Printer Friendly

Distribution reform and labour productivity in China: an evaluation of performance and public policy.

Introduction

The Chinese economic reform strategy emphasizes competition over privatization. The loss-making state enterprises are replaced not by extensive privatization in the short run, but by being gradually outcompeted and outgrown by other non-state enterprises in the long run. In addition, its associated distribution reform is also different from the retail development of other developing countries. In contrast to East European countries, which also carried out extensive economic reforms recently, China's distribution reform adopted a gradual pace with an emphasis on competition over privatization. Her successful experience may provide valuable insight for the East European governments.

Because of China's active bid for membership of the World Trade Organization (WTO), the dream of selling in the Chinese market has finally become reality. In late 1992, the Chinese government pushed its distribution reform to a new stage with the announcement that foreign businessmen could establish joint ventures in the retail sectors of Beijing, Tianjin, Shanghai, Guangzhou, Dalian, Qingdao and the five special economic zones of Hainan, Shenzhen, Zhuhai, Shantou and Xiamen. The Chinese government will negotiate the cooperation with overseas corporations on a case-by-case basis[1]. These rapid changes have motivated overseas retailers to knock on the door of China.

Although business in China is developing rapidly, research in Chinese retailing and distribution has not grown proportionately. Until now, little has been published about the distribution system in China. Furthermore, the limited literature has been made outdated by the rapid development in China[2,3]. Wortzel and Wortzel[4] provide a detailed analysis of the development of free-market retailing and its associated social functions. Mun[5] and Qiang and Harris[6] only review the development of retailing reform in general. Chow and Tsang[7] study the relative changes in the performance of major players, namely state enterprises, collective enterprises, joint enterprises and individual enterprises subject to the changes in the consumption patterns during the economic reform.

This paper extends Chow's[8,9] analysis to investigate the socioeconomic factors which determine the performance of retailing enterprises across 30 provinces during the period of 1985-1992. The approach adopted here follows the macro-marketing research developed by Bucklin[10,11], Takeuchi and Bucklin[12], Ingene and Lusch[13] and Ingene[14]. Sales per employee of retailing enterprises which also measure labour productivity are chosen as an indicator of retail performance (dependent variable). Moreover, China has been suffering from inflation frequently, sales per employee, per capita provincial national income and average wage are further deflated by the corresponding provincial retail price indices to obtain the values in real terms. A regression analysis is conducted to examine which socioeconomic factors have statistically significant impact on the dependent variable.

Methodology

The effects of predictor variables on retail performance are presented as follows.

Household-specific variables

The household factors used as predictor variables are income and mobility. In this analysis, per capita real national income (PCNI) is used. It is expected that income will have a positive effect on the sales according to Engel's law. Raising per capita real income has two effects:

(1) it raises the demand of existing products; and

(2) as residents' incomes grow, their consumption patterns also change.

Such changes allow markets for new products or services to emerge.

Mobility is not measured by the number of automobiles owned by households because most Chinese households still cannot afford automobiles. It is measured by the total number of civil and private vehicle seats available divided by the corresponding provincial population, so as to obtain a per capita index of mobility. The factor of mobility should have a positive effect on the sales per employee. Because retail outlets may be located in city or free market areas which may not be close to the residential areas, convenient transportation reduces the travelling time. Shoppers, therefore, can have a better comparison of their demanded products and their potential choice sets should move closer to their ideal ones. They would like to buy more under less restriction.

Environmental variables

Environmental variables include population density, population growth rate, dummy variables accounting for city-province difference and coastal-inland difference. Bucklin[11] suggests that higher population density is likely to increase the level of competition through larger and more technologically advanced stores in the region. Thus, a higher degree of urbanization is likely to reduce the sales per employee.

The effects of population growth on sales per employee are more complicated. Rapid growth can raise the sales as those areas are understored. However, Takeuchi and Bucklin[12] point out that rapid growth areas decrease entry barriers and bring in more new stores and, in turn, more aggressive competition. Keener competition should reduce the sales per employee. Therefore, the final effect on sales is ambiguous.

A dummy variable accounting for city-provincial difference is used because Beijing, Tianjin and Shanghai are autonomous cities which have a more homogeneous urban structure compared with other provinces which have both urban and rural sectors. Another dummy variable for coastal-inland difference is used because coastal provinces usually have better infrastructure and better access to overseas businesses and investors. It is, therefore, reasonable to include this effect in the analysis. However, its effect on sales per employee is not certain. On the one hand, better infrastructure and access to overseas businesses and information should improve operational efficiency of retailers which increases sales. On the other, such convenience would also invite more aggressive competition with more sophisticated tools and reduce sales per employee. Thus, the final effect on sales is ambiguous. A summary of the definitions of variables is presented in Table I.

