The Economy Through the Process of Globalization in China and Economic Transformation - Research Paper Example

Chapter 1 Introduction Globalization Trend Globalization is an ongoing process of interaction whereby integrating the domestic economy, culture, technology, etc. into the worldwide scope through activities such as Foreign Direct Investment, Foreign trade. Globalization is a significant but not a new economic phenomenon. World Bank (2002) throws out that there were three main globalization waves since 1870 where the global economy, capital flow and migration have been increasing dramatically. The first two waves appeared within 1870-1979. Due to the lower cost of transportation and reduced barriers of international trade, the globalization has notably motivated the world’s economy. The third wave happened in 1980 which exhibited different feature, its scale and pace was the largest and the most critical change is that the developing countries are gradually merging into global market. In general thought, expanding international integration makes people’s life better. As some activities like trade liberalization, financial and production integration lead more efficient production, higher investment and output, therefore, the standard of life should be improved. This is supported by Dollar and Kraay (2002), point out that the globalization movement has improved the global inequality and reduce poverty. However, it is considered that the prosperity cannot bring to all people in the world. The Ex-Malaysian Prime Minister, Mohamad ( 1996 ) said that the globalization does not mitigate poverty, but leads world inequality. The process of globalization widened the income gap between developing and developed countries. China steps into the global market There are substantial studies to investigate that whether economic integration has been widening the gap between countries. However, there is relatively smaller number of literatures which has examined the effect on national disparity. The impact of globalization on inequality within the country is one of the most controversial topics which has been debated for long. In this paper, rather than looking at the world inequality, I am going to concentrate on examining the relationship between globalization and economic inequality in China. China is the fastest growing country among all the developing countries and dominates an important role in the most recent globalization wave. From the figure 1, we can see that the GDP of China has been significantly increasing to an unprecedented rate and the rate has maintained 12% growth annually in the global economic integration decade. The rapid economic growth in China captures the world’s attention especially after the economic reform and openness policy in 1978 as the Trade, exports and imports have enormously gone up (Refer to Figure 2). Perhaps like other developing countries, China has been experiencing an increase in economic inequality accompany with the global economic integration. As shown in figure 3, the Gini coefficient from 1978-2008 presents that the inequality within the country has been highly raised. Growing on the intra-country inequality not only affect the stability of the country, but also hamper the potential economy growth. Moreover, the inequality problem is quite complicated in China as the wage inequality, uneven income distribution and regional disparity are all needed to take into account when we concern about the economic inequality of a country. The following chapter will briefly review the current literature to see what scholars have discovered on this topic and how they argue about the story. In chapter 3, the empirical model will be constructed and displays the details of data resource, data description and OLS assumptions. Chapter 4 will access the tests for the model to check its validity, and then find out the relationship between the four independent variables and the dependent variable. Critical analysis about the globalization impact on the inequality in China will also applied in Chapter 4. The conclusion will be drawn in Chapter 5 to summarize what evidences have been found from the results of this paper. Finally, chapter 6 will outline the limitation of this study and suggestions for further research. Chapter 2 Literature Review The effect of the openness of the economy through the process of globalization on the income inequality of country has always been the topic of curiosity among the academic professionals and also the policy makers of the economy. Various active literatures exist that study the relationship between the openness of the economy of a country and the economic growth of the same with a link to the income inequality of the country. However since there are various methods of calculating the income inequality in countries, different methods may sometimes evolve different results and contradictory pattern of the relationship between the two concerned variables for the same time period. A study was made by the international monetary fund in the year 2002 for empirical finding of the impact of globalization on the income disparity of the country. (Wei and Wu, 2002, pp 2 – 3) Economy of china before and after 1978 The economy of china before the year 1978 was characterized by Autarky and an economy closed from the rest of the world. The foreign investments in the economy of the country were restricted as the investors were regarded with hostility. The legal system of the country could not provide the basis for foreign investment in the country. The domestic market of the country was totally under developed with very low per capita income. More than 80 percent of the total population of the country depended on agriculture for their livelihood receiving income mainly in kind and living in the rural communes of the country. The country adopted the policy of ‘open door’ in the year 1978 with the encouragement of foreign investment in the economy in unrestricted numbers. (Kaplinsky, 2005, pp 17-18) Fiscal incentives from the domestic government along with institutional infrastructure were provided to the investors. The investment policies of the country were liberalised to enhance the potentiality of the emerging economy of the country. At recent period the country holds the ranking of being the sixth top ranking nations in terms of exports and imports considered worldwide. Moreover the economy is at present one of the largest receiver of the foreign direct investment in the world, which contributed to the improvement in the living standards of the residents of the country. The largest developing country of the world experiencing integration with international trade and the economy of the world only in the last quarter of the twentieth century, reported an increase in the trade to GDP ratio of the country to 36.5 percent in the year 1999 from a mere percentage of 8.5 just two years after the openness of the economy. The most recent impact of globalization of the economy of the country is its integration with the World Trade Organization as a member country in the year 2001. The openness of the economy and hence increase in the investment flow in the country contributed hugely to the economic transformation of the country. (Davies, 2003; Wei and Wu, 2002, p 4; Lozada, n.d) Defining the relation using Heckscher–Ohlin model