Impact of Digital Supply Chain on Sustainable Trade Credit Provision: Evidence from Chinese Listed Companies

Abstract

Given the trend of digitization, it is imperative to ascertain the role of the digital supply chain on sustainable trade credit provision. Based on data from Chinese listed firms from 2008 to 2020, we utilized the TF-IDF algorithm to measure the digital supply chain and ascertained its impact on trade credit. We found that the digital supply chain was positively associated with trade credit provision. Specifically, we arrived at the following conclusions: (1) the digital supply chain strengthens trade credit provision, including to customers and suppliers; (2) top management team power positively and significantly moderates the effect of digital supply chain; (3) among the sub-indicators of the digital supply chain, the dimensions of logistics, products and information have significant and positive impacts, while cash is insignificant; (4) curbing financialization and enhancing asset specialization are the mechanisms of the effect of the digital supply chain; and (5) the effect is more pronounced in firms with higher agency costs and lower supply chain collaboration and non-state ownership, and it is more salient in industries with higher competition and non-national support. We extend the theory of trade credit and enrich the literature on the digital supply chain. Our study offers managerial insights into the digital supply chain for emerging countries and enterprises.

1. Introduction

Driven by non-financial motives such as operational efficiency and market competition, or financial motives such as alleviating financing constraints, sustainable trade credit (TC) has been one of the most important decisions with respect to short-term financing behaviors in the supply chain [1,2,3]. It is a commercial contract that favors supply chain partners and is offered by large enterprises, as well as an act of capital reallocation and risk sharing [4,5]. Particularly in countries with strict regulation and conservative mode in financial development, due to severe information asymmetry, SMEs have difficulty accessing adequate bank credit [3]. Therefore, most SMEs are heavily dependent on TC. In supply chain systems, to cope with the uncertainty of financing links external and operating conditions internal, there is often a prominent demand for TC financing—that is, a need for sustainable and high-quality TC provision. There is no doubt that enterprises with outstanding capabilities can provide more TC for their collaborators. However, the TC competitiveness hypothesis and the buyer-market theory suggest that enterprises with market competitive advantages tend to obtain TC rather than provide TC [6,7]. Therefore, how to motivate excellent enterprises to provide sustainable TC is a challenge faced by researchers.

Sustainable TC provision originated from scholars’ assumptions of the “signaling effect”, which helped firms to quickly establish reputation [8,9]. Under these circumstances, some scholars investigated the motives and determinants of enterprises to provide sustainable TC from various perspectives. Some studies suggested that sustainable TC provision was a signal of product quality assurance. Along with their ideas, it could be motivated via transaction motives, such as market-environment changes, reducing transaction costs and maintaining customer relationships [1,10,11]. These motives are associated with the operational and transaction needs of firms. In addition, other studies indicated that sustainable TC provision reflected the role of financial allocation played by firms with high liquidity, especially in contexts with information asymmetry and severe credit constraints [2,10,12,13]. Therefore, these motives are also called financial motives. Although it is commonly assumed that firms with low financing costs and large firms are more likely to provide TC, the opposite situation also occurs. This may be due to customers’ positions in a buyer’s market or market competition motives such as the nature of the product, long-term relationship, etc. [6,7]. Furthermore, sustainable TC provision could be influenced by other factors such as the economic crisis and monetary policy tightening at the level of macro [14,15], the market competition and the differentiation degree at the level of industry [16,17] and bargaining power and financial constraints at the level of firm [18,19]. Meanwhile, in recent years, given the application and transformation of digital technology in enterprises’ behavior, relationship and welfare [20], scholars have also focused on the role of digital technology application in the development mode of sustainable TC provision and considered it as an approach to address the current challenges of TC [21,22]. Specifically, the latest research also attended to the potential impact of the digital supply chain (DSC) on TC [23]. However, the exploration of this perspective has been simplistic, and it lacks empirical evidence support.

