ANALYSES ON FDI AND CHINA ’ S EXPORTS : 1995-2002

Based on data for the years 1995 to 2002, this paper has established a panel data model that reflects the relationship between China’s foreign direct investment and China’s exports, and regarding this, empirical analysis is made. The selected countries and regions include: Hong Kong of China, China’s Taiwan, Japan, South Korea, the European Union, and the United States. We have found that the relationship between the accumulated FDI (FDE stock) of different countries and regions in China and Chinese exports to the target countries is quite strong.


INTRODUCTION
Since the 1990s, in 1994 in particular, the combination of exchange rates and FDI (Foreign Direct Investment) has played a significant role in China's economic development.In recent years, research by economists on the relationship between FDI and China's foreign trade has been considerable.Researchers hold that FDI has improved the domestic investment environment and corporate governance standards, which have accordingly accelerated China's economic growth.Almost all studies have focused on the contribution made by current FDI inflows to the current economic growth (Liu, 2004) or the contribution made by the FDI lagged for one year to exports (EX) (Li et al., 2003).However, these studies have basically ignored the relationship between the growth situation of a country's stock of capital in China (accumulation of FDI inflows) and China's export trade, for which results are incomplete and inconclusive.
The objectives of this paper are as follows: first, using a number of countries (or regions) as research targets, to build a Panel Data Model on the basis of the consideration of national strength.Second, using our constructed model based on a number of countries' GDP and FDP stock in China, to make an empirical analysis of the

MODEL
In effect, utilizing Panel Data to analyze and research the macroeconomy is a relatively recent matter, which is wonderfully manifested in the most successful example of the utilization of Panel Data for research on the international economy and foreign investment.
The problems we are most likely to encounter in using Panel Data are heteroscedasticity of cross sections and autocorrelation of the sequence.Because both of these two phenomena have broken through the classical regression model assumptions, to use the least-squares procedure (OLS) at this time is inappropriate.In order to eliminate the influence of such phenomena, this paper tries to adopt the Seemingly Unrelated Regression (SUR) to make calculations when such a situation is encountered.
The general form of the single equation Panel Data is given as Equation 1.
Here, x is the vector of 1XK, β the vector of KX1, and K the number of explanatory variables.According to the normalized expressions, x and β should be written as matrix X it and B. This paper has adopted the above written forms.The equalizing value of error item it u is zero, and the variance is 2 u σ .

Models in common use are shown in Equations
Equation 2 states no individual difference in cross sections and variances in structure are assumed (namely all the trading partners are considered totally the same), and the estimation made by the least-squares procedure (OLS) has given the consistent and effective estimation of α and β .At this time, it equals putting together cross sections data during various periods as sample data.
Equation 3 is the changing intercept model.The individual influence in cross sections is different, which manifests itself in the variable impact which reflects the ignored individual difference.Such influence can be divided into two situations: fixed effects and random effects.This paper mainly makes quantitative studies on such two effects.Equation 4 is a variable coefficient model.Apart from the existent individual influences, there are changing economic structures in cross sections, and consequently structural parameters in the different cross sections units are different.The commonly used models include the Fixed Effects Model and the Random Effects Model.

Fixed Effects Model
Through F check and changing intercept model, its format can be showed as Equation 5.
Data Science Journal, Volume 6, Supplement, 9 June 2007 Here, it x is the vector of 1XK, β the vector of KX1, i α the individual influence and the ignored influence reflecting individual differences; it u are the random error item and the ignored factor influence changing with cross sections and time.Similarly, its equalizing value is zero and its variance is constant.it u and it x are irrelevant.

Random Effects Model
When the cross section units are the whole units of the population, the fixed effects model is a reasonable model.
If the cross section units are randomly selected from a large population, such models are merely applied to the selected cross section data units but not to the other units outside the sample.Under such situations, to regard the individual differences in the population as being subject to random distribution will be more appropriate, hence its form can be written as Equation 6.
When international trade is analyzed, it is only to select China's major export trading partners.Therefore, sample units are relatively smaller, and random effects changing cross sections are more effective at this time.In order to make further judgment between these two methods, the Hausman Check can be adopted.For the convenience of calculation, models are built under the following premises: using dollars as unified measurement units and taking the natural logarithms of the original data when calculating.The selected countries and regions include Hong Kong of China, China's Taiwan, Japan, South Korea, the EU, and the United States.Respectively, its variables are GDP, one hundred billion dollars; F1, ten thousand dollars; EX, ten thousand dollars.

As an explanatory variable, export (EX) denotes the volume of exports from
The adopted initial model is given as Equation 7.