The huge population of the country made it a labour abundant economy with innumerable availability of unskilled cheap labours. According to the Heckscher–Ohlin model of international trade with the increase in the free international trade in a country, the return to the relatively abundant factor of the country will tend to increase more in comparison to the relative return to the scarce factor. Following this model the income of the unskilled labours of China would likely to increase with the increase in the country’s participation in international trade relatively more than the rest of the population of the country. Generally the income levels of the unskilled workers are lower and hence this assumption implies that their income is likely to increase with trade thus reducing the inequality of income in the economy. As the economy of the country adopted liberalisation policies and opened up its economy to the global market industrialization of the labour intensive sectors of the economy expanded that facilitates a large number of the country’s workers. A significant proportion of the population of the country have experienced a lift from the situation of poverty through the expansion of the country’s exports of mainly manufacturing goods that are labour intensive in nature. Poverty has been estimated to decrease dramatically for the economy of china mainly after liberalisation. Since there several other models defining the impact of international trade, the relation between the openness of the economy and inequality are ambiguous in the short run. (Wei and Wu, 2002, pp 5-6; Bardhan, 2007; NBER Working paper, 2001) Defining the relation using Gini Co-efficient The impact of globalization and hence increased trade on the inequality of the country become ambiguous to be proved with the models of international trade. Therefore focuses have been made by the economists to measure it directly with the measurement methods of income inequality mainly using the Gini Co-efficient. Gini Co-efficient with a value of zero implies equitable distribution with the inequality increasing with larger value of the same. Thus 100 representing the largest value of the Gini Co- efficient resemble a situation of complete inequality. With globalization there have been an unbelievable growth of the economy of the country, but the fruit of the economic growth have not been evenly shared by every class of the society. Though the living standards of the people of the country have experienced significant improvement, there have been declines in the level of the social welfare of the country. The Gini Co-efficient of China as measured by Riskin and Khan in the year 1998 proved to have increased to 45.2 for the year 1995 from 38.2, just before seven years. The Gini Co- efficient of China as measured by the World Bank in the year 1997 for a different data set using a different methodology estimated an increase to 38.8 for 1991 from 28.8 after three years of the ‘open door’ policy of the country. The value of the Gini Co-efficient increased further from 39.7 percent in the year 1999 to 41.7 percent within the span of a single year. The value of the Gini Co- efficient continued to increase with the highest value of 49.6 percent for the year 2006. The huge prevalence of the inequality of the country can be easily depicted from the values of the co–efficient. Thus these measures emphasize on the increase in the equality of the country due to globalization. (Wei and Wu, 2002, p 6 : Yue, 2010, p4 : Harrison, 2007, p627) Globalization and inequality The consequences of globalization are generally very unlikely and unintended. The conventional wisdom about globalization states that the interconnectedness of the country with world economy resulted in the decrease in the poverty rates of the country but with the consequent increase in the inequality of the country. Most of the developing countries of the worlds always are in the worry that the liberalization of the economy of the country and hence its openness to the international market have adverse effect on the poor population of the country with the poor becoming poorer and benefiting the rich section of the society by making them richer. Citing the example of China many economists have argued that the benefits of globalization of the country’s economy have been more disproportionately for the richer section of the society. (Wei, 2002; Wei & Wu, 2004) However according to some economists’ internal policies and change in the domestic population of the country often contributed to these changes. According to the Asian development bank the inequality of the country have exceeded that of the United States of America and soon likely to reach the countries of Brazil, Chile and Mexico with the highest income inequalities. (Collins and Graham, 2004, p.34) There are different forms of income inequalities that existed in the country that are particularly being reflected in three major aspects. They are the inter-industrial inequality, inter-provincial inequality and the perhaps the most important being rural urban inequality. The problems of the inequality of income are reflected on different aspects in each one of them. The ratios of the difference in the labour wages that existed in these different sectors are the measure of the existing inequality of income. Inter provincial inequality or often called regional inequality is measured by the ratio of the difference that existed in the wages of the labours belonging to different regions. This inequality has been experiencing a trend of considerable increase with globalization. Measured in respect to the income per capita of the eastern region to that of the west the ratio increased from 1.26 in 1990 to 1.39 in 1999. Difference in relative wage of the workers in different industries as measured by the inter-industrial inequality has also been reported of being expanded with significant changes in the relative wages in different industries. The rural –urban inequality of the country have also experienced a considerable increase with the ratio increasing from 2.57 in the year 1978 to 3.11 in 2002 with the increased globalization of the economy. (Yue, 2010, pp 5-6) Another study reveals that the inequality of the country is comparatively less in the coastal provinces of the country that are exposed more to the global markets than the regions of the interior China. However the study reveals that the inequality in income that exists between the urban areas and that of the rural areas of the country has been reduced by the impact of globalization of the economy. The inequality of income existing within the rural areas experienced a reduction with globalization associated with an increase of the same within the urban areas. (Bardhan, 2007) A survey was made by the international monetary fund for the period 1988 to 1993 to report the impact of globalization on the income inequality that exists between the rural areas of the country with that of the urban areas. The data studied clearly depicts the pattern that the cities of the country that have opened up more quickly to the global market have the experience of a faster decline of the inequality on an average basis. Refining the study considering the inequality of a city at the initial level and including the local GDP growth rate and the measure of the