The notion of DSC originated from the idea of intelligent supply chain, which was IT-enabled organization, as proposed by Rai et al. [24]. Subsequently, most scholars agreed that a key trend was the integration between digital technology and supply chain management [21,25]. Enterprises and their partners of in the supply chain enhance the exchange of digital applications in systems of logistics, production, information and capital, and they improve the supply chain resilience [26,27]. In the process, they leverage digital technology to restructure the processes, coordination, organization and value of business in both the external and internal organization of supply chains, which is what DSC means [28]. Based on the theories of resource-based and dynamic capabilities [29,30], and applying various methods, which have included case studies, QCA, DID and questionnaire survey, scholars found the partial effects of DSC on enterprises, such as competitive advantage [31,32], supply chain management modeling and supply chain efficiency [25,33]. Due to the lack of empirical measurements for DSC, few studies could offer empirical evidence for the effects of DSC [34,35]. TC is the reflection of the collaboration relationship between enterprises and their partners [5,36]. Some studies found that digital transformation may influence the sustainability of TC provision, adopting the lens of bank credit substitution [23,37]. In conclusion, numerous surveys demonstrated that DSC enhanced the competitive advantage of firms, but they overlooked the phenomenon that both competitive advantage and TC provision increased along with the development of DSC. This clearly diverged from the existing theoretical explanations. Some studies observed this phenomenon from the viewpoint of bank credit substituting TC financing [38]. However, they could only indicate that firms had the ability to forgo occupying TC, and they could not connect the supply chain perspective and the relevant theory of TC to address the phenomenon from the perspective of the motives and capabilities. Therefore, we believe there are still gaps in the systematic research on the relationship between DSC and sustainable TC provision.

Meanwhile, based on the principal-agent problem, upper echelons theory argued that top management team (TMT) power was a crucial factor influencing the strategic orientation of enterprises [39]. Scholars suggested that TMT power ensured that enterprises were consistent in their strategic formulation and execution [40,41]. However, scholars focused on how TMT power resulted in related-party transactions that harm shareholders’ interests [42], which represented a significant challenge. In practice, TMT is the initiator and implementer of DSC strategy, as well as the key to whether firms provide sustainable TC. Therefore, TMT power may play a significant role in our research. We did not, however, find that this issue had been adequately addressed in the existing studies, which are still essential to discuss.

To investigate the relationship between DSC and sustainable TC provision, we gathered the financial data and reports of Chinese listed firms from 2008 to 2020. We utilized Python to create a feature-word corpus based on market-strategy and technology-strategy orientations and to assess the degree of DSC with the TF-IDF method. To ensure the robustness of our study, we used five methods to verify the outcomes of our baseline regression, including alternative measures of variables, alternative measures of models and alternative measures of samples. We also tackled the endogenous issue by using the IV method.

Compared with the extant literature, our study makes three innovative contributions. First, we examined the micro-effects of DSC applications and extended the competitive hypothesis of TC and buyer-market theory. Unlike previous studies [23,37,38], we adopted a risk perspective and linked existing theories with enterprise capability and motivation. We investigated the impact and mechanism of DSC and sustainable TC provision systematically, which helped us to uncover the economic benefits of DSC [43] and, furthermore, enriched the research angle and scope of TC provision motivation. Second, we broadened the application domain of the upper echelon theory [44]. The extant literature highlighted the role of TMT in supply chain management [45], but its overlooked the power of TMT. We elucidated the effect of TMT power on the relationship between DSC and TC provision and offered a foundation for future research. Finally, we developed a measurement method for DSC via a methodological framework that could include both the market-strategy orientation and the technology-strategy orientation of DSC [32,34], and we utilized the methods of text analysis and TF-IDF to develop a comprehensive assessment for DSC, which could address the challenges in the existing research.

Then this paper remaining is as follows: Section 2 develops the hypotheses; Section 3 describes the data and methods; Section 4 reports the empirical results and robustness tests; Section 5 presents the extended analysis; and Section 6 concludes the paper and presents it implications.