S366
First of all, utilizing the current FDI to do a regression analysis on EX and adopting E-VEWS 3.1 Statistical software, we can make estimates using the models.The following results can be gained through calculation as shown in Table 1.≤ , it can be judged that it has auto-correlation.Thus, new variables need to be brought in.If we adopt the accumulated value of FDI (F1) and add its current value (FDI) and at the same time make simulations to each country's GDP data, the result will be more satisfying as shown inTable 2. The difference between different countries and regions can be indicated in quotient 0 β ) .Results can be improved if we make use of the fixed effects method as in Table 4.It can be found that in the random effects model the elasticity of China's accumulated FDI to the target country or region is 0.4.However, there is a relative auto-correlation in the above model.
The Asian economic crisis broke out in the years 1995 to 2002, and for the years 1997 to 1999, a dummy Data Science Journal, Volume 6, Supplement, 9 June 2007 variable can be considered to make a relative statistical adjustment.The definition is shown in Equation 8.
1，year 1997 to 1999 Simulation results of these fixed effects are given in Tables 6 -8.The simulation results of random effect without weighting are given in Table 9.
Models can be chosen through the utilization of STATA 8.0 and the Hausman Test as given in Tables 10 and 11.
Therefore, model analysis results approach more closely the acceptance of the Random Effect Model.

CONCLUSION
In the past, a good many studies held that FDI and exports are inextricably intertwined, which is mainly based on the relative analysis of the impact of FDI on the sum foreign trade.These studies adopted FDI as the explanatory variable and made a quantitative analysis by placing the total sum of exports as the explanatory variable to research their correlation.For instance, using data from the years 1983 to 2002, the multiplicator effect of FDI to exports trade can be calculated: .Analysis shows that China's exports are inextricably intertwined with foreign direct investment, and each additional dollar of foreign direct investment in one year will lead to the increase of 3.81 U.S. dollars in the next year (Li, Song, & Liu, 2003).
Different from the original study work, we are not directly utilizing the sum data to analyze the relationship between FDI and exports.We believe that such a relationship has hidden a number of problems.Through research on the Panel Data Model, we present new discoveries in this paper.If we make an analysis based on the relationship between the FDI of an individual country and its exports, the relationship between FDI and exports is not notable, and auto-correlation is serious.However, based on the Panel Data Model, it is found that the relationship between accumulated FDI (or FDE stock) of different countries and regions in China and exports from China to the target countries is quite notable, which not only can indicate the difference between countries and regions but also can effectively eliminate the issues related to sequence.
Data in this paper has its root in the Statistical Yearbook of the World Economy for the years 1995 to 2002, the Statistical Yearbook of China's Economy of the State Statistical Bureau for the years 1995 to 2002, the International Statistical Yearbook, the Statistical Yearbook of China's Foreign Trade, and eight-year data in the World Bank Data Base.
China to the target country.The data come from the Project of the Total Sum of China's Customs Exports to Every Country (Region) in the Statistical Yearbook of China's Economy; GDP refers to the gross domestic product of every country or region, whose data come from the International Statistical Yearbook, the Statistical Yearbook of China's Foreign Trade, and the World Bank Data Base.F1 refers to the FDI stock of every country in China.Export being an explanatory variable, its data have root in the Project of the Actual Use of Foreign Investment and Other Amount of Investment (according to different countries and regions) and in the Statistical Yearbook of China's Economy.
(1% Critical Value).In accordance with the rule of sequence-correlation and judgment principles, at 1% Critical Value because DW L d 1980 to 2000, placing the gross account of China's exports (EX) as the explanatory variable and the sum of foreign direct investment one year lagging behind as the explained variable,

Table 1 .
Current FDI and GDP to China's EX (HK: Hong Kong of China; TAIWAN: China's Taiwan; JAPAN: Japan; KOREA: South Korea; EU: European Union; US: United States)It can be showed that the elasticity of the current FDI to China's EX is 0.79.The coefficients in the equation are notable.At the same time, for 48 observations and 2 explanations:

Table 2 .
Accumulated FDI, Current FDI, and GDP to China's EX

Table 3 .
Accumulated FDI, Current FDI, and GDP to China's EX (fixed effects weighting effect)

Table 4 .
Accumulated FDI, Current FDI and GDP to China's EX(SUR)

Table 5 .
Accumulated FDI, Current FDI, and GDP to China's EX(random effects without weighting)

Table 6 .
Accumulated FDI, GDP, and D to China's EX (weighting is not included)

Table 7 .
Accumulated FDI, GDP, and D to China's EX (weighting is included)

Table 8 .
Accumulated FDI, GDP, and D to China's EX (SUR)

Table 9 .
Accumulated FDI, GDP, and D to China's EX (random effect without weighting) are all disturbance terms in the Fixed Effects Model and the Random Effects Model.In measuring analysis, Hausman's Test is often used to judge whether Fixed Effects or Random Effects are more effective.The check forms are as follows: H= ε

Table 10 .
Accumulated FDI, Current FDI, and GDP to China's EX Thus, model analysis results approach more closely the acceptance of the Random Effects Model.

Table 11 .
Accumulated FDI, GDP, and D to China's EX