A Comprehensive Evaluation of and Policy Recommendation to Foreign Direct
Investment Environments in Western China
Ever since its reform and opening-up, the pace of China’s integration into
the economic and financial globalization has been faster and faster. Now China has become the largest recipient country of foreign direct investment (FDI) among all developing nations. During 1979—2002, foreign investment in China totaled
US$623.4 billion out of which $446.3billion was FDI. Yet there lies a serious imbalance as to the actual spread of FDI amongst the country’s different regions.
In 2002, FDI into China was $52.743bil., out of which 86.1% went to the eastern region, 9.5% to the central and the remaining 5.71% to the western region. That means that per capita FDI in the western region was only $8.30 compared with the eastern region’s $95.60. Then what are the determinants or factors that
affected the regional distribution of foreign investment in China? What places do the various provinces in the western region hold in terms of their FDI environments? What are the disparities and causes? Through its FDI environment assessment system, this paper conducts a comprehensive evaluation and cluster analysis (CA) of the different investment environments among China’s different regions by using statistical data and quantitative models.
Academic research in the past 20 or 30 years on the choice of locations of foreign investment has been focused on: 1. increased analysis of location factors in international FDI theory to explain the influence of geographical locations on the choice of FDI recipient countries; 2. site investigation on investors to find out the decision-making process of their FDI locations; and 3. quantitative methods to determine the differences of FDI destination locations or the factors deciding on the choice thereof.
With the ever-increasing foreign investment into China, research findings on FDI to China has proliferated in both Chinese and English. Both foreign scholars and
; Ningxia Academy of Social Sciences, China, firstname.lastname@example.org I am
grateful to Professor Kitahara, Professor Hirakawa and staffs at Economic Research Center, Graduate School of Economics, Nagoya University. I would like to thank Professor Arayama especially for his helpful comments. Mr. Duan Qinglin and Mr. Yu Dahai have provided excellent research assistant work.
Chinese scholars working or studying overseas have contributed in English to the field of research. Due to the growing geographical imbalance, choice of FDI destination locations in China has become a hot topic of academic research, a major area of which has been to examine the determinants leading to the choice of FDI locations in the country by relying on the basic principles of modern geographical location theory and by using all kinds of econometric tools, such as the analysis done by Minghong Lu(1997) on the GDP, labor cost and other data from 29 provinces during 1988—1995 in their influence on the FDI
locations in the country. Houkai Wei and Feican He(2002) researched on the same by further analyzing the relationships between the choice of locations and different industry groups, the methodology of entry, differences in economic development stages in China and the different country origins of FDI.
Kevin Honglin Zhang(2002) is representative of recent research in English in the field. He used data from 29 provinces during the 1987—1990, ’91—’94 and ’95—’98
periods and analyzed the influence of such factors as market scale, labor cost, labor quality, business concentration, transportation cost, stimulation policy and cultural link on the choice of FDI destinations. Then he compares the results from his regression analyses of the three above-mentioned periods with those of the panel estimate from the 12-year period between 1987—1998. Changhui Zhou, Andrew Delios and Jingyu Yang(2002) used
data from 28 provinces during 1980—1998 for the analysis of Japanese businesses in their
decision of investment locations. One feature standing out in their research is the use of the number of businesses and the number of employees as a variable. In their study the accumulative number of businesses was used to explain the degree of economic concentration in the variable whereas the same was explained by the development stage of industry in other studies.
Though different variables and data years were used in the above researches, the quantitative methodology remained the same. The main purpose of their research was to try to find out what factors, and to how large a degree, influenced the inflow of FDI into China or from which country, i.e., the relationship between FDI and certain determining variables(determinants). Based on the assessment of the FDI environments in western China, this paper intends to: 1. set up an assessment system of indicators by using the above-mentioned FDI and its relevant factor analysis methodology to determine the certain factors most relevant to FDI in China; 2. to ascertain the combined index of FDI in 30 provinces in China in order to discover where western provinces lie in the index; and 3. to carry out a cluster analysis in the hope of finding out the commonality of the FDI environments in the 12 western provinces and their disparities with that of their eastern counterparts by objectively analyzing the internal types of the FDI environments in China’s 30 provinces and regions.