In 1988, the Chinese economy was hit by runaway inflation. The government imposed an austerity programme to slow the economy down in the later half of 1988. A structural break created by this shift in public policy is expected. To test the effect of 1988 austerity programme on the retail structure, a Chow test[15] is implemented on the equation of SALEPE. If the austerity programme really changes the retail structure, then the estimated parameters of predictor variables before the programme was implemented should be different from those after the [TABULAR DATA FOR TABLE I OMITTED] programme's implementation. By dividing the samples into two groups, 1985-1988 and 1989-1992, and by running two separate regressions on these two subsamples, this should lead to two separate sets of parameters. Therefore, the hypothesized model is as follows:

[Mathematical Expression Omitted] (1)

* dependent variables:

Y - N x 1 vector;

[Y.sub.1] - [N.sub.1] x 1 vector;

[Y.sub.2] - [N.sub.2] x 1 vector;

* predictor variables:

X - N x 2 k matrix;

[X.sub.1] - [N.sub.1] x k matrix;

[X.sub.2] - [N.sub.2] x k matrix;

* parameters:

B - 2k x 1 vector;

[B.sub.1] - k x 1 vector;

[B.sub.2] - k x 1 vector;

* error:

U - N x 1 vector;

[U.sub.1] - [N.sub.1] x 1 vector;

[U.sub.2] - [N.sub.2] x 1 vector;

where N = [N.sub.1] + [N.sub.2] is the number of observations, k is the number of estimated parameters. If the austerity programme has no effect on the retail structure, the two sets of parameters should be identical, i.e. [B.sub.1] = [B.sub.2]. Running regression using the whole sample should be equivalent to running two separate regressions as there is no structural break. As a result, the null hypothesis is that the 1988 austerity programme has no effect on the retail structure:

[H.sub.0] : [B.sub.1] = [B.sub.2].

The test statistics are as follows: RSSR, [USSR.sub.1] and [USSR.sub.2] are the restricted sum of squared residuals, unrestricted sum of squared residuals of 1985-1988 subsample and 19891992 subsample respectively. F is distributed as an F(k, N - 2k) distribution:

F = (RSSR - [USSR.sub.1] - [USSR.sub.2])/k / ([USSR.sub.1] + [USSR.sub.2])/(N - 2k). (2)

In estimating the retail structure, the following log-linear structural equations are considered:

SALEPE = [a.sub.0] + [a.sub.1]DC + [a.sub.2]DIC + [a.sub.3] PCNI + [a.sub.4] SPOP + [a.sub.5]PD + [a.sub.6]POPG + [a.sub.7] SEAT + e (3)

All variables are log-transformed except DC, DIC and POPG. The purpose of these regression exercises is to find out how much the socio-economic factors can explain the dependent variables.

Empirical results

The data of 1985-1992 are chosen for the analysis. The sales data of Tibet from 198789 are not available, probably due to the political instability in that period. Since Hainan province was established in 1988, there is no information available from 1985 to 1987. Altogether, with 30 provinces and seven years, the pooled data set has 234 observations. The multicollinearity problem can be seen in Table II.

It is particularly severe between PCNI and SEAT, between PCNI and AW, between AW and SEAT. Although ordinary least square (OLS) is preferred in analysing the data, given that OLS is a reasonable and straightforward procedure, the high multicollinearity raises the possibility that the OLS regression estimates "(1) are subject to large errors, (2) are likely to have erroneous algebraic signs, and (3) are highly sensitive to small changes in the database" [16, p. 671]. The ridge regression method is adopted as an improved estimation procedure. The statistical essence of ridge regression may be described simply in this formula, B = [(X[prime]X + mI).sup.-1] X[prime]Y, where I is the identity matrix and 0 [less than or equal to] m [less than or equal to] 1. At m = 0 ridge and OLS estimates are identical. As m increases, stability is imposed on the regression estimates. The values of regression estimates appear to stabilize in the m = 0.3 to m = 0.4 range. Thus, m = 0.333 is chosen and the results are reported in Table III. The estimates of intercept, DC and SEAT during the period of 1985-1988 are affected considerably by the problem of multicollinearity as their signs are reversed after imposing stability to estimated parameters.
Table II Correlation matrix of all independent variables

         PCNI       SPOP      PD      POPG       AW     SEAT

PCNI     1.000
SPOP     0.173     1.000
PD       0.442     0.092     1.000
POPG    -0.058    -0.147    -0.042    1.000
AW       0.550     0.320    -0.154    0.116    1.000
SEAT     0.722     0.055    -0.017    0.091    0.592    1.000


