Do Internet Development and Urbanization Foster Regional Economic Growth: Evidence from China’s Yangtze River Economic Belt

Abstract

Internet development and urbanization are widely perceived to have a significant impact on the economy and sustainability in China. However, existing studies fail to consider their interaction patterns and directions with economic growth in China’s Yangtze River Economic Belt (YREB). This study applies a bootstrap panel Granger causality test to examine the causal relationships between Internet development, urbanization, and regional economic growth for the YREB‘s 11 provinces. The main findings are as follows: (1) Internet development, urbanization and regional economic growth exhibit cross-sectional dependence and province-specific heterogeneity in the YREB. (2) Granger causality from Internet development to economic growth exists in most provinces, while Granger causality from economic growth to Internet development and two-way Granger causality are only observed in economically developed provinces, such as Shanghai, Jiangsu, Zhejiang, and Chongqing. (3) Granger causality from urbanization to economic growth exists in nine provinces, except for Hubei and Guizhou. However, the reverse Granger causality is only confirmed in Jiangsu, Anhui, and Hubei. This suggests that urbanization significantly improves economic growth efficiency, but the effect of economic growth on urbanization is restricted by regional bias policies. Therefore, the local government should implement tailored economic policies and establish an interactive mechanism to help China leverage its potential for economic growth and sustainability.

1. Introduction

China has undergone a remarkable economic transformation over the past two decades. The advancement in Internet technology, together with a rapid increase in urbanization, has propelled China to the forefront of global economic growth. On one hand, Internet development has been shown to have a profound impact on the economic system, including economic productivity [1,2], export growth [3], firm performance [4], poverty reduction [5,6], and economic well-being [7]. The popularization of the Internet has penetrated into nearly every aspect of regional economic development, and its influence is continuously expanding. According to the China Internet Network Information Center (CNNIC) [8], China has the largest number of internet users in the world, reaching 1.067 billion by the end of December 2022. On the other hand, China’s urbanization process has reached a stage in which a new type of urbanization is taking place. This trend is expected to have a significant impact on the sustainable development of the regional economy and society. China’s urbanization rate reached 65.22% by the end of 2022, which is more than double the rate of 31.91% in 1997. In this context, many theoretical and empirical studies have attempted to shed light on the relationship between Internet development, urbanization, and regional economic growth in China.

The popularization of Internet technology can stimulate network economy and industrial structure upgrading, which may lead to economic growth [9,10]. Conversely, economic growth can accelerate the construction of information and communication technology infrastructure and promote the development of high-tech industries, thus speeding up Internet development [11]. At the same time, urbanization has a close link with economic growth, by forming agglomeration economies and economies of scale. Economic growth may generate spillovers and change the economic structure by promoting the transfer of rural labor to urban regions [12,13]. As a result, research has focused on the regional economic effects of Internet development and urbanization [14,15]. China is still in the process of developing, with a wide range of economic growth levels across regions. The Yangtze River Economic Belt (YREB) spans 11 provinces across three prominent regions in China, accounting for over 45% of the country’s overall economic growth [16]. It has emerged as the most active and powerful region within China’s economic landscape. Recently, the Chinese government has put forward the concept of constructing the “digital Yangtze River” and the new-type urbanization with the aim of fostering sustainable economic development in the YREB. Therefore, how can the YREB accelerate its economic growth amidst the development of internet technology and urbanization? What are the interaction patterns among Internet development, urbanization, and economic growth? Answering these questions is crucial for China to formulate effective policies that increase the country’s Internet competitiveness and promote sustainable urbanization growth.

In this study, we investigate the causal links between Internet development (INT), urbanization (URB), and regional economic growth (REG), based on the province-level panel data from 1997 to 2019 in China’s YREB. There are two Granger causality hypotheses that need to be examined. First, we examine the “INT-REG growth” nexus, which covers the hypotheses of “INT-led REG growth” and “REG-led INT growth”. The INT-led REG growth hypothesis highlights that Internet development will positively contribute to economic growth. In this situation, the government should prioritize investments aimed at enhancing Internet infrastructure, such as broadband networks. The REG-led INT growth hypothesis implies that any increase in regional economic growth will lead to a corresponding increase in Internet penetration. Therefore, promoting economic growth should be the government’s top priority policy. Second, we investigate the Granger causality between URB and REG, which includes “URB-led REG growth” and “REG-led URB growth” hypotheses. The URB-led REG growth hypothesis suggests that urbanization is considered as a significant factor in economic growth. The increasing migration of individuals to urban areas has contributed to the expansion of urban economies, generating a heightened demand for a wide range of goods and services. In contrast, the REG-led URB growth hypothesis indicates that economic growth’s effect on urbanization is restricted without a rational movement of rural labor to urban regions. Therefore, we use the bootstrap panel Granger causality test to find out how and in which direction INT, URB, and REG contribute to each other.