difference in the investment pattern of the rural areas of the country and that of the urban areas also proved the same pattern of the obtained result. (Wei, 2002) According to a study made by Wan in the year 2005 concluded that the integration of the economy of the country with the global one accelerated with the accession of the country to the World Trade Organization. This accession is aggravating the impact of the integration through globalization on the increasing inequality of the country mainly in income between regions. The study further concludes that a substantially positive share of inter regional income inequality is contributed by globalization and also its increase over time. (Wan, 2005; Fan, 2009, p121) Thus most likely it can be concluded that the liberalization of the economy and hence its integration with the global market whether resulted in the reduction of the overall inequality of the country is an issue of controversy among different economists. (Wei and Wu, 2002, p 6) Is globalization of the economy, the only factor affecting income inequality? The effects of the ongoing forces acting within the economy of the country are often difficult to be disentangled from the effects of globalization on the same. Progress of the economy through the technical advancement of the skill based sectors like the information technology as well as certain macroeconomic policies undertaken by the government of the country is often similar to that of globalization. Moreover changes in the demographic and structural patterns of the country also impose certain impact on the country’s economy. Moreover the structure of the capital of a country is also an important contributor to the increasing inequality among the different regions of the country. Privatization of the economy undertaken through policies of economic reform also has significant contribution to regional inequality of the country. However with the increasing globalization the impact of urbanization, educational policies and dependency ratio on the inequality of the country has been declining. (Bardhan, 2007; Wan, 2005; Wan, Lu and Chen, 2007, p35) Thus the literatures of the post globalization period of the country of China confirms that the inequality of the income of the country have more or less increased with the advent of the process of globalization in the economy since the year 1978 though in certain cases there are some internal factors associated with it. (Wu, 2006, p231) References 1. Bardhan, P, (October, 2007), Inequality in India and China: Is Globalization to Blame? Yale Global, available at http://yaleglobal.yale.edu/content/inequality-india-and-china-globalization-blame (accessed on 28th April, 2011) 2. Collins, S, M, and C, Graham, (2004), Globalization, Poverty and inequality, United States: Brookings institution press 3. Davies, K, (July, 2003), China’s economy: Still Way to go, OECD Observer No 238, available at http://www.oecdobserver.org/news/fullstory.php/aid/1016/China_92s_economy:_Still_some_way_to_go.html (accessed on 28th April, 2011) 4. Fan, S, (2009), Regional Inequality in China, United Kingdom: Taylor & Francis 5. Harrison, A, E, (2007) , Globalization and Poverty, Chicago: University Of Chicago 6. Kaplinsky, R, (2005), Globalization, Poverty and inequality, United Kingdom : Polity 7. Lozada, C, (n.d), Globalization Reduces Inequality in China, NBER, available at http://www.nber.org/digest/mar02/w8611.html (accessed on 28th April, 2011) 8. NBER Working paper, Wei S, J and Y, Wu, (2001), Globalization and Inequality: Evidence from Within China, available at http://www.nber.org/papers/w8611 (accessed on 28th April, 2011) 9. Wei, S.J. & Y. Wu. (2004) Globalization & Inequality: Intra-China Evidence, Working Paper, SSRN, available at http://papers.ssrn.com/sol3/papers.cfm?abstract_id=603541 (accessed on 28th April, 2011) 10. Wan, G, (2005), Globalization and Regional Income Inequality-Evidence from within China, Working Papers UNU-WIDER Research Paper , World Institute for Development Economic Research (UNU-WIDER). Available at http://ideas.repec.org/p/wpa/wuwpem/0511014.html (accessed on 28th April, 2011) 11. Wan, G, Lu, M, and Z, Chen, (March, 2007) GLOBALIZATION AND REGIONAL INCOME INEQUALITY: EMPIRICAL EVIDENCE FROM WITHIN CHINA, Review of Income and wealth, Vol-53, no 1, pp. 35-60 , available at http://profluming.com/english/UploadFiles_2370/200703/20070302221441850.pdf (accessed on 28th April, 2011) 12. Wei S, J and Y, Wu, (2002) Globalization and Inequality without Differences in Data Definition, Legal System and Other Institutions, International Monetary Fund, pp 1-54, available at http://www.brookings.edu/~/media/Files/rc/papers/2002/1024globaleconomics_wei/20021024.pdf (accessed on 28th April, 2011) 13. Wei, S, J (September, 2002), Is Globalization Good for the Poor in China?, Finance and Development, Vol -39, No-3, International Monetary fund, available at http://www.imf.org/external/pubs/ft/fandd/2002/09/wei.htm (accessed on 28th April, 2011) 14. Wu, Y, (2006), Economic growth, transition, and globalization in China, United Kingdom: Edward Elgar publishing 15. Yue, L, (2010) , Globalization and Inequality in china, pp 1-13, available at http://www.u-shimane.ac.jp/36near/41kenkyu/file/18-19_01_li.pdf (accessed on 28th April, 2011) Chapter 3 Methodology In the empirical investigation part, a set of variables will be employed to reflect the impacts of economic globalization on income inequality in China from 1978 to 2007. Data Resources The data using in the model are collected from different sources across the period I investigate which are displayed in Table 1. The Gini Coefficients are from WIID (World Income Inequality Database) and the working paper of Chen, Dai, Pu, Hou and Feng(2010). The GDP per capita growth rate, FDI inflow of GDP and total trade of GDP are taken from World Bank. The data of percentage of government expenditure on social security are obtained from China Statistical Yearbook. Model The model used to investigate the relationship between economics globalization and inequality in this study is based on the previous empirical literature. The cointegration model shown below is suggested by Wang, Tian and Dayanandan (2008) which is specified as: where Gini is the annual Gini coeffieient of the country. CYG is defined as the growth rate of GDP per capita. FDI is expressed as the percentage of Foreign Direct Investment over GDP. Trade is denoted the proportion of total trade of GDP. GX is the percentage of government spending on social insurance. Date description There are four independent variables in this model to explain the dependent variable, GINI coefficient which are the following: Gini Coefficient Gini coefficient is widely used to measure the economic inequality in a society. The range is from 0 to 1, the higher the value, the higher the inequality. The Gini Coefficient higher than 0.4 indicating the income distribution is highly uneven. In this study, the Gini Ratio provides important information for evaluating the income disparity in China. China does not officially publish the yearly GINI coefficient. Thus, many scholars such as Zhao, Li and Riskin (1999), Cheng (2007) and Ravallion and Chen (2007) have calculated the Gini index for China. However, Chen and Hou (2008) point out that the estimations of Gini Coefficients from different