?;An Empirical Assessment of FDI Determinants
2.1 Model specification and variables
Based on calculations, the following model is constructed:
FDI=α0+αX+ε i ii
where =1,…,30; FDI is the amount of FDI inflow into the provinces in a given ii
period of time; X denotes a set of independent variables that vary across i
provinces and over time; and ε denotes stochastic disturbance. The variables used in this analysis are defined below(See Table 1).
FDI: A dependent variable referring to the share of FDI inflow into various regions, its unit being 100 million yuan(RMB) at the average annual exchange rate with US dollars listed by the Ministry of Commerce of China.
Following are independent variables:
GDP: Gross Domestic Product, the total amount of production and services of a certain region in a given year, and a substitutive variable of the market volume in this study, its unit being billions of RMB. Theoretically, its expected impact should be positive.
LOCA: A dummy variable with the eastern region=3, the central=2, and the western=1. Geographically, Guangxi Province belongs to the eastern coastal region and Inner Mongolia the central, but for the Great Development of Western China, both Guangxi and Inner Mongolia are considered western province and region. It should have a positive impact on FDI inflows,
STA: The state-owned industrial output, or ratio of the state-owned and state-held majority industrial enterprises’ output to the gross industrial output.
Reflecting the degree of maturity of China’s market economy, it should be
negatively related to FDI inflows. Its unit is the percentage (%). TER: Ratio of the value-added of the tertiary in GDP, mainly indicating the stage of development of finance, transportation, information services, etc. It’s expected
impact should be positive. Its unit is in percentage.
CITY: Urbanization level, representing the ratio of cities with an urban population of 500,000 out of the total number of cities in a given region. Theoretically, this variable should be positive on FDI, its unit being in percentage. POLI: A dummy variable. With reference to Sylvie Démurger’s(2002) approach,
this paper measures and tests favorable policy index based on the types of the special economic zones established by each province(the weight varies from three to one and that of non-open regions is null) and the open-door policy(Western Development region=0.5). Representing preferential policies, and it should be
positively related to FDI inflows.
WAGE: A dummy variable. It is 1 if the labor cost of the region is higher than that of the national average, or it will be null. In theory, it is inverse to FDI. LI: This variable refers to the proportion of light industry above a certain size relative to the gross industrial output value. With the priority change from heavy industry to the compensatory development of light industry since China’s reform
and opening-up, FDI has been affected by the regional distribution of light industry locations in China. Its unit is in percentage.
FTD: Foreign trade dependency degree is equal to the ratio of total imports and exports in GDP. Reflecting the openness of the economic development of the region concerned, it’s impact should be positive theoretically. Its unit is in
FI: The proportion of the amount of foreign enterprises’ imports out of the total
local imports and exports, which reveals the degree to which local governments control the imports by foreign enterprises. It should be positively related to FDI inflows, its unit being the percentage.
HC: Illiteracy rate of the population at or over the age of 15 in the region concerned, which represents the accumulation of the local human resource. It should be theoretically negative related on FDI inflows, its unit being the percentage.
We conduct a multi-regression analysis with the comprehensive data of 30 provinces (excluding Tibet) from the various years between 1998--2002 and the cross-sectional data of the year of 2002 respectively.
Time frame for the selection of the data is based on two considerations: one is that Chongqing, which has attracted quite an enormous amount of FDI in recent years, began to have its own statistics in 1998, and the other is that the on-going Great Western Development started in 1999, so statistics from 1998—2002 can better reflect the changes that took place after the Development began.