There are three marginal contributions in this study. Firstly, we provide empirical evidence on the interaction patterns and directions between INT or URB and REG in the YREB. To our best knowledge, it is the first study that compares the causality associations among these variables while considering regional differences by applying the bootstrap panel Granger causality approach. Secondly, we investigate the cross-sectional dependence and slope heterogeneity among provinces in the YREB, contrary to previous empirical studies that have only considered independent countries or regions. In addition, the panel Granger causality method can also increase estimation accuracy by identifying time dimensions [17]. Thirdly, we provide policy implications and motivate regional governments to accelerate economic development reform, improve Internet penetration, and promote sustainable urbanization growth in the YREB.

The rest of this empirical study is structured as follows: Section 2 provides a literature review and identifies research gaps, Section 3 details the data information of China’s YREB and the methodology of the bootstrap panel Granger causality approach, Section 4 presents and discusses the empirical results and policy implications, and the main conclusions are provided in Section 5.

2. Literature Review

In recent years, a substantial number of empirical studies have investigated the relationship between Internet development (INT), urbanization (URB), and regional economic growth (REG). In terms of the linkage between INT and REG, Bertschek et al. [18] and Vu et al. [14] have provided an overview of the quantitative research on broadband’s economic impacts. As for the research into the causality between URB and REG, Pradhan et al. [19] have proposed four testable hypotheses, which are validated by numerous researchers. However, there is still no consensus on the interaction patterns and directions regarding the INT–REG nexus and URB–REG nexus. In particular, there has not been sufficient research conducted on specific regions or economic belts within one country.

Table 1. Summary of the studies on the relationships between INT or URB and REG.

Literature	Data	Methods	Results
(a) The nexus between INT and REG
Czernich et al. [20]	1996–2007; OECD countries	FE	INT→(+) REG
Ghosh [21]	2001–2014; 15 MENA countries	DID	INT→(+) REG
Appiah-Otoo and Song [9]	2002–2017; 123 countries	IV-GMM	INT→(+) REG; Poor countries gain more
Haldar et al. [22]	2000–2018; 16 emerging economies	FE, IV-GMM	INT→(+) REG in the lower and middle-income countries
Mayer et al. [23]	2008–2012; OECD countries	GMM	INT→(/) REG; Broadband speed→(+) REG
Aldashev and Batkeyev [24]	2006–2016; 121 rural districts in Kazakhstan	OLS, FE	INT→(/) REG; INT→(+) Retail sector
Nabi et al. [25]	2000–2018; N11 countries	ARDL	INT expansion→(−) REG
Jin and Li [26]	2007–2015; 31 provinces in China	LQ	REG→(+) INT
Arvin and Pradhan [27]	1998–2011; G-20 countries	Granger causality test	REG→(+) INT INT→(/) REG
Belloumi and Touati [28]	1995–2019; 15 Arab countries	ARDL	REG→(+) INT
(b) The nexus between URB and REG
Pradhan et al. [19]	1961–2016; G-20 countries	VECM	URB⇔REG;
Solarin and Shahbaz [29]	1971–2009; G-20 countries	ARDL, VECM	URB⇔REG
Dzator et al. [30]	1960–2019; Australia	FMOLS, DOLS	URB→(−) REG
Nguyen [31]	1993–2014; ASEAN countries	Granger causality test, FE, PMG	URB→(+) REG URB→(−) REG after a threshold
Liu et al. [32]	1952–2011; 28 provinces in China	Bootstrap panel Granger causality test	URB→(/) REG in 1/4 provinces REG→(+) URB in southern region
Shaban et al. [33]	1971–2020; India	Bootstrap panel Granger causality test	REG→(+) URB
Brückner [34]	1960–2007; 41 African countries	OLS, FE	REG→(/) URB

Note: (1) Abbreviations used for the data: OECD: Organization for Economic Cooperation and Development; MENA: Middle East and North Africa; N11: the Next 11 countries, including Egypt, Vietnam, Bangladesh, Pakistan, Turkey, Mexico, Nigeria, South, Korea, Iran, Philippines, and Indonesia; G-20: 19 member countries plus the European Union; ASEAN: Association of Southeast Asian Nations. (2) Abbreviations used for the methods: FE: Fixed Effect; DID: Differences-in-Differences; IV-GMM: Instrumental Variable Generalized Method of Moments; OLS: Ordinary Least Squares; ARDL: Autoregressive Distributed Lag; LQ: Location Quotient; VECM: Vector Error Correction Model; FMOLS: Fully Modified Ordinary Least Squares; DOLS: Dynamic Ordinary Least Squares; PMG: Pooled Mean Group. (3) Abbreviations used for the results: → denotes a relationship; →(+) implies a positive relationship; →(−) means a negative relationship; →(/) indicates no a relationship; ⇔ represents a bidirectional relationship.

summarizes the literature that examines the causality between INT or URB and REG.

2.1. The Nexus between Internet Development and Regional Economic Growth

The theoretical links between Internet development and economic growth can be traced back to the Solow growth model and endogenous growth theory, which state that Internet technology is a kind of technological progress. Consistent with these traditional theories, many studies have found positive results and noted that Internet development has a significant economic impact at the country level. For example, Czernich et al. [20] estimated the impact of broadband infrastructure on economic growth in 25 OECD countries, and the results showed that broadband penetration significantly increases GDP per capita. Ghosh [21] conducted a similar investigation for countries in the Middle East and North Africa (MENA) countries, further confirming the significance of Internet penetration in driving economic growth. Similarly, in their study, Appiah-Otoo and Song [9] found that broadband Internet exerts a positive effect on economic growth in 123 countries, especially in poor countries. Haldar et al. [22] suggested that Internet penetration increases economic growth monotonically in 16 emerging countries.

