How Digitalization Shapes Export Product Quality: Evidence from China

Abstract

Digitalization has imposed new production and trade requirements on enterprises. Quality upgrading of export products, a requirement of international consumption upgrading, is also a product of enterprise industrial upgrading. We examine the impact of the digitalization of Chinese listed firms on the quality of their export products from 2011 to 2015 using the panel fixed effects model and chain mediating effects model. The results demonstrate that digitalization has an inverted U-shaped effect on the quality of export products. Further, digitalization can promote the quality upgrading of enterprises’ export products. However, the supporting capacity of digital infrastructure and the decreasing consumption upgrading of foreign customers limit this process. Empirical results show that digitalization can alleviate the financing constraints to improve the quality of export products. Because digitalization can improve human capital quality, this effect can improve the export product quality. A chain mediation effect also occurs. Digitalization alleviates financing constraints to improve human capital and enterprises’ export product quality. By clarifying digitalization⇒, alleviating financing constraints⇒, improving human capital quality⇒ improve export product quality relationships, executives can improve the competitiveness of enterprises in the international market.

1. Introduction

In the era of the digital economy, digitalization marks the arrival of the fourth industrial revolution. Digital technologies, such as 5G, information and communication technology, the Internet of Things, and big-data systems, have profoundly impacted all sectors of human life [1]. The digital behavior of enterprises has a profound impact on environmental and economic performance [2]. However, the concept of digitalization still needs to be clarified for many businesses. Furthermore, digitalization content is broad: it includes production technology, marketing behavior, corporate governance, and customer management of enterprises, among other areas [3]. Digitalization is not only a technology choice but also a strategy choice [4]. Digitalization has broken the original organizational boundaries. Digital production, management, and trade fall under digitalization [5]. Specifically, digitalization involves introducing data resources into all links of enterprises. Enterprises can utilize digital technology to achieve value chain reconstruction [6]. Digitalization has facilitated a new value chain model, namely, a data value chain connected by data nodes, creating new opportunities and threats for enterprises to compete. China’s international trade volume has always been globally dominant. Most of the export products of Chinese industrial enterprises rely on low prices to gain international competitiveness [7]. Product quality shows the competitive advantage of the product. High quality reflects the high competitiveness of international trade. Most companies aim to take advantage of external demand. The high export quality will be famous for international consumers. Digitalization is the only method of transforming and upgrading industrial structures [8]. Digitalization offers a significant opportunity to improve the quality of enterprises’ export products. By facilitating boundary crossing, digitalization can enhance the sustainable competitive advantage of enterprises [9]. Through digitalization, enterprises can promote their internationalization strategies [10,11]. It improves marketing flexibility and expands enterprises’ financing channels or methods [12]. Moreover, it improves the enterprise’s ability to resist risks [13].

Scholars initially observed the impact of new technologies on exports. Some suggest that applying digital technology to factory production can improve the quality of enterprises’ export products. For example, enterprises can increase total factor productivity by applying robots to production [14].

The number of telecommunications subscriptions and available bandwidth speed in ICT affect export performance by adjusting transaction costs for both supply and demand [15]. Moreover, the Internet can improve the quality of export products [16] and their profitability to promote export upgrading [17]. However, a consensus has yet to be reached. Hong et al. (2022) believed that although using robots can reduce costs and improve resource allocation optimization, the mismatch between robots and existing production materials at the initial stage will inhibit the quality upgrade of exported products [18].

The adoption of new technologies is not synonymous with digitalization. In addition to digital technology, digitalization also includes employees’ digital skills and strategies [7,19]. Research into the impact of digitalization on export product quality is still in its infancy. Because digitalization only occurs after some time, success is not guaranteed. However, some enterprises have gained a leading edge in digitalization. The difficulties associated with transformation may result in enterprises falling behind their competitors [20]. Some scholars believe digitalization positively impacts the quality of Chinese enterprises’ export products.

Nevertheless, the impact path may differ. Digitalization can enhance the positive influence of home country institutional constraints on export intensity [21]. Digitalization can improve the quality of export products in three ways: quality improvement of intermediate products, the transformation of export product selection, and innovation. Researchers [7] further confirmed that digitalization could improve export product quality. Digitalization promotes export product quality improvement by facilitating innovation. However, the influence of digitalization on export product quality is U-shaped [18,19]. The mediating effect of digitalization on innovation is positively U-shaped [19]. In particular, digitalization will improve the competitiveness of enterprises in the domestic market. It may transform enterprises’ export sales into domestic sales and eliminate the dependence on export. Therefore, researchers are split on the effect of digitalization on improving the quality of export products.

Although some scholars have made a causal analysis of the relationship between digital transformation and product quality, some valuable conclusions have also been drawn. However, only some scholars analyze it from the angle of financing constraint and human capital accumulation. Their analysis of mediations is also parallel and uses fewer chain mediations. Accordingly, the impact of digitalization on production and trade should be determined. Will digitalization affect the quality of export products? Is there a nonlinear relationship between them, and is the U-shape positive or negative? This paper aims to clarify the relationship between digitalization and export product quality. The digitalization concept is analyzed, and the indicators of digitalization are calculated via machine analysis. To some extent, this avoids the computing bias associated with using the Internet or artificial intelligence alone. In addition, the impact of digitalization on the quality of export products may be nonlinear. The mechanism by which digital empowerment affects the quality of export products is relatively complex. This paper analyzes this path from two aspects: enterprise financing constraints and human capital quality. In addition, the chain mediation model is used to explore the relationship between financing constraints and human capital quality and comprehensively analyze the relationship between digitalization and export product quality. Our research is dedicated to unlocking the black box of causality between digitization and export quality. We strive to provide valuable insights for the management of enterprises and provide beneficial supplements for scholars’ research.

This study makes four contributions. First, we theoretically analyze and summarize the main digitalization content. Second, we contribute to the nonlinear research on the relationship between digitalization and export product quality. Third, to find a more specific and realistic intermediate path for the impact of digitalization on export product quality, chain, and parallel mediation effects are also analyzed, which complements the existing literature. Fourth, we use the panel fixed effect model and chain intermediary effect model to test the existence of chain mediation. In the robustness test, applying the IV-2SLS and double-difference method and the u-shaped relationship approach to the initial results ensures that the empirical results are scientific, robust, and reasonable.

2. Materials and Methods

The need for digitalization is no longer in doubt, nor is it a choice. It is an inevitable trend in the development of enterprises. Digitalization involves technology, management, communications, and customers [22]. Unsurprisingly, digitalization also profoundly impacts export product quality from these aspects.

In the following, we will discuss the impact of digitalization on the quality of export products. We will also detail the two channels of alleviating financing constraints [23] and human capital accumulation [24] and elaborate on the logical relationship between variables. A theoretical model is presented in Figure 1.