Model based on the comprehensive data from 1998—2002 is as follows:
Model I: ln(FDI)=a+aln(GDP)+aSTA+aTER+aLOCA+aWAGE+ε 12345
Model II: ln(FDI)=a+a ln(GDP)+aLI+aFTD+aFI+aHC+ε 12345
And the model based on the cross-sectional data from 2002 is:
Model III: ln(FDI)=a+a ln(GDP)+aLOCA+aPOLI+aCITY +ε 1234
2.2 The estimation results
Factors affecting FDI inflows are estimated by the ordinary
least-squares(OLS) techniques of the SAS statistic analytic software package. Through repeated measurements and calculations, 11 factors remarkably affecting FDI are established. They are lnGDP, TER, CITY, FTD, LI, POLI, LOCA, FI, STA, WAGE and HC. The results are illustrated in Table 2.
The overall performance of three estimates is satisfactory. Values of
2 adjusted Rin the three cases were from 71 percent to 85 percent, indicating a strong explanatory power of the models, and the significance level of F test is p<0.0001, indicating that the significance of the model regression as a whole is high.
The determinant model of FDI in 2002 is purposefully designed as Model III in order to test the impacts of various factors in pure cross-sectional data, the results of which denote that factors In (GDP), LOCA, CITY and POLI affect FDI significantly but both factors of POLI and LOCA are significant at 5 percent.
Specifically, some coefficient estimates in the models appear to be low. But in effect, when the statistic position of independent variable to dependent variable is horizontal-logarithmic value, the interpretation of coefficient a should
be: %?y=(100a)?x, i.e. the coefficient should be multiplied by 100. For a better understanding, factors in question are discussed as follows.
GDP refers to the economy and market size of a region. In Model I, the impacts of GDP on FDI is significant and the elastic coefficient is 1.147, denoting that when GDP between provinces increases at 1%, FDI will correspondingly increase by 1.147%. In addition, we also tested the relationship between FDI and per capita GDP, but it failed to pass the t test. As other researchers concluded, FDI is mainly to capture the markets of all provinces, municipalities and autonomous regions where the average individual consumption level remains low but its total amount is enormous. However, the elasticity estimated by cross-sectional data is far lower than that by comprehensive data. Provinces with less GDP values in the western region do see their FDI inflows affected, but it is relatively more favorable to those regions with larger GDP’s such as Sichuan, Shaanxi (Xi’an), Chongqing, etc.
Three belts of the eastern, central and western regions can fully illustrate the natural and economic environment variability of China. Regional factors mainly affecting foreign investment policy-making are transportation costs. Particularly, eastern coastal regions are endowed with naturally convenient conditions for
export. Other economic advantages are the closeness between these regions and the nearby investor countries, the consanguinity with their overseas Chinese investors, and the geographical factors that offer them superior agriculture, resources, human capital and so on and so forth. The disturbance coefficient of LOCA on FDI is 0.733, i.e., the regional variation of the eastern, central and western regions affects FDI inflow to a certain extent and the regional disadvantage of western China is unfavorable to attracting FDI inflow. 3) POLI
Establishing regional variations by offering preferential economic policies is crucial in order to attract foreign investment into China. It is well known that the reform and opening-up of China began with the preferential policies granted to Guangdong and Fujian provinces for their economic and foreign trade activities. In 1980, four special economic zones(SEZ), typical of which was the Shenzhen SEZ, were set up as a pilot scheme; in 1984, 14 more eastern coastal cities were opened up to the outside world. In 1985, the pace of the coastal regions’
opening-up hastened. Then in 1988 the whole island of Hainan was established as a SEZ. In 1990, Pudong in Shanghai was opened up for development, followed by the opening-up of the inland cities along the Yangtze River, land frontier cities and inland provincial capital cities after 1992. When it comes to the western region, its opening-up was by far later than any other region. It has enjoyed no special preferential policies, either. That is the single serious and institutional reason as to why there exists this tremendous disparity of the total FDI between the eastern and western regions. This study shows a significant impact of the preferential policy factor upon FDI inflow.