studies are contradictory as they believe many calculations disregard the unequal income between each decomposed income groups which made the Gini data less accurate. To overcome the shortcomings on these current studies, Chen, Dai, Pu, Hou and Feng(2010) decompose the Gini coefficient from different parts such as rural and urban, and then compute the latest national Gini coefficient. From the estimation of Chen, Dai, Pu, Hou and Feng(2010), we can see the Gini ratio has been increased from 0.3029 in 1978 to 0.4448 in 2007. Total Trade % of GDP Total trade is an important factor to include in the model to investigate the impact on economic inequality and it is also considered as one of the proxies of globalization. The ratio of trade of GDP has dramatically increased from 14% in 1978 to 70% in 2007. Heckscher-Ohlin theorem states that if a developing country like China has labour-abundant resource, it exports large amount of this kind of products such as agricultural products. As a result, increased the trade between countries results in higher productivity of the low-skilled workers and their wages would therefore increased. The theorem suggests that trade liberalization could help to reduce the income inequality as the income gap between rich and poor becomes narrower. Wade (2004) supports the theorem and discovers that liberalization of trade causes the increase in trade ( % of GDP) which has benefited the economy in China and influence the inequality to halt. However, Henderson (2002) argues that the trade is only focused in the coastal region, and therefore, the income gap between coastal and interior areas in China would be widened, thus the inequality would increase because of uneven trade. In this case, the sign of the coefficient is expected to be positive (negative) if the trade/GDP worsens (mitigates) the inequality in China. Foreign Direct Investment of GDP FDI is representing the percentage of net inflow of the foreign investment over GDP. Since the economic reform of China in 1978, the FDI (% of GDP) has increased steadily. FDI is considered as one of the key elements of economic reform as the “Open-door” policy attracts a substantial amount of foreign capital inflow. Velde (2003) suggests that there is a close relationship between FDI and economic growth as the increasing FDI in China is caused by the globalization wave. Theoretically, increased in FDI should drive the growth in income because the inward foreign investment could help to improve the productivity, technology and capital accumulation, etc. As a result, the income inequality should be reduced. However, Feenstra and Hanson (1997) argue that the FDI would raise the income inequality because the increasing FDI only benefits the skilled employees rather than the unskilled workers. Consequently, the former would get higher wages, which would widen the income inequality within the country. Statistical evidence from Choi (2006) proves that the increased in FDI would bring the negative impact to national inequality based on 119 countries 1993 to 2002 panel data. Chintrakarn, Herzer and Nunnenkamp (2010) examine that there is no short run relationship between FDI and inequality, but the FDI could deteriorate the income inequality in the long run. Therefore, the sign of FDI is expected to be positive in the model according to the studies above. Government Spending on Social insurance This independent variable is the political factor in the model. The proportion of government spending on social insurance of GDP reveals the welfare spending from government. The rate remained steady and stay between 1.5% and 2%. It started to increase up to 10.9% in 2007 since 1996. According to the Keynesian theory, increases the government expenditure would stimulate the national economy through the multiplier effect. Subsequently, the overall income increase and the living standard should then be improved. Stack (1978) supports the theory and points out the greater the direct government involvement could lead less degree of income inequality. As a result, the government spending on social insurance should have a positive impact on the income inequality. Nevertheless, Tian, Wang and Dayanandan (2008) contend that although the Chinese government has introduced some policies and raised the welfare expenditure to help the less development regions in order to reduce the inequality, the evidences prove that the policies are not too effective to solve the problem of regional disparity. The sign of the GX is expected to be positive if the government policy in China has no effect on easing the inequality in this case. Otherwise, the sign is negative. Growth rate of GDP per capita Neutel and Heshmati (2006) find that there is a positive relationship between growth of GDP per capita and globalization index. The GDP per capita growth rate is considered as the main indicator of economic growth. The trend of the China GDP per capita growth rate has been fluctuating from year 1978-2000. In 2001, the movement of growth rate steadily increases from 7.52% to 12.41% in 2007 as China joined the World Trade Organization in 2001; the economic performance has been improved since then. According to the table 2, the ratio of GDP per capita between coastal and inland area is 1.699 in the year of 1986-1991. The ratio has risen to 2.051 in 1992-1998. The figure indicates that although the overall GDP per capita has grown rapidly in the past decades, the income gap between coastal and inland region has widened. The growth rate of GDP per capita may deteriorate the inequality level. On the other hand……………………………………………………………………………………… ………………………………………………………………………………………………….. Assumptions and their validity The model is explained by the OLS estimators. OLS is relied on a set of assumptions. In order to use this method to estimate the regression, the certain assumptions must to be satisfied. The OLS is not BLUE unless these assumptions are fulfilled which are the following: 1) The regression model is liner in the parameters 2) The random error has an expected value of zero 3) The independent variables has no multicollinearity 4) Errors have constant variance 5) The disturbances are not autocorrelated 6) The error is normally distributed In the following part, the assumptions will be explained and examined before applying the OLS regression. Some tests will be taken to ensure whether all the assumptions are hold in this model. Wooldridge (2008) states if the regression is under assumptions 1 to 3, the model is unbiased. The OLS is a BLUE under assumptions 1 to 5. If the all the assumptions from 1 to 6 are satisfied, we could rely on the statistical inference for any sample size. 1) Linear in parameters This is the basic assumption of the OLS model which states the parameters of the independent need to be linear. The OLS model in this study does not violate this rule. 2) Zero conditional Mean nder this assumption, the expected mean of error, given any explanatory variables needs to be zero. The disturbance term is required to be uncorrelated to any explanatory variables. In this study, most independent variables are not considered as strictly exogenous which means the error is not independent to the variables. It is because the macroeconomic and political variables in our model are relatively affected by what have happened to the outcome variable in the past. Therefore, the OLS may lead to be bias if this assumption did not hold. 3) Multicollinearity One of the assumptions of OLS is the independent variables are linearly independent. Multicollinearity occurs if variables are highly correlated to each other. Multicollinearity would not distort the regression results, but the estimation of one of the explanatory variables would become less precise and the variance would tend to be large. The exact statistical inference would be more difficult to perform under this problem. The correlation matrix is introduced to detect the problem. The result is shown in Table 3 to represent that the independent variables have no mulitcollinearity. 