2.1. Digitalization and Export Product Quality

Digitalization critically affects the production activities of enterprises. The application of digital technology has upended the entire production process. When enterprises introduce digital technology into production, the digital monitoring system can increase the real-time monitoring of the production process and improve the frequency and intensity of product quality inspection [19]. Establishing a digital factory system deepens enterprises’ agile production and intelligent production to a certain extent. It can help enterprises effectively manage the production process. Enterprises can track raw materials, semi-finished products, and packaging. The whole production chain can ensure that the enterprises not only achieve fast delivery but also control the production error [25], reduce the cost and loss of each part, and increase the qualified product quality rate. Enterprises apply digital technology to sophisticated product production. Digital technology can help the enterprise supervision site achieve digital, transparent, and complete process traceability. Introducing big-data elements in the production process can improve the technical complexity of products, reduce the probability of product repair, and increase the qualification rate. Digital employees will collect and analyze production data through the system. Further, digital systems can help enterprises manage nonconforming products and production anomalies.

Enterprise digital operation can enhance total quality management. Management digitalization enhances the total quality management of products by employees and achieves the closed-loop tracking of quality improvement. In addition, enterprises can visualize quality management for their customers, whereas customers can enhance their confidence in the quality of the enterprises’ products. Digital systems rely on reliable data, algorithms, and methods to improve the ability of employees to analyze and make accurate statistical decisions, thus helping enterprises to realize the correct fault liability management in production quickly. Under the guidance of a digital system, enterprises can give feedback to employees on the types and locations of quality problems and predict product quality using machine learning. The application of management platforms and software improves the efficiency of product quality management.

Digitalization negatively impacts the quality of export products to a certain extent. With advancing digitalization, the difficulty of digital transformation exhibits a geometric growth trend. Although digitalization has the practical effect of increasing cost and efficiency, full play is required to achieve its comprehensive ability. Developing this comprehensive ability requires the formation of supporting projects at the organizational, strategic, and industrial levels. However, enterprises need help with digitalization. Some enterprises need help even to support digitalization. The market’s digital ecosystem could be better, but it is still in its infancy. Infrastructure links are still under homogeneous low-price competition, which requires enterprises to increase capital investment. In addition, digitalization support remains at the senior management level, without digital transformation from top to bottom, which prevents expansion to different businesses. Therefore, phased evaluation of digitalization is challenging. Because the demand of some foreign customers remains traditional, these customers cannot accept the more advanced digitalization products. Considering the market competition, companies must adapt to customers and provide export products of different quality to meet their customers’ requirements in less developed countries.

H1. 

Digitalization has an inverted U-shaped effect on the quality of export products.

2.2. Digitalization, Human Capital Quality, and Export Product Quality

Human capital is the core element of enterprise operation and production. Human capital is also a prerequisite for the relevant material elements of the enterprise. It can increase the marginal output of other essential production factors of the enterprise. Digitalization has introduced new requirements for the structure of human capital. Digital systems, represented by industrial robots, significantly impact the labor market, and artificial intelligence will have a substitution effect on the labor force, thereby reducing the employment rate [26]. However, digitalization imposes higher requirements for human capital. Only with sufficient human capital participation can the digital system of the enterprise be fully utilized. The application of digital technology will increase machine and intelligent production. Enterprises must undertake digital transformation to adapt to big-data processing technology, cloud computing, 5G, and other technologies. However, technology upgrading and technology spillover requires a large number of R&D personnel. Employees must transfer their work and study the digital environment during digitalization. Further, employees need to upgrade their knowledge level to use digital technologies, so they need to acquire new knowledge and skills to meet new job demands [27]. When enterprises conduct digitalization, employees are required to possess digital thinking, which increases the chances of successful innovation of enterprises [28]. To realize innovation, enterprises require the participation of R&D and innovation personnel.

On the one hand, digitalization imposes higher requirements for employees’ knowledge capital and sets new goals for the management’s leadership. On the other hand, digitalization upgrades human capital management from quantitative to qualitative [29]. The human resource management digital transformation promotes the repositioning and division of labor and the accumulation of human capital, thus improving employees’ working efficiency. Moreover, digitalization will increase the social group’s recognition of the enterprise’s brand value, increasing the attractiveness of the enterprise’s brand to the labor market.

The application of digital technology can effectively alleviate the ex-post information asymmetry derived from the principal–agent relationship between enterprise owners and managers. Moreover, this reduces the two-way moral hazard facing enterprises, improving their human capital quality and that of their export products. Using digital technologies and having employees with digital capabilities is conducive to increasing the export propensity of enterprises [30].

Human capital promotes enterprise value [31]. Human capital can help enterprises to improve their production efficiency. Higher total factor productivity can improve the quality of exports. Human capital positively promotes enterprises’ export performance [32]. Human capital improves the international trade level of enterprises. Further, its expansion can significantly improve the quality of export products of Chinese manufacturing enterprises. The accumulation of human capital promotes investment in fixed assets and enterprise innovation, further improving the quality of export products. Through digitalization, enterprises can improve export product quality by improving human capital quality.

H2. 

Digitalization can improve export product quality by improving human capital quality.

2.3. Digitalization, Financing Constraints, and Export Product Quality

Digitalization can alleviate financing constraints [33]. On the one hand, enterprises can break the information island using digital technology. Realizing the multichannel access to financing information illustrates enterprises’ development ability to banks and other financial institutions, from whom enterprises obtain funds through multiple channels. High-quality financial performance allows enterprises to win the favor of more investors. Digitalization improves the quality of information disclosure. Enterprises can collect capital from external and potential investors by creating a positive image. Showcasing their high value can help them increase financing channels and enhance their financing ability. Moreover, with the help of digital technologies, enterprises can realize resource sharing, gather big-data resources through data platforms, enhance their standing in the industry, and attract outside investors.

On the other hand, digitalization involves production, R&D, sales, and packaging. Through online crowdsourcing, enterprises deeply integrate their industrial chain with external stakeholders, thus expanding their production boundaries and reducing their financing needs and capital pressure. Digitalization expands knowledge boundaries related to innovation, promotes the application of digital technologies in production [34], contributes to green innovation and resource conservation, and helps attract government subsidies.

Credit financing constraints hurt export sales. Financing constraints will shrink the exporters’ total output, inhibit the destination country’s market competitiveness, and reduce export sales [35]. In contrast, easing financing constraints will release financial pressure on enterprises. Enterprises will have sufficient space and time to conceive collaborative digitalization applications. Digitalization promotes continuous optimization and improvement of products and export product quality. In addition, companies can upgrade traditional information technologies and use digital technologies and algorithmic platforms to efficiently collect, store, and process massive amounts of data, giving them insight into overseas markets, consumers, and competitors. Through big-data value mining, enterprises can improve product innovation ability, fully release innovation vitality, and improve the quality of export products.

Easing financing constraints can help enterprises eliminate the dilemma of human capital upgrading. On the one hand, easing financing constraints can quickly reduce the digital transformation pressure by improving wages and benefits and attracting high-quality employees. With the support of digital technology, consumers can also participate in product concept design, product conception, product improvement, and other related value-creation activities through the virtual environment, allowing them to obtain external human capital at a lower cost [36]. Enterprises support employees and consumers to participate in product development and realize collaborative innovation, which is also conducive to improving export quality. Enterprises can use capital to establish a digital ecosystem. Through colleges and universities, research institutes, cooperative-enterprise human capital, the rapid accumulation of human capital, the formation of new product ideas and innovation concepts, product innovation, and iteration, technical product complexity can be increased, thus improving the quality of export products.

