The effect of the Internet on regional integration: evidence from China Maoliang BU * 1. Assistant Professor School of Business, Nanjing University, Hankou Road 22, Nanjing 210093, P. R. China 2. Post doctoral researcher Courant Research Center Poverty, Equity and Growth, University of Göttingen, Wilhelm-Weber-Str. 2, D-37073 Göttingen, Germany Zhibiao LIU 1. Professor School of Business, Nanjing University, Hankou Road 22, Nanjing 210093, P. R. China 2. Dean Jiangsu Provincial Academy of Social Science, Nanjing 210013, P.R. China Sanfeng ZHANG 1. PhD and Assistant Professor, School of economics and management, Nanjing University of Information Science & Technology, Nanjing 210044, P.R. China *corresponding author, E-mail: bumaoliang@hotmail.com. The first author appreciates the research grant from China s Natural Sciences Foundation (No. 70972040 and No. 71003046) and the financial support from the Erasmus Mundus project. 1
The effect of the Internet on regional integration: evidence from China Abstract: This study investigates the effect of Internet on the regional integration in China for the period of 2001 2009. To measure regional integration, we constructed a panel data on inter-regional disintegration index based on the dataset of retail price indices of commodities by region. Regression results have shown that the Internet plays a positive and significant role in regional integration after GDP and population have been used as control variables. Keywords: Internet, Regional integration, Gravity equation, China I. Introduction The Internet has evidently influenced every aspect of the economy. However, studies on the effects of the Internet on the macroeconomic level have been relatively few in economics literature; in empirical studies, they are even fewer. Some papers have looked into the effects of Internet on economic growth [e.g., (Choi & Hoon Yi, 2009)] and trade [e.g., (Blum & Goldfarb, 2006; Choi, 2010; Freund & Weinhold, 2004)]. However, there are no studies on how the Internet affects regional economic integration within one country. Thus, this study attempts to contribute in this space by focusing on one country and using sub-national data. One of its significant advantages is that it avoids the difficulty of controlling institutional differences across countries in international studies (Bacchetta, Rose, & Van Wincoop, 2001; Davis, Weinstein, Bradford, & Shimpo, 1997). China has been chosen as the research target, as it is undergoing a lengthy transition process of into market economy. Additionally, China is very large and its domestic market is highly fragmented. According to (Poncet, 2005), the border effect (i.e., reflecting the extent of market disintegration) of China is close to that of European Union and Canada-US border. In recent years, the regional integration of China has attracted numerous studies [for reviews, see (Young, 2000) and (Naughton, 1999)] and substantial efforts have been devoted to this knowledge field. However, the effect of the Internet has been largely ignored. The succeeding segments are as follows. In Section 2, we derive a modified gravity equation incorporating the Internet variable. In Section 3, we perform several estimations for the gravity equation. Finally, Section 4 concludes the paper. II. Model and data Following (Choi, 2010), we have examined the effect of the Internet on regional integration. A modified gravity equation has been used for this purpose. The Internet 2
variable is included in the right side of the equation. Other independent variables include GDP and population; these control for the income effects and region size, respectively. The dependent variable is regional integration. Notably, finding a credible measure of integration is the most difficult process in conducting empirical studies. We constructed the panel data of the inter-regional disintegration index based on the method of (Parsley & Wei, 1996, 2001; Lu & Chen, 2009), which has been founded theoretically on the iceberg model (Samuelson, 1954). The primary data are the retail price indices of commodities by region obtained from the China Statistical Yearbook across various years. Disintegration = β + β ln Internet + β ln GDP + β ln Population + η + υ + ε it 0 1 it 2 it 3 it i t it (1( where subscript i represents a province and subscript t represents year t. υ individual (province) effect and t ε is the time effect. it ηi is the is distributed independently and identically among provinces and years. The logarithm is taken in all independent variables. Excluding the dependent variable, the rest of the data source is from the China Economic Information Network (CEIN). Statistics for the variables are listed in Table 1. [Insert Table 1 about here.] III. Empirical results [Insert Table 2 about here.] Table 2 lists the regression results. We estimated Eq. (1) by using various estimation methods: (1) pooled ordinary least squares (OLS), (2) fixed effects, (3) random effects, (4) panels corrected standard errors (PCSEs), and (5) generalized method of moments (GMM) estimation. According to the benchmark pooled OLS regression [see column (1) in Table 2], the estimated coefficient of Internet is -3.204 and significant at the 1% level as expected. This means that when the Internet user ratio increases by 1% point, the disintegration index tended to decrease by 3.204% points. The estimated coefficient of GDP is 3.175 and significant at the 1% level. This means that when the income level increases, the regional disintegration index tends to decrease as well. The estimated coefficient of the population is -4.847 and significant at the 1% level. When the population increases by 1% point, the regional disintegration index tends to decrease by 4.847% points. Given that panel data has been used in our regressions, we re-estimated Eq. (1) by using panel data regression methods such as fixed effects, random effects, and panels corrected standard errors (PCSEs) [see columns (2), (3), and (4) in Table 2, 3
respectively). The estimated coefficients of Internet are from -4.559 to -3.204 and significant at the 1% level in columns (2) and (3) and at the 5% level in column (4). This means that when the Internet user ratio increases by 1% point, the regional disintegration index tend to decrease by between 3.204 and 4.559% points. The estimated coefficients of GDP are very similar to pooled OLS estimation in column (1). The estimated coefficients of population are negative and significant at the 5% level in column (2) and at the 1% level in columns (3) and (4). Inasmuch as explanatory variables like GDP and population can be influenced by regional integration, we performed GMM estimation to take into account any endogeneity of the explanatory variables [see column (5) in Table 1]. The coefficient of Internet is -4.625 and significant at the 1% level. The coefficient of GDP is 4.530 and significant at the 1% level. The coefficient of population is -6.186 and significant at the 1% level. Sargan test of overid is 15.18 with a p-value of 0.438, suggesting that the model is well specified. In summary, the Internet has a negative and significant relationship across the regional disintegration indices in all the regressions. Thus, it can be inferred that the effect of the Internet on regional integration is positive and significant. Furthermore, the regression coefficients of GDP and population are mostly consistent with the standard results in literature. GDP has a positive effect on regional integration while population has a negative impact on regional integration. This suggests that the result is quite robust against different estimation methods. IV. Conclusion The Internet may reduce the communication costs across different regions. Therefore, the increase in the use of the Internet in a country is hypothesized as having a positive effect on regional integration. Using panel data on Chinese provinces from 2001 to 2009, we found evidence that the Internet plays a positive and significant role in regional integration after GDP and population are used as control variables in the estimation equation. References Bacchetta, P., Rose, A., Van Wincoop, E., 2001. Intranational economics and international economics. Journal of International Economics 55, 1. Blum, B.S., Goldfarb, A., 2006. Does the internet defy the law of gravity? Journal of International Economics 70, 384-405. Choi, C., 2010. The effect of the Internet on service trade. Economics Letters 109, 102-104. Choi, C., Hoon Yi, M., 2009. The effect of the Internet on economic growth: Evidence from 4
cross-country panel data. Economics Letters 105, 39-41. Davis, D., Weinstein, D., Bradford, S., Shimpo, K., 1997. Using international and Japanese regional data to determine when the factor abundance theory of trade works. The American Economic Review 87, 421-446. Freund, C.L., Weinhold, D., 2004. The effect of the Internet on international trade. Journal of International Economics 62, 171-189. Ming, L. and C. Zhao (2009). "Fragmented Growth : Why Economic Opening May Worsen Domestic Market Segmentation?" Jingji Yanjiu (in Chinese) 3. Naughton, B., 1999. How much can Regional Integration do to Unify China's Market. University of California at San Diego Mimeo. Parsley, D.C., Wei, S.-J., 1996. Convergence to the Law of One Price Without Trade Barriers or Currency Fluctuations Quarterly Journal of Economics 111, 1211-1236. Parsley, D.C., Wei, S.-J., 2001. Limiting Currency Volatility to Stimulate Goods Market Integration: A Price Based Approach. NBER Working Paper No. W8468. Poncet, 2005. A Fragmented China: Measure and Determinants of Chinese Domestic Market Disintegration. Review of International Economics. Samuelson, P., 1954. Theoretical Note on Trade Problem. Review of Economics and Statistics 46, 145-164. Young, A., 2000. The Razor'S Edge: Distortions And Incremental Reform In The People'S Republic Of China. Quarterly Journal of Economics. 115, 1091-1135. 5
Table 1. Statistics Variable Obs Mean Std. Dev. Min Max Disintegration 279 4.319713 5.432589 1.112404 85.36004 Internet (users per 1000 people) GDP (current 100 million RMB) 279 126.2143 118.5819 4.644485 628.4901 279 4164.692 2644.244 263 9717 Population (10 thousand) 279 6975.996 6826.657 139.16 39482.56 Table 2. Regression Results (1) (2) (3) (4) (5) a Pooled OLS Fixed effect Random effect PCSE Panel GMM Internet -3.204*** -4.559*** -3.459*** -3.364** -4.625*** (1.022) (1.340) (1.084) (1.560) (-1.265) Lnpop -4.847*** -21.89** -5.148*** -5.107*** -6.186*** (1.171) (9.522) (1.293) (1.936) (-1.317) Lngdp 3.175*** 6.642 3.470*** 3.256** 4.530*** (1.143) (5.319) (1.252) (1.530) (-1.313) Constant 32.54*** 146.6 33.75*** 34.905*** 37.97*** (5.478) (92.60) (6.011) (12.971) (-4.561) R 2 0.271 0.250 0.239 0.271 Observations 279 279 279 279 279 ***, **, and * indicate significance at the 1%, 5%, and 10% levels, respectively. Standard errors are in parentheses. a Instrumental variables include (internet) t-1, (lnpop) t-1, and (lngdp) t-1. Arellano-Bond test for AR(1) in first differences: z =-1.93, Pr > z =0.053. Arellano-Bond test for AR(2) in first differences: z = 0.14, Pr > z=0.889. Sargan test of overid. restrictions: chi2(15)=15.18, Prob> chi2=0.438. 6