4) Constant variance ( No Heteroskedasticity ) In the OLS regression, the error should have the constant variance given any value of the explanatory variable, which is  . The error does not have the same variance is known as Heteroskedasticity. Although heteroskedasticity does not cause bias or inconsistency estimates, the minimum variance is a no longer valid even using large sample size. The White test is conducted to detect heteroskedasticity which is shown in Table 4. Since the critical  value of 5% significance level is 9.488 which is less than the  statistic of 24.437, we reject the null hypothesis of no heteroskedasticity. The result indicates the OLS is not a BLUE in the presence of non-constant variance. 5) Error autocorrelation The  is the time series error or disturbance in the model. The error correlated to the others across observations is known as error autocorrelation. Under error autocorrelation, the variance would be underestimated which would lead the result of variance to be bias. The t-value would consequently become less efficient and reliable. Therefore, it is subtle to detect the autocorrelation. Durbin-Watson test is used to test the problem in this regression. If the DW value is close to 2, the error has no relationship with each other in the observations. The DW statistic shown in Table 5 is 0.69 which means it has a quite strong positive correlation between the disturbances. In this case, OLS estimator is not a BLUE and the probability of causing bias would increase. Such regression may be considered as spurious regression. Further evaluation of this problem will be discussed in the “Result and Discussion” part. 6) Normality The last assumption of using OLS estimators is that the population error is normally distributed as Normal (0, ). Jarque-Bera test is used to test the normality, the null hypothesis is the data of samples are following the normal distribution. The descriptive statistics in the Table 6 present that the skewness and kurtosis are 0.68 and 2.62 respectively. Under the normal distribution, skewness is expected to be zero and kurtosis is three. Therefore, the samples are not normally distributed. Consequently, the statistical results of the OLS model could be less accurate. As we can see from the above tests, some of the assumptions are not satisfied to produce a BLU estimator. Subsequently, continue using this model need to be cautioned as it would raise the probability of biasness and inconsistency of the regression result. Chapter 4 Empirical Results Tests for time-series data It is important to know whether a time series is stationary before analyzing the regression result of the model illustrated in Table 5. The main reason of having the stationary test is that we could be easily trapped in a situation on obtaining the likely significant results from the uncorrelated data when the series is not stationary. This type of problem caused by non-stationary time-series is known as spurious correlation. The typical symptoms of spurious correlation is the regression would overestimate the , -statistics and F-value. In order to see whether the series are stationary, the residuals plot illustrated below gives us the crude message. The residuals graph below do not tend to show a systematic pattern as they appear to wander slowly upward or downward. This is a characteristic of nonstationary series. Augment Dicky-Fuller Test Although the graph indicates the nature of the series, it is desirable to have a econometrical stationary test to detect whether the variables have unit roots. One of the common methods to implement is to conduct a Augmented Dicky-Fuller test for each of the series. The null hypothesis of this test is the series contain unit roots. Failure to reject the hypothesis means the regression is stationary. The ADF test result is shown in Table 7. The ADF test of all variables cannot reject the hypothesis of a unit root in level except the growth rate of GDP per capita. This outcome demonstrates that most of the series in the regression are non-stationary. According to the econometrics book written by Hill, Griffiths and Lim (2008), series can be made stationary by taking the first-difference. Refer to table 3, the ADF test is run again to test the unit root in first difference instead of in level. The null hypothesis of a unit root is rejected proves that the series become stationary in first difference form. All sequences have one unit root and is said to be integrated of order one I (1) (except CYG is I (0)). Johansen Cointegration Test The regression result is compromised if the variables have unit root unless they are cointegrated. Variables are cointegrated if there is a long-run relationship between them. them. When the linear combination of the variables is stationary, they move together in long term relationship even the series are non-stationary. The Johansen cointegration test is introduced to test whether the variables are cointegrated. Enders (2009) mentions the Johansen tests in Eviews would be able to detect the different degree of integration, however, it is desirable to only include the series with the same orders of integration. Therefore, I conduct the Johansen test by using the variables of order one I (1) (without CYG). The result is depicted in Table 8 and reflects that both trace and Max-eigenvalue indicates no cointegration at the 5% level. Granger and Newbolds (1974) have pointed out that the regression may be meaningless if the variables are non-stationary and not cointegrated. The OLS regression therefore still contains the spurious correlation problem, implying the model is an incorrect OLS estimation of this regression, which can yield the results completely wrong. Hill, Griffiths and Lim (2008) states that if the variables have unit and not cointegrated, the model needs to be changed to first differences form, as the specification shown below. The first difference equation is summarized as  where  denotes the change from T to T+1 Data analysis Since the original model is proved to be invalid through the above tests, the high value of  in the regression result ( Table 5) does not represent that the model fits the data well. The model is then estimated in first difference. Table 9 shows the regression results of the first difference model. Before actually looking at the statistical inference, the OLS assumptions are needed to go through again. Firstly, the sample data has passed the White test implying there is no heteroskedasticity of the regression, refer to the table 10. Secondly, the Durbin Watson test is now 1.64 which is relatively close to 2, indicating there is no positive or