To What Extent and How Does Internet Penetration Affect a Firm’s Upgrading in the Global Value Chain? Evidence from China

Abstract

Clarifying the upgrading implications of Internet penetration is essential in the digital transformation era. This paper examines to what extent and how Internet penetration affects Chinese firms’ GVC positions. We find that Internet penetration significantly improves firms’ GVC positions by lowering transaction costs and improving resource allocation efficiency. However, this improvement effect of Internet penetration varies among firms, wherein foreign-owned, private, medium, and large-sized, high-tech-intensive firms tend to benefit more from Internet penetration. This finding will help us improve firms’ value chain position in developing countries by promoting the popularization of the Internet, narrowing the gap between developing and developed countries, and enabling us to achieve the United Nations Sustainable Development Goals 8 and 9 as much as possible.

1. Introduction

Since the 1970s, Internet communication technology has undergone dramatic changes. The popularity of personal computers, the widespread use of the Internet, and the rise of the mobile Internet have significantly changed the business models of global companies [1,2]. The use of the Internet has dramatically altered our production and lifestyle. Research generally believes that the increase in Internet penetration has enhanced the welfare and productivity of individuals and firms and promoted disruptive innovation [3,4]. The development of Internet communication technology and disruptive innovation (such as digital technology and artificial intelligence) has brought new productivity to firms and promoted firms’ digital transformation and upgrading [5].

Today, digital transformation is recognized as the core of the Fourth Industrial Revolution (Industrial 4.0), significantly enhancing the productivity of investors [6]. Meanwhile, Industry 4.0 is disrupting industries across developing countries at an exponential rate [7]. Research indicates that Industry 4.0 has profoundly altered the competitive dynamics of multiple industries and corporate strategies [8]. Numerous scholars have explored the economic implications of Industry 4.0, including its effects on production efficiency and income inequality [9,10,11]. Delera et al. (2022) [12] argue that participation in global value chains (GVCs) has significantly fostered the development of Industry 4.0, while Manda and Ben (2019) [13] highlights that digital transformation has become a key strategy for achieving inclusive growth in developing countries.

However, the transformation brought about by Industry 4.0 has not only presented opportunities for developing countries but also posted significant challenges. Rodrik (2018) [14] presents a counterargument, suggesting that adopting new technologies makes it more difficult for developing countries to offset their labor cost advantages with technological advantages, thereby exacerbating the gap between developing and developed countries. Similar views are expressed by [15,16]. Matthess and Kunkel (2020) [17] argue that while digital transformation may initially facilitate labor transfer, it could also result in a shift of labor towards low-tech industries, thus slowing down the technological upgrading process. Particularly in African and other developing countries, there is a significant competitive disadvantage relative to developed nations [7].

Nevertheless, existing research often overlooks a critical factor: Internet penetration. All information and communication technologies (ICTs), digital technologies, and artificial intelligence (AI) rely on the foundational infrastructure of the internet [18]. As of 2017, the global Internet penetration was less than 50%, and even by 2024, the global Internet penetration was only 67.5% [19,20].

Therefore, exploring the impact of Internet penetration on firms’ positions in global value chains is crucial for understanding how internet access enhances corporate competitiveness and innovation capacity. As Fuente and Lampón (2020) [21] pointed out, upgrading the firm GVC position often depends on the local innovation infrastructure. For example, Nyagadza (2019) [22] believes that in South Africa, where the Internet penetration rate has reached 59%, there is a solid foundation for responding to Industry 4.0. However, we have to say this is still significantly lower compared to the nearly 100% Internet penetration rate in developed countries.

In the context of global value chains, the position of a firm within the value chain is closely related to its competitiveness, innovation capabilities, and market performance. According to Gereffi et al. (2005) [23], the governance structure of global value chains determines how firms allocate value and resources in the global economy. Baldwin (2016) [24] points out that the advancement of information technologies has enabled global firms to optimize their production and distribution models, enhancing their competitive advantage worldwide. As a study by Lampón et al. (2024) [25] shows, many traditions are not fields belonging to the same GVCs position, as the use of new information technologies becomes equivalent. The increase in Internet penetration can reduce transaction costs and improve resource allocation efficiency, enabling firms to occupy a more advantageous position in GVCs.

Further research shows that digital transformation increases production efficiency and influences firms’ profitability in developing countries in global value chains. Sako and Zylberberg (2019) [26] argue that firms participating in GVCs can enhance their GVCs position through technological innovation, strengthening their international competitiveness. Kummritz et al. (2017) [27] add that not all firms benefit equally from GVCs, with those in higher-value positions typically reaping more significant profits. As a key infrastructure driving this transformation, Internet penetration has become necessary in determining whether firms can experience rapid transformation. Hence, it is critical to comprehend the impact of Internet penetration on the position of firms in the GVCs and its mechanism.

This paper uses data from Chinese firms from 2000 to 2013 to analyze the impact of Internet penetration on a company’s position in the GVCs. The goal is to better reflect the reality of developing countries and provide essential insights for the global South. By studying China’s experience, we offer valuable lessons to other developing countries, supporting their economic growth and helping to close the development gap with advanced economies. Furthermore, Internet penetration enhances firms’ ability to innovate and integrate into more sustainable and high-value-added segments of GVCs, contributing to the achievement of SDG 9 on fostering inclusive and sustainable industrialization. It also creates opportunities for improved job quality and economic productivity, directly advancing SDG 8 by promoting inclusive growth and decent work.

Firstly, this paper demonstrates that Internet penetration improves firms’ positions in the global value chain, deepening our understanding of its impact on firms. This provides a foundation for government agencies to develop more effective industrial planning strategies.

Secondly, the paper highlights that Internet penetration enhances firms’ resource allocation capabilities and reduces information and transaction costs, further strengthening their position in the global value chain.

Thirdly, the study uncovers the heterogeneous effects of Internet penetration on different types of firms. It finds that foreign-funded, private, medium-sized, large-sized, and high-tech-intensive firms benefit more from Internet penetration.

Additionally, this study offers a theoretical basis for government policy formulation in the Internet industry, enabling developing countries better to leverage the opportunities and challenges of Industry 4.0. It also contributes to the existing literature by providing a new theoretical perspective on the relationship between the Internet and the global value chain. Thus, it offers a helpful reference for future research and policymaking.

This paper will explore the causal relationship between Internet penetration and firms’ positions in global value chains. The structure is as follows: Section 2 presents the hypothesis, Section 3 describes the model settings, Section 4 describes the data, Section 5 provides empirical analysis, and Section 6 concludes with recommendations.

2. Hypothesis

As Internet technology develops, its impact on firms’ access to resources and cost reduction profoundly impacts their position in the GVCs. This section will examine how the increase in Internet penetration affects firms’ position in the GVCs and through what mechanisms. Through these studies, we propose hypotheses to explain how Internet penetration affects firms’ position in the GVCs.

Firstly, we argue that the development of Internet technology has created excellent opportunities for firms to reduce costs and enter new markets. Firms that use these technological advances will gain more excellent benefits globally. A higher Internet penetration, especially in developing countries, enables firms to better integrate into the GVCs. Therefore, Internet penetration may improve firms’ operational efficiency and allow them to occupy a more advantageous position in the global value chain.