H3a. 

Digitalization can improve the quality of exports by easing financing constraints.

H3b. 

Digitalization can improve the quality of export products by alleviating financing constraints and improving human capital quality.

3. Methodology

3.1. Sample

The sample comprised panel data observations from 2011 to 2015. The enterprise’s financial data and basic information were from the Database of Chinese Listed Companies (CAMAR). The quantity, total value, and destination countries of the import and export products of the enterprises were obtained from the China Customs database. The foreign investment, GDP, and average employees’ wages in the cities where the enterprises are located were sourced from China’s Urban Statistics database. We obtained panel data matching the listed enterprise, customs, and urban statistics database.

We chose these samples because they contained financial statements. Financial statements give an accurate picture of the company’s financial position. Because of data unavailability, China Customs data up to 2015 comprised the import and export data of enterprises’ products published by the China Customs administration department. Chinese city data comes from Chinese local government statistics. These figures are substantial and credible. After matching these data, we removed abnormal enterprise data from the financial indicators.

3.2. Dependent Variable: Export Product Quality

The measurement methods of export product quality mainly include using the export price to measure product quality [37], product-specific characteristics to construct product quality indicators [38], the summation method of supply and demand information [39], and the total product price and quantity information to measure product quality [40,41]. This study utilizes Khandelwal et al. (2013)’s method of calculating export product quality [41] to construct a product demand function in the dimension of year, firm, and exporting country:

$$x_{isvt}=Y_{vt}{p}_{isvt}^{-{\sigma }_{is}}{q}_{isvt}^{{\eta }_s}{P}_{isvt}^{{\sigma }_s-1}$$

(1)

where i, v, s, and t, represent the firm, export destination country, 8-digit HS products, and year, respectively; Y represents the total income of the export destination country; P represents the exogenous industry price index; p represents the price of product s; q represents the quality of product s; ${\sigma }_s$ represents the elasticity of substitution between product s and various commodities, ${\sigma }_s>1$; and ${\eta }_s$ represents the index for measuring the range of quality differences of product s, ${\eta }_s>0$. After taking the natural logarithm of both ends of the product demand function (1), we can obtain the following:

$$ln{x}_{isvt}+{\sigma }_{is}{lnp}_{isvt}={\mu }_{vt}+{\epsilon }_{isvt}$$

(2)

where ${\mu }_{vt}$ is a dummy variable with fixed effects of destination country and year, representing macro-characteristics such as price index and total income of export destination country, and ${\epsilon }_{isvt}$ is the residual term. Referring to Fan et al. (2015) [40] and using Broda and Weinstein (2006) [42] estimation results of ${\sigma }_s$, we can obtain the firm–product–export country–year export product quality:

$$q_{isvt}=\widehat{{\epsilon }_{isvt}}/({\sigma }_s-1)$$

(3)

The value of $q_{isvt}$ was then standardized:

$$st\_q_{isvt}=(q_{isvt}-{\mathrm{m}\mathrm{i}\mathrm{n}\_q}_{isvt}/({\mathrm{m}\mathrm{a}\mathrm{x}\_q}_{isvt}-{\mathrm{m}\mathrm{i}\mathrm{n}\_q}_{isvt})$$

(4)

${\mathrm{m}\mathrm{a}\mathrm{x}\_q}_{isvt}$ and ${\mathrm{m}\mathrm{i}\mathrm{n}\_q}_{isvt}$ are the maximum and minimum product quality of s, respectively, and $st\_q_{isvt}$ is the export product quality.

3.3. Independent Variable: Digitalization

Digitalization means using new digital technologies, such as mobile, artificial intelligence, cloud, blockchain, and ICT technologies, to enable significant business improvements to enhance customer experience, simplify operations, or create new business models [43]. The company’s financial statement disclosure does not publish the specific value of the degree of digitalization, so scholars have developed different measurement methods. Moreover, because of its size, the words Internet, robotics, smart manufacturing, network technology, 3D technology, and data elements alone cannot be used to replace digitalization. We used machine learning methods [44] to search and match the word frequency of a series of keywords related to digitalization in the company’s financial statements [45]. We used a natural logarithm of the sum of the number of occurrences of these terms in the financial statements to represent the level of digitalization, such as, digital, digital marketing, digital technology, digital technology, digital currency, digital operations, digital terminals, digital economy, trade, the digital system, the digital supply chain, data integration, data integration, data information, data management, data assets, network, Internet of things, in the information age, informatization, information technology, information integration, information and communications, automation, 5G, edge-computing, cloud computing, cloud services, cloud, big data, blockchain business, the age of wisdom, intelligence construction, wisdom, intelligence, Internet, e-commerce, cross-border electricity, electric business platform, 3D printing, 3D technology 3D tools, AI, electronic technology, electronic technology, online and offline, robots, machine learning, computer technology, O2O, B2B, C2C, P2P, C2B, and B2C. In addition, we manually collected the details of the intangible assets in enterprises’ financial statements in the current year. We selected the keywords related to digitalization and those referring to intangible assets, such as software, network, client, management system, and intelligent platform. The proportion of intangible assets in the current year can be calculated by adding up these words. We use those proportions to represent the degree of enterprises’ digitalization.

3.4. Mediator Variable: Corporate Financing Constraints

We can utilize critical corporate financial indicators to calculate the degree of financing constraint of enterprises. Kaplan and Zingales (1997) were the first to operate the net cash flow, cash holdings, cash payout level, debt degree, and growth [46]. They considered five classification indicators as proxy variables to characterize financing constraints and constructed a comprehensive index through regression analysis to measure the degree of financing constraints of enterprises. Since then, this approach has become popular in the research field of financing constraints [47]. In this study, we calculate the financing constraint index using the method of Kaplan and Zingales (1997) for reference based on the financial data of listed enterprises in China [46]. For example, operating net cash flow, cash dividends, cash holdings divide the total assets of the previous period, asset-liability ratio, and Tobin Q. We then established a set of medians and a dummy variable. If the categorical index is lower than the median, the value is 1; otherwise, it is 0. The dummy variables of these categorical indicators were then added. We use ranked logistic regression to estimate the regression coefficients of each variable by taking the financing constraint index as the dependent variable. Finally, using the estimation results of the above regression model, we can calculate the financing constraint index representing each listed company’s financing constraint. The larger the financing constraint index, the higher the degree of financing constraint faced by the listed company.

3.5. Mediator Variable: Human Capital Quality

The number of highly educated employees determines the quality of human capital. We utilized the human capital accumulation level to measure human capital quality. We measured human capital accumulation from the natural logarithm of the total number of people with college degrees and above employed by enterprises based on the years of education. The data on the educational structure of the employees employed by the enterprises were from CAMAR. We eliminated the number of employees with a college degree or below by manual screening.

