Chapter 5 Empirical Analysis and Results
5.1 Regional Telecommunications Development and Regional
國
立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Chapter 5 Empirical Analysis and Results
This study presents the empirical analysis which is conducted comprising a series of socio-economic variables. This chapter is organized as follows:
section one presents the regional growth disparity among provinces. Section two provides empirical evidence and discusses the regression analyses findings. The final section concludes.
5.1 Regional Telecommunications Development and Regional Economic Growth
As described in Chapter 3, at the beginning of the reforms the
telecommunications industry was one of the least developed industries in China. In 1980, China had only 43 telephones per 1,000 residents. Most residents did not have telephone services. But now telecommunications development has been the one of the fastest-growing industries during the past two decades. Since the first analog mobile phone network was
established and put into operation in China in 1987. After reforms in this country, three large telecommunication operators (China Mobile, China Telecom and China Unicom) now dominate the Chinese market. The number of mobile phone subscribers in China surpassed the number of fixed-line subscribers in 2003. Into 2010, China Mobile continued to be the world’s largest mobile operator in terms of number of subscribers.
As of the end of 2003, China had more than 270 million mobile phone subscribers, 277 million fixed line connections, and 79 million Internet
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
subscribers. Till 2009, there had been 680 million mobile phone subscribers.28 As brief review of the average growth rate of main telephone line from 1988 to 2003 double the rate of GDP growth and accelerated above 30% after 1992 (Ding, 2005). The growth rate of telephone lines has exceeded GDP
development since 1985. With an average annual GDP growth rate above 10%, China already became the second largest economy ahead of Japan after the United States in 2009.29
Additionally, the rapid growth has widened the economic gap among regions and appears the large disparities in per capita income between regions.
It explores the phenomenon of regional inequality and large difference in growth performance accompanied by the rapid GDP growth. The economic development in coastal provinces has generally been more rapid than in the interior. Besides, rising inequalities have resulted from increasing income disparities between urban and rural areas and across regions.30 There is a concentration of regions in GDP per capita in 2008 with higher figures in the coastal areas. Regions in western and central China usual have much lower GDP per capita. The region with biggest and lowest figures can be found in Shanghai with RMB 72,536 and Guiehou with RMB 8,789, respectively (see Figure 7). As Figure 8, 9 and 10 shows, there is a high correlation between GDP per capita in natural logarithm and three different telecommunication subscribers . It is obvious that the level of telecommunications development in regions is closely correlated to its economic growth level.
According to the conditional convergence hypothesis, it states that relative regional earnings will not converge toward unity, but towards a stable
28 http://www.china.org.cn/business/2009-07/22/content_18183326.htm 29 http://en.wikipedia.org/wiki/Economy_of_the_People's_Republic_of_China 30 http://www.adb.org/media/Articles/2007/12084-chinese-economics-growths
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
differential. If countries possess the same technological possibilities and population growth rates but differ in savings propensities and initial
capital-labor ratio, then there should still be convergence to the same growth rate, but just not necessarily at the same capital-labor ratio.31 It means that the telecommunications development will converge by controlling for a set of variables reflecting regional specific effect differences in the stay-state equilibrium (Barro, 1991).
Figure 7 GDP per capita of Chinese provinces in 200832
31 http://homepage.newschool.edu/het/essays/growth/neoclass/solowconv.htm 32 http://en.wikipedia.org/wiki/File:GDP_per_capita_of_Chinese_provinces.PNG
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Figure 8 GDP and Telephone Subscribers per Capita (2008) Source: Author’s calculation based on China Statistical Yearbook
Figure 9 GDP and Mobile Phone Subscribers per Capita (2008) Source: Author’s calculation based on China Statistical Yearbook
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Figure 10 GDP and Internet Subscribers per Capita (2008) Source: Author’s calculation based on China Statistical Yearbook
5.2 Empirical Analysis and Results
As outlined in Chapter 4, the study follows Barro-type conditional convergence framework based on growth equation to investigate the
determinants of regional economic growth. A set of initial conditioning variables and production factors are included in Equation 4.2.