Based on the above arguments, we hypothesize that as Internet penetration increases, firms can use digital technology better to access international markets and resources, thereby improving their position in the GVCs. Therefore, Hypothesis 1 is that an increased Internet penetration will improve firms’ position in the GVCs.

Secondly, another thing that needs to be considered is the role of transaction costs in shaping firms’ participation in the GVCs. Transaction costs are a key factor that affects whether firms can integrate into GVCs. Transaction costs, such as information delays, information asymmetry, and intermediary fees, often hinder the efficiency of firms in the global market. An increased Internet penetration can reduce firm transaction costs by speeding up communication, improving information transparency, and simplifying interactions with suppliers and customers. Reducing firm transaction costs enables them to integrate more advantageously into the GVCs.

Therefore, we assume that Internet penetration reduces transaction costs by improving communication efficiency between firms and global partners. Through more efficient communication, firms can reduce the costs associated with communication and negotiation. Reducing transaction costs will improve firms’ competitiveness and enhance their ability and position to participate in the GVCs. Therefore, Hypothesis 2 is that Internet penetration can improve firms’ positions in the GVCs by reducing transaction costs.

Lastly, the ability of firms to allocate resources is also a key to whether they can effectively participate in the GVCs. Resource allocation includes managing human, financial, material, and information resources, which is very important for the competitiveness of firms. The popularization of the Internet enables firms to improve their resource allocation capabilities by providing information. With an increased Internet penetration, firms can more effectively combine their businesses, optimize resource utilization, and quickly reply to market demands. Improved resource allocation can improve efficiency and innovation capabilities and promote firms to occupy a higher position in the GVCs.

Based on the above analysis, we hypothesize that Internet penetration can improve firms’ resource allocation capabilities, enabling them to obtain a higher position in GVCs. Firms can use a higher Internet penetration to optimize operations, integrate firm resources more effectively, and produce competitive products. This enhanced resource allocation can help firms improve their global value chain position. Therefore, Hypothesis 3 is that Internet penetration can strengthen the position of firms in the GVCs by improving their resource allocation capabilities.

This section explores the relationship between Internet penetration and a firm’s position in the GVCs. We discuss two mechanisms through which Internet penetration can improve a firm’s position in the GVCs: lower transaction costs and improved resource allocation capabilities. Through these two mechanisms, firms can better integrate into the GVCs, which is particularly important for promoting the position of firms in GVCs. The hypotheses proposed in this section will serve as a theoretical basis for further empirical testing and help to understand how increased Internet penetration affects a firm’s position in the global value chain.

3. Model Settings

Our model setting is mainly divided into three parts. The first is the setting of the basic model, the measurement model of the firm’s value chain position, and the last is the measurement model of the intermediary utility. We use these models widely used in economic research to ensure the research method’s reliability and the empirical results’ comparability. This can effectively support us in verifying the impact of Internet penetration on improving the GVCs position of Chinese firms and provide empirical evidence and inspiration for developing countries.

This study extended the previous provincial-level research to the city level. This will help us to find out the differences between cities more deeply and make the research more comprehensive so that more targeted and reliable policy recommendations can be put forward.

3.1. The Basic Model

To analyze the impact of Internet technology adoption on the position of firms within the GVCs, we construct the following econometric model:

$${gvc}_{ijt}={\alpha }_0+{\alpha }_1{int}_{jt}+{\alpha }_2{X}_t+v_i+c_j+y_t+{\epsilon }_{ijt}$$

(1)

where ${gvc}_{ijt}$ represents the GVCs position of firm $i$ in city $j$ during period $t$; ${int}_{jt}$ denotes the level of Internet technology adoption in city $j$ during period $t$; $X_t$ is a vector of control variables; $v_i$, $c_j$, and $y_t$ are fixed effects for the firm, city, and time period, respectively, and ${\epsilon }_{ijt}$ represents the error term. Here, the fixed effects of firms can reduce the interference of constant factors between firms, the fixed effects of cities consider the possible heterogeneity between different cities, and the fixed effects of time control the impact of sample time changes on the dependent variable. Introducing these fixed effects can more accurately estimate the impact of Internet penetration on the position of enterprises in GVCs. This method is similar to that of Kurtulus and Tomaskovic-Devey (2012) [28], Xue et al. (2021) [29] and Park et al. (2023) [30].

We adopt the method proposed in Chor et al. (2021) [31] to measure firm’s GVC position, which can be seen below, as follows:

$$P_{it}^X={\sum }_{j=1}^N{\displaystyle \frac{X_{ijt}}{X_{it}}}{OU}_{jt}$$

(2)

Here, $P_{it}^X$ represents the GVCs position of firm $i$ in year $t$, and ${OU}_{jt}$ represents the GVCs position at the sector level.

3.2. The Measurement Model of the Firm’s GVCs Position

First, we measure the GVCs position at the sector level. Consider a world consisting of G countries (or regions), where each region has N sectors. The output of each sector can be allocated in two distinct ways. It may either flow directly to the final demand or serve as an intermediate input for subsequent sectors. The basic structure of these global input–output tables is outlined in

Table 1. The basic structure of input–output tables.

		Intermediate Use				Final Use				Total Output
		Region 1	Region 2	$\cdots$	Region G	Region 1	Region 2	$\cdots$	Region G
Intermediate Input	Region 1	${\mathbf{Z}}^{11}$	${\mathbf{Z}}^{12}$	$\cdots$	${\mathbf{Z}}^{1G}$	${\mathbf{f}}^{11}$	${\mathbf{f}}^{12}$	$\cdots$	${\mathbf{f}}^{1G}$	${\mathbf{x}}^1$
	Region 2	${\mathbf{Z}}^{21}$	${\mathbf{Z}}^{22}$	$\cdots$	${\mathbf{Z}}^{2G}$	${\mathbf{f}}^{21}$	${\mathbf{f}}^{22}$	$\cdots$	${\mathbf{f}}^{2G}$	${\mathbf{x}}^2$
	$\cdots$	$\cdots$	$\cdots$	$\cdots$	$\cdots$	$\cdots$	$\cdots$	$\cdots$	$\cdots$	$\cdots$
	Region G	${\mathbf{Z}}^{G1}$	${\mathbf{Z}}^{G2}$	$\cdots$	${\mathbf{Z}}^{GG}$	${\mathbf{f}}^{G1}$	${\mathbf{f}}^{G1}$	$\cdots$	${\mathbf{f}}^{GG}$	${\mathbf{x}}^G$
Value Added		${\left({\mathbf{v}}^1\right)}^{\prime }$	${\left({\mathbf{v}}^2\right)}^{\prime }$	$\cdots$	${\left({\mathbf{v}}^G\right)}^{\prime }$
Total Input		${\left({\mathbf{x}}^1\right)}^{\prime }$	${\left({\mathbf{x}}^2\right)}^{\prime }$	$\cdots$	${\left({\mathbf{x}}^G\right)}^{\prime }$

.

