Cross-Sector and Time Series Analysis of Firm Performance

In Chapter 5, we created a model to explain some of the characteristics observed in the development of the private sector in the PRC. Generally speaking, this model, though relatively simple, seems to be sufficient to explain our observations of the private sector. However, to gain a more detailed understanding, it would be helpful to have cross-sectional and time series analysis of the data surveyed by CASS on SOEs, collective enterprises and private enterprises. In the PRC, as indicated above, more radical SOE reform (especially of SME enterprises) started from 1995. By correlating economic performance indicators of different types of enterprises with factors dependent on reform, we can gain deeper insight into private sector development. To carry out such as analysis, we classified the surveyed sample firms into three groups according to their original registration status (see Appendix 1), though this status may have changed with the progress of reform. The first group included 79 SOEs, the second 74 collective enterprises, and the third 347 non-public enterprises. Here, non-public enterprises include all those other than SOEs and collective enterprises. In addition, 20 private enterprises were investigated separately.²²

6.1 Descriptive Statistics by Sector and Time

6.1.1 Export Dependence and Efficiency Indices

Different types of enterprises are exposed to the outside world to differing degrees. On average, from 1994 to 1999, the collective enterprises were most dependent on exports for value added, followed by non-public enterprises, SOEs, and private enterprises. Looking at the trend, both SOEs and collective enterprises continue to decline and non-public enterprises to rise. In 1999, non-public enterprises had the highest dependence on exports, and private enterprises the least.

Row 3 in Table 6.1 [PDF 64KB | 1 page] shows that non-public enterprises and private enterprises are more efficient than collective enterprises and SOEs in terms of profit ratio to value added. The average profit ratios for these enterprises are 16%, 11.6%, 8.9% and 4.4%, respectively. However, through the economic transition and perhaps due to the stimulative macroeconomic policy, both SOEs and collective enterprises seem to have halted their profit declines. For example, in 1994 when the PRC economy was overheating, the profit rate of SOEs was 10%; this rate fell to -4.6% in 1997. However, it then recovered to 7.3% in 1999. The table also shows that the profit rate of nonpublic enterprises declined. Perhaps because of firm size, SOEs make the largest contribution to government tax, followed by non-public enterprises and collective enterprises. The contribution of private enterprises is the lowest. This may be one of the reasons why the government did not actively encourage private sector development. Table 6.1 also shows that the non-public sector, which presumably has higher marketization and privatization experiences than SOEs and collective enterprises, is also more efficient in terms of the profit rate. This follows from proposition 2 in Chapter 5, i.e. the higher the privatization, the larger the profit²³.

6.1.2 Proportion of Equity Owned by the State and Collectives by Enterprise Type

Table 6.2 [PDF 64KB | 1 page] shows that the equity assets owned by state and collectives shrunk for all types of enterprises. The ratio of SOEs fell from 0.92 in 1994 to 0.85 in 1999; in the case of collective enterprises the ratio falls from 0.78 in 1994 to 0.54 in 1999, for non-public enterprises, it fell from 0.64 to 0.25 during the same period. For private enterprises the ratio was stable or equal to zero. It is clear that until 1999, the equity of state and collectives dominated enterprises under reform, in turn giving local governments an excuse for intervening in the internal affairs of firms. This indicates that there were still much work to do for SOEs and collective enterprises in pursuing privatization.

6.1.3 Interest Payments and Wage Rate

The main production costs faced by enterprises are capital costs and labor wages. In terms of capital costs, Table 6.3 [PDF 110KB | 1 page] shows the median (med) for the interest/liability ratio and median, maximum and standard deviation for wage rates. From it, we find the following characteristics. Firstly, the interest/liability ratio fluctuates widely from a low level to high level and again to a low level. One of the reasons for this is that interest rates fluctuate with the macroeconomic environment. During the period from 1994 to 1999, the PRC experienced shocks both from overheating investment around 1994 and the Asian crisis in 1997.

Secondly, due to the lack of incentive mechanisms linking responsibility and rewards for the management, as well as the ambiguity of property rights in traditional SOEs, both SOEs and state-owned banks lacked sensitivity to investment risk. Soft budget constraints led some SOE managers to fail to consider the repayment of interest and even principal of banking loans. Thirdly, a considerable number of insolvent firms went bankrupt during the reform.

Median wage levels differ among different types of firms. Non-public enterprises enjoy the highest wages among the four, followed by SOEs, collective enterprises and private enterprises. Taking into account the fact that collective enterprises and private enterprises enjoy less social security and welfare, their real income is even lower.

Lastly, if we compare wage difference within SOEs, collectives and non-public enterprises, we find that, in addition to the relatively low wage in collectives and private enterprises, the wage rate gap is also smaller in collectives and private enterprises than in SOEs and non-public enterprises²⁴.

6.1.4 Structure of Employment

The size of enterprises measured by number of employees is listed in Table 6.4 [PDF 64KB | 1 page]. It shows that with the progress of reform, SOEs, collectives and non-public enterprises became smaller, in accordance with Proposition 4 in chapter 5. However, if we look at private enterprises separately, we find that they enjoyed an increase in the number of employees, although the size remained small.