negative correlation between disturbance terms. Thirdly, the correlation matrix in Table 11 states this model has no perfect multicollinearity too. All the assumptions tend to be fulfilled in this model except the normality of the disturbance term. The Jarque-Bera test result of normality is graphed on figure 12 showing the sample is not normally distributed. The model in first difference is more likely to be a BLUE and fit the data better. However, in the first difference model, the is very low and all the variables are insignificant as shown in table 5. In order to examine why the variables are not significant, I construct the Actual Fitted Residual Graph below to see if there is any outliner exists. The graph exhibits that the data from 1981 and 2004 are abnormal. These two values differ substantially from others which might distort the regression results. Therefore, the data from these two years are likely to be removed in order to have a more accurate investigation. The results of the new regression with the adjusted sample data are presented in Table 13. Apparently, the new model has a big improvement after omitting the outliers. The outliers seem to mislead the results from the previous model. Compare to the model with and without the outliers, the latter is more suitable to use to investigate the relationship between inequality and economic globalization because it exhibits a better empirical result as the following: The first-difference model Adjusted  Sum Squared Resid F-statistic Prob(F-statistic) Durbin-Watson stat 0.137675 -0.006045 56.19682 0.957936 0.448374 1.639126 The adjusted model with the omitted data Adjusted  Sum Squared Resid F-statistic Prob(F-statistic) Durbin-Watson stat 0.523411 0.423077 20.36816 5.1666 0.005230 2.233586  is a statistical measurement of the proportion of variation in the dependent variables explained by the independent variables. It is considered as an indicator on how well the model can predict the sample data.  is measured between 0 and 1. The higher the , the better the model fits the data. The  from the result is 0.523. It represents that there is 52.3% of the data can be explained by the regression. This figure is relatively not very high, but it is much better than the previous model. Another similar technique to define the “goodness of fit” is the sum squared residual. SSR is the sum of squared errors of prediction which is used to investigate the total variation between the data and the regression. A smaller value of SSR demonstrates that model has smaller random error and is more successful to predict the data. The SSR in the adjusted model is 20.3816 which is nearly three times less than the first difference model. These two statistical measures express that the model with the omitted data has provided a better observation compared to the previous one. F-statistic is used to determine the overall significance of the regression. The null hypothesis is that the values of all parameters are zero, indicating the independent variables have no influence on dependent variables. Failure to reject the null hypothesis would result that the economic growth variables do not have significant effect upon the inequality. Selecting the 99% significance level, the 𝝆-value of F-statistic is 0.005 which is less than 0.01. As a result, it is the evidence to reject , representing that all of the explanatory variables help to explain the dependent variable GINI in this case. The Durbin-Watson Statistic of this new model is 2.23 which is very close 2 and reveals that there is no autocorrelation between errors. The more adequate statistical inference will be generated in this model rather than the one without omitted the variables. The most important improvement of this model is that all the variables became significant at 90% significance level. Therefore, the interpretation of the results is likely to rely on the latter regression. Results Interpretation Independent Variables Coefficient Std. Error t-Statistic constant 0.044334 0.267656 0.165636 D(Trade) 0.246552** 0.070521 3.496156 D(FDI) 0.689340* 0.376330 1.831744 D(GX) 0.495147* 0.265721 1.863411 D(CYG) -0.156132* 0.076027 -2.053654 **, * significant at 0.05, 0.1 critical value respectively The total trade of GDP is significant at 95% level. The high t-statistic and low standard error indicating the coefficient can be estimated accurately. This variable has a significant influence on the dependent variable. The coefficient is 0.246 representing 1% increasing total trade of GDP would increase the GINI coefficient by 0.246. The result expresses that the substantial increase of trade in the past 30 years has not helped to ease the economic inequality instead it has worsened the problem. This result is inconsistent with the Heckscher-Ohlin analysis. On the other hand, Chan and Kulkarni (2006) propose that the problem of inequality within China is likely to be caused by increasing amount of trade in past three decades, based on the Kuznets inverted U curve hypothesis which states that the economic inequality increases while the country is in the process of developing as the most wealth is hold in the entrepreneur’s hands. The income distribution would then be improved after the economy reaches to a certain level because the wages increase gradually. Although the economy in China is growing very fast, it is still on the early stage of development. If this is the case, the increased in total trade does not seem to decrease the inequality within the country. The variable FDI is significant at 90% level. Although 90% significance level is not too rigorous, the range is still acceptable. The result suggests that the FDI has positive impact on the GINI coefficient as expected. 1% increased in the percentage of FDI in GDP results in 0.689 increases in the GINI coefficient. The result supports the suggestion from Feenstra and Hanson (1997) that the FDI could not reduce the inequality level because of the income disparity between skilled and unskilled workers. In addition, although China is now the largest recipient country of FDI in the world, the inflow investments from foreign countries/investors are not even distributed geographically. Ali and Guo (2005) express that about 90% of total FDI is highly concentrated in the eastern and southern regions. One of the main reasons causing inequality in China is the imbalance development among the country, announced by Wang, Tian, Dayanandan (2008). Therefore, it has been argued that the rapid increased in the volume of FDI may consider as one of the main reason to cause the regional inequality. Thus the overall economic inequality raises. GX is also significant at 90% level. The coefficient is 0.495 indicating 1% increased in the share of government spending on social insurance could bring the GINI coefficient to go up 0.945. The positive relationship between this variable and the independent variables denotes that fiscal policy on social insurance in China did not contribute to reduce the inequality between regions. The outcome violates the findings of Stack (1978). In fact, in order to allocate the social insurance expenditure efficiently and fairly into to various provinces, China government should base on the proportion of national population of different provinces. However, the social insurance does not contribute base on this rule. Refer