When the market is clear, the input–output model (Miller and Blair, 2009) [32] can be expressed as follows:

$$\left[\begin{array}{c}{\mathbf{x}}^1\\ ⋮\\ {\mathbf{x}}^G\end{array}\right]=\left[\begin{array}{ccc}{\mathrm{A}}^{11}& \cdots & {\mathrm{A}}^{1\mathrm{G}}\\ ⋮& \ddots & ⋮\\ {\mathrm{A}}^{\mathrm{G}1}& \cdots & {\mathrm{A}}^{\mathrm{G}\mathrm{G}}\end{array}\right]\left[\begin{array}{c}{\mathbf{x}}^1\\ ⋮\\ {\mathbf{x}}^G\end{array}\right]+\left[\begin{array}{c}{\mathbf{f}}^1\\ ⋮\\ {\mathbf{f}}^G\end{array}\right]$$

(3)

Let $\mathbf{x}$ be the vector of the global total output, where all outputs of economy k are recorded as ${\mathbf{x}}^k$ (N*1 vector). The direct input coefficient matrix is denoted as A, and the global final demand vector as f, where ${\mathbf{f}}^k$ represents the final demands of economy k. Formula (3) can thus be reformulated as follows:

$$\mathbf{x}=\mathbf{A}\mathbf{x}+\mathbf{f}={\left(\mathbf{I}-\mathbf{A}\right)}^{-1}\mathbf{f}=\mathbf{L}\mathbf{f}$$

(4)

Here, ${\mathbf{L}=\left(\mathbf{I}-\mathbf{A}\right)}^{-1}$ is the famous Leontief inverse matrix.

According to Miller and Temurshoev (2017) [33], and Antràs and Chor (2018) [34], the position index (upstreamness) of a particular sector of a country can be calculated as follows:

$$\begin{array}{c}{OU}_m^i=1·{\displaystyle \frac{f_m^i}{x_m^i}}+2·{\displaystyle \frac{{\sum }_{n,j}{a}_{mn}^{ij}{f}_n^j}{x_m^i}}+3·{\displaystyle \frac{{\sum }_{n,j}{\sum }_{s,k}{a}_{mn}^{ij}{a_{ns}^{jk}f}_s^k}{x_m^i}}\\ +4·{\displaystyle \frac{{\sum }_{n,j}{\sum }_{s,k}{\sum }_{t,l}{a}_{mn}^{ij}{a}_{ns}^{jk}{a}_{st}^{kl}{f}_t^l}{x_m^i}}+\cdots \end{array}$$

(5)

In addition, Formula (5) can be expressed in the form of a matrix as follows:

$$\mathbf{O}\mathbf{U}={\widehat{\mathbf{x}}}^{-1}\left(\mathbf{f}+2\mathbf{A}\mathbf{f}+3{\mathbf{A}}^2\mathbf{f}+\cdots \right)={\widehat{\mathbf{x}}}^{-1}{\mathbf{L}}^2\mathbf{f}$$

(6)

Formula (6) is the calculation formula for the position index. It represents the distance between the sector m in region i and the final consumption end, thereby indicating its position in the global value chain.

Next, to further expand the global value chain position from the sector level to the firm level, this paper will refer to Chor (2021) [31], and give the calculation formula as follows:

$$P_{it}^x={\sum }_{j=1}^N{\displaystyle \frac{x_{ijt}}{x_{it}}}{OU}_{jt}$$

(7)

In Formula (7), $P_{it}^x$ represents the GVC position of firm $i$ in $t$ year, $x_{ijt}$ is the total export value of sector $j$ of firm $i$ in $t$ year, $x_{it}$ is the sum of all exports of firm $i$ in the $t$ year, and ${OU}_{jt}$, as calculated above, represents the upstreamness index of sector $j$ in the region in the $t$ year. That is, upstreamness of a firm is the weighted upstreamness index of the sector in which the firm’s various export products are located, and the weight is the proportion of the firm’s products in each sector. This calculation method considers and reflects the overall export structure of the firm in a relatively comprehensive manner, so we can calculate the GVC position of the firm.

3.3. The Measurement Model of the Intermediary Utility

In addition, based on Atkeson and Burstein (2008) [35] and Czernich et al. (2011) [36], we believe that the popularization of the Internet may promote the position of firms in the GVCs by reducing transaction costs and improving resource factor allocation. Therefore, we set up mediation effect detection models (8) and (9) based on the baseline model.

$$M_{jt}={\alpha }_0+{\alpha }_1{int}_{jt}+{\alpha }_2{X}_t+v_i+c_j+y_t+{\epsilon }_{ijt}$$

(8)

$${gvc}_{ijt}={\alpha }_0+{\alpha }_1{int}_{jt}+{{\alpha }_{2}M}_{jt}+{\alpha }_3{{int}_{jt}\ast M}_{jt}+{\alpha }_4{X}_t+v_i+c_j+y_t+{\epsilon }_{ijt}$$

(9)

Among them, $M_{jt}$ is the mediating variable. Model (8) detects the impact of the Internet penetration (int) on the mediating variable, and model (9) detects the mediating effect of the mediating variable in the process of the Internet penetration rate affecting the position of the firm GVCs.

4. Data Description and Treatment

4.1. Data Description

The data used in this study is obtained from three authoritative sources. The China Industrial Enterprise Database and China City Statistical Yearbook are both provided by the National Bureau of Statistics of China (NBS) https://www.stats.gov.cn/english/. The China Customs Database is sourced from the General Administration of Customs of the People’s Republic of China (Customs Statistics) http://stats.customs.gov.cn/indexEn. The World Input–Output Database (WIOD) provides detailed input–output data for global economic analysis https://www.rug.nl/ggdc/valuechain/wiod/wiod-2016-release (accessed on 10 December 2024). The merged dataset’s temporal scope is restricted to 2000 to 2013, corresponding to the period for which the China Industrial Enterprise Database provides data. The descriptive statistics of each variable are shown in

Table 2. Descriptive statistics.

Variable Name	Variable Meaning	Samples	Mean	SD	Min	Max
lngvc	Global value chain position	640,840	1.024	0.187	0.688	1.473
int	Urban Internet Penetration Rate	3434	0.090	0.118	0.004	2.004
lnage	Company age	640,840	2.128	0.653	0.000	5.136
lnwage	Human capital level	640,840	2.543	2.042	−11.118	9.963
lncap	Capital Intensity	640,840	3.830	1.452	−5.908	11.646
lnsize	Firm asset size	640,840	10.742	1.460	7.847	14.890
liquidity	Liquidity	640,840	0.620	0.219	0.088	1.000
mobilephone	Urban mobile Internet penetration rate	3434	0.576	0.705	0.054	9.255
prop_int	Proportion of Internet employees in cities	2934	0.001	0.003	0.000	0.045
lntelebusiness	Telecommunications business Volume per capita in cities	3433	−3.125	0.924	−6.758	1.137
lnpop_city	Urban population density	3500	5.740	0.897	1.609	7.904
lngdp_city	Gross urban product	3500	15.627	1.050	12.669	19.191

Note: The different number of data samples at the city level in the table is due to missing data in the China City Statistical Yearbook.

.

The control variable group includes the following:

• Firm age (age), representing the time the firm has survived, which is expressed by the natural logarithm of the current year minus the year of establishment plus one.
• Firm human capital level (wage), reflecting the human capital situation of the firm, which is expressed in the logarithm of per capita wage.
• Firm capital intensity (cap), representing the relative situation of capital and labor of the firm, which is expressed in the logarithm of the ratio of fixed assets to employment.
• Firm asset scale (size), reflecting the firm’s total asset size, which is expressed in the logarithm of the firm’s total asset.
• Firm asset liquidity (liquidity), reflecting the operation of the firm’s assets, which is expressed in the ratio of the firm’s current assets to its total assets.