The F-test statistics for the effect of the 1988 austerity programme are 3.4932 and 3.2361 respectively. They strongly reject the null hypothesis ([H.sub.0]: [B.sub.1] = [B.sub.2]) that the retail structure is invariant to the austerity programme implemented in late 1988. The implementation of the austerity programme has slowed down the economic growth rate from a double-digit figure to a single-digit figure. The retail sector was undoubtedly affected by this programme. Since the retail structure is sensitive to the shift in government policy, splitting the sample into two groups (1985-1988 and 1989-1992) is necessary to yield unbiased estimates. This observation also suggests that a researcher who [TABULAR DATA FOR TABLE III OMITTED] engages in studying retail structure in China should consider the effect of any public policy change.

To conserve space, only the statistically significant estimates of ridge regression results will be studied. In Table III, both equations have high goodness of fit (high adjusted [R.sup.2]). Most of the parameters estimated are close to what was expected in the previous section and their values are interpreted as follows: 1 per cent increase in, let us say, per capita national income (PCNI) leads to, on average, 0.15 per cent increase in sales per employee (SALEPE) during 1985-1988 period reported in Table III. Coefficients of PCNI and store per capita (SPOP) reveal the predicted signs; in addition, they are consistent and highly significant for explaining the variations in real sales per employee. It suggests that income and store per capita are the two important factors of the retail structure. The high explanatory power of per capita national income is expected according to Engel's law.

Average wage (AW) is also strongly significant for explaining the equation of real sales per employee (SALEPE). In terms of magnitude, AW is the largest factor in explaining the variations in the dependent variable. The significance of wage and store per capita has some important implications in the organizational dynamics of enterprises which are experiencing the distribution reform. Before the economic reform, there was no correlation between workers' wage rates and the enterprise performance. The wage rate was not a factor to be considered in explaining the changes in real sales per employee at that time.

Because the centrepiece of economic reform involved decentralization of decision making to enterprises, reflected most clearly in the higher autonomy in allocating resources within enterprises, there were significant changes in the organizational structure of enterprises. Since the economic reform, enterprises have been becoming more and more market oriented externally, and acting or reacting according to the changes in their markets. Keener competition in this sector is captured by the significant estimates of SPOP.

Internally, managers of enterprises have been implementing better remuneration schemes, albeit imperfect, to match an employee's pay with his/her productivity. For example, bonuses are commonly used to motivate their employees. Since wage rates are now more flexible, they do serve as a significant factor in improving the operational efficiency of enterprises. In fact, average wage is the largest factor in affecting sales per employee (SALEPE). This piece of evidence actually suggests that, by introducing better incentive schemes, enterprises are actually making progress in increasing employees' productivity.

The estimated parameter of average wage before the austerity programme is larger than its counterpart after the austerity programme. There are at least two reasons:

(1) The austerity programme simply reduced demand for all output of the retail sector (in terms of demand). Therefore, the effects of wage rates on sales were reduced accordingly.

(2) The austerity programme weakened the reform that was being carried out within enterprises (in terms of supply).

Prior to the austerity programme, the central and provincial governments showed greater enthusiasm in encouraging enterprises, especially state enterprises, to introduce different incentive schemes to improve worker productivity. After the implementation of the austerity programme, the governments, especially the central government, showed less interest in continuing the reform in the enterprises although such efforts were later resumed in 1991. Such a slowdown is sufficient to hinder the productivity growth.

Mobility, measured by per capita civil and private vehicle seats, is also a significant factor in accounting for the sales. However, in terms of magnitude, mobility has a relatively small effect on sales per employee. The coefficients of dummy variable of city-province difference, DC, are all significant, thus, suggesting that sales in cities like Beijing, Shanghai and Tianjin are relatively higher than those of other provinces. Population density and coastal-inland difference are only significant in explaining the SALEPE variable during 19851988 period, but their impact on sales is minor. The rest of the predictor variables are insignificant in explaining the dependent variable.

Conclusion

The economic reform pursued by the Chinese government is remarkably different from that which the East European countries have implemented. The Chinese model emphasizes competition over privatization. The loss-making state enterprises are replaced not by extensive privatization in the short run, but by being gradually outcompeted and outgrown by other non-state enterprises in the long run[17]. Such a reform strategy has been working well in China although it is not advocated by many western economists and business people. The purpose of this paper was to study the changes in the Chinese retail sector induced by the distribution reform.

Before the economic reform, the Chinese economy was a centrally planned economy. Distribution was monopolized by the state enterprises. Managers merely took orders from the ministries in meeting output targets and were not concerned with consumer preferences. Marketing obviously had a very limited role to play in such an economy.