3.6. Control Variables

This paper selects control variables from different levels. At the enterprise level, the financial indicators, operating conditions, and characteristics affect the quality of export products. This study considered years of establishment, enterprise size, the ratio of capital to labor, and asset liability ratio. They are the control variables at the enterprise level. The asset-liability ratio is the total liabilities divided by total assets. Therefore, we took the quality of imported intermediate goods as the control variable. At the regional level, we considered the level of opening to the outside world, economic development, and the average wage of employees in the region where the enterprise is located. The level of opening to the outside world refers to the amount of foreign investment in the local area and the level of economic development from the region’s GDP per capita. At the industry level, we chose the Herfindahl index to measure the degree of industry competition.

3.7. Method of Analysis

We set up models (5) and (6) according to the nonlinear relationship between digitalization and export product quality.

$$st\_q_{isvt}={\alpha }_0+{\alpha }_1{\left(Digital\right)}_{it}+{\gamma \left(Control\right)}_{it}+{\epsilon }_a$$

(5)

$$st\_q_{isvt}=b_0+b_1{\left(Digital\right)}_{it}{+b}_2{\left(Digital\right)}_{it}^2+{\gamma \left(Control\right)}_{it}+{\epsilon }_b$$

(6)

To further verify the interaction mechanism between digitalization and export product quality, we set chain mediation models:

$${Finance}_{it}=c_0+c_1{\left(Digital\right)}_{it}+{\gamma \left(Control\right)}_{it}+{\epsilon }_c$$

(7)

$${Human}_{it}=d_0+d_1{\left(Digital\right)}_{it}{+d}_2{\left(Finance\right)}_{it}+{\gamma \left(Control\right)}_{it}+{\epsilon }_d$$

(8)

$$st\_q_{isvt}=e_0+e_1{\left(Digital\right)}_{it}{+e}_2({Finance)}_{it}{+e}_3{\left(Human\right)}_{it}+{\gamma \left(Control\right)}_{it}+{\epsilon }_e$$

(9)

$$st\_q_{isvt}=f_0+f_1{\left(Digital\right)}_{it}{+f}_2{\left(Digital\right)}_{it}^2+{\left(Finance\right)}_{it}{+f}_4{\left(Human\right)}_{it}+{\gamma \left(Control\right)}_{it}+{\epsilon }_f$$

(10)

where subscript i, t, and $\epsilon$ denote enterprise, year, and disturbance term, respectively, and $a,b,c,d,e,f$ are the coefficients to be estimated. This study adopted the panel fixed effects model after the Hausman test eliminated the influence of changes over time and excluded the characteristics of products, export destination countries, individual firms, industries, and regions. The panel data fixed effect model can effectively control the time trend of samples and the differences among individuals. The intermediary effect model describes well the channel of digitization affecting product quality.

4. Results

4.1. Descriptive Statistics

There are 290,000 sample observations in this paper, involving 905 enterprises, 2560 commodities, and 203 export destination countries. As presented in

Table 1. Descriptive statistics of regression variables.

Variables	Mean	Std.	Min	Max	Median	Year	Industry	Region	Firm	Product	Country	N
Export Product Quality	0.470	0.240	0.000	1.000	0.460	5	60	144	905	2560	203	290,000
Digitalization	2.800	1.190	0.000	7.120	2.770	5	60	144	905	2560	203	290,000
Open	12.13	1.410	5.730	14.56	12.32	5	60	144	905	2560	203	290,000
PGDP	1.880	3.010	0.060	11.42	0.720	5	60	144	905	2560	203	290,000
Lnpwage	10.42	2.320	0.000	11.64	10.90	5	60	144	905	2560	203	290,000
Lnage	2.700	0.370	1.100	3.890	2.770	5	60	144	905	2560	203	290,000
Lnsize	22.35	1.520	17.39	27.70	22.05	5	60	144	905	2560	203	290,000
Lnklratio	12.22	0.960	3.220	17.52	12.20	5	60	144	905	2560	203	290,000
Asset Liability Ratio	0.430	0.210	0.010	2.860	0.440	5	60	144	905	2560	203	290,000
R&D	0.400	1.200	0.000	6.850	0.000	5	60	144	905	2560	203	290,000
Intermediate Product Quality	0.540	0.300	0.000	1.000	0.600	5	60	144	905	2560	203	290,000
HHI	0.110	0.110	0.010	1.000	0.080	5	60	144	905	2560	203	290,000
Financing Constraints	0.380	2.550	-8.230	12.05	0.550	5	60	144	905	2560	203	270,000
Human Capital Quality	6.340	1.430	1.500	11.04	6.080	5	60	144	905	2560	203	280,000

, the value of export product quality was within the [0,1] range.

The minimum value of digitalization was 0, the maximum value was 7.120, and the standard deviation was 1.190. This index indicates a significant difference in the level of digitalization among enterprises. The standard deviations of foreign direct investment, regional economic development level, and the average wage of employees were 1.410, 3.010, and 2.320, respectively. The result reveals a significant development gap between different regions in China. In other words, regional development needs to be balanced. The related indicators of other enterprises all exhibit specific enterprise heterogeneity. In general, these variables should be controlled during regression analysis. The statistical results of specific variables are listed in Table 1. Before the regression analysis, we conducted a multicollinearity test. The variance inflation factor (VIF) of all dependent variables in the test results was all less than 10, with mean VIF = 2.07, indicating that multicollinearity was absent in the regression analysis in this paper. Specific results are presented in

Table 2. Cross-sectional correlation matrix of regression variables.

Variables	VIF	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)
(1) Export Product Quality	——	1
(2) Digitalization	1.29	0.023 ***	1
(3) Open	1.65	0.028 ***	0.139 ***	1
(4) PGDP	1.71	0.029 ***	0.043 ***	0.592 ***	1
(5) Lnpwage	1.09	0.056 ***	−0.173 ***	0.015 ***	0.142 ***	1
(6) Lnage	1.19	0.014 ***	−0.017 ***	−0.115 ***	−0.168 ***	−0.089 ***	1
(7) Lnsize	5.88	0.017 ***	0.007 ***	0.004 **	0.160 ***	0.048 ***	0.091 ***	1
(8) Lnklratio	1.30	0.019 ***	−0.295 ***	−0.001	0.097 ***	0.079 ***	0.014 ***	0.172 ***	1
(9) Asset Liability Ratio	2.61	0.005 ***	−0.088 ***	−0.029 ***	0.062 ***	0.004 **	0.214 ***	0.594 ***	0.133 ***	1
(10) R&D	1.07	−0.047 ***	0.158 ***	−0.034 ***	−0.029 ***	−0.135 ***	−0.065 ***	−0.091 ***	−0.126 ***	−0.087 ***	1
(11) Intermediate Product Quality	1.10	0.028 ***	0.034 ***	0.025 ***	−0.110 ***	0.006 ***	0.125 ***	0.119 ***	0.009 ***	0.014 ***	−0.015 ***	1
(12) HHI	1.09	0.011 ***	−0.099 ***	0.040 ***	0.045 ***	0.053 ***	−0.034 ***	0.115 ***	−0.092 ***	0.034 ***	−0.018 ***	0.080 ***	1
(13) Financing Constraints	1.73	−0.016 ***	−0.215 ***	−0.042 ***	0.062 ***	0.073 ***	−0.069 ***	0.052 ***	0.161 ***	0.485 ***	−0.108 ***	−0.118 ***	−0.118 ***	1
(14) Human Capital Quality	5.13	0.018 ***	0.146 ***	−0.065 ***	0.003	0.000	0.236 ***	0.864 ***	0.002	0.488 ***	−0.060 ***	0.230 ***	0.114 ***	−0.069 ***	1

Notes: Mean VIF = 2.07, Significance levels: *** p < 0.01, ** p < 0.05.