This paper also employs a 1% trimming method for all continuous variables to minimize the influence of outliers.

4.2. Data Treatment

This paper mainly relies on three databases when calculating GVC position. The first is the WIOD input–output database, which calculates the global value chain position at the manufacturing sector level mentioned above. The second is the China Customs Database. This paper needs each firm’s import and export trade information provided in the customs database. The third is the China Industrial Enterprise Database. For the needs of subsequent research, this paper selects firm samples that appear in both the Customs and Industrial Enterprise databases to ensure the integrity of their data information.

In the data treatment process of this study, the most critical link is to connect the three databases effectively. We will connect these three databases and calculate the GVCs position in three steps: integrating the databases, cleaning and adjusting the data, and calculating the GVCs position at the firm level.

Step one: the Integration of the Databases

Firstly, this involves connecting the three databases effectively. Firms from China Customs and Industrial Enterprise databases are matched using their unique Chinese names. This method ensures a high accuracy because firm names are unique when registered. The approach has been validated by previous studies, such as Upward et al. (2013) [37].

Next, the above integrated database is linked to the WIOD database, which operates at the sector level using the ISIC classification system. However, since the China Customs database uses HS codes to define commodities, we match HS codes to ISIC classifications through a mapping table (

Table A1. Sector classification, ISIC classification, and HS code matching table.

Sector Name	ISIC	HS Code
Food, beverage, and tobacco products industry	C10-12	Chapter 2, 03.03-06 (only until 03061990), Chapter 4 except 04.09-10, 07.10-12, 08.11-12, 08.14, 09.01, 10.06, Chapters 15-21, Chapter 22 except 22.09 (including 22.09.02), Chapter 23, 24.02-03.
Textile raw materials and textile products, leather and footwear manufacturing	C13-15	Chapters 50-53, 56-65, 41-43 (except 41.01-03, 43.01), 67 except 67.02, 94.04.
Wood, cork, and products thereof	C16	Chapter 44, 45.01-04, Chapter 46.
Paper Industry	C17	Chapter 47, and Chapter 48 except 48.20.
Printing Industry	C18	Chapter 49.
Coke, petroleum refining, and nuclear fuel processing industries	C19	27.04, 27.06-08, 27.10-13, 27.15.
Chemical raw materials and chemical products (including chemical fibers)	C20	15.18, 15.20, Chapters 28-29, Chapters 31-38, 39.01, 40.02, Chapters 54-55.
Pharmaceutical Manufacturing	C21	Chapter 30.
Rubber and plastic products	C22	39.02, Chapter 40 except 40.01-02.
Non-metallic mineral products	C23	Chapters 68-70, 90.03-04.
Basic metal manufacturing	C24	26.18-19, Chapter 72, 74.01-74.10, 75.01-75.06, 76.01-76.07, 78.01-78.04, 79.01-79.05, 80.01-80.05, Chapter 81.
Metal products industry	C25	66.01, Chapter 73, 74.11-19, 75.07-08, 76.08-16, 78.05-06, 79.06-07, 80.06-07, Chapters 82-83, 94.06.
Electronic communications, optical instruments and equipment	C26	84.23, 84.69, 84.70-73, 85.17-29, 85.40-43, 90.01-02, 90.05-17, 90.23-33, Chapter 91.
Electrical equipment	C27	84.15, 84.50, 85.01-16, 85.30-39, 85.44-48, 94.05.
Machinery and equipment manufacturing industry	C28	84.01-14, 84.16-22, 84.24-49, 84.51-68, 84.74-79, 84.80-85, 90.18-22.
Automotive Manufacturing	C29	Chapter 87.
Other transportation manufacturing	C30	Chapter 86, Chapter 88-89.
Furniture and other manufacturing industries	C31-32	48.20, Chapters 92-93, 94.01-94.03, Chapter 95

in Appendix A).

Step two: Data Cleaning and Adjustments

Data cleaning and adjustments are essential for ensuring the accuracy and reliability of the analysis. Our process follows the methodology outlined by Ahn et al. (2011) [38].

In the China Customs database, records with missing critical information, such as firm names, HS codes, or transaction amounts, were removed. Additionally, monthly trade data were aggregated into annual figures. A two-step adjustment process was applied to address the impact of intermediary firms. Firstly, intermediary firms were identified based on keywords in their names (e.g., “trade” and “economic and trade”), and their trade volumes were aggregated. Secondly, the export volumes of each sector were adjusted using a formula that accounts for the share of intermediary trade as follows:

$${IE}_{ijt}^{adj}={IE}_{ijt}/(1-{share}_{jt}^E)$$

(10)

Among them, ${IE}_{ijt}^{adj}$ represents the actual export value of firm $i$ in sector $j$ after correction, ${IE}_{ijt}$ represents the export value of firm $i$ in sector $j$ in the database before correction, and ${share}_{jt}^E$ represents the proportion of export value through trade agency to the total export value of the sector, which can be obtained from the calculation in the first step. The above two steps complete the actual estimation of the firm’s export value.

In addition, the China Industrial Enterprise database retained only active manufacturing firms with valid data. Firms with critical financial indicators (e.g., sales and assets) reported as zero or negative or fewer than eight employees were excluded as outliers.

Step three: Calculating Firm-Level GVC Positions

In GVC analysis, the heterogeneity of firms is a crucial factor influencing their position within the GVCs. This study presents a multidimensional classification of firm heterogeneity. It integrates existing formulas (A5) to suggest an enhanced calculation method for assessing enterprise position in the GVCs as follows:

$$P_{k\_t}^X={\sum }_{j-1}^N{\displaystyle \frac{X_{k\_jt}}{X_{k\_t}}}{OU}_{jt}$$

(11)

Corresponding to the symbols in Formula (7), $X_{k\_jt}$ is the total export volume of all firms of this type in sector j under the k firm heterogeneity condition, $X_{k\_t}$ is the total export volume of all firms of this type under the k firm heterogeneity condition, and ${OU}_{jt}$ has the same meaning, representing the upstreamness index of sector j in the country or region at period t. Therefore, this paper can analyze the global value chain position of various firms under different firms’ heterogeneity by calculating and summarizing them.

To better analyze the position of different types of firms in the GVCs, this paper selects five dimensions, namely, the region of the firm, and the ownership, trade pattern, firm scale, and technology intensity, for research based on firms’ heterogeneous characteristics. These classification methods of heterogeneity have been widely used in econometric models in applied economics [39,40,41,42].

Specifically, the firm’s region is distinguished by the place of firm registration to examine the impact of regional differences on the position of GVCs. At the ownership level, firms are divided into three categories: state-owned, private, and foreign-funded. The aim is to explore the potential differences in the position of firms with different ownership structures in GVCs. Trade patterns are divided into general trade and processing trade to analyze the differences in the position of firms in GVCs under different transaction modes. Firm scale is divided into three categories, large, medium, and small, to examine the changes in the position of firms of various sizes in GVCs. Finally, according to the technology intensity of the sector in which the firm is located, the firm is divided into three categories: high, medium, and low technology intensity, to study the impact of technology intensity on the position of the firm in GVCs.

The firm heterogeneity classification criteria are presented in Appendix A.

5. Empirical Analysis

In the empirical analysis, we mainly include five parts, namely the following: the Benchmark Results, Robustness Test, Endogeneity Test, Mechanism Test, and Heterogeneity Analysis.