2. Hypothesis Development

2.1. Digital Supply Chain and Sustainable Trade Credit Provision

After 2020, data emerged as a new factor for production, which has a decisive impact on the evolution of supply chain management [32,46]. According to supply chain-management theory, DSC is a vital means of ensuring the effective integration of external and internal resources [34], functioning in logistics, product, information and cash, which can enhance the overall efficiency and competitive edge of the supply chain [27]. The competitive hypothesis of TC and the buyer-market theory suggested that enterprises with a higher position in the market and better efficiency would leverage their competitive edge to occupy TC [6,7]. Thus, theoretically, DSC could enable enterprises to occupy TC, rather than offer sustainable TC. However, as stated above, there was a prevalent contrary phenomenon in practice; that is, both DSC and TC provision of firms increase simultaneously. This indicates that new perspectives are needed to extend existing theories. Considering that financial risk and operational risk are important considerations for TMT in deciding TC strategies [18,47], the assessment of risk influences firms’ confidence in the market [48]. We argue that DSC may induce enterprises to offer more TC sustainably due to the source of risks shift, which is a critical perspective from which to discuss the phenomenon above. Based on the risk transfer mechanism, we explore the effect of DSC on providing sustainable TC from the two aspects, complementing the theoretical framework of the competitive hypothesis of TC and buyer-market theory.

2.1.1. Analysis from Financial Risk

The financial risk of enterprises is a significant factor influencing sustainable TC provision [48]. Previous studies demonstrated that financialization would enlarge financial investments of enterprises, and that the value volatility of financial assets would amplify the financial risk and financial distress of enterprises [49]. To balance the cash flow volatility induced by financial risk, enterprises may proactively adopt the operational strategy of lowering TC provision, and some may even exploit their dominant competitive position to occupy TC [47].

The existing literature indicated that the core motive for firms in financialization was that the return rates of financial assets were much higher than those of real industry, and that the opportunities of investment decline in real industry [19]. Accounting for these two aspects, we argue that DSC can effectively restrain firm financialization, mitigate financial risks, unleash the “reservoir” function of financial assets and encourage firms to increase TC provision. First, DSC collects key data of procurement, sales and transportation in the supply chain [31], reduces the cost between production line and market and enhances the dynamic capability of firms [26,30]. Hence, it can curb financialization motives and financial risks and avoid the occupancy of TC. Through the analysis of information, TMT can optimize the decision-making process of firm production, procurement and sales, and strengthen the supervision of implementation of the supply chain strategy and management [25,27]. With the help of multi-level information nodes to depict the product portrait of firms, firms reduce sales uncertainty. This process will improve the operating efficiency of core businesses and the return rate of real industry. It would restrain the incentive of firm financialization and thus avoid the possibility that firms refuse to provide sustainable TC because of this incentive. Second, DSC would provide more potential opportunities in investment, restrain the process of financialization and financial risks and thus enable TC provided by firms to be more sustainable. Based on a large amount of operational data in the supply chain, DSC provides information on the external environment to TMT. By exploiting information from the external environment, firms can integrate and make optimal use of internal and external resources and improve external collaboration network [29]. Then, DSC will provide firms with more potential real economy investment opportunities, further unleashing the firm’s financial “reservoir”. It will reduce the probability of financial risk outbreaks for firms and lead firm to the provision of sustainable TC.

In this context, the higher-volatility attribute of financial investment compared with real assets will induce TMT to reduce financial assets, thereby reducing financial risks resulting from the fluctuation of financial assets and lowering the capital demand of non-core businesses. It is evident that DSC enhances the capability of firms to provide sustainable TC for collaborators.

2.1.2. Analysis from Operational Risk

Operational risk is a factor that influences the strategic decision of sustainable TC made by TMT. Previous studies found that the increase in specific assets would enhance the “lock-in effect” of both firms and their partners in the supply chain and raise the operational risk [50]. The operational investment of specific assets often requires a high level of cooperation from supply chain partners [51]. To avoid the risk of losing specific assets caused by the deterioration of cooperative relations, enterprises may establish favorable commercial terms with supply chain partners and provide them with more TC [18,52].