. Concurrently, we also conducted the correlation test of variables. The correlation test results demonstrated that the relationship between dependent and independent variables was significant at 1%. Variables such as enterprise R&D and financing constraints were negatively correlated with export product quality. The specific results are listed in Table 2.

4.2. Analysis of Regression Results

Table 3. Regression results of fixed effects models estimation.

Variables	(1)	(2)	(3)	(4)	(5)	(6)
	Export Product Quality	Export Product Quality	Financing Constraints	Human Capital Quality	Export Product Quality	Export Product Quality
Digitalization	0.0011 *	0.0141 ***	−0.1505 ***	0.1015 ***	0.0015 **	0.0162 ***
	(1.778)	(8.271)	(−32.331)	(61.197)	(2.262)	(8.766)
Digitalization-squared		−0.0022 ***				−0.0025 ***
		(−8.051)				(−8.468)
Open	0.0030 **	0.0030 **	−0.0355 ***	0.0096 ***	0.0025 *	0.0029 **
	(2.366)	(2.356)	(−5.481)	(4.520)	(1.900)	(2.177)
PGDP	−0.0022	−0.0026	0.3323 ***	0.1870 ***	−0.0051 ***	−0.0049 ***
	(−1.322)	(−1.580)	(28.929)	(39.358)	(−2.771)	(−2.693)
Lnpwage	−0.0008 ***	−0.0010 ***	0.0006	0.0047 ***	−0.0006 ***	−0.0009 ***
	(−3.917)	(−5.082)	(0.357)	(9.507)	(−2.914)	(−4.034)
Lnage	−0.0083 ***	−0.0080 ***	−0.4573 ***	0.1458 ***	−0.0106 ***	−0.0096 ***
	(−3.981)	(−3.826)	(−27.116)	(23.874)	(−4.609)	(−4.155)
Lnsize	0.0017 **	0.0015 **	−0.4648 ***	0.7540 ***	−0.0039 ***	−0.0044 ***
	(2.484)	(2.150)	(−89.964)	(358.876)	(−3.678)	(−4.085)
Lnklratio	0.0018 ***	0.0019 ***	0.0292 ***	−0.2114 ***	0.0027 ***	0.0028 ***
	(2.704)	(2.824)	(5.605)	(−96.176)	(3.519)	(3.650)
R&D	−0.0019 ***	−0.0018 ***	0.1185 ***	−0.0151 ***	−0.0016 ***	−0.0014 ***
	(−4.750)	(−4.568)	(34.325)	(−14.079)	(−3.836)	(−3.327)
Asset Liability Ratio	−0.0428 ***	−0.0422 ***	9.3053 ***	0.5097 ***	−0.0101 *	−0.0089
	(−10.028)	(−9.849)	(278.576)	(34.316)	(−1.795)	(−1.579)
Intermediate Product Quality	0.0213 ***	0.0200 ***	0.1054 ***	0.2240 ***	0.0132 ***	0.0121 ***
	(9.960)	(9.264)	(6.861)	(42.009)	(5.645)	(5.149)
HHI	0.0560 ***	0.0594 ***	−0.6448 ***	−0.3477 ***	0.0438 ***	0.0481 ***
	(5.733)	(6.065)	(−9.734)	(−15.474)	(4.102)	(4.496)
Financing Constraints				−0.0119 ***	−0.0021 ***	−0.0024 ***
				(−14.469)	(−6.685)	(−7.433)
Human Capital Quality					0.0067 ***	0.0069 ***
					(6.545)	(6.789)
Constant	0.4079 ***	0.3986 ***	7.8534 ***	−9.3942 ***	0.4880 ***	0.4719 ***
	(18.212)	(17.655)	(51.396)	(−156.277)	(19.015)	(18.264)
N	291,041	288,819	270,045	257,530	257,530	255,609
R2	0.2221	0.2234	0.7305	0.9128	0.2302	0.2317
adj. R2	0.2071	0.2083	0.7250	0.9109	0.2137	0.2151
F	29.29	32.50	7659.88	20,214.27	21.47	25.35

Notes: Fixed effect nested within the cluster–robust. Models control for fixed effects of year, industry, region, enterprise, export destination, and product. The t statistics are in parentheses. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.

lists the panel fixed effect model and chain mediation effect model.

Column (1) shows that the estimated coefficient of digitalization is significantly positive (p < 0.10). The results in column (2) show that the estimated coefficient of the first term of digitalization is significantly positive (p < 0.01). Moreover, the coefficient of the digitalization square is significantly negative (p < 0.01). Therefore, the relationship between digitalization has a significant inverted U-shaped impact on export product quality. Thus, hypothesis H1 was confirmed.

The results in column (5) demonstrate that the estimated coefficient of digitalization is significantly positive (p < 0.05). In contrast, the results in column (6) reveal that the digitalization coefficient is significantly positive (p < 0.01). However, the coefficient of the digitalization square is significantly negative (p < 0.01). This result indicates that despite adding mediating variables, the digitalization of enterprises still has a significant inverted U-shaped influence on the quality of export products. Hypothesis H1 is still valid.

Column (4) shows that the digitalization coefficient is significantly positive (p < 0.01). That means digitalization can promote human capital quality. Columns (5) and column (6) present the impact of human capital accumulation on the quality of export products, and the results demonstrate that the estimated coefficient of human capital accumulation is significantly positive (p < 0.01). It shows that the accumulation of human capital improves the quality of export products. Therefore, the mediating effect of human capital quality exists, confirming research hypothesis H2.

Column (3) shows that the digitalization coefficient is significantly negative (p < 0.01). The result indicates that digitalization can alleviate financing constraints. Columns (5) and column (6) present the influence of Financing Constraints on the quality of export products. Moreover, the results demonstrate that the estimated coefficient of financing constraints is significantly negative (p < 0.01); therefore, the weakening of Financing Constraints can promote the quality of export products, so the mediating effect of Financing Constraints exists. Thus, research hypothesis H3a was confirmed. Column (4) shows that the estimated coefficient of financing constraint is significantly negative (p < 0.01), which indicates that easing financing constraints will improve the human capital quality of enterprises. Thus, the moderating effect of Financing Constraints and Human Capital Quality exists. Thus, research hypothesis H3b was confirmed.

Under the comprehensive view, three paths exist Digitalization⇒ Financing Constraints ⇒Export Product Quality; Digitalization⇒ Human Capital Quality ⇒Export Product Quality; and Digitalization Financing Constraints ⇒ Human Capital Quality ⇒Export Product Quality. We used the bootstrap method for additional tests, sampled 1000 times, and confirmed the persistence of the above three mediating effects. In addition, enterprise size, capital–labor ratio, foreign direct investment, quality of imported intermediate goods, and industry concentration were set as the control variable. Those variables can improve the quality of export products of enterprises. The years of enterprise establishment, asset–liability ratio, R&D and innovation, the average salary of local employees, and the level of economic development decrease the quality of export products. The results of control variables are consistent with those of previous studies.