5.1. Benchmark Results

The benchmark result of Internet penetration on the position of firms in the GVCs are shown in

Table 3. Benchmark regression results.

	(1)	(2)	(3)	(4)
VARIABLES	lngvc	lngvc	lngvc	Lngvc
int	0.0522 ***	0.0132 ***	0.0461 ***	0.0128 ***
	(0.0008)	(0.0006)	(0.0008)	(0.0006)
lnage			0.0166 ***	0.0027 ***
			(0.0004)	(0.0004)
lnwage			−0.0017 ***	−0.0002 **
			(0.0001)	(0.0001)
lncap			0.0294 ***	0.0010 ***
			(0.0002)	(0.0001)
lnsize			0.0070 ***	−0.0043 ***
			(0.0002)	(0.0002)
Liquidity			0.0545 ***	−0.0020 ***
			(0.0012)	(0.0006)
Constant term	1.0095 ***	1.0191 ***	0.7586 ***	1.0572 ***
	(0.0003)	(0.0002)	(0.0019)	(0.0026)
Time-fixed effects	N	Y	N	Y
Firm-fixed effects	N	Y	N	Y
City-fixed effects	N	Y	N	Y
N	640,494	610,891	640,482	610,879
R-squared	0.0052	0.9248	0.0692	0.9249

Note: The values in parentheses are standard errors, and *** and ** indicate significance at the 1% and 5% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “N” in the fixed effect means no control; “Y” means control.

. In column (1), we report the results obtained by including only the main explanatory and dependent variables. Subsequently, we incorporate fixed effects and control variables, with the results displayed in columns (2), (3), and (4). Our findings indicate that regardless of how we adjust the control variables and various fixed effects, the increase in the urban Internet penetration rate significantly benefits the position of firms in the GVCs, with a confidence level of 1%. This suggests that improving the Internet penetration can help firms improve their standing in the GVCs.

5.2. Robustness Test

We further conducted a series of robustness tests to verify whether the research results are reliable and consistent. Firstly, to assess the stability of the results, we used proxy variables (the mobile Internet penetration rate and urban telecommunications business volume) to replace the main explanatory variables. Then, we replaced the main variables explained with substitute variables. Following the method outlined by Wang et al. (2022) [43], our substitute variables use the ratio of upstream import degree to downstream import degree as an indicator of global value chain (GVC) status.

The results shown in

Table 4. Robustness test results (I).

	(1)	(2)	(3)	(4)
	Replace The Main Explanatory Variable			Replace the Explained Variable
VARIABLES	lngvc	lngvc	lngvc	lngvc_pos
mobilephone	0.0025 ***
	(0.0001)
prop_int		0.3273 ***
		(0.0278)
lntelebusiness			0.0018 ***
			(0.0003)
int				0.0093 ***
				(0.0006)
Control variables	Y	Y	Y	Y
Time-fixed effects	Y	Y	Y	Y
Firm-fixed effects	Y	Y	Y	Y
City-fixed effects	Y	Y	Y	Y
N	610,879	568,370	610,860	610,879
R-squared	0.9249	0.9254	0.9248	0.9152

Note: The values in parentheses are standard errors, and *** indicates significance at the 1% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “Y” means control.

show that the coefficients of the main explanatory variables are consistent across different specifications and are statistically significant at the 1% level, which enhances the robustness of our research results.

We use a two-dimensional fixed effects (FE) model that accounts for firm- and time-specific variations to mitigate the impact of regression methods further, sample selection errors, and omitted variables. Meanwhile, we also include city-specific fixed effects to account for city-level heterogeneity. This method helps to control for unobserved factors varying across cities and time, thereby providing more precise estimates. In addition, to address potential heteroskedasticity issues, we use Poisson quasi-maximum likelihood (PPML) estimation. This estimation approach can address potential heteroskedasticity effects [44]. The results in columns (1) and (2) of

Table 5. Robustness test results (II).

	(1)	(2)	(3)	(4)
	FE Test	PPML Test	Considering Sample Selection Bias	Considering Omitted Variables
VARIABLES	lngvc	lngvc	lngvc	lngvc
int	0.0128 ***	0.0130 ***	0.0172 ***	0.0131 ***
	(0.0008)	(0.0006)	(0.0008)	(0.0006)
lnpop_city				−0.0060 ***
				(0.0022)
lngdp_city				0.0008
				(0.0012)
Control variables	Y	Y	Y	Y
Time-fixed effects	Y	Y	Y	Y
Firm-fixed effects	Y	Y	Y	Y
City-fixed effects	N	Y	Y	Y
N	640,482	610,879	544,228	610,879
R-squared	0.3581		0.9255	0.9249

Note: The values in parentheses are standard errors, and *** indicates significance at the 1% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “Y” means control.

prove that the coefficients of the main explanatory variables remain statistically significant at the 1% level, therefore validating the robustness of the initial model specification.

Then, to consider the possible impact of outliers or changes in statistical properties over time, we exclude data collected after 2010, which may be influenced by changing reporting standards or statistical anomalies. The results in column (3) of Table 5 show that there has not been any significant change in the direction or statistical significance of the coefficients, additionally supporting the long-term stability of our results.

Finally, to examine the potential impact of other key city-level factors, we incorporate additional control variables in column (4) of Table 5, specifically the logarithm of city output and population density. Once again, the coefficients of the main explanatory variables remain consistent in both direction and statistical significance, further strengthening the overall robustness of our results.

Given that the data utilized in this study is constrained to enterprise-level data from 2000 to 2013, concerns may arise regarding the timeliness of the research findings. To address this issue and verify the robustness of our conclusions, we conducted an additional robustness test using sector-level data from 2008 to 2023. The input–output table data comes from ADB Multiregional Input–Output Tables at constant prices (https://kidb.adb.org/globalization). Specifically, we employed input–output tables to calculate the GVCs’ positions of 35 industries in China, which serve as the dependent variable. Since the Internet penetration rate in the enterprise-level dataset is a city-level indicator and is not applicable at the sector level, we use the output value of the telecommunications sector utilized by various industries as an alternative explanatory variable to assess the impact of Internet use on sector-level GVC positions. Based on the sector classification in the input–output tables, we selected two telecommunications-related industries, “Electrical and Optical Equipment” and “Post and Telecommunications.” We used their respective output values and their combined total as core explanatory variables in the regression analysis.

The results of the robustness test (see Columns (1), (2), and (3) of

Table 6. Robustness test results (III).

	(1)	(2)	(3)	(4)	(5)	(6)
VARIABLES	lngvc	lngvc	lngvc	lngvc	int lngvc	lngvc
E and EO Output	0.1943 ***			0.1827 ***
	(0.0477)			(0.0432)
P and T Output		0.1453 ***			0.1247 ***
		(0.0422)			(0.0413)
Total Output			0.2167 ***			0.2041 ***
			(0.0474)			(0.0453)
Constant term	1.1134 ***	1.5996 ***	0.8206 **	1.7844 **	2.9870 ***	2.7465 ***
	(0.3672)	(0.2907)	(0.3921)	(0.6885)	(0.6575)	(0.6046)
	N	N	N	Y	Y	Y
Control variables	N	N	N	Y	Y	Y
Time-fixed effects	Y	Y	Y	Y	Y	Y
Sector-fixed effects	Y	Y	Y	Y	Y	Y
N	560	560	560	560	560	560
R-squared	0.4682	0.4349	0.4912	0.4820	0.4373	0.4908

Note: The values in parentheses are standard errors, and *** and ** indicate significance at the 1% and 5% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “N” in the fixed effect means no control; “Y” means control.