From two perspectives, we found that DSC would increase the investment of specific assets and the operational risk, and would then provide more sustainable TC. First, DSC can enhance the capability of the supply chain system in dynamic coordination, meeting the requirements of reinvestment for the timeliness and adequacy of resources, and expanding the possibility of reinvestment in specific assets. According to the existing literature, DSC improves the communication foundation and trust foundation between firms and supply chain collaborators [21]. Based on supply chain information sharing from the foundations upward, firms and partners conduct full communication in terms of processes, products and information [53]. This not only helps firms to fully integrate resources and knowledge of the supply chain [54,55]; it also improves the operating capability, enabling firms to coordinate production and operation activities more rapidly, reducing related costs and improving product quality [26]. In this context, firms will expand specific investments to gain competitive advantage from scale and differentiated production. Second, specific-asset investments will increase the new production and operation requirements of partners in the supply chain. Then, the requirements ask partners to expand their specific investments in parallel and crowd out the operational capital of collaborators. At this time, in order to obtain the support of collaborators, sustainable TC provided by firms plays a role in compensating the financial demand of specific assets for partners [52].

In this context, to guarantee that the specific assets driven by DSC have the congruent structure in supply chain organization and managerial modes [20], enterprises need to obtain the collaboration of partners to avoid the wastage of resources, inefficiency and the deterioration of competitiveness, which make them offer more TC to partners to ensure the stability of supply chain. It is evident that DSC boosts enterprises’ motivation to provide TC for partners.

In conclusion, DSC reduces financial risk and increases operational risk by curbing financialization and enhancing investment in specific assets. Then, by increasing sustainable TC to ensure stable cash flows, DSC plays a role in the sustainable development of the supply chain. Consequently, we propose the first hypothesis:

Hypothesis 1. 

DSC is positively associated with TC provision.

2.2. The Moderating Effect of TMT_POWER

Drawing on the upper echelons theory, we investigated how TMT power moderated the relationship between DSC and sustainable TC provision, which contributed to the advancement of supply chain management theory and TC theory [39]. Some studies suggested that enterprises decision-making was a process of negotiation and compromise among TMT, shareholders and employees, and that the characteristics of TMT would have a significant influence on enterprises decisions [45]. The development and execution of DSC strategy is the outcome of effective interactions and collaboration among TMT members, while TC provision is a major financial decision for the enterprise. The magnitude of TMT power plays a role in the direction and strength of corporate strategic planning [40]; hence, we would take this as a starting point from which to explore whether there is any potential impact of executive power on our study.

We contend that the magnitude of TMT power is associated with the abundance of the resources they possess, and also with the competence of the TMT. First, higher-power TMT allocate more human, financial and material resources for DSC, not only to facilitate internal organizational transformation but also to leverage external influence of TMT in order to motivate supply chain partners to collaborate with enterprise reform [44,56]. Second, in the corporate power structure, higher-power TMT implies that shareholders and employees relinquish more oversight power [41,57]. This not only indicates that TMT can earn the trust of shareholders and employees; it also suggests that TMT has superior abilities. Based on the two aspects discussed, we argue that TMT with higher power will not only facilitate the formulation and execution of DSC strategy more effectively; it will also convince shareholders and employees to modify their strategy of TC provision. Consequently, we put forward the second hypothesis:

Hypothesis 2. 

TMT power can positively moderate the relationship between DSC and TC provision.

Figure 1 reports the logical relationship between the key variables in our hypotheses.

3. Research Design

3.1. Sample and Data

Our empirical data are based on the economic data of enterprises listed on China’s A-share market from 2008 to 2020. On this basis, we excluded enterprises in real estate, the financial industry and those with special treatment (ST; PT). Our sample includes 3038 listed enterprises and 17,248 enterprise year observations, which mainly come from the CSMAR and CNRDS databases. The data of DSC are derived from enterprises’ annual reports. We used Python and Stata to conduct text analysis and statistics.

3.2. Measures of Variable

3.2.1. Dependent Variable

Trade credit provision (TC). We studied TC from the perspective of provision, which can be defined as “sustainable provision of TC”. We measured TC via the ratio of the sum of notes receivable, accounts receivable and prepayments to operating revenue [38]. In addition, we defined TC_S as sustainable TC provided to suppliers and TC_C as sustainable TC provided to customers. Moreover, some scholars found that the adoption of digital technology could also increase TC financing [23]. To mitigate the endogenous issues of “dual growth”, we defined net the TC provision (TC1) of enterprises measured by subtracting TC financing from TC provision. We measured TC financing via the ratio of the sum of accounts payable, notes payable and advances from customers to operating revenue. These measures were widely adopted in the extant literature [3,37].