4.3. Analysis of Robustness Tests

Upgrading export product quality may affect the level of the digital infrastructure. Moreover, export product quality even affects the digitalization process of enterprises. In addition, enterprises’ financing constraints and human capital may affect the levels of corporate governance. Then affects the successful enterprises achieve in digitalization. Therefore, endogeneity due to bidirectional causality should be tested. Because topographic features are exogenous variables separated from the economic system, we chose topographic feature variables as exogenous instrumental variables. The flatter the terrain is, the more convenient the local transportation, the better the foundation of economic development, the more complete the digital infrastructure, and the higher the digital level achieved by local enterprises. The primary sample used in this study was the data that changes with time and city in two dimensions, whereas the terrain relief of each region as a variable does not change with time. China’s total number of software and information technology service enterprises highly correlates with regional digital economy development. Therefore, their multiplier is considered the instrumental variable of digitalization.

In addition, the quality of export products can affect the enterprises’ human capital demand and financing environment. Therefore, it is necessary to consider the bidirectional causality between financing constraints and export product quality. Furthermore, we tested the bidirectional causality between human capital and export product quality. We considered the railway density of each region as an instrumental variable. It represents the level of local transportation infrastructure construction, economic development, and inter-regional communication. It can alleviate the degree of labor and capital market segmentation to a certain extent, which is also related to human capital and financing constraints. However, no causal relationship was identified between regional rail density and export product quality.

We test the model using the instrumental variable and the two-stage least squares method. The estimated coefficients of digitalization in

Table 4. Results of the endogenetic test.

Variables	(1)	(2)	(3)	(4)	(5)
	Export Product Quality	Financing Constraints	Human Capital Quality	Export Product Quality	Export Product Quality
Digitalization	0.0546 ***	−1.6184 ***	0.1338 ***	0.0197 ***	0.0087 ***
	(4.294)	(−19.819)	(6.670)	(9.612)	(4.016)
Digitalization-squared				−0.0042 ***	−0.0031 ***
				(−8.824)	(−9.344)
Financing Constraints				−0.0386 ***
				(−5.189)
Human Capital Quality					0.1001 ***
					(4.967)
Open	−0.0044 **	0.1176 ***	0.0081 **	0.0005	0.0030 **
	(−2.277)	(7.859)	(2.378)	(0.370)	(2.226)
PGDP	−0.0085	2.5812 ***	0.1360 ***	0.0083 ***	−0.0185 ***
	(−0.965)	(33.816)	(7.052)	(2.760)	(−5.170)
Lnpwage	−0.0012 ***	0.0169 ***	0.0012 **	−0.0010 ***	−0.0013 ***
	(−5.610)	(8.220)	(2.462)	(−4.396)	(−5.890)
Lnage	0.0003	−0.4358 ***	0.2768 ***	−0.0237 ***	−0.0228 ***
	(0.111)	(−18.949)	(45.062)	(−5.942)	(−6.373)
Lnsize	−0.0029 *	−0.5704 ***	0.6813 ***	−0.0161 ***	−0.0736 ***
	(−1.949)	(−55.765)	(211.706)	(−4.599)	(−4.840)
Lnklratio	0.0079 ***	−0.0618 ***	−0.1993 ***	0.0029 ***	0.0213 ***
	(4.281)	(−4.933)	(−55.743)	(3.838)	(5.218)
R&D	0.0013 *	0.0448 ***	−0.0099 ***	0.0029 ***	−0.0004
	(1.721)	(6.936)	(−6.578)	(2.866)	(−0.760)
Asset Liability Ratio	−0.0405 ***	9.5761 ***	0.1523 ***	0.3233 ***	−0.0754 ***
	(−7.882)	(210.372)	(11.237)	(4.695)	(−8.509)
Intermediate Product Quality	0.0235 ***	−0.0397 **	0.1350 ***	0.0190 ***	−0.0053
	(9.954)	(−1.980)	(25.805)	(7.847)	(−1.041)
HHI	0.0862 ***	−1.2467 ***	−0.1336 ***	0.0302 ***	0.0806 ***
	(6.269)	(−12.551)	(−5.245)	(2.595)	(6.905)
N	254602	235369	244312	267924	275817
F statistic	28.22	5138.53	10923.34	22.07	27.14
Kleibergen–Paap rk LM statistic	569.814	747.037	770.871	553.785	719.593
	(0.0000)	(0.0000)	(0.0000)	(0.0000)	(0.0000)
Kleibergen–Paap rk Wald F statistic	629.848	849.150	872.220	503.058	606.933
	[16.38]	[16.38]	[16.38]	[16.38]	[16.38]
Endogeneity test	19.314	537.868	7.866	24.211	22.520
	(0.0000)	(0.0000)	(0.0050)	(0.0000)	(0.0000)

Notes: Fixed effect nested within the cluster–robust. This table controls for fixed effects of year, industry, region, enterprise, export destination, and product. t statistics in parentheses. Significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01. Stock–Yogo weak ID test critical values: 10% maximal IV size = 16.38.

are all significant at the 1% level, and the direction of influence is consistent with the theoretical assumptions and empirical results mentioned above.

We refer to the method of Kleibergen and Paap (2006) to test the validity of instrumental variables [48]. The F statistic values in Table 4 are all larger than 10, and the Kleibergen–Paap RK LM test rejects the null hypothesis of insufficient identification of instrumental variables. The Kleibergen–Paap RK Wald F-test values were all greater than the maximum value of Stock–Yogo weak ID at the 10% level of 16.38, which rejects the null hypothesis that the instrumental variables are weak. Therefore, the selection of instrumental variables is reliable, and the regression results of the two-stage least square method are credible. (See Table 4).

In addition to the endogeneity test, we also conducted other robustness tests. Columns (1) and (2) in

Table 5. Results of other robustness tests.