) indicate that Internet use exerts a significant and positive effect on sector-level GVCs positions at the 1% confidence level, reinforcing the core findings of the baseline regression. Furthermore, drawing on the methodologies of Wang et al. (2017) [45], Lee and Yi (2018) [46], and Zhao et al. (2022) [47], we incorporated four additional control variables into the model: the sector value added, national GDP, sector wage levels, and sector education levels. The data sources for these control variables are obtained from the China Securities Markets and Accounting Research Database (CSMAR) (https://data.csmar.com). The regression results, after accounting for these control variables (see Columns (4), (5), and (6) of Table 6), confirm that Internet use continues to play a statistically significant role in enhancing sector-level GVC positions at the 1% confidence level. These findings suggest that even within a more recent timeframe, the role of Internet use in facilitating sector upgrading within GVCs positions remains significant, strengthening the relevance and applicability of our research conclusions.

In summary, across various alternative specifications and robustness checks, our results consistently demonstrate that the key explanatory variables are robust, significant, and reliable, supporting the conclusions drawn in the primary analysis.

5.3. Endogeneity Test

As previously discussed, our primary explanatory variables are derived from city-level data, while the explained variables are based on firm-level microdata. This distinction helps mitigate the potential issue of endogeneity to some extent, as using city-level variables reduces the likelihood of reverse causality or omitted variable bias between the explanatory and dependent variables.

Firstly, we use the one-period lagged urban Internet penetration rate as an instrumental variable and eliminate endogenous interference using the second order least squares method, which was also used in Lin and Huang’s 2023 study [48].

Secondly, given the complexity of the relationships in the study and the potential for residual endogeneity, we take additional steps to address this concern more rigorously. To further control for potential endogeneity, we employ two distinct instrumental variables: the number of fixed-line telephones and the number of post offices. Both variables will likely affect economic activity and communication infrastructure in the cities. Still, they are less likely to be directly influenced by the firm-level outcomes we are examining, making them suitable candidates for instrumental variables.

In

Table 7. Endogeneity test results.

	(1)	(2)	(3)	(4)	(5)	(6)
	Phase 1	Phase 2	Phase 1	Phase 2	Phase 1	Phase 2
VARIABLES	int	lngvc	int	lngvc	int	lngvc
L.int	0.4830 ***
	(0.0049)
fixphone			0.1058 ***
			(0.0147)
postoffice					0.0314 ***
					(0.0006)
int		0.0174 ***		0.0286 ***		0.0264 ***
		(0.0014)		(0.0029)		(0.0041)
Control variables	Y	Y	Y	Y	Y	Y
Time-fixed effects	Y	Y	Y	Y	Y	Y
Firm-fixed effects	Y	Y	Y	Y	Y	Y
City-fixed effects	Y	Y	Y	Y	Y	Y
Kleibergen–Paap rk LM Test	20,864.60 [0.0000]		4989.14 [0.0000]		8045.08 [0.0000]
Kleibergen–Paap Wald rk F Test	9572.08 {16.38}		52.05 {16.38}		2927.17 {16.38}
N	456,749		568,371		610,879

Note: The values in parentheses are standard errors, and *** indicates significance at the 1% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “Y” means control. The values in [ ] are the p-values of the test statistics, and the values in { } are the critical values of the Stock–Yogo test at the 10% level.

, we present the first-stage regression results and the Kleibergen–Paap Wald F test, commonly used to assess the strength of instrumental variables. The F-statistic for the first stage exceeds ten, indicating that the instruments are not weak and are highly correlated with the endogenous regressors. This suggests that the instruments are appropriate and valid for addressing endogeneity concerns in our model.

In addition, both instrumental variables pass the over-identification test and reach statistical significance at 1% in each stage. This confirms that the number of landlines and post offices are valid instrumental variables. They effectively alleviate the potential endogeneity problem in the second-stage regression. The instrumental variable regression results are significant, strengthening our confidence in them and confirming that this study’s results are not caused by endogeneity bias.

In short, by using instrumental variables and related diagnostic tests, we have reduced the impact of endogeneity on the results. The strength and validity of the instruments used further support the robustness of our findings, allowing us to draw more reliable conclusions about the relationships explored in this paper.

5.4. Mechanism Test

Internet penetration enhances firms’ position in GVCs by reducing transaction costs and improving resource allocation.

Firstly, as shown in

Table 8. Mechanism analysis results (I).

	(1)	(2)
VARIABLES	Cost	lngvc
int	−0.0080 ***	0.0106 ***
	(0.0008)	(0.0008)
cost		−0.0063 ***
		(0.0013)
int × cost		0.0125 ***
		(0.0030)
Control variables	Y	Y
Time-fixed effects	Y	Y
Firm-fixed effects	Y	Y
City-fixed effects	Y	Y
N	610,866	610,866
R-squared	0.7203	0.9249

Note: The values in parentheses are standard errors, and *** indicates significance at the 1% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “Y” means control.

, an increased Internet penetration significantly reduces transaction costs, and the findings are robust and consistent at the 1% significance level. This result suggests an increased Internet penetration enhances firms’ position in GVCs by reducing transaction costs. Specifically, an increased Internet penetration promotes a more efficient information exchange, enabling firms to streamline processes, negotiate better terms, and reduce costs associated with contract coordination, monitoring, and enforcement [49]. Through digital platforms, firms can quickly obtain information about suppliers, customers, and market conditions, reducing transaction-related uncertainty and friction, especially in cross-border transactions [50] Reduced transaction costs can improve firms’ cost advantages and competitiveness in national markets. In this way, firms can better integrate and thrive in GVCs and enhance their position in GVCs.

Secondly, the regression results show that the increase in Internet penetration positively impacts the labor income share within the firm. As shown in

Table 9. Mechanism analysis results (II).

	(1)	(2)	(3)
VARIABLES	Employees	ls	lngvc
int	−0.0889 ***	0.0030 ***	0.0089 ***
	(0.0040)	(0.0007)	(0.0008)
ls			−0.0117 ***
			(0.0019)
lnt × ls			0.0261 ***
			(0.0039)
Control variables	Y	Y	Y
Time-fixed effects	Y	Y	Y
Firm-fixed effects	Y	Y	Y
City-fixed effects	Y	Y	Y
N	610,879	610,879	610,879
R-squared	0.9160	0.7294	0.9249

Note: The values in parentheses are standard errors, and *** indicates significance at the 1% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “Y” means control.

, the relationship between the Internet penetration and labor income share is significantly positive at the 1% significance level. This suggests that as firms’ Internet penetration increases, their operating efficiency improves, and the income share allocated to labor increases. There are several possible explanations for this phenomenon.

Firstly, Internet penetration often leads to enhanced productivity, as firms utilize digital tools and platforms to streamline operations, increase output, and reduce inefficiencies [51]. This increase in productivity can result in higher profits, which may be distributed to workers through higher wages, benefits, and job security.