3.2.2. Independent Variable

Concerning the digital supply chain (DSC), Saberi et al. classified the developmental model of DSC into two forms: inter-organizational form and intra-organizational form [28]. Some studies found that DSC was not only the process of applying digital technology but also the process of linking and transforming the supply chain organizational form via core enterprises [35,43]. Based on these papers, we define DSC as the process of information integration and optimization via supply chain enterprises with digital technology, mainly achieving four market-strategy objectives—logistics monitoring; product flow development; information flow sharing; and capital flow tracing [27]—with the two dimensions of market-strategy orientation and technology-strategy orientation [20,32,43]. In this context, financial indicators are inadequate to measure DSC. Annual report content communicates the future optimization direction and industrial upgrading mode of listed enterprises to supply chain partners and investors [58]. This method has become the dominant method for measuring the digital domain of enterprises. We believe that it can measure DSC with applicability, practicability and accuracy.

We constructed an objective corpus of DSC guided by policies and enterprises via the following steps. First, by using the Jieba-word-segmentation method on the “Digital Economy and Its Core Industry Statistical Classification (2021)” guideline document issued by the National Bureau of Statistics, we obtained 144 policy-guided feature words, which constituted our first feature corpus, guided by policies. Second, Fang et al. and Zhou and Li used the text-analysis method to extract digital-transformation feature words from annual reports’ management discussion sections [38,59], which formed our second feature corpus, guided by enterprises. Then, based on the definition of DSC, by filtering and integrating the feature words above, we determined 158 feature words related to DSC and report them in Figure 2. Finally, we categorized the market-strategy-oriented feature words of DSC into four orientations of logistics, product flow, information flow and cash flow [27]. We eliminated Chinese stop words, numbers and punctuation.

To mitigate the overestimation of common-word weights and underestimation of key-word weights in the method of frequency-based calculation, drawing on the literatures of TMT tone manipulation [60], we used the TF-IDF algorithm to further refine the measurement of DSC. The TF-IDF algorithm and the measurement of DSC are illustrated as following:

TF-IDF_i,j = n_i,j/∑_kN_i,j × log[M_j/(m_j + 1)];

(1)

DSC_N_i,j = DSC-n_i,j × DSC-t_i,j × 100;

(2)

DSC_i,j = ∑DSC_N_i,j (N = 1, 2, 3, 4).

(3)

In model (1), n_i,j indicates the frequency of feature n in the annual report of company i for year j; while ∑_kN_i,j indicates the total frequency of all valid words in the same report; M_j indicates the total number of annual reports for year j; and m_j indicates the number of reports that include feature n. In model (2), DSC-n_i,j (n = 1~4) represent the TF-IDF values of market-strategy-oriented features for logistics flow, product flow, information flow and cash flow, respectively; and SDC-t_i,j represents the TF-IDF value of technology-strategy-oriented features. DSC_N_i,j (N = 1~4) represent digital logistics flow (DLF) (DSC_1_i,j), digital product flow (DPF) (DSC_2_i,j), digital information flow (DIF) (DSC_3_i,j) and digital cash flow (DCF) (DSC_4_i,j), respectively; and DSC_i,j in model (3) represents the digital supply chain. To make the regression coefficients more readable, DSC_N_i,j is measured by multiplying the result of the initial calculation by 100.

3.2.3. Moderator Variable

The moderator variable was TMT power (TMT_POWER). Finkelstein suggested that TMT power consisted of structural power, ownership power, expert power and prestige power [61]. Building on this framework, Breuer and Knetsch categorized TMT power into formal authority and informal authority [62]. Accordingly, based on the sources of TMT power, we classify TMT power into direct power and indirect power. Direct power stems from formal authority, which is the foundation for TMT involvement in enterprise decision-making [40]. Indirect power originates from informal authority, which is the power entrusted to the TMT by shareholders, employees and society for trust reasons [41].