In contrast, some scholars have argued that Internet development does not have a significant effect on economic growth. Mayer et al. [23] made a reassessment of the causal link between broadband Internet and economic growth, proving that existing studies overestimate the economic effects of broadband Internet. Another study conducted by Aldashev and Batkeyev [24] suggested that broadband Internet access does not foster economic growth in Kazakhstan, and has no improving effect on the agricultural or manufacturing sector. In a recent study, Nabi et al. [25] found a negative and significant effect of INT expansion on economic growth in N11 countries.

Furthermore, some researchers have demonstrated that there is a causal relationship between economic growth and Internet development. Jin and Li [26] documented that Internet development in China is largely stimulated by economic growth. Economically developed regions, such as the eastern provinces, usually have a better Internet infrastructure than the less-developed western regions. Arvin and Pradhan [27] revealed an unidirectional causality from economic growth to broadband penetration in the developing countries within the G-20 countries. Belloumi and Touati [28] focused on the Arab countries and showed that the per capita GDP has a positive impact on the long-term development of INT.

2.2. The Nexus between Urbanization and Regional Economic Growth

The interaction between urbanization and regional economic growth has been continually investigated; however, a consensus regarding the direction of causality between these two variables has not been reached. While previous research has demonstrated that urbanization is a crucial factor in regional economic growth, the impact of urbanization on economic growth is ambiguous due to variations in sample periods, methodologies, and other factors that can impact the results [19]. Most scholars have recognized the importance of urbanization as a driver of sustainable economic growth [35,36,37]. Urbanization provides countries with advantages in terms of transaction costs, economies of scale, and internal specialization [31].

From the perspective of causality, Solarin and Shahbaz [29] found evidence that urbanization reduced economic growth during the civil war in Angola. The findings of Dzator et al. [30] demonstrated that urbanization has a significant negative effect on Australia’s economic growth. Nguyen [31] proved that the relationship between urbanization and economic growth is non-linear in ASEAN countries for the period 1993–2014. Urbanization would impede the economic growth when it reached beyond the threshold. Liu et al. [32] proved no Granger causality relationship between economic growth and urbanization in one-fourth of China’s provinces, and that economic growth can drive urbanization in the southern coastal region. Shaban et al. [33] found that a majority of the states in India show a unidirectional Granger causality from economic growth to urbanization, which suggests that economic growth accelerates urbanization in India. In contrast, Brückner [34] found that GDP per capita growth, which is generated by the change share of agricultural value added, does not improve the urbanization rate in 41 African countries.

2.3. Literature Gaps

Overall, the findings of the existing literature highlight the complex and multifaceted relationships between Internet development or urbanization and regional economic growth. The disparities in the obtained results can be attributed to the periods and indicators utilized, the employed econometric methods, and the selected regions. These differing outcomes indicate the need for the further investigation and exploration of these issues. From the mentioned literature review, we find that there are still three gaps in the existing empirical literature. First, although the previous literature has investigated the relationships between Internet development and economic growth, as well as between urbanization and economic growth, there is a gap regarding the consideration of all three indicators simultaneously. Second, the majority of studies have focused on the country level, and little is known about how their relationships differ spatially in China’s YREB. Third, the current literature has not clearly identified the interaction patterns and directions between Internet development, urbanization, and regional economic growth. While considering cross-sectional dependence and heterogeneity issues, it may possibly lead to the distortion of the causality and policy implications. Therefore, we aim to provide further evidence of the nexus of Internet development or urbanization and regional economic growth by employing the bootstrap panel causality test. Using this method means that no pretesting for unit roots and co-integration is needed; it is thus considered to be the most suitable method with which to deal with the spatial spillover effects of provinces or regions.

3. Materials and Methods

3.1. Data Sources and Description

In this study, we use annual data of the China’s Yangtze River Economic Belt (YREB) to investigate the causality between INT or URB and REG from 1997 to 2019. The YREB is a major national strategic development region and plays a crucial role in the country’s economy. It contributes to more than 46% of the country’s total GDP and has a population of approximately 600 million [38]. Promoting its development of technological innovation and new urbanization is of great strategic significance to China’s sustainable economic development. It covers 11 provinces (municipalities) along the Yangtze River (Figure 1), including the upper reaches (Yunnan, Sichuan, Guizhou and Chongqing), the middle reaches (Hubei, Hunan, Jiangxi) and the lower reaches (Anhui, Jiangsu, Zhejiang, Shanghai) [16].

Three indicators are considered in this study, including Internet development (INT), urbanization (URB) and regional economic growth (REG). Following the previous literature, we use the internet penetration rate as a proxy variable to represent INT [17,39]. Therefore, INT is defined as the proportion of internet users to the total population. URB is measured by the ratio of urban population to total population. REG is expressed by gross domestic product per capital in CNY at the 1997 constant price. All the data are published by the National Bureau of Statistics of the People’s Republic of China.