With the implementation of the distribution reform came rapid changes in the retail structure. Enterprises became more concerned with the external changes in their markets and the internal control of their operations. This paper examined the changes of retail structure in the 30 provinces of China during the distribution reform. Due to the presence of a multicollinearity problem, ridge regression was used to stabilize the values of regression estimates. The following empirical results were obtained:

(1) The Chinese retail structure is sensitive to changes in the government policy. The 1988 austerity programme which was launched to cool down the overheated economy at that time has significantly altered the retail structure. It suggests that the implementation of recent restrictive policy for combating the overheated economy may also affect the retail structure. Researchers in studying Chinese retailing should take heed of this.

(2) The per capita national income, store per capita, average wage and mobility measured by the availability of transport are important predictors of the retail performance measured by sales per employee.

Although these results are not unexpected, they do provide valuable insight into understanding the retail structure in China. Overseas retail executives can use these results to predict the retail structure in different provinces and formulate their development plans.

References

1 Financial Times, 28 January 1993, p. 4.

2 Wortzel, H.V., "Equity and efficiency in the distribution of non-food consumer goods in China", Asian Survey, Vol. 23 No. 7, 1983, pp. 845-57.

3 Reeder, J.A., "Entrepreneurship in the People's Republic of China", Columbia Journal of World Business, Fall 1984, pp. 43-51.

4 Wortzel, H.V. and Wortzel, L.H., "The emergence of free market retailing in the People's Republic of China: promises and consequences", California Management Review, Vol. 29 No. 3, 1987, pp. 59-76.

5 Mun, K.C., "Chinese retailing in a changing environment", in Kaynak, E. (Ed.), Transnational Retailing, Walter De Gruyter & Co., Berlin, 1988, pp. 211-26.

6 Qiang, Z.W. and Harris, P., "Retailing reform and trends in China", International Journal of Retail & Distribution Management, Vol. 18 No. 5, 1990, pp. 31-9.

7 Chow, K.W. and Tsang, W.K., "Distribution reform in China: an analysis of the private sector development", International Journal of Retail & Distribution Management, Vol. 22 No. 2, 1994, pp. 27-33.

8 Chow, K.W., "Distribution reform and retail structure in China: an empirical analysis of entries and exits of enterprises", Asia Pacific International Journal of Marketing and Logistics, 1995.

9 Chow, K.W., "Evaluating small business development in China's retail sector: an empirical analysis", Journal of Small Business Management, Vol. 33 No. 1, 1995, pp. 87-92.

10 Bucklin, L., Competition and Evolution in the Distributive Trades, Prentice-Hall, Englewood Cliffs, NJ, 1972.

11 Bucklin, L., Productivity in Marketing, American Marketing Association, Chicago, IL, 1978.

12 Takeuchi, H. and Bucklin, L., "Productivity in retailing: retail structure and public policy", Journal of Retailing, Vol. 53, Spring 1977, pp. 35-46.

13 Ingene, C.A. and Lusch, R.F., "A model of retail structure", in Sheth, J. (Ed.), Research in Marketing, Vol. 5, JAI Press, Greenwich, CT, and New York, NY, 1981, pp. 101-64.

14 Ingene, C.A., "Labour productivity in marketing", Journal of Marketing, Vol. 46, Fall 1982, pp. 75-90.

15 Chow, G.C., "Tests of equality between sets of coefficients in two linear regressions", Econometrica, Vol. 28 No. 3, 1960, pp. 591-605.

16 Churchill, G., "A regression estimation method for collinear predictors", Decision Sciences, Vol. 6, October 1975, pp. 670-87.

17 Weitzman, M.L., "Economic transition: can theory help?", European Economic Review, Vol. 37, April 1993, pp. 549-55.

The author

Clement Chow Kong Wing is Lecturer in the Department of Marketing and International Business, Faculty of Business, Lingnan College, Teun Mun, New Territories, Hong Kong.
COPYRIGHT 1996 Emerald Group Publishing, Ltd.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1996 Gale, Cengage Learning. All rights reserved.

 
Article Details
Printer friendly Cite/link Email Feedback
Author:Wing, Clement Chow Kong
Publication:International Journal of Retail & Distribution Management
Date:Sep 1, 1996
Words:3275
Previous Article:Smart cards and retailers - who stands to benefit?
Next Article:Sales in UK petrol forecast shops reach 2.9 billion pounds sterling in 1995, boosted by National Lottery.
Topics:

Terms of use | Privacy policy | Copyright © 2018 Farlex, Inc. | Feedback | For webmasters