Based on the analytical framework of Bertrand and Mullainathan [63] and Chen et al. measured TMT power by using the enterprise’s pay gap [64]. We argue that the pay gap is only a part of the direct power of executives and cannot adequately measure the direct and indirect power of the TMT. Following the framework of Muttakin et al. [57], Gao et al. and Zhang et al. [40,56], we adopted principal component analysis to construct a system for evaluating TMT power from multi-dimensional perspectives, and we used seven variables related to TMT power as the principal components. The variables and measurement methods are as follows: TMT equity power, measured via the shareholding ratio of TMT; TMT compensation, measured via the total compensation of the top three TMT; non-institutional investors’ power, measured via the shareholding ratio of non-institutional investors; major shareholders’ power, measured via the ratio of the shareholding ratios of the second to fifth largest shareholders to that of the largest shareholder; board power, measured via the board size; supervisory-board power, measured via the supervisory board size; independent-director power, measured via the percentage of non-independent directors; and media power, being the number of analysts who track and analyze enterprises annually.

We calculated the industry median for the above variables. We assigned 1 to the samples with the first five components greater than the industry median and assigned 1 to the samples with the last two components lower than the industry median. TMT equity power and TMT compensation are used to measure the direct TMT power, and the other items are used to measure the indirect TMT power. All of these variables are in the form of “0–1” variables and are calculated in principal-component analysis.

3.2.4. Control Variable

To be consistent with the existing literature [3,37], we controlled for the following variables in the model. First, we controlled for various financial indicators reflecting the firm’s operating performance [38], such as enterprise size (SIZE), enterprise age (AGE), leverage ratio (LEV), net profit growth rate (PROFIT) and cash flow from operations (CFO); second, we controlled for several aspects of corporate governance [23], such as internal control index (ICI), overseas experience of executive team (MOVS), and financial expertise of executive team (MFIN); last, we selected per capita GDP (PCGDP) as a region-level control variable. The measurement definitions of these variables are presented in

Table 1. Definitions of variables.

Variable	Definition	Measurement
TC	Trade credit provision	Measured by the method in Section 3.2.1
DSC	Digital supply chain	Computed by (1), (2) and (3)
TMT_POWER	TMT power	Measured by the method in Section 3.2.3
Size	Size of firms	The natural logarithm of enterprise size plus 1
Age	Age of firms	The natural logarithm of enterprise age plus 1
Lev	The ratio of leverage	The ratio of total liabilities to total assets
Profit	The rate of growth of net profit	The growth rate of net profit in the current year relative to the previous year
CFO	Cash flow from operations	The ratio of cash flow from operating activities to net interest-bearing debt
ICI	The index of internal control	Dibo China’s internal control index for listed companies
MOVS	The overseas background of TMT	A dummy variable that equals 1 if the TMT has overseas experience, and 0 otherwise
MFIN	The financial background of TMT	A dummy variable that equals 1 if the TMT has financial experience, and 0 otherwise
PCGDP	Per capita GDP of province	The ratio of GDP to population in the provincial region where the listed company is located

.

3.3. Model

To examine the effect of DSC on TC provision, we constructed the following model to test Hypothesis 1:

TC_i,t = α₀ + α₁DSC_i,t−1 + ∑αCVs_i,t−1 + Year + Ind + Prov + ɛ.

(4)

In model (4), CVs include all control variables. The i and t denote enterprises and years, respectively. Year, Ind and Prov are time, industry and province fixed effects, respectively. We lag the independent variables and control variables of the model by one period to mitigate reverse causality concerns.

To examine the moderating effect of TMT power on the relationship between DSC and TC provision, we draw on previous studies and construct the following model to test Hypothesis 2:

TC_i,t = α₀ + α₁DSC_i,t−1 + α₂DSC_i,t−1 × TMT_POWER_i,t−1 + α₃TMT_POWER_i,t−1 + ∑αCVs_i,t−1 + Year + Ind + Prov + ɛ.