to Table 14, we can discover that the percentage of social insurance assigned into different regions according to the population is not equal. The proportion of population of Bejing and Shanghai are only 1.2% and 1.4% respectively, but these two cities have occupied 6.4% of overall social expenditure, whereas the population of Guizhou and Guongxi are 2.9% and 3.6%, they merely have 1.3% and 2.1% of government social insurance respectively. Theoretically, government increases the welfare expenditure which should bring an efficient effect on improving the living standard of citizens. Nevertheless, if government did not implement the social insurance policy efficiently, the regional economic disparity is hardly to be improved. The coefficient of CYG is -0.156 at 90% significant level. The growth rate of GDP per capita has negative relationship with economic inequality representing 1% increased in GDP per capita growth rate can mitigate the GINI coefficient by 1. ……………………………. Although Neutel and Heshmati (2006) point out that growth of the GDP per capita is a good indicator on economic growth as it has a positive correlation with globalization, Chan and Kulkarni (2006) argue that GDP per capita only expresses the small picture, it does not demonstrate the whole distribution of the population. Therefore, we need to be caution when using the GDP per capita growth rate to measure the relationship as it may not tell us the truth whether increased in GDP per capita growth rate indicates the reduced in inequality. Chapter 5 Conclusion Chapter 6 Bibliography 1. Chen, J., Dai D., Pu, M., Hou, W and Feng, Q. (2010) The trend of the Gini coefficient of China, The University of Manchester Brooks World Poverty Institute 2. Tian, X., Wang, B. and Dayanandan, A. (2008) The impact of Economic Gobalization on Income Distribution: Empirical Evidence in China, Economics Bulletin, Vol.4, No. 35 pp.1-8 3. Choi, C. (2006) Does foreign direct investment affect domestic income inequality? Applied Economics Letters, 13, 811-814 4. Chan, K, Y. and Kulkarni, K, G. (2006) A Test of the Kuznets Inverted U Hypothesis: Income Inequality Behind the Rapid Economic Growth in China, Indian Journal of Economic and Business 5. Zhao, R., Li, S. and Riskin, C. (1999) A Re-study on Income Distribution of 6. Chinese Residents, China Financial and Economic Publishing House, Beijing. 7. Cheng, Y. (2007) Overall Gini coefficient of China and its decomposition by 8. rural and urban areas since reform and opening-up, Social Sciences in China, 4, 45-60. 9. Ravallion, M. and Chen, S. (2007) China’s (uneven) progress against poverty, Journal of Development Economics, 82, 1–42. 10. Chen, J. and Hou, W. (2008) New Approach to Estimate the Chinese Gini Coefficients. [online] Chinese Economy Association (UK), Available from: 11. http://www.ceauk.org.uk/2008-conference-papers/Chen-Jiandong-Wenxuan-Hou-Estimate-the-Chinese-Gini-Coefficients.pdf [Accessed 06 March 2011]. 12. WADE, R, H. (2004) On the Causes of Increasing World Poverty and Inequality, or Why the Matthew Effect Prevails, New Political Economy, Vol. 9, No. 2 13. Velde, D, W. (2003) Foreign Direct Investmentand Income Inequality in Latin America. Experiecne and policy implications, Overseas Development Institute 14. Reenstra, R, C. and Hanson, G, H. (1997) Foreign direct investment and relative wages: Evidence from Mexico’s maquiladoras, Journal of International Economics 42, 371-393 15. Chinatrakarn, P., Herzer, D. and Nunnenkamp P. (2001) FDI and Income Inequality: Evidence from a Panel of US States, Working paper, Kiel Institute for the World Economy, No1579 16. Stack, S. (1978) The Effect of Direct Government Involvement in the Economy on the Degree of IncomeInequality: A Cross-National Study, American Sociological Review, Vol. 43, No. 6 pp. 880-888 17. Neutel, M. and Heshmati, A. (2006) Globalisation, Inequality and Poverty Relationship: A Cross Country Evidence, Discussion Paper, The Institute for the Study of Labour (IZA), No. 2223 18. Wooldridge, J. (2008) Introduction to Econometrics: A Modern Approach, Fourth Edition, Cengage Learning EMEA, 19. Hill, R. C., Griffiths, W, E. and Lim, G, C. (2008) Principle of Econometrics, Third Edition, John Wiley & Sons Inc 20. Enders, W. (2009) Applied Econometric Times Series - Wiley Series in Probability and Statistics, Third Edition, John Wiley & Sons Inc 21. Granger, C.W. J. and Newbold, P. (1974) Spurious Regression in Econometrics, Journal of Econometrics 2, 111-120 22. Ali, S. and Guo, W. (2005) Determinants of FDI in China, Journal of Global Business and Technology, Volumn 1, No 2 23. World Bank (2002) Globalization, growth and poverty, Building an Inclusive World Economy, Oxford, UK, Oxford University Press 24. Dollar, D. and Kraay, A. (2001) Trade, Growth, and Poverty, Working Paper, World Bank Policy Research Working Paper, No. 2041 25. Mohamad, M. (1996 ) Globalization: What it means to small nations? Text of speech 26. delivered at the Inaugural Lecture of the Prime Ministers of Malaysia Fellowship Exchange Programme, Kuala Lumpur 27. Henderson, V. (2002) Urbanization in China, Notes for North Holland Volume, Paper with Fujitam, Kanemoto, and Mori on China and Janpan 28. Zhang and Zhang (2003) How does Globalization Affect Regional Inequality within A Developing Country? Evidence From China, The Journal of Development Studies, Vol. 39, No. 4, pp 47-67 Appendix Figure 1 Source: World Bank Figure 2 Source: World Bank Figure 3 Source: World Income Inequality Database and the working paper of Chen, Dai, Pu, Hou and Feng(2010) Table 1 Year Trade CYG FDI GINI GX 1978 13.67 10.22 0 30.29 1.685248064 1979 17.96 6.17 3.04E-05 30.73 1.55485688 1980 21.66 6.46 0.018596 31.44 1.631633342 1981 24.64 3.86 0.090182 27.98 1.611018877 1982 22.37 7.51 0.14558 25.91 1.533358266 1983 20.73 9.31 0.291129 26.02 1.584227255 1984 22.62 13.7 0.447137 26.89 1.549070558 1985 23.99 11.96 0.632841 26.45 1.554197331 1986 26.33 7.19 0.737225 29.2 1.613671306 1987 32.48 9.82 0.701385 28.9 1.630727882 1988 35.22 9.52 0.772468 29.5 1.640166827 1989 34.25 2.52 0.737864 31.78 1.634688255 1990 34.6 2.29 0.862091 31.55 1.784932498 1991 38.01 7.72 1.029513 33.1 1.987822667 1992 42.81 12.81 2.202122 34.24 1.775693442 1993 48.68 12.7 4.292045 36.74 1.621394567 1994 47.31 11.83 5.795298 37.6 1.642434684 1995 43.94 9.7 4.956738 36.53 1.692038946 1996 38.06 8.85 4.677676 35.05 2.301465818 1997 39.01 8.19 4.594409 35 3.556807992 1998 36.39 6.77 4.349674 35.37 5.516022144 1999 37.97 6.67 3.662753 36.37 9.079996694 2000 44.24 7.55 3.413277 38.49 9.552576087 2001 43.08 7.52 3.560617 39.45 10.51390868 2002 47.7 8.37 3.627331 43.21 11.95393855 2003 56.91 9.32 3.246805 44.67 10.77450461 2004 65.35 9.45 3.130903 43.94 10.93864581 2005 69.28 9.75 3.14437 43.97 10.90135419 2006 72.03 10.98 2.615769 44.48 10.79041421 2007 70.28 12.41 2.413701 46.9 10.90175337 Table 2 　 GDP per capita ( in Chinese Yuan) 　 1986-1991 1992-1998 Coastal areas 1449 3055 Inland areas 853 1489 Note: GDP per capita is expressed in 1986 constant prices Sources: Calculated by Zhang, Zhang (2010) Table Correlation Matrix FDI TRADE GX CYG FDI  1.000000  0.609908  0.432342  0.252358 TRADE  0.609908  1.000000  0.762750  0.306464 GX  0.432342  0.762750  1.000000  0.090223 CYG  0.252358  0.306464  0.090223  1.000000 Table 4 Heteroskedasticity Test: White F-statistic 4.706822     Prob. F(14,15) 0.0026 Obs*R-squared 24.43727     Prob. Chi-Square(14) 0.0405 Scaled explained SS 13.72607     Prob. Chi-Square(14) 0.4703 Test Equation: Dependent Variable: RESID^2 Method: Least Squares Date: 04/19/11 Time: 01:36 Sample: 1978 2007 Included observations: 30 Variable Coefficient Std. Error t-Statistic Prob.   