Urban structure is an indicator of the urbanization scale and density of a given province or region. In general, cities in the western region are small in number and sparsely scattered in scale. Their urban functions are far from perfect and are inadequate in co-operating with each other. Besides, they are in shortage of light industry bases and manpower reserves. In contrast with rural areas, their dyadic economy is deeply rooted. All of these factors are unfavorable to attracting foreign investment. The coefficient of the impact of the urban structural index on FDI is 0.019, indicating that when the proportion of the number of large- and medium- scale cities with a population of more than 500,000 increases by one percentage point, FDI will correspondingly go up by 1.9%. The higher the proportion, the more capable it is to attract foreign investment. In addition, we have investigated on the impact of the urbanization of 2002 on FDI but failed to get significant results, which could mean that it is the urban structure but not the urbanization
level that affects the FDI more.
The ratio of state-owned economy reflects the degree of the market-oriented reform and the maturity of market economy environment. It represents the structure of ownership and the structure of competition in the market. In western China, the state-owned economy still remains a high proportion, while in its eastern region the non-public ownership in the economy has been growing swiftly and actively. It is concluded from Model I that when the proportion of state-owned industrial economy increases one percent, FDI will decrease by 3.5%, which indicates that foreign investment prefers those regions whose growth rate of the market economy is higher. Shuyun Chen et al.(1995) calculated that higher market economy regions mainly lie in the eastern coastal areas whereas those lower regions are in the west. The stagnation in the development of the market economy in the western region is a key factor in restricting the FDI inflow. 6) TER
The improvement of the proportion of tertiary industry is a necessity for the national economic growth to a certain level. The higher the proportion, the higher the requirements for the division of labor the economic development needs for its services and infrastructures. During the period of 1978--2002, the ratio of the tertiary industrial structure in the eastern 11 provinces has shifted from 21.5:59.3:19.2 to 10.2:48.9:40.9 while that in the western 12 provinces has changed from 37.2:43.1:19.7 to 20.1:41.3:38.6. In appearance, the difference between the eastern and the western regions mainly lies in the development of the first and second industries. However, the tertiary industry in the western areas consists of services mostly from the ideological Party and administrative departments of the government, which controls more and offers less, especially in such areas as modern service and information industries, thus limiting the amount of FDI inflows. Given that when the TER proportion is increased by one percent, FDI will increase by 8.3%, TER is the biggest coefficient factor. 7) Wage & HC
Seeking for cheap labor is also an important factor contributing to the FDI inflows. As the models postulate, when the labor cost in an area is higher than the national average, FDI decreases by 1.19%; when illiteracy goes up by one percent, FDI goes down by 7.6%, suggesting that it is necessary for FDI to make a choice between the decrease of labor cost and increase of human capital. Since the Chinese government offered a preferential wage treatment for the workers and staff members in the border ethnic minority regions and impoverished districts that are at the rudimentary stage of reform, the variation of regional labor cost in
the whole nation is perfectly indistinctive. Thus it is natural for foreign enterprises to invest in the eastern regions where a large sum of human resources has been accumulated. With the development of non-public ownership in the economy and change in pay policy, additional income beyond wages in the eastern regions is increasing by a large extent and the regional labor cost variation has begun to show up. But in contrast to developed countries, the impact of regional labor cost in China on FDI is comparatively small. With more and more farm workers as well as talents from the western region moving to the eastern and southern regions in seek of better pay, the impact of WAGE and HC on FDI is getting weaker and weaker.