(5)

In model (5), TMT_POWER is TMT power. To facilitate the interpretation of the regression coefficients, we center the interaction term of DSC and TMT_POWER.

3.4. Descriptive Statistics

Table 2. Descriptive statistics.

Variable	N	Mean	S.D.	Min	Max
TC	17,248	0.321	0.255	0.008	1.272
DSC	17,248	0.023	0.083	0.000	0.604
TMT_POWER	17,248	0.978	0.516	0.080	2.085
Size	17,248	22.332	1.270	20.100	26.231
Age	17,248	2.075	0.745	0.693	3.258
Lev	17,248	0.416	0.191	0.058	0.836
Profit	17,248	0.083	4.216	−23.572	20.402
CFO	17,248	3.675	13.978	−3.026	110.945
ICI	17,248	0.589	0.492	0	1
MOVS	17,248	0.701	0.458	0	1
MFIN	17,248	669.601	105.308	0.000	904.590
PCGDP	17,248	11.054	0.516	9.085	12.013

reports the descriptive statistics of our variables. TC has a maximum of 1.272, a minimum of 0.008, a mean of 0.321 and a standard deviation of 0.255, indicating heterogeneity in TC provision among enterprises. DSC has a minimum of 0, a maximum of 0.604, a mean of 0.083 and a standard deviation of 0.083. The left-skewed mean suggests that DSC is still in its infancy, and there is a large disparity in DSC. TMT_POWER has a maximum of 2.085, a minimum of 0.08, a mean of 0.978 and a standard deviation of 0.516, indicating substantial variation in TMT power. The means and standard deviations of other variables are within reasonable ranges, which provides a solid basis for the next phase of our research.

4. Results and Discussion

4.1. Baseline Regression Results

Table 3. The effect of SCD and TMT_POWER on TC.

Variables	TC	TC	TC	TC1	TC1
	(1)	(2)	(3)	(4)	(5)
SCD	0.446 ***	0.216 ***	0.195 ***	0.107 ***	0.069 **
	(0.024)	(0.028)	(0.027)	(0.029)	(0.029)
SCD × TMT_POWER			0.107 **		0.225 ***
			(0.042)		(0.046)
TMT_POWER			0.045 ***		0.040 ***
			(0.004)		(0.004)
SIZE		−0.034 ***	−0.032 ***	−0.028 ***	−0.026 ***
		(0.002)	(0.002)	(0.002)	(0.002)
AGE		−0.033 ***	−0.026 ***	−0.016 ***	−0.010 ***
		(0.003)	(0.003)	(0.003)	(0.003)
LEV		−0.007	−0.003	−0.363 ***	−0.359 ***
		(0.012)	(0.012)	(0.012)	(0.012)
PROFIT		0.002 ***	0.002 ***	0.001	0.001 *
		(0.000)	(0.000)	(0.000)	(0.000)
CFO		−0.002 ***	−0.002 ***	−0.002 ***	−0.002 ***
		(0.000)	(0.000)	(0.000)	(0.000)
ICI		−0.000 ***	−0.000 ***	0.000	0.000
		(0.000)	(0.000)	(0.000)	(0.000)
MOVS		0.013 ***	0.011 ***	0.021 ***	0.019 ***
		(0.004)	(0.004)	(0.004)	(0.004)
MFIN		−0.005	−0.006	−0.007 *	−0.007 *
		(0.004)	(0.004)	(0.004)	(0.004)
PCGDP		0.054 ***	0.051 **	0.048 **	0.045 **
		(0.020)	(0.020)	(0.020)	(0.020)
Constant	0.311 ***	0.652 ***	0.609 ***	0.440 **	0.154
	(0.002)	(0.212)	(0.222)	(0.212)	(0.226)
YEAR/Prov/IND	YES	YES	YES	YES	YES
R2	0.021	0.274	0.277	0.244	0.252
N	17,248	17,248	17,248	17,248	17,248

Note: *** P < 0.01; ** P < 0.05; * P < 0.1. Standard errors in parentheses.