Variables	Replace the Calculation Method of Digitalization		Change the Parameters for Calculating Export Product Quality		Take out the Observations of the Processing Trade		PSM-DID	Inverted U-Shape Test
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Export Product Quality
Digitalization	0.0138 **	0.0681 ***	0.0034 ***	0.0198 ***	0.0026 ***	0.0116 ***
	(2.157)	(5.466)	(5.561)	(11.447)	(3.967)	(6.272)
Digitalization-squared		−0.0865 ***		−0.0028 ***		−0.0015 ***
		(−4.813)		(−9.900)		(−4.999)
DID							−0.0114 ***
							(−5.664)
Digitalization_low								0.0042 ***
								(4.961)
Digitalization_high								−0.0096 ***
								(−6.677)
high								0.0050 ***
								(2.889)
Open	0.0031 **	0.0029 **	−0.0000	−0.0001	0.0039 ***	0.0039 ***	−0.0022	0.0033 ***
	(2.426)	(2.256)	(−0.032)	(−0.093)	(2.884)	(2.826)	(−1.181)	(2.606)
PGDP	−0.0023	−0.0025	−0.0003	−0.0006	−0.0026	−0.0028	0.0022	−0.0025
	(−1.360)	(−1.492)	(−0.147)	(−0.367)	(−1.278)	(−1.378)	(0.708)	(−1.474)
Lnpwage	−0.0008 ***	−0.0008 ***	−0.0010 ***	−0.0012 ***	−0.0012 ***	−0.0014 ***	−0.0001	−0.0011 ***
	(−3.977)	(−4.068)	(−4.616)	(−5.886)	(−5.570)	(−6.415)	(−0.285)	(−5.196)
Lnage	−0.0084 ***	−0.0086 ***	−0.0065 ***	−0.0057 ***	−0.0065 ***	−0.0064 ***	−0.0245 ***	−0.0075 ***
	(−4.048)	(−4.126)	(−3.031)	(−2.636)	(−2.837)	(−2.779)	(−7.633)	(−3.586)
Lnsize	0.0016 **	0.0016 **	0.0018 **	0.0017 **	0.0019 **	0.0015 *	0.0027 **	0.0017 **
	(2.370)	(2.312)	(2.558)	(2.374)	(2.338)	(1.850)	(2.413)	(2.431)
Lnklratio	0.0018 ***	0.0019 ***	−0.0012 *	−0.0011 *	0.0001	0.0002	0.0068 ***	0.0019 ***
	(2.625)	(2.816)	(−1.772)	(−1.646)	(0.129)	(0.284)	(5.564)	(2.845)
R&D	−0.0020 ***	−0.0021 ***	−0.0018 ***	−0.0017 ***	−0.0016 ***	−0.0016 ***	0.0027 ***	−0.0017 ***
	(−4.954)	(−5.123)	(−4.272)	(−4.052)	(−3.762)	(−3.692)	(3.299)	(−4.094)
Asset Liability Ratio	−0.0396 ***	−0.0390 ***	−0.0553 ***	−0.0549 ***	−0.0502 ***	−0.0493 ***	−0.0254 ***	−0.0430 ***
	(−9.154)	(−9.023)	(−12.562)	(−12.420)	(−10.704)	(−10.449)	(−3.832)	(−10.087)
Intermediate Product Quality	0.0207 ***	0.0203 ***	0.0280 ***	0.0264 ***	0.0168 ***	0.0161 ***	0.0160 ***	0.0200 ***
	(9.643)	(9.463)	(12.769)	(11.917)	(7.414)	(7.061)	(4.545)	(9.336)
HHI	0.0557 ***	0.0565 ***	0.0633 ***	0.0676 ***	0.0574 ***	0.0605 ***	0.0781 ***	0.0577 ***
	(5.708)	(5.783)	(6.460)	(6.873)	(4.920)	(5.167)	(5.118)	(5.907)
Constant	0.4116 ***	0.4132 ***	0.4299 ***	0.4137 ***	0.4070 ***	0.4053 ***	0.4184 ***	0.4128 ***
	(18.321)	(18.392)	(18.770)	(17.912)	(16.514)	(16.293)	(11.653)	(18.383)
N	290645	290645	291041	288819	234845	233423	134049	291041
R2	0.2223	0.2224	0.2562	0.2574	0.2366	0.2379	0.2764	0.2223
adj. R2	0.2073	0.2073	0.2418	0.2430	0.2191	0.2202	0.2515	0.2073

Notes: Fixed effect nested within the cluster–robust. This table controls for fixed effects of year, industry, region, enterprise, export destination, and product. t statistics in parentheses, significance levels: * p < 0.10, ** p < 0.05, *** p < 0.01.

show the results of re-accounting digitalization.

We utilize the economic degree of digitalization to represent this index. Column (1) shows that the estimated coefficient of digitalization is still significantly positive. The results in column (2) show that the estimated coefficient of the primary term of digitalization is still significantly positive. The estimated coefficient of the quadratic term of digitalization was still significantly negative. Columns (3) and (4) present the results of re-accounting the quality of the export products of the explained variables. Let the parameter σ = 10 in the process of export product quality accounting represent re-accounting of the quality of the enterprise’s export products. Column (3) shows that the estimated coefficient of digitalization is still significantly positive. The results in column (4) demonstrate that the estimated coefficient of the primary term of digitalization was still significantly positive. The estimated coefficient of the quadratic term of digitalization was still significantly negative (see Table 5 for details). These conclusions are not contradictory because of the accounting methods of the explained and explanatory variables.

Columns (5) and (6) present the regression results from varying sample sizes. Because a part of the sample processing trade is generally processed abroad, the quality and standards of the main export products are affected by foreign brands but less by their digitalization. Therefore, to elucidate digitalization, the real impact of product quality should be determined by excluding samples of processing trade, thus reducing the sample size. The results in column (5) show that the estimated coefficient of digitalization has a positive significance at the 1% level. The results in column (6) show that the estimated coefficient of the first term of digitalization has a positive significance at the 1% level. In contrast, the estimated coefficient of the quadratic term of digitalization has negative significance at the 1% level (see Table 5 for details). The results in column (6) are consistent with the previous estimation results. It demonstrates that digitalization will have an inverted U-shaped effect on the quality of export products.

In addition, considering that digitalization may come from business choice behavior, we perform score matching and use the double difference method. To avoid sample selection bias, we considered digitalization as a shock and compared the transformation in digital export quality before and after the change. Suppose the digitalization index is lower than the upper third of the digitalization index. We expected those enterprises to need more digitalization or report successful transformation. These enterprises were the control group. If the enterprise’s digitalization value is in the median upper third of the sample, we consider that the enterprise has achieved digitalization. These enterprises are the treatment group. We utilized the propensity score matching method to find more qualified control objects for the enterprises in the treatment group by testing matching variables relating to the digitalization of enterprises. We selected enterprise size, establishment years, R&D and innovation level, capital–labor ratio, asset-liability ratio, and quality of imported intermediate goods as matching variables. We utilized those variables to match these characteristic enterprise variables. Propensity scores were then estimated for the different years of the sample and matched in a ratio of 1:3. After obtaining the matched samples of the treatment and control groups, we used the difference method for regression. The results in column (7) show that the estimated coefficient of the interaction term of the differential is significantly negative (see Table 5 for details), which indicates that the implementation of digitalization will hurt the quality of exported products. This implies that digitalization has increased the competitiveness of enterprises in the domestic market. Moreover, improving competitiveness drives enterprises to switch from exports to domestic sales.

Finally, we further examined the effect of digitalization and the inverted U-shape of export product quality via the quadratic regression. Based on the method [49], we first found the maximum point Digitalization_max on the inverted U-shaped curve and then calculated the upper and lower values of digitalization indicators:

$$\begin{gathered} Digitalization\_low=Digitalization-Digitalization\_max, \\ if Digitalization<Digitalization\_max. \end{gathered}$$

(11)

$$\begin{gathered} Digitalization\_high=Digitalization-Digitalization\_max, \\ if Digitalization>Digitalization\_max. \end{gathered}$$

(12)

$$\begin{gathered} high=1, \\ if Digitalization>Digitalization\_max. \end{gathered}$$

(13)

Then, Digitalization_low, Digitalization_high, and high were substituted into the panel fixed effect model for regression analysis. The results in column (8) of Table 5 demonstrate that the estimated coefficient of Digitalization_low is significantly positive (p < 0.01), whereas the estimated coefficient of Digitalization_high was significantly negative (p < 0.01). (see Table 5). Therefore, combined with the breakpoint regression results, an inverted U-shaped relationship between digitalization and export product quality is further verified.