In Table 6.5 [PDF 87KB | 1 page], the employees of the surveyed enterprises are classified into five categories, i.e. ordinary workers, managers, technicians, temporary staff and laid-off workers. In terms of the number of ordinary workers, the average proportions of SOEs between 1994 and 1999, collectives, non-public and private enterprises are 70.8%, 68.2%, 71.9% and 62% respectively. The proportion of ordinary workers in SOEs shows a declining trend, while the number of laid off workers shows an increasing trend; this is the outcome of the policy called "reduce the number of staff in order to promote efficiency." In collective enterprises, the composition of employment is relatively stable, fluctuating at between 66% and 68%. An outstanding characteristic of nonpublic enterprises is that more ordinary and temporary workers were laid off while the proportion of technicians rose. In comparison, private enterprises had fewer ordinary workers and more temporary workers than the other types.

6.2 Cross-Sector and Time Series Regressions of Output Growth

In this section, by carefully selecting variables that may be correlated, we explore the main factors that influenced the output of the surveyed firms between 1994 and 1999. This analysis is based on a general framework of a generalized production function and pooled data regressions. This approach allows us to understand how output was affected by what factors. Based on the regression equations, we can identify positive and negative factors that affect enterprise performance, and eventually uncover evidence supporting some of our propositions put forward above.

The empirical study that follows is based on a revised neoclassical production function. As is well known, most neoclassical production functions use a Cobb-Douglass production function with constant returns to scale and double factors, i.e. labor and capital (Zhao Zhijun, 2002 [42]). In Chapter 5, we cited some defects of this type of production function. Economists are usually classified into two schools depending on their views on economic growth momentum, i.e. the exogenous school and endogenous school. Some economists, such as Romer, 1986 [24], attempt to explain economic growth by introducing endogenous variables. Others, such as Mankiw, Romer, and Weil, 1992 [21] and Barro and Sala-i-Martin, 1995 [4], emphasize that neoclassical models with exogenous technology can interpret economic growth once human capital is introduced into the model as an exogenous variable.

Following the ideas of Mankiw, Romer, and Weil, 1992 [21], here, in addition to fixed assets, we consider ordinary workers, managers and technicians as different production factors, and place them into the model as factors influencing output. Given that the PRC is experiencing reform and pursuing privatization and marketization and that its capital structure is constantly changing in a way that is considered to be important momentum for economic growth (Zhao, 2002 [42]), we introduce a proxy to represent privatization. We take the proportion of equity owned by state and collectives to total equity as a proxy for the capital structure variable. Under these assumptions, we specify the production function as follows:

output = F(labor, fixedassets, manager, technician, structure)

It is well known that economic time series with level variables often produce "unit roots" and cause false regressions. In order to avoid this, we assume that the logarithm of the production function is approximately a linear logarithm function of the factors and the growth rate of output can be expressed as a difference of the linear logarithm function:

Where "dlnoutput" is the dependent variable, representing the difference of the logarithm of value added; "dlnlabor", "dlnfixedassets", "dlnmanager", "dlntechnician", and "dlnstructure" represent differences in the logarithms of labor, fixed assets, managers, technicians and structure, respectively. The constant "C" can be regarded as an effect of other exogenous variables. The subscript i identify cross section variables of firm i, and time t. Correspondingly, these differences with respect to time can be treated as the growth rate. ε_it is the effect of random shocks from unobserved factors.

Next we turn to the empirical outcome of this equation for SOEs, collective and private enterprises, as shown in Table 6.6 [PDF 80KB | 1 page]. First, the rates of change of labor, fixed assets and managers have significant effects on output at the T-value significance level of 1%, but the impacts of the rates of change of technicians and structure on output growth are mixed. Concretely, for state owned enterprises, the growth rates for all variables other than technicians have significant effects on output growth at the 1% significance level. For collective enterprises, all growth rates, such as labor, fixed assets, managers, technicians and structure, have significant effects on output growth at the 1% significance level. For non-public enterprises, with the exception of a negative effect of the technician growth rate, the growth of the independent variables has significant effects on output growth. For private enterprises, the growth rates of labor, fixed assets and manager have positive effects on output growth at the 1% level; the growth rate of technician has no significant effect on output growth, but has a positive effect at the 5% level; however, change in structure has no significant effect on output growth at even the 5% level, meaning that changing it is not as urgent as the other factors.

Secondly, factors such as labor, fixed assets, managers, and structure, all have positive impact on output growth, which is in conformity with the prior theoretical analysis. Surprisingly, however, the growth of the number of technicians, who are seen as important human capital resources, is negatively related to output growth in such sectors as SOEs, non-public enterprises (including private enterprises) and private enterprises alone. In collective enterprises, although a positive effect of technicians is identified at the 1% significance level, the elasticity coefficient (0.005388) is very small. We can offer two possible explanations for this phenomenon. Firstly, in the transition process, even if technicians are widely recruited by firms, they are not necessarily placed in important positions and do not play an important role in the enterprises. The other reason may be that the abilities of technicians may not suit the demands of enterprises due to the backward educational system.

Thirdly, as it is already mentioned above, it is worth emphasizing that change in structure, meaning more privatization or a smaller public equity share, has a major negative impact on output growth (or a positive impact on profits) for all categories. This finding again supports our proposition in the previous chapter.

Finally, Table 6.6 shows that both R-squared and DW statistics are satisfactory. The intercepts, whether significant or not, seem too small. To better estimate the above regression equations, we delete those variables without significant effects, and re-estimate them. The estimation results are listed in Table 6.7 [PDF 85KB | 1 page].

In summary, all types of enterprises in the PRC should attach great importance to the privatization process when increasing capital and labor input. Firms should not only increase the numbers of technicians, but also use them efficiently. The educational sector should be strengthened to produce students more suited to the market demand. The progress of privatization may be beneficial for the rational use of human capital. The privatization process should be sped up in order to make economic growth sustainable.

Comment(s)

There are [0] comment(s) for this entry. Post a comment.