics from 1980, 1985 and 1988 for these enterprises are presented in Appendix 7B. The total annual gross output value of the 99 enterprises amounted to 81, 79 and 78 per cent of the total gross output value for the whole industry for 1980, 1985 and 1988 respectively. This sample is used as a proxy to represent the situation in the whole industry.
In the following analysis, the observations are separated by size, based on the magnitude of their gross output value. Enterprises which had annual gross output values above 300 million yuan were taken to be the sub-group of large-scale enterprises. Enterprises which had annual gross output values between 50 to 300 million yuan were
medium-scale enterprises. The remaining enterprises comprised small-scale enterprises. This division is equivalent to the enterprise size division by steel output levels in Table 4.3 (p.68).
The variables used in the estimated production function are gross output, fixed capital, labour force, working capital and enterprise-size dummies. Gross output is the dependent variable and the remaining variables are independent
The independent variables cover almost all production factors directly used in production. Fixed capital stock and labour force are the standard production factors in most production function analyses. Working capital is a proxy for the cost of purchased industrial materials in enterprises' accounting frameworks.
The enterprise-size dummy needs more explanation. In the last chapter, a scale parameter was specified in the final estimation equation of the time series analysis. The scale parameter was expected to capture the effect of scale economies in the whole industry. The negative values of the estimates obtained for the scale parameter suggested that the iron and steel industry suffered further loss of scale economies during the reform period. This finding supported the hypothesis that the relatively higher growth rate of the local medium- and small-scale enterprises which had a cost disadvantage compared with the large-scale enterprises, had led to a further loss of scale economies. As far as cross sectional analysis is concerned, the difference in enterprise size can sometimes have a substantial effect on a enterprise's productivity (van den Broeck 1988). The effect of enterprises' scale is captured directly by the enterprise-size dummies. Specifically, the following estimation model makes medium- scale enterprises the benchmark and contains dummy variables for large- and small-scale enterprises. It is then expected that the cost advantage of the large-scale enterprises due to economies of scale will be captured by a positive coefficient for their size dummy variable, and, conversely, the cost disadvantage
of the small-scale enterprises due to the lack of economies of scale will be captured by a negative coefficient of their size dummy variable. If this is the case, the finding from the time series estimation of a loss of scale economies will be supported by the cross sectional analysis.
It should be noted that the estimate of scale effects from the enterprise-size dummy is somewhat different from the estimate of enterprise-specific scale efficiency defined by equation (7.11). Enterprise-specific scale efficiency as defined by equation (7.11) was derived from comparison of output price and marginal cost. By equating marginal cost to output price, every enterprise has a specific optimal production scale. Hence, a enterprise in an industry might be relatively more efficient in scale than other enterprises but might still not be able to achieve its full scale economy potential. The enterprise-size dummy captures enterprises' relative efficiency in scale but the estimate of scale efficiency from equation (7.11) indicates whether a enterprise has achieved its optimal scale.
In the derivation of allocative efficiency, the cost of working capital and fixed capital are compared with the cost of labour. As stated in chapter 5, since reforms in industrial finance in 1984, most working capital has been in the form of bank loans, which are subject to repayment and carry a cost in the form of interest. Thus, there is an opportunity cost to enterprises for use of their working capital. As enterprises retained part of their profit under the contract system, and continuously added new investment to their fixed capital stock from retained profits, the fixed capital stock might gradually have come more or less under the control of enterprises. This is because enterprises' total capital formation would have been adjusted towards the realization of profit maximization. Fixed capital stock might have become more or less sensitive to enterprises' output as an increasing part of fixed capital stock came to have an opportunity cost for enterprises. The estimates of allocative efficiency allow empirical investigation of the above situation.
The estimates for allocative and scale efficiencies are sensitive to the input and output prices used in the estimation. Considering that factor and goods markets were seriously distorted in China, actual input and output prices and hypothesized market clearing prices were used in the estimation of allocative and scale efficiencies. Comparative analysis can therefore be conducted using the different price-based estimates. The hypothesized market clearing prices for the average allocative efficiency of all sample enterprises were estimated by an iteration process: different prices were tried and the set finally chosen was that under which the average allocative efficiency of the 99 enterprises became zero, that is, on average allocatively efficient. As small-scale enterprises are hypothesized to be the most scale inefficient in the iron and steel industry, the hypothetical market clearing price for scale efficiency is simulated for small-scale enterprises. Under the simulated output prices small-scale enterprises on average, are scale efficient, that is, their marginal cost equals the hypothetical prices. Although this method is more or less an arbitrary means of identifying a correct set of the potential market clearing prices, it can, nevertheless, correctly indicate whether the actual input and output prices were lower or higher than the potential market clearing prices. Hence, the policy implications of the estimates of allocative and scale efficiency are easily verified.
The actual price for working and fixed capital is taken from loan rates set by the China's Industrial and Commercial Bank and were 4.8 per cent for 1980, 7.3 per cent for 1985 and 7.8 per cent for 1988 (see Appendix 3A). The average annual income of workers was calculated from workers' average wage. Average welfare expenditure in the industry was used as a proxy for the cost of employing labour. Considering that 'bonus spiral' activity among enterprises led to a more or less equivalent income level among workers within a enterprise as well as across enterprises, this industrial average income is a reasonable proxy for the cost of additional labour for each enterprise. These annual labour costs were 1,200 yuan, 1,550 yuan and 2,200 yuan for 1980, 1985 and 1988