5. Discussion and Conclusions

5.1. Contributions to the Literature

Our research contributes to digitalization and export product quality research in three main ways.

First, our study contributes to improving the understanding of the role of digitalization in helping firms achieve a competitive advantage for their products. In particular, our work adds to the conceptual corpus on digitalization. However, it is not pertinent to the network layout of digitalization [50], the defining stage of digitalization [51], and the type of digitalization [52]. Our work analyzes the role played by digitalization in production, management, communication, and customer satisfaction. This effect is neither technical nor modal [53,54,55]. Digitalization has broken the original organizational boundaries. Digital production, management, and trade are all part of digitalization [5]. Therefore, in discussing the role of digitalization, we need to introduce digitalization into every aspect of the enterprise. We theoretically and empirically analyze the influence of digitalization in improving the quality of export products. The direction of influence is nonlinear, and the specific performance variable. However, the path chosen should consider digitalization. We not only refer to the literature but also choose the specific path according to the actual situation in digitalization.

Second, our work complements the existing literature on the quality of export products [24,40,41]; we explored the effects of digitalization on the quality of export products in the context of the digital economy. Here, we reviewed the controversies in existing research [7,19] and conducted a restudy. We highlight the inverted U-shaped impact of digitalization on the quality upgrading of export products. We also explained the mechanism of financing constraints and human capital theoretically. In conclusion, our study contributes to a better understanding of upgrading the quality of export products. Moreover, we utilize a new approach to examine the relationship between digitalization and export product quality. Unlike the previous parallel path [7,19], we establish a specific logical framework by adopting the chain mediation effect model. In addition, for the endogeneity test, we used the two-stage least squares method with the instrumental variables method and propensity score matching with the difference method, which ensured the robustness of the results. In addition, we conduct a further test of the inverted U-shaped relationship. This test will guide similar subsequent research.

Finally, our work contributes to the study of the sustainable development of enterprises’ products. In the process of strategic layout, digitalization is not only an application tool but also the requirement of sustainable development by upgrading the quality of export products. Under the impact of information, the sustainable development of products can feed back the sustainable development of society. Moreover, the enterprise digital transformation strategy is also a sustainable development strategy. The goal of digitalization is to enhance environmental protection, resource conservation and total quality management. Therefore, this study can provide a new guideline for sustainability.

5.2. Practical Implications

Our research is of great significance to enterprise management. First, we sought to elucidate the relationship between digitalization and the quality of export products. Enterprise managers should consider combining digitalization with a sustainable development strategy. Moreover, managers should build a system to guarantee the development of digitalization [56]. Regarding the development direction of digitalization, enterprises should protect both the production and trade ends. Accordingly, digitization should significantly improve the quality of export products. When the degree of digitalization is low, or the enterprise is in transition difficulty, it should increase digitalization-related investment to achieve smooth digitalization. Concurrently, it may squeeze investment in other aspects and ensure the stability of the human capital quality. Managers need to gain experience exploring digitalization to integrate technology and staff effectively initially. The need for more digital staff limits digitalization. Enterprises need to establish a good brand value and ease the financing constraints in the capital market. Increasing human capital accumulation through training or recruitment can improve the quality of export products.

Second, our work highlights the importance of capital and labor in digitalization. Human resources help enterprises unlock digitalization, and capital assistance can ensure rapid digitalization. The enterprise alleviates the financing constraint to widen the enterprise competition boundary. Especially in the field of export trade, digitalization provides opportunities for information exchange and cooperation in international trade. Moreover, easing financing constraints can help enterprises expand exports. Improving human capital quality promotes the company’s digital innovation. Digitalization releases the data elements, and human capital’s processing of the data elements helps achieve digital innovation. The role and function of digitalization in information exchange and cooperation are crucial to product quality upgrading.

Finally, we derive practical implications from the digital empowerment perspective. The development of the digital economy stimulates enterprises to take digital actions. Thus, it is still possible for enterprises to be aware of the importance of enormous data resources. This increased awareness may help managers enhance their understanding of data value and data innovation in digitalization. Data transformation requires enterprises to realize the value of data elements, increasing the demand for labor and capital in this process. To give full play to the value of data elements, enterprises need to achieve data innovation.

Furthermore, enterprises place high requirements on the existing enterprise management and working mode. Data innovation also introduces significant challenges to the original digitalization path. To avoid mistakes during transformation, enterprises should re-evaluate their export competition strategies from a digitalization perspective.

5.3. Limitations and Future Research

Our study has some limitations. First, we examined the role of financing constraints and human capital quality. However, we find that the role of financing constraints and human capital quality is comparable to that found in other studies. Notably, the sample from the Chinese Customs database on financing constraints and human capital quality needs to be improved. Therefore, we encourage further investigating the role of financing constraints and human capital quality on export product quality using more extensive data sets in the future. It is also necessary to consider the quality characteristics of export products under different stages of digitalization. Moreover, the relationship between digitalization and export product quality offers much scope for further investigation. For nonlinear research, the panel threshold model or panel smooth transition model can be utilized in further research.

Second, we have often mentioned that the essence of digitalization is to break the market segmentation caused by information asymmetry. However, as our primary research direction is export product quality, we have yet to discuss this aspect sufficiently. We will introduce the quality of information disclosure and discuss it in the future. Digitalization can promote passive information disclosure, but active information disclosure is still affected by corporate governance. Accordingly, the path from digital organizational agility to enterprise digitalization can be further explored [57]. Enterprise management digitization and employee digitization are also good research objectives.

5.4. Conclusions

We have conducted an in-depth study of the relationship between digitalization and export product quality. We used data on listed Chinese enterprises from 2011 to 2015 and the import and export data of China Customs to conduct a test using the panel fixed effects model. Our study demonstrates that the impact of digitalization on the quality of export products is nonlinear and that the impact is an inverted U-shaped relationship. We also investigate how digitalization affects export product quality. Using a chain mediation model, we theorize and validate the two paths of financing constraints and human capital. Digitalization can affect export product quality through two parallel mediators: financing constraints and human capital. In the digitalization process affecting the quality of export products, a chain mediation path of financing constraints and human capital also exists. Our study outlines the intrinsic relationship between a firm’s digitalization and the quality of its export products. This study opens new avenues of limited investigation because the literature separately examined the parallel effects of different mediators while ignoring the chain effects. Our research provides a theoretical basis for firm digitalization and production and export decisions, which has important guiding significance for firm positioning in global value chain reconstruction. In conclusion, our work implies that digitalization is equally essential for production and trading. Therefore, enterprises need to embrace digitalization in their total quality management. Enterprises should carry out scientific management of material and human capital in digital transformation. In the future, our research will pay more attention to analyzing the internal logic of enterprise digital transformation. To guide researchers to explore the digital path of total quality management in enterprises.