The reason to design and test the relationship between FI and FDI is because we think that when it comes to the investment structure, FDI tends to favor real estate and light industrial production and pays more attention to those regions with light industrial bases when selecting investment locations. In China, eastern regions have always been regarded as the major target areas for light industries to locate while western regions are characteristic of heavy industry sites. To a large extent, FDI further lowers the position of western regions when it comes to the locating of light industries. During the period of the centrally planned economy, China placed its priority on the development of heavy industries. Since its reform and opening-up, the country has experienced a rapid period of great development in the production of the means for livelihood. FDI-invested enterprises and villages and township businesses, especially those in the eastern and coastal areas, seized the rare opportunity to quickly develop themselves, which in turn greatly stimulated FDI. Our models indicate that when the proportion of light industry increases by one percent, FDI goes up by 3.8%. 9) FTD & FI
Model IV postulates that FDI correspondingly increases by one percent when trade dependency degree increases by one percent, and 2.1% when the import ratio of FDI increases by one percent. It is noted that the import of FDI into China was initiated from coastal port cities. Thus such provinces and metropolises as Guangdong, Beijing, Shanghai and Tianjin well-known for their leading positions in imports and exports have been the earliest and biggest beneficiaries of FDI. The implementation of the open-door policy and the SEZ policy also originated from coastal port cities with rich international experiences and then slowly extended to the central and western regions under such circumstances as had been gradually permitted by administrative decisions. The FTD level is an important resource and precious experience in the attraction of FDI. The
development of FDI needs the importation of technology and equipment and labelling business. Just as the direct correlation of FI to FDI indicates, it is easier to attract more FDI for regions in which FDI-invested enterprises enjoy a higher position in the total number of imports.
This research has also measured and tested the factors related to the per capita GDP and the total sum of investment in the nation’s infrastructure. But
these factors were eventually eliminated due to the failure of result confirmations. The establishment of the assessment system on FDI environments should emphasize the comprehensiveness and stability of the indicators. Our analysis indicators are highly general and representative since they take into consideration all factors ranging from macro, to location, to policy and to micro factors.
III. A Comprehensive Evaluation of Foreign Investment Environments by Principal Componential Analysis
3.1 Calculation of the comprehensive estimation index for foreign investment
environments by Principal Componential Analysis
Following section intends to make a componential analysis of the k number of affecting factors ascertained through the regressional analysis by using the SAS statistic analytic software package, to calculate the characteristic value and vector of the correlation matrix R, contribution ratio and the like, to reduce the k number of variables to a lesser m number of new variables by a further selection of factors and reduction of dimensions, and to interpret the economic significance of the selected principal components.
The analysis is done through the 11 selected interpretative variables of lnGDP, TER, LI, CITY, FTD, POLI, LOCA, FI, N-STA, N-WAGE and N-HC. In accordance with the comprehensive analysis, three indicators of STA, WAGE and HC are alternated whereby N-STA refers to the proportion of non-state-owned industrial production value, N-WAGE=--WAGE and N-HC stands for the ratio of the population with an education beyond the elementary school.
Suppose the previous m as the number of principal components are, respectively:
Y=f[ln(GDP), LOCA, CITY …] 11
Y=f[ln(GDP), LOCA, CITY …] 22
Y=f[ln(GDP), LOCA, CITY …] mm
then, when the various interpretative variables of the standardized process of the ith region are placed in the above model, the values of Y, Y, …,Y will be 12m
achieved. When the characteristic values corresponding to the m principal components are weight-imposed, the comprehensive index Y of foreign investment environments can be obtained as follows.
where b is the characteristic value corresponding to the previous m principal components. In the operation, the weight is thoroughly determined by the model and thus the subjective deviation caused by artificially controlled weighting is avoided.
3.2 The estimation result of the foreign investment environments
We select the 11 factors of ln(GDP), TER, LI, CITY, POLI, LOCA, FI, N-STA, N-WAGE and N-HC for the componential analysis. As illustrated in Table 2, with the exception of the correlation of TER to FI and N-STA and LI to N-HC being weak, the correlation between all the key factors, especially that between those and GDP, are of medium correlation. There is no factor that can be rejected due to its extremely low level of correlation(<10%) or extremely high correlation(>95%). As shown in Table 3, the representative can be as high as 94% if m=6 factors is selected. According to Tables 4 and 5, the six principal components are, respectively, as follows:
Y=5.4603 Y+2.0224 Y+ 1.2153Y+ 0.6348Y+ 0.5614Y+ 0.4477Y 123456