presents the regression results of model (4) and model (5). In column (1), we can see that the coefficient of DSC from the univariate regression is 0.446, and that it is significant at the 1% level. In column (2), after controlling for all other variables and fixed effects, the coefficient of DSC remains positive and significant at the 1% level. These coefficients suggest that DSC is a key determinant of TC provision among enterprises. In column (3), we examined the moderating effect of TMT power. In addition, both the coefficient of DSC and the coefficient of its interaction with TMT_POWER are significantly positive. This implies that TMT power enhances the positive impact of DSC on TC provision. In columns (4) and (5), we used TC1 to replace TC in model (4) and model (5). The coefficient in column (4) indicates that DSC still facilitates TC provision when accounting for TC financing. The interaction term in column (5) is significantly positive, indicating that TMT power still exerts a positive moderating effect. The results indicate that Hypothesis 1 and Hypothesis 2 pass our test of regression.

Table 4. The tests of the sub-indicators of SCD and TC.

Variables	TC_S	TC_C	TC
	(1)	(2)	(3)	(4)	(5)	(6)
SCD	0.011 **	0.204 ***
	(0.005)	(0.026)
SCD_1			0.895 ***
			(0.083)
SCD_2				1.233 ***
				(0.125)
SCD_3					0.230 ***
					(0.064)
SCD_4						0.147
						(0.395)
SIZE	−0.001	−0.033 ***	−0.034 ***	−0.035 ***	−0.035 ***	−0.035 ***
	(0.000)	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)
AGE	−0.003 ***	−0.031 ***	−0.033 ***	−0.033 ***	−0.033 ***	−0.033 ***
	(0.001)	(0.002)	(0.003)	(0.003)	(0.003)	(0.003)
LEV	0.016 ***	−0.022 **	−0.004	−0.004	−0.006	−0.004
	(0.003)	(0.011)	(0.012)	(0.012)	(0.012)	(0.012)
PROFIT	0.000	0.002 ***	0.002 ***	0.002 ***	0.002 ***	0.002 ***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
CFO	−0.000 ***	−0.001 ***	−0.002 ***	−0.002 ***	−0.002 ***	−0.002 ***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
ICI	−0.000 ***	−0.000 ***	−0.000 ***	−0.000 ***	−0.000 ***	−0.000 ***
	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)	(0.000)
MOVS	0.001	0.012 ***	0.013 ***	0.013 ***	0.013 ***	0.013 ***
	(0.001)	(0.003)	(0.004)	(0.004)	(0.004)	(0.004)
MFIN	0.002 ***	−0.008 **	−0.005	−0.005	−0.006	−0.006
	(0.001)	(0.003)	(0.004)	(0.004)	(0.004)	(0.004)
PCGDP	−0.013 **	0.066 ***	0.054 ***	0.054 ***	0.053 ***	0.052 ***
	(0.005)	(0.018)	(0.020)	(0.020)	(0.020)	(0.020)
Constant	0.276 ***	0.355 *	0.693 ***	0.704 ***	0.720 ***	0.730 ***
	(0.057)	(0.188)	(0.214)	(0.214)	(0.215)	(0.215)
YEAR/Prov/IND	YES	YES	YES	YES	YES	YES
R2	0.113	0.302	0.277	0.275	0.271	0.270
N	17,248	17,248	17,248	17,248	17,248	17,248

Note: *** P < 0.01; ** P < 0.05; * P < 0.1. Standard errors in parentheses.

further tests Hypothesis 1 by adding the four sub-indicators of DSC and the two sub-indicators of TC to model (4). From columns (1) and (2), we find that DSC enhances the TC provision of enterprises to both upstream suppliers and downstream customers. In columns (3) to (6), we report the coefficients of the sub-indicators of DSC. The coefficients and significance levels of DSC_1, DSC_2, DSC_3 and DSC_4 show that DLF, DPF and DIF increase the TC provision of enterprises, but DCF cannot have such an impact. We attributed this to several reasons: First, China’s Digital RMB plan is a new economic policy proposed by the State Council of China in 2020, aiming to address the capital traceability difficulties in the process of economic operation. Due to the short time spanning, DCF is still under construction, and only a few enterprises have implemented DCF. Second, in the initial stage, DCF requires enterprises to