C 20.23897 13.24090 1.528520 0.1472 FDI -7.444684 3.831171 -1.943188 0.0710 FDI^2 -0.034537 0.510577 -0.067644 0.9470 FDI*TRADE 0.207644 0.160818 1.291171 0.2162 FDI*CYG -0.347100 0.625566 -0.554859 0.5872 FDI*GX 0.798109 0.751675 1.061775 0.3051 TRADE -0.937472 0.439478 -2.133146 0.0498 TRADE^2 0.016271 0.010222 1.591770 0.1323 TRADE*CYG -0.005896 0.048227 -0.122257 0.9043 TRADE*GX -0.166122 0.059072 -2.812202 0.0131 CYG 0.772832 1.673790 0.461726 0.6509 CYG^2 -0.082171 0.062086 -1.323487 0.2055 CYG*GX 0.611735 0.236289 2.588930 0.0205 GX -1.057121 3.973053 -0.266073 0.7938 GX^2 0.053752 0.211709 0.253894 0.8030 R-squared 0.814576     Mean dependent var 2.871919 Adjusted R-squared 0.641513     S.D. dependent var 3.715153 S.E. of regression 2.224403     Akaike info criterion 4.743707 Sum squared resid 74.21955     Schwarz criterion 5.444306 Log likelihood -56.15561     Hannan-Quinn criter. 4.967835 F-statistic 4.706822     Durbin-Watson stat 2.165307 Prob(F-statistic) 0.002560 Table 5 Dependent Variable: GINI Method: Least Squares Date: 04/17/11 Time: 11:47 Sample: 1978 2007 Included observations: 30 Variable Coefficient Std. Error t-Statistic Prob.   C 23.20907 1.274566 18.20938 0.0000 FDI 0.648742 0.240274 2.700006 0.0123 TRADE 0.202611 0.040183 5.042227 0.0000 GX 0.596608 0.135725 4.395702 0.0002 CYG -0.059481 0.131279 -0.453089 0.6544 R-squared 0.923667     Mean dependent var 34.72500 Adjusted R-squared 0.911454     S.D. dependent var 6.238670 S.E. of regression 1.856422     Akaike info criterion 4.226191 Sum squared resid 86.15756     Schwarz criterion 4.459724 Log likelihood -58.39286     Hannan-Quinn criter. 4.300900 F-statistic 75.62827     Durbin-Watson stat 0.690146 Prob(F-statistic) 0.000000 Table 6 Table 7 Augmented Dickey-Fuller Test Variable IN LEVEL Augmented Dickey-Fuller Test Statistic Critical Values – 10% level t-statistic Prob. GINI -2.622989  0.705352  0.9902 TRADE -2.625121 -0.648551 0.8438 FDI -2.625121 -1.865623 0.3428 CYG -2.629906 -3.629345 0.0121 GX -2.625121 -0.580568 0.8597 Augmented Dickey-Fuller Test Variable IN 1st DIFFERENCE Augmented Dickey-Fuller Test Statistic Critical Values – 10% level t-statistic Prob. GINI -2.625121 -3.804905 0.0076 TRADE -2.625121 -3.164764 0.0331 FDI -2.625121 -2.838152 0.0658 GX -2.625121 -2.762537 0.0766 Table 8 Unrestricted Cointegration Rank Test (Trace) Hypothesized Trace 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None  0.463175  37.42886  47.85613  0.3276 At most 1  0.369742  20.01056  29.79707  0.4222 At most 2  0.221813  7.085011  15.49471  0.5678 At most 3  0.002245  0.062941  3.841466  0.8019  Trace test indicates no cointegration at the 0.05 level  * denotes rejection of the hypothesis at the 0.05 level  **MacKinnon-Haug-Michelis (1999) p-values Unrestricted Cointegration Rank Test (Maximum Eigenvalue) Hypothesized Max-Eigen 0.05 No. of CE(s) Eigenvalue Statistic Critical Value Prob.** None  0.463175  17.41831  27.58434  0.5442 At most 1  0.369742  12.92555  21.13162  0.4591 At most 2  0.221813  7.022070  14.26460  0.4865 At most 3  0.002245  0.062941  3.841466  0.8019  Max-eigenvalue test indicates no cointegration at the 0.05 level  * denotes rejection of the hypothesis at the 0.05 level  **MacKinnon-Haug-Michelis (1999) p-values Table 9 Dependent Variable: D(GINI) Method: Least Squares Date: 04/17/11 Time: 11:48 Sample (adjusted): 1979 2007 Included observations: 29 after adjustments Variable Coefficient Std. Error t-Statistic Prob.   C 0.194411 0.367119 0.529559 0.6013 D(TRADE) 0.120030 0.083556 1.436523 0.1638 D(FDI) 0.645608 0.525811 1.227833 0.2314 D(GX) 0.294993 0.401777 0.734222 0.4699 D(CYG) -0.045616 0.108412 -0.420768 0.6777 R-squared 0.137675     Mean dependent var 0.572759 Adjusted R-squared -0.006045     S.D. dependent var 1.525603 S.E. of regression 1.530207     Akaike info criterion 3.844269 Sum squared resid 56.19682     Schwarz criterion 4.080010 Log likelihood -50.74190     Hannan-Quinn criter. 3.918100 F-statistic 0.957936     Durbin-Watson stat 1.639126 Prob(F-statistic) 0.448374 Table 10 Heteroskedasticity Test: White F-statistic 0.242153     Prob. F(14,14) 0.9940 Obs*R-squared 5.653445     Prob. Chi-Square(14) 0.9745 Scaled explained SS 6.765831     Prob. Chi-Square(14) 0.9434 Test Equation: Dependent Variable: RESID^2 Method: Least Squares Date: 04/19/11 Time: 22:57 Sample: 1979 2007 Included observations: 29 Variable Coefficient Std. Error t-Statistic Prob.   C 3.016034 1.884786 1.600200 0.1319 D(FDI) 1.804372 3.950966 0.456691 0.6549 (D(FDI))^2 -2.890492 3.217353 -0.898407 0.3842 (D(FDI))*(D(TRADE)) 0.431186 0.904466 0.476730 0.6409 (D(FDI))*(D(GX)) -3.449061 21.11841 -0.163320 0.8726 (D(FDI))*(D(CYG)) -0.519779 1.281385 -0.405639 0.6911 D(TRADE) -0.033885 0.602952 -0.056199 0.9560 (D(TRADE))^2 0.008571 0.110791 0.077363 0.9394 (D(TRADE))*(D(GX)) 0.361648 1.259922 0.287040 0.7783 (D(TRADE))*(D(CYG)) -0.047552 0.183681 -0.258882 0.7995 D(GX) -1.603911 7.334753 -0.218673 0.8301 (D(GX))^2 -0.379858 5.731851 -0.066271 0.9481 (D(GX))*(D(CYG)) -0.493292 2.445534 -0.201711 0.8430 D(CYG) -0.211916 0.489920 -0.432552 0.6719 (D(CYG))^2 -0.005286 0.112867 -0.046837 0.9633 R-squared 0.194946     Mean dependent var 1.937821 Adjusted R-squared -0.610107     S.D. dependent var 3.686714 S.E. of regression 4.678072     Akaike info criterion 6.229893 Sum squared resid 306.3810     Schwarz criterion 6.937115 Log likelihood -75.33345     Hannan-Quinn criter. 6.451386 F-statistic 0.242153     Durbin-Watson stat 1.970259 Prob(F-statistic) 0.993958 Table 11 Correlation matrix D(CYG) D(GX) D(TRADE) D(FDI) D(CYG)  1.000000 -0.081549  0.205862  0.140969 D(GX) -0.081549  1.000000 -0.214542 -0.349949 D(TRADE)  0.205862 -0.214542  1.000000  0.157196 D(FDI)  0.140969 -0.349949  0.157196  1.000000 Table 12 Table 13 Dependent Variable: D(GINI) Method: Least Squares Date: 04/20/11 Time: 02:18 Sample (adjusted): 1980 2007 Included observations: 24 after adjustments Variable Coefficient Std. Error t-Statistic Prob.   C 0.044334 0.267656 0.165636 0.8702 D(TRADE) 0.246552 0.070521 3.496156 0.0024 D(FDI) 0.689340 0.376330 1.831744 0.0827 D(GX) 0.495147 0.265721 1.863411 0.0779 D(CYG) -0.156132 0.076027 -2.053654 0.0540 R-squared 0.523411     Mean dependent var 0.675417 Adjusted R-squared 0.423077     S.D. dependent var 1.363139 S.E. of regression 1.035378     Akaike info criterion 3.090463 Sum squared resid 20.36816     Schwarz criterion 3.335891 Log likelihood -32.08555     Hannan-Quinn criter. 3.155575 F-statistic 5.216666     Durbin-Watson stat 2.233586 Prob(F-statistic) 0.005230 Table 14 Population(10000) % of national population Government expenditure on Social insurance (0.1 billion Yuan) % of total Government expenditure on Social insurance Provinces Beijing Tianjin Hebei Shanxi Neimenggu Liaoning Jilin Heilongjiang Shanghai Jiangshu Zhengjiang Anhui Fujian Jiangxi Shandong Henan Hubei Hunan Guangdong Guangxi Hainan Chongqin Sichuan Guizhou Yunnan Xizang Shanxi Gansu Qinghai Nixia Xinjiang http://www.imf.org/external/pubs/ft/fandd/2002/09/wei.htm Read More