Additionally, the growing demand for skilled labor, particularly in areas such as IT, data analytics, and digital marketing, may contribute to a larger share of the firm’s income directed toward its workforce. As firms integrate more advanced technologies, they may need to hire or retain a more highly skilled workforce, which can command higher wages, thus increasing the labor income share. In the globalized environment, firms prioritizing high-skilled workers and equitable compensation are better positioned to attract top talent, innovate, and maintain a developable competitive edge.

In summary, a greater Internet penetration strengthens firms’ positions in GVCs by optimizing transaction models and improving factor allocation, particularly labor income. The Internet enhances firms’ operational efficiency and market integration by reducing transaction costs and increasing the labor income share, fostering a more sustainable and equitable growth model within GVCs. These factors combined contribute to firms’ competitiveness and ability to secure more substantial roles in global supply networks.

5.5. Heterogeneity Analysis

We classify firms based on geographical location, ownership type, trading methods, size, and technology intensity, as detailed in the results in

Table 10. Heterogeneity analysis results (I)—region.

	(1)	(2)
	Eastern Region	Midwest Region
VARIABLES	lngvc	lngvc
int	0.0126 ***	0.0021
	(0.0006)	(0.0047)
Control variables	Y	Y
Time-fixed effects	Y	Y
Firm-fixed effects	Y	Y
City-fixed effects	Y	Y
N	557,766	53,108
R-squared	0.9219	0.9455

Note: The values in parentheses are standard errors, and *** indicates significance at the 1% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “Y” means control.

,

Table 11. Heterogeneity analysis results (II)—firm ownership.

	(1)	(2)	(3)
	Private Firms	State-Owned Firms	Foreign-Invested Firms
VARIABLES	lngvc	lngvc	lngvc
int	0.0169 ***	−0.0031	0.0142 ***
	(0.0008)	(0.0108)	(0.0012)
Control variables	Y	Y	Y
Time-fixed effects	Y	Y	Y
Firm-fixed effects	Y	Y	Y
City-fixed effects	Y	Y	Y
N	369,535	8198	217,644
R-squared	0.9326	0.9588	0.9338

Note: The values in parentheses are standard errors, and *** indicates significance at the 1% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “Y” means control.

,

Table 12. Heterogeneity analysis results (III)—trade pattern.

	(1)	(2)
	General Trade	Processing Trade
VARIABLES	lngvc	lngvc
int	0.0047 ***	0.0095 ***
	(0.0009)	(0.0012)
Control variables	Y	Y
Time-fixed effects	Y	Y
Firm-fixed effects	Y	Y
City-fixed effects	Y	Y
N	370,396	119,180
R-squared	0.9275	0.9465

Note: The values in parentheses are standard errors, and *** indicates significance at the 1% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “Y” means control.

,

Table 13. Heterogeneity analysis results (IV)—firm size.

	(1)	(2)	(3)
	Small Business	Medium Business	Large Business
VARIABLES	lngvc	lngvc	lngvc
int	0.0088 ***	0.0109 ***	0.0161 ***
	(0.0013)	(0.0011)	(0.0010)
Control variables	Y	Y	Y
Time-fixed effects	Y	Y	Y
Firm-fixed effects	Y	Y	Y
City-fixed effects	Y	Y	Y
N	191,599	192,298	199,756
R-squared	0.9322	0.9377	0.9295

Note: The values in parentheses are standard errors, and *** indicates significance at the 1% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “Y” means control.

and

Table 14. Heterogeneity analysis results (V)—firm technology intensity.

	(1)	(2)	(3)
	Low Tech	Medium Tech	High Tech
VARIABLES	lngvc	lngvc	lngvc
int	−0.0016	0.0021	0.0229 ***
	(0.0010)	(0.0014)	(0.0007)
Control variables	Y	Y	Y
Time-fixed effects	Y	Y	Y
Firm-fixed effects	Y	Y	Y
City-fixed effects	Y	Y	Y
N	183,971	74,693	337,447
R-squared	0.9569	0.9532	0.9246

Note: The values in parentheses are standard errors, and *** indicates significance at the 1% levels, respectively. The coefficient of variables, standard errors, and R² results are all retained to four decimal places. “Y” means control.

.

Firstly, geographical location is critical for the benefits of Internet penetration. Firms located in the eastern coastal areas of China experience significant advantages from Internet penetration. These firms are generally more integrated into global markets, have better access to infrastructure, and are closer to major economic hubs, which facilitates the adoption of new technologies. The Internet penetration ability to reduce transaction costs, streamline operations, and improve access to global markets significantly enhances the competitiveness of these firms. In contrast, inland firms show only a minor positive effect from Internet penetration. This difference is likely due to limited technological support, less developed digital infrastructure, and fewer linkages to global networks in these regions. While these companies benefit from Internet penetration, the overall impact is minor compared to coastal regions. This highlights the importance of regional transportation facilities and geographic location in determining the extent of Internet penetration benefits.

Secondly, ownership type is another important factor explaining Internet penetration’s impact. Private and foreign-invested firms are generally better at exploiting Internet penetration and gaining more from it than state-owned firms. Private and foreign-invested firms tend to be more flexible and quicker to adopt new technologies and innovative business models. This enables them to use Internet penetration to optimize operations and strengthen their position in GVCs. On the other hand, state-owned firms gain negligible or even adverse effects from increased Internet penetration. This may be because they are slower to adapt to technological changes, have more rigid organizational structures, and rely more on traditional operating methods. Therefore, state-owned firms are often less able to take full advantage of increased Internet penetration, which limits their ability to upgrade in GVCs.

Thirdly, the mode of trade, especially the nature of business operations, is a key factor determining the benefits of Internet penetration. Firms engaged in processing trade (i.e., firms involved in manufacturing or assembling products and exporting them) have gained incredible benefits from an increased Internet penetration. These firms tend to have longer production chains and are more integrated with high-tech industries, which makes them more dependent on efficient communication and rapid information exchange. Increased Internet penetration helps these firms simplify their supply chains, reduce transaction costs, and improve coordination with global suppliers and customers, enabling them to gain tremendous competitive advantages in the international market. Therefore, processing trade firms are one of the biggest beneficiaries of Internet penetration.

In addition, the firm’s size will also affect the extent of the benefits of Internet penetration. Larger firms will feel more significant positive effects from increased Internet penetration. Larger firms have the ability and resources to invest in R&D and recruit high-tech talents, enabling them to take full advantage of increased Internet penetration. These firms can use the increase in Internet penetration to expand the scale of their operations, rationalize resource allocation, and improve their productivity. However, we have to say even small firms will experience favorable effects from Internet penetration. Their more significant market share and ability to integrate into GVCs make them more capable of benefiting from the increase in Internet penetration, further upgrading their position in GVCs.

Finally, a firm’s technology intensity is one key factor determining its ability to benefit from an increased Internet penetration. Firms with a higher technology intensity are more likely to use Internet penetration to innovate, improve product development, and enter new markets. The synergy between high-tech industries and the Internet fosters more significant growth opportunities, enabling these firms to leverage digital platforms, big data, and automation to gain a competitive edge. In contrast, medium-tech and low-tech firms experience only minor effects from Internet penetration. These firms may use Internet penetration for operational efficiencies. Still, they are less likely to realize transformative gains, as their technological capabilities may not be as advanced or as profoundly integrated into their core business operations. As a result, while Internet penetration still provides some benefits, it does not have the same profound impact as it does for high-tech firms.