Model Specification
Following the traditional operation on the initial conditioning variables under the conditional convergence framework, this study conducts an initial screening test on the significance levels of all possible independent variables highlighted in Equation 4.2 which are considered as growth factors if it is statistically significant at the 10% level (Ding, 2005; Liu, 2008). To keep on watch the model specification and characteristic, we will not remove these insignificant variables. Table 7 shows a list of variables and its mean value and standard value in the final model, respectively. It is expected that there are
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
positive interrelationships between regional growth and other variables, which are considered that they always are expected signs and are highly significant during the test.
Table 8 shows the regression results from modeling. In model A estimated by using a fixed effects model, almost 80 percent of the variation in the growth rates can be explained by this model. For employment rate (EMP), human capital (HC), SOE industrial output (SOE), transportation percentage (TRANS), the number of fixed line telephone user per 100 inhabitants (TEL) and the number of Internet user per 100 inhabitants (NET) are not significant. It is noteworthy that POP, SOE, TRANS, TEL and NET show a negative sign for its coefficient. It may be due to multicollinearity,33 low data quality or other
possible issues.
The coefficients of most of the variables of interest are significant (most at 1 percent level). It confirms that China’s regional growth rates are positively related to fixed investment, foreign direct investment, urbanization, lagged growth rates in previous year and mobile phone user per 100 inhabitants; and are negatively related to the initial level of real GDP per capita and population growth.
33 Multicollinarity is an inevitable issue in any multivariate regression analysis. Thecorrelation between independent variables can reduce reliability and robustness of the resultsin terms of values and signs of coefficients as well as significance of some independentvariables (Liu, 2008).
‧
Table 7 Variables of Its Expected Signs, Mean and Standard Error Value
Variables Explanation Expected
Sign
INV Total investment in fixed assets in GDP (%)
+ 52.70 0.48E-01
FDI FDI divided by total investment in fixed assets (%)
+ 3.06 0.14
POP Annual population growth rate (%)
- 0.67 0.15
EMP Total employed persons to total population (%)
+ 51.57 0.16
HC Average years of schooling for the population aged 6 and above (%) enterprises in total industrial output (%)
- 0.46 3.15
TRANS Transportation percentage (%) + 5.63 0.20 TEL Number of telephone
subscribers per 100 inhabitants
+ 28.43 0.11
MOB Number of mobile phone subscribers per 100 inhabitants
+ 37.45 0.75E-01
NET Number of Internet subscribers per 100 inhabitants
+ 12.56 0.69E-01
Source: Author’s calculation based on data from various sources.
Note: The expected sign of POP is negative since a lower population growth rate relates to
growth rate
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
The result on Table 8 shows the MOB variable is positive and significant at the 1% level in Model A, suggesting a positive relationship between
telecommunication development and economic growth. Most of the previous studies use number of fixed lines as the proxy for telecommunications infrastructure (e.g. Hardy, 1980; Savage et. al, 2003, and etc.), but we have found the number of mobile phone users to be the new proxy for
telecommunications infrastructure in China. As discussed before, the MOB variable has been separated from the whole telephone users (MOB+TEL) to investigate the economic growth effect. According to the report published by Ministry of Industry and Information Technology of China, it appears a net increase of 9.367 million mobile phone users and a decrease of 1.497 million fixed line telephone users in February 2010. The new data report states that China had a phone user base of 1.07 billion with a net increase of 7.8 million. It revealed that the number of the country's mobile phone users has been ranked the first in the world.
In terms of causality study, previous discussion and some investigations have indicated a two-way causation between telecommunications
development and economic growth. We test the causality relationship using lagged values of MOBt-1 to replace current value of MOB on Equation 5.1. If the positive effect of MOB on economic growth is a result of reverse causality, the coefficient of the lagged values of MOB should be insignificant. Model B on Table 8 report the results using lagged mobile phone user density to replace current index to check whether the positive effect is totally caused by reverse causality.
‧
As shown on Table 8 in Model B, the coefficients of the lagged values of MOB, MOBt-1 are significant at 1 percent level and have an almost same magnitude when compared to the current value. These results give support to the argument that the positively relationship is not merely due to reverse causality. Telecommunications infrastructure for mobile communication systems does have positive impact on the regional economy of China.Regarding the linearity, some researches indicate the relationship between telecommunications development and economic growth might not be linear.
Roller and Waverman (2001) revealed that investment in telecommunications development would not significantly affect economic growth until the
infrastructure development is achieved a critical mass.34
In our research, we add MOBSQ variable as a square of MOB as Equation 5.2 to examine the character of returns to scale to
telecommunications investment. The purpose of introducing a square term is to check whether the relationship between economic growth and
telecommunications is linear.
it If the coefficient of MOBSQ (β13) is negative and significant while the
34 Roller and Waverman (2001), the level of “critical mass” is about 40 mainlines per 100.
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
coefficient of MOB (β12) is positive and significant ,((β13) <0 and (β12) >0), then we have support for a “diminishing returns” hypothesis. Hence, a unit increase in MOB density would have a smaller magnitude of growth for a region with greater level of MOB density. In contrast, positive signs for both coefficients, (β12) and (β13) would present “increasing returns”, ((β13) >0 and (β12) >0). If, however, the coefficient of MOBSQ (β13) is positive and
significant while the coefficient of MOB(β12) is negative and significant, ((β13)
>0 and (β12) <0), then we have evidence in support of a “critical mass” theory, as investment in telecommunications infrastructure would not significantly affect economic growth until a critical mass of telecommunication infrastructure is achieved (Ding, 2005).
According to the report published by Ministry of Industry and Information Technology of China, China’s mobile phone penetration rate has surpassed 40 percent in 2007. Furthermore, it has reached 54.3 percent in the end of 2009.
Newly-added mobile phone subscribers in west and central China kept rising.
By the end of June 2009, mobile phone penetration rate was 43.2% in central China and 43.8% in west China comparing with the 72% in east China which is higher than the country's average level.35
As Table 8 in Model C shows, it evidences that the results support the assumption of diminishing returns of telecommunication investment in China because the coefficient of MOBSQ is negative and significant at a 1 percent level while the coefficient of MOB is positive and significant at a 1 percent level as well. Furthermore, the results are matched with Ding’s findings in 2005; the size of the effect of telecommunication infrastructure on the economic growth is inversely related to its prior level. This indicates that the higher the level of
35 http://www.chinaeconomicreview.com/china-eye/
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
telecommunications infrastructure, the smaller the magnitude of the effect of a marginal increase in mobile density on economic growth.
This suggests diminishing returns and implies that regions at an earlier stage of development in China are likely to gain the most from further investment in telecommunications infrastructure. An investment in
telecommunications infrastructure in those regions would lead to a higher growth rate than that in already developed regions. The positive incremental effect decreases for those regions with more developed telecommunications infrastructure. Since regions in the central and western China usually have a lower level of telecommunications infrastructure, a strategy for investment in these regions should have a larger impact on growth (Ding et al., 2006). Figure 11 shows the relationship between mobile phone user density and annual growth rate of real GDP per Capita.
‧
Table 8 Regression Results from Modeling
Variables Coefficient
Residual 801.05 801.65 722.95
Source: Author’s calculation based on data from various sources.
Notes:
1. Definition of each variables is listed in Chapter 4.1
2. t-statistics in parentheses;
3. *** significant at 1% level; ** significant at 5% level; *significant at 10% level;
4. Number of groups: 31
5. Period: 2003~2008
‧
國立 政 治 大 學
‧
N a
tio na
l C h engchi U ni ve rs it y
Figure 11 Mobile Phone User Density and Annual Growth Rate of Real GDP per Capita (2003-2008)
Source: Author’s calculation based on China Statistical Yearbook