Table 2. Summary statistics for INT, URB and REG in China’s YREB.

Province	Variable	Mean	SD	Skewness	Kurtosis	J–B
Shanghai	INT	43.04	26.56	−0.22	1.56	7.01 **
	URB	83.50	6.53	−0.70	1.83	5.23 *
	REG	20,370	1607	0.45	2.03	2.63
Jiangsu	INT	27.74	22.83	0.10	1.31	13.79 ***
	URB	53.32	12.23	−0.41	2.08	2.07
	REG	10,519	1456	0.02	1.43	9.74 ***
Zhejiang	INT	33.53	26.36	−0.02	1.27	15.42 ***
	URB	57.10	8.37	−0.50	2.42	1.48
	REG	11,308	930	−0.15	1.54	7.41 **
Anhui	INT	18.62	18.32	0.42	1.53	8.14 **
	URB	39.09	10.17	−0.12	1.83	3.36
	REG	4760	828	0.12	1.29	14.65 ***
Jiangxi	INT	17.65	17.17	0.6	1.96	3.88
	URB	40.45	9.82	−0.05	1.78	3.93
	REG	4761	779	−0.06	1.34	12.80 ***
Hunan	INT	17.90	17.79	0.54	1.68	6.14 **
	URB	39.84	8.97	−0.15	1.91	2.56
	REG	5474	926	−0.13	1.36	11.99 ***
Hubei	INT	22.86	20.38	0.33	1.50	8.45 **
	URB	46.65	8.73	−0.08	2.06	1.29
	REG	6184	1224	0.06	1.32	13.46 ***
Chongqing	INT	22.89	20.61	0.27	1.42	10.43 ***
	URB	48.84	10.78	−0.09	1.77	3.97
	REG	6043	727	−0.24	1.48	8.71 **
Yunnan	INT	16.60	15.79	0.43	1.58	7.24 **
	URB	33.04	8.63	0.20	1.84	3.48
	REG	4513	575	0.00	1.41	10.43 ***
Guizhou	INT	16.55	17.32	0.57	1.79	5.02 *
	URB	31.65	8.19	0.62	2.01	3.72
	REG	3172	933	0.24	1.45	9.57 ***
Sichuan	INT	18.62	16.94	0.42	1.63	6.41 **
	URB	37.56	8.44	0.27	1.73	4.71 *
	REG	4780	741	0.04	1.28	15.20 ***

Note: J–B denotes the Jarque–Bera test for normality. *, **, *** indicate rejection of the null hypothesis at the 0.1, 0.05 and 0.01 levels of significance, respectively.

presents the statistical description of INT, URB and REG in China’s YREB. All the variables show positive means. INT for Shanghai and Zhejiang has the largest standard deviation but, on the other hand, Jiangsu’s URB and Shanghai’s REG, respectively, show the highest volatility. This indicates that INT, URB and REG are very uneven in the different reaches of China’s YREB. Similarly, the skewness of URB and REG is negative for most provinces in the middle and lower reaches, which implies that their time series distributions are skewed to the left. The kurtosis values are less than 2.5 for the variables, suggesting that almost all provinces show platykurtic distributions and have a huge fluctuation. Most importantly, the Jarque–Bera test shows that URB is normally distributed, except for Shanghai and Sichuan. In contrast, except for Shanghai’s REG and Jiangxi’s INT, we find that INT and REG in other provinces are non-normally distributed. These results make it clear that the traditional method cannot be employed to Granger causality analysis. When the variables are not normally distributed, the asymptotic methods lose power when estimating parameters. Consequently, we take advantage of the bootstrap approach to obtain reasonable and precise parameters.

3.2. Bootstrap Panel Granger Causality Test

Granger causality is widely applied to study the causality between economic variables, and exploits the knowledge of the past values of one variable (X) to predict the current information of another variable (Y). If provinces present heterogeneity along with cross-sectional dependency, the method utilized to test causality should reflect these features. Although some panel causality methods have been inspired to detect panel-data Granger causality, Kónya [40] and Su et al. [41] suggest the bootstrap panel Granger method to consider and examine the cross-sectional dependency and heterogeneity. The bootstrap panel Granger causality approach is performed on the basis of the Wald tests for each province using bootstrap critical values and the seemingly unrelated regression (SUR) system. In addition, this method is proven to be robust to the unit root and cointegration properties of variables, which implies that it does not require pretesting for the stationarity of variables. Therefore, we follow this method and implement it by estimating the SUR system specified below:

$$\begin{array}{c}{Y}_{1,t}={\beta }_{1,1}+{\displaystyle \sum _{k=1}^{ml{Y}_1}}{\gamma }_{1,1,k}{Y}_{1,t-k}+{\displaystyle \sum _{k=1}^{ml{X}_1}}{\delta }_{1,1,k}{X}_{1,t-k}+d_{1,1}{Z}_{1,1,t}+{\epsilon }_{1,1,t}\\ Y_{2,t}={\beta }_{1,2}+{\displaystyle \sum _{k=1}^{ml{Y}_1}}{\gamma }_{1,2,k}{Y}_{2,t-k}+{\displaystyle \sum _{k=1}^{ml{X}_1}}{\delta }_{1,2,k}{X}_{2,t-k}+d_{1,2}{Z}_{1,2,t}+{\epsilon }_{1,2,t}\\ ⋮\\ Y_{N,t}={\beta }_{1,N}+{\displaystyle \sum _{k=1}^{ml{Y}_1}}{\gamma }_{1,N,k}{Y}_{N,t-k}+{\displaystyle \sum _{k=1}^{ml{X}_1}}{\delta }_{1,N,k}{X}_{N,t-k}+d_{1,N}{Z}_{1,N,t}+{\epsilon }_{1,N,t}\end{array}$$

(1)

and

$$\begin{array}{c}{X}_{1,t}={\beta }_{2,1}+{\displaystyle \sum _{k=1}^{ml{Y}_2}}{\gamma }_{2,1,k}{Y}_{1,t-k}+{\displaystyle \sum _{k=1}^{ml{X}_2}}{\delta }_{2,1,k}{X}_{1,t-k}+d_{2,1}{Z}_{2,1,t}+{\epsilon }_{2,1,t}\\ X_{2,t}={\beta }_{2,2}+{\displaystyle \sum _{k=1}^{ml{Y}_2}}{\gamma }_{2,2,k}{Y}_{2,t-k}+{\displaystyle \sum _{k=1}^{ml{X}_2}}{\delta }_{2,2,k}{X}_{2,t-k}+d_{2,2}{Z}_{2,2,t}+{\epsilon }_{2,2,t}\\ ⋮\\ X_{N,t}={\beta }_{2,N}+{\displaystyle \sum _{k=1}^{ml{Y}_2}}{\gamma }_{2,N,k}{Y}_{N,t-k}+{\displaystyle \sum _{k=1}^{ml{X}_2}}{\delta }_{2,N,k}{X}_{N,t-k}+d_{2,N}{Z}_{2,N,t}+{\epsilon }_{2,N,t}\end{array}$$

(2)

where Y and X refer to the dependent and independent variable; Z represents the control variable; the index $i$ ($i=1,\dots ,N$) denotes the province and index $t$ ($t=1,\dots ,T$) is the time period; $k$ is the lag length, and $ml{Y}_1$, $ml{X}_1$, $ml{Y}_2$, and $ml{X}_2$ indicate the maximal lags of Y and X by choosing from 1 to 4 according to the Schwarz Bayesian Criterion (SBC); $\beta$ is the intercept item, $\gamma$ and $\delta$ are the regression coefficients; and the residual error terms ${\epsilon }_{1,i,t}$ and ${\epsilon }_{2,i,t}$ are supposed to be white noises.

In this study, the regression system is estimated by each equation and it has different predetermined variables. However, error terms may be contemporaneously correlated (i.e., cross-sectional dependence) within this system. We perform the SUR model instead of the traditional VAR model and province-by-province OLS estimation. Meanwhile, we compare the Wald statistics with the province-specific critical values that are calculated using the bootstrap sampling procedure. Finally, with respect to this system, four possible Granger causal relations can be examined, including one-way Granger causality from X to Y or Y to X, two-way Granger causality, and no Granger causality.

4. Results and Discussion

Our empirical study first examines the existence of cross-sectional dependency and slope heterogeneity across the 11 provinces along the YREB. It is a crucial step before conducting the bootstrap panel causality test. To test the existence of cross-sectional dependence, we carry out three different methods (LM, CD_lm, LM_adj) and report the subsequent results in

Table 3. Results for cross-sectional dependence and homogeneity tests.

Study	Test	INT	URB	REG
Cross-sectional dependence test
Breusch and Pagan [42]	LM	531.132 ***	632.623 ***	638.961 **
Pesaran [43]	CDlm	19.981 ***	21.130 ***	23.490 ***
Pesaran et al. [44]	LMadj	114.015 ***	138.701 ***	140.121 ***
Homogeneity test
Swamy [45]	S	22.932 ***	35.201 ***	192.210 ***
Pesaran and Yamagata [46]	Δ	9.357 ***	12.034 ***	12.464 ***
Pesaran and Yamagata [46]	Δadj	10.069 ***	12.949 ***	13.412 ***

Note: Bootstrap critical values are obtained from 10,000 replications. ** and *** indicate significance at the 0.05 and 0.01 levels, respectively.

. It reveals that all variables reject the null hypothesis of no cross-sectional dependence at the 0.01 significance level and confirm the advantage of the SUR method over a province-specific OLS estimation. The conventional OLS method ignores the cross-sectional information and the estimation results are less reliable.

Results from the three slope homogeneity tests (S, Δ, Δ_adj) show that the null hypothesis of the slope homogeneity is also rejected at the 0.01 significance level. This finding supports the province-specific heterogeneity and indicates that the panel causality analysis will present misleading inferences if it imposes a homogeneity restriction on the variables. Overall, the existence of cross-sectional dependence and heterogeneity across the 11 provinces proves the rationality and suitability of the bootstrap panel causality approach. Accordingly, this method can be applied to explore the Granger causality relationship between INT and REG or URB and REG.

4.1. Bootstrap Panel Causality between INT and REG

Table 4. Results for panel Granger causality between INT and REG.

Provinces	H0: INT Does Not Cause REG				H0: REG Does Not Cause INT
	Wald Stat.	Bootstrap Critical Value			Wald Stat.	Bootstrap Critical Value
		1%	5%	10%		1%	5%	10%
Shanghai	15.45 *	36.39	17.38	10.51	38.06 **	44.15	22.38	15.02
Jiangsu	23.16 **	38.80	19.74	13.11	18.88 *	44.07	23.66	15.66
Zhejiang	21.11 **	37.01	18.46	12.27	31.43 **	50.09	27.95	19.25
Anhui	15.83 *	40.76	21.69	14.68	6.39	46.76	26.34	18.51
Jiangxi	5.89	55.71	28.69	18.92	1.6	61.62	35.63	25.6
Hunan	30.08 *	88.56	39.51	23.97	9.69	36.23	20.04	13.89
Hubei	8.63	36.02	18.06	11.32	11.37	66.70	41.03	29.37
Chongqing	42.24 **	58.76	28.52	19.02	47.90 **	76.36	44.52	30.90
Yunnan	30.20 *	73.91	31.29	20.76	14.88	44.80	25.22	16.61
Guizhou	18.19 **	34.37	17.04	12.00	13.29	70.00	42.93	31.66
Sichuan	8.91	44.25	23.56	16.08	10.89	63.06	37.82	26.80

Note: In this table, the Wald statistics and Bootstrap critical value of each province are reported. Bootstrap critical values are obtained from 10,000 replications. * and ** indicate significance at the 0.10 and 0.05 levels, respectively.

provides the empirical results regarding the causality between INT and REG when URB is controlled. The Wald test statistics and bootstrap critical values for the null hypothesis reported on the left column, which reveals that one-way Granger causality from INT to REG is found in most provinces, except for Jiangxi, Hubei and Sichuan. It provides empirical support for the INT-led REG growth hypothesis in China’s YREB and indicates that Internet development has made a great contribution to economic growth in the YREB. Since the 1990s, China has implemented a series of informatization development strategies to promote Internet development. We have witnessed the booming development of the digital economy and Internet technology in recent years in China [47]. By the end of 2022, China’s Internet penetration rate has reached 75.6% and the number of Internet users has reached 1.067 billion, making China the country with the most Internet users in the world. The popularization and application of Internet technology has greatly accelerated economic growth in China, especially in the YREB. However, we do not find empirical evidence of Granger causality in Jiangxi, Hubei and Sichuan, which illustrates that INT in these provinces cannot provide enough kinetic energy for economic growth. Thus, the government in these provinces should take measures and formulate policies to further develop Internet technology so as to stimulate economic growth.

Moreover, one-way Granger causality from REG to INT, as well as two-way Granger causality between them, are confirmed in Shanghai, Jiangsu, Zhejiang and Chongqing. These provinces are among the four most developed provinces in the YREB. Shanghai, Jiangsu, and Zhejiang are in the lower reaches of the Yangtze River, and are located in the coastal area. In particular, Shanghai is a well-known global center for finance, innovation and technology, possessing the largest GDP per capital in the YREB, which reached CNY 135,000 (USD 20,400) in 2019. The city has made significant strides in the construction of broadband Internet infrastructure and has attained great achievements, such as the fastest broadband speed in China. Its new consumption drivers are the strongest across the country. Meanwhile, Jiangsu and Zhejiang are two economically strong provinces, with respective GDP values per capita of CNY 115,000 (USD 17,400) and CNY 99,000 (USD 15,000) in 2019. These provinces are home to numerous small and medium-sized enterprises, with the network economy and private economy exhibiting a high level of activity and prosperity. INT has affected regional economies through spillover effects, which are more pronounced in developed provinces [1]. In addition, despite Chongqing’s location in the upper reaches of the Yangtze River, as one of the four municipalities in China, it has invested heavily in infrastructure to attract investment. Consequently, the city has been recognized as the “Economic and Technological Development Zone” and the “Hi-Tech Industry Development Zone” since 2013. These findings suggest that the REG-led INT growth hypothesis is supported by the four economically developed provinces, namely Shanghai, Jiangsu, Zhejiang, and Chongqing, and an interaction mechanism between REG and INT has been established. However, evidence to support this claim was not found in other less developed provinces in the YREB. Hence, these provinces should prioritize economic development to achieve the condition of REG-led INT growth. When formulating policies, the government should consider the bilateral dynamic nexus of INT and REG and strive to achieve a favorable state of sustainable economic growth and Internet development.

Finally, no Granger causality is detected between INT and REG in Jiangxi, Hubei and Sichuan. These provinces are in the middle or upper reaches of the Yangtze River, which are far away from the coast. The results imply that INT and REG in these areas are still independent, with no mutual promotion between them. One possible explanation for this finding is the lagging policy response or insufficient governance. Although these provinces have experienced rapid economic development and the deployment of broadband infrastructure between 1997–2019, there still exists a lack of sufficient dynamic development between INT and REG. For instance, the rate of Internet penetration in Jiangxi is only 56.2%, compared with 88.1% in Shanghai in 2019, and its GDP per capita accounts for less than one-third of Shanghai. Consequently, these regions’ governments should pay attention to dealing with the relationship between INT and REG, establishing close ties and promoting mutual development.

4.2. Bootstrap Panel Causality between URB and REG

The results of the Granger causality test for the relationship between URB and REG are presented in

Table 5. Results for panel Granger causality between URB and REG.

Provinces	H0: URB Does Not Cause REG				H0: REG Does Not Cause URB
	Wald Stat.	Bootstrap Critical Value			Wald Stat.	Bootstrap Critical Value
		1%	5%	10%		1%	5%	10%
Shanghai	27.42 **	34.97	15.75	10.24	0.24	42.03	20.63	13.43
Jiangsu	27.73 **	47.03	22.92	15.56	29.65 **	47.19	24.71	17.05
Zhejiang	33.67 *	34.70	18.04	12.44	9.17	34.85	18.93	13.35
Anhui	26.77 *	61.64	29.61	20.66	23.89 **	29.18	15.59	10.10
Jiangxi	15.61 *	43.45	25.53	15.27	2.4	28.19	15.14	9.79
Hunan	49.36 **	74.13	38.08	26.39	0.61	28.29	16.33	11.03
Hubei	4.02	46.74	22.62	15.07	23.68 *	48.70	24.99	17.06
Chongqing	24.85 **	40.18	19.49	13.16	9.36	28.43	14.93	9.98
Yunnan	24.23 *	48.82	26.71	18.95	3.67	24.72	12.76	9.14
Guizhou	0.41	39.20	19.28	12.62	7.47	24.97	13.25	9.77
Sichuan	15.05 *	41.30	21.59	14.69	4.48	27.73	14.40	9.52

Note: In this table, the Wald statistics and Bootstrap critical value of each province are reported. Bootstrap critical values are obtained from 10,000 replications. * and ** indicate significance at the 0.10 and 0.05 levels, respectively.

, while controlling for the influence of INT. The first column of the table indicates a statistically significant one-way Granger causality from URB to REG in most of the provinces along the YREB, with the exception of Hubei and Guizhou. This finding suggests that the transfer of rural labor to urban areas optimizes resource allocation and contributes to economic growth in most provinces in the YREB. Moreover, it indicates that the migration of labor from less efficient regions to more efficient ones can significantly enhance the efficiency of economic growth. China is currently one of the fastest urbanizing countries in the world, and it is predicted that the urbanization rate will reach 65% to 70% by the end of 2030 [32]. The results of the study demonstrate that the Chinese government’s promotion of urbanization has been effective in the YREB. However, it is important to note that the development of China’s urbanization is not balanced. The slow urbanization rate of Hubei and Guizhou has limited their ability to drive sustainable economic growth due to the lack of a free flow of surplus labor from rural to urban regions. Overall, the findings of the study for most of the provinces in the YREB support the urbanized economy theory proposed by Black and Henderson [48], which asserts that urbanization has played a critical role in economic growth. It supports the URB-led REG growth hypothesis.

The second column of the table confirms that there exists one-way Granger causality from REG to URB only in Jiangsu, Anhui and Hubei. The results suggest that the REG-led URB growth hypothesis does not hold for most provinces in the YREB, and there is no evidence to support it. Since the initiation of reform and opening up, China has encouraged and developed various forms of economic growth. Economic factors are increasingly gathering in urban cities as foreign investment and investment in advanced technologies is attracted and a large amount of surplus rural labor is transferred to urban regions. However, some regional policies have presented obstacles to urbanization, such as the household registration system and the urban–rural dual structure division [49]. These barriers hinder the free flow of labor and prevent the optimization of resource allocation, which is crucial for economic growth. Overall, the findings highlight that the government should further clear institutional barriers and relax household registration control to obtain REG-led URB growth in these provinces.

Third, there is a significant two-way Granger causality between URB and REG in Jiangsu and Anhui. This implies that the government policies implemented in these regions have had positive results in recent years. Jiangsu, an eastern province with a forward economy, boasted the highest REG in the Yangtze River Economic Belt (YREB) in 2019. The province has embraced an export-oriented economy and implemented laws that promote market development in the process of urbanization. The government of Jiangsu has introduced several preferential industrial policies, particularly in the service industry sector, resulting in a transformation of the employment structure and attracting rural populations to urban areas. As a consequence, Jiangsu’s urbanization rate increased from 29.9% in 1997 to 69.6% in 2019. Anhui, which borders Jiangsu to the east, lies across the basins of the Yangtze River and has a comparative advantage in manufacturing. It hosts the largest center of household electrical appliances in China, comprising the largest number of manufacturing centers of many renowned manufacturing enterprises. Hefei, the capital city of Anhui, is one of the largest intelligent manufacturing centers globally and attracts numerous rural workers to work in urban regions. The stable relationship observed between URB and REG in Jiangsu and Anhui suggests that economic growth and urbanization reinforce each other.

Last but not least, no Granger causality between URB and REG is found in Guizhou. It implies that URB did not contribute to REG, nor did REG contribute to URB in Guizhou over the period of 1997 to 2019. On one hand, Guizhou is a landlocked province with the lowest GDP per capita and urbanization level in the YREB. Compared to other provinces in China, Guizhou is relatively poor and has not benefitted substantially from Chinese technological progress and economic reform. On the other hand, the interaction mechanism between URB and REG is relatively complex, making it difficult for Guizhou to achieve URB-led REG growth or REG-led URB growth. However, Guizhou is relatively enriched in natural, environmental, and cultural resources, which the government could leverage to accelerate economic development and urbanization. Therefore, it is recommended that the government effectively harness these resources to boost regional development in Guizhou.

5. Conclusions and Policy Implications

5.1. Conclusions

This study utilizes a panel dataset covering 11 provinces in China’s YREB from 1997 to 2019 to investigate the Granger causality links among Internet development (INT), urbanization (URB) and regional economic growth (REG). We specifically adopt a bootstrap panel Granger causality test to explore the interaction patterns and direction between INT or URB and REG. The main conclusions obtained are as follows: (1) The analysis of Internet development, urbanization, and regional economic growth in the YREB reveals the presence of cross-sectional dependence and province-specific heterogeneity. It confirms the rationality and suitability of the bootstrap panel causality approach used in this study. (2) The results of the tested “INT–REG growth” nexus show that Granger causality from INT to REG exists in most provinces in China’s YREB, except for Jiangxi, Hubei and Sichuan. In turn, Granger causality from REG to INT, as well as two-way Granger causality between them, are confirmed in Shanghai, Jiangsu, Zhejiang and Chongqing. Finally, no Granger causality is detected between INT and REG in Jiangxi, Hubei and Sichuan. (3) In the investigation of the “URB–REG growth” nexus, we have identified that Granger causality from URB to REG exists in nine provinces along the YREB, with the exception of Hubei and Guizhou. Conversely, the causation direction only exists in Jiangsu, Anhui, and Hubei. Moreover, two-way Granger causality between URB and REG is only found in Jiangsu and Anhui, while no Granger causality between URB and REG is detected in Guizhou.

5.2. Policy Implications

Based on the above research conclusions, we provide the following management insights:

Firstly, the cross-sectional dependence of INT, URB, and REG suggests that the Chinese government should create a unified economic development management system to promote regional cooperation. Meanwhile, the province-specific heterogeneity of INT, URB, and REG implies that each province in the YREB should implement differentiated economic policies based on local conditions.

Secondly, the Granger causality from INT to REG reveals that INT has made a great contribution to REG in most YREB’s provinces. Thus, to foster economic growth, it is crucial that these provinces prioritize investment in the construction of broadband infrastructure. Moreover, the Granger causality from REG to INT and their two-way Granger causality in Shanghai, Jiangsu, Zhejiang and Chongqing indicate that provincial governments outside them should take into account the bilateral dynamic nexus between REG and INT. These governments should make concerted efforts to achieve a mutually beneficial state of sustainable economic growth and Internet development. Furthermore, the lack of Granger causality between INT and REG in Jiangxi, Hubei and Sichuan suggests that efforts should be made to bridge the gap and establish a connection between INT and REG in these three provinces.

Thirdly, the Granger causality from URB to REG in nine provinces suggests that the efficiency of REG can be significantly improved by URB, particularly when labor flows from a low-efficiency region to a relatively higher efficiency region. In addition, local governments in Hubei and Guizhou should further accelerate urbanization development. In addition, the impact of REG on URB is limited by regional bias policies, such as the household registration system. Therefore, local governments should aim to clear institutional barriers and relax household registration controls to promote economic growth. Finally, the two-way Granger causality and lack of Granger causality between URB and REG implies that virtuous cycle relationships between URB and REG have not been achieved in most provinces in the YREB. Hence, it is crucial to coordinate their relationships and achieve mutually reinforcing links between them through appropriate policy interventions.

There are several limitations to consider in this study, which open avenues for future research. First, although the Granger causality test provides insights into the relationships between INT, URB, and REG, it does not provide a comprehensive understanding of the underlying mechanisms. Future studies could explore the specific channels through which INT or URB influences REG and vice versa, shedding light on the intricate dynamics and processes involved. Second, this study focuses on the YREB, limiting the generalizability of the findings to other regions or economic belts in China. It would be valuable for future research to compare the relationships between INT, URB, and REG in different economic contexts, such as the Pearl River Delta economic belt or the Beijing–Tianjin–Tangshan economic belt. Additionally, the study primarily relies on existing indicators to assess Internet development. Future research could expand the scope of analysis by incorporating more comprehensive measures of Internet development. These could consider indicators related to Internet infrastructure, digital adoption rates, e-commerce activities, and digital innovation ecosystems.