These heterogeneous analyzes are consistent with the history of China’s economy. In the early stages of China’s development, the government directly introduced or encouraged the introduction of relatively advanced foreign technologies and carried out many reforms of state-owned firms, which significantly improved firms’ technological level and international competitiveness [52]. China has rapidly become an industrial power from a traditional agricultural country through a series of policy measures. At the same time, we can intuitively observe that the development speed of cities in eastern China is faster than that of cities in central and western China, and the economic growth is more rapid.

According to the theory of international business competition, developing countries are usually in the factor-driven stage, relying mainly on primary factors such as cheap labor or abundant natural resources to reduce production costs or improve product quality [53]. For example, in the early days, China took advantage of these advantages and introduced many light industrial industries from Japan to learn advanced manufacturing technology. Then, it innovated and finally achieved brilliant achievements in sectors such as home appliance manufacturing. Therefore, based on these competitive advantages, if developing countries can introduce high-tech industries that foreign governments gradually withdraw from, imitation and learning can improve their technological level and enhance their competitiveness in international trade.

On the contrary, if only low-tech firms are encouraged to be established, although it may promote the domestic economy in the short term, it will always be in a relatively disadvantageous position in the GVCs. This will further widen the gap between developing and developed countries, which is detrimental to their long-term economic development and poses a potential risk to the sustainable development of the global economy.

However, based on our heterogeneity analysis, it is worth noting that foreign direct investment enterprises and private enterprises are more profitable than state-owned enterprises in improving their position in the global value chain. Therefore, the government should subsidize the scientific and technological innovation and technology introduction of private firms and actively introduce foreign direct investment to improve the country’s scientific and technological level and R&D capabilities.

In summary, the impact of Internet penetration on firms’ positions in GVCs is influenced by several factors, including geographical location, ownership type, trading methods, firm size, and technological intensity. Firms in coastal regions, private and foreign-owned firms, processing trade companies, larger firms, and high-tech firms are more likely to experience significant benefits from Internet adoption. These findings highlight which firms can more effectively leverage Internet penetration to improve their competitiveness and enhance their position in the global supply chain. They also provide some ideas for which companies in developing countries should prioritize, helping to promote the more sustainable and equitable development of global value chains.

6. Conclusions and Policy Recommendations

The rapid increase in Internet penetration can significantly improve firms’ positions in GVCs by reducing transaction costs and optimizing resource allocation, which is conducive to the sustainable development of firms. In an increasingly interconnected world, the Internet can enable firms to operate more efficiently, integrate into global supply networks, and maintain the competitiveness of sustainable development. This paper explores how Internet penetration affects a firm’s position in the global value chain. It emphasizes the role of Internet penetration in reducing transaction costs, facilitating labor resource allocation, and upgrading firms’ positions in GVCs.

Our analysis shows that Internet penetration significantly reduces transaction costs. This helps firms improve their position in GVCs, allowing them to allocate resources more efficiently. It can also manage relationships with suppliers and customers. Notably, the impact of Internet penetration is particularly significant in coastal areas. Similarly, foreign-owned and privately owned firms and large and medium-sized firms benefit the most from Internet penetration, as they are generally more flexible and better able to take advantage of opportunities brought about by Internet penetration.

In addition, the most significant advantage that high-tech firms gain from Internet penetration is that they are often at the forefront of technological innovation and can better use it to improve operational efficiency, promote innovation, and seize new market opportunities. Internet penetration supports high-tech firms’ ongoing R&D actions, enabling them to scale their businesses and employ higher-value-added activities more effectively. Thus, it is a key driver of sustainable growth and can help high-tech firms enhance their competitive advantage and consolidate their position in the GVCs.

To upgrade the position of firms within GVCs and ensure that the benefits of Internet penetration are more broadly distributed, we recommend three key policies that can support sustainable development in this context:

Firstly, governments should prioritize investment in Internet infrastructure, especially in underdeveloped regions, to increase Internet penetration and narrow the digital divide. For example, some Southeast Asian and African countries have successfully improved Internet access in rural and remote areas through government and foreign direct investment (FDI) cooperation, providing the basic conditions for local companies to integrate into the global supply chain [54,55]. Specific implementation paths can include providing Internet access subsidies to firms, providing low-cost digital technology training, and promoting public–private partnerships (PPPs) to achieve cross-sector and cross-regional digital economic integration. On this basis, firms can use digital technology to improve production efficiency, strengthen market connections, and ultimately enhance their position in the GVCs. Especially in the early stages of digital transformation, the government can also encourage firms to adopt Internet technology by formulating tax incentives and promoting the transformation of companies to high-value-added and technology-intensive sectors.

Secondly, the government should support the growth of high-tech industries through more precise policy interventions, especially in high-tech manufacturing. Specifically, the government can rely on existing special economic zones or industrial parks to create a regional development model that integrates digital infrastructure, technological innovation, and industrial chain integration. For example, China’s “Pearl River Delta” and “Yangtze River Delta” regions have successfully transformed their manufacturing sector to high-tech content through digital infrastructure construction and policy support from local governments, which has enhanced the bargaining power of enterprises in the region in the global supply chain [56]. Similar strategies can be customized according to the industrial characteristics of different areas, providing specific digital infrastructure subsidies, tax incentives, innovation funds, etc., to help high-tech industries obtain the basic conditions for leapfrog development.

Thirdly, the government can encourage the construction of digital platforms, especially in industries involving long production chains or high-tech manufacturing, such as smart device manufacturing [57]. Through cooperation between the government and enterprises, building digital platforms that meet local market needs can provide enterprises with convenient production scheduling, market docking, and supply chain management services, further improving firms’ global competitiveness. In the long run, these support measures will help enhance firms’ position in the GVCs and promote sustainable economic development.

In addition to policy recommendations, future research could further explore Internet penetration in other market environments, especially in other developing countries. Comparing the impact of digital transformation in different market structures will help to gain a deeper understanding of how Internet penetration affects firms in various regions and economic contexts. In addition to policy recommendations, future research could further explore Internet penetration in other market environments, especially in other developing countries. Comparing the impact of digital transformation in different market structures will help to gain a deeper understanding of how Internet penetration affects firms in different regions and economic contexts. In addition, given that different industries play various roles in global value chains, future research may not be limited to industrial firms by studying other sector (e.g., service sector) mechanisms, providing more targeted policy recommendations to firms to upgrade their position in GVCs.

Finally, we acknowledge that the period of our data, which ranges from 2000 to 2013, limits our ability to capture recent developments, particularly the rapid rise of the Internet and the digital economy. Consequently, while our findings provide valuable insights, they are based on data that may not fully reflect current trends. However, it is worth stressing that the early 21st-century Chinese economy closely mirrors the current economic transformation in many developing countries. Therefore, the conclusions of this study have strong practical significance for developing countries and will help improve the positions of firms in the GVCs of developing countries. Future research can be explored if the enterprise-level data is updated, primarily covering more ICT-related data. Further analysis of the impact of ICT on improving the GVCs’ positions improves this. Such research will help us better understand the role of ICT innovation in improving the position of the GVCs.

In conclusion, dramatically increased Internet penetration significantly improves firms’ positions in the GVCs, reducing transaction costs and optimizing resource allocation. If the government makes targeted policy interventions, the benefits of Internet penetration will be distributed more equitably, promoting sustainable economic growth and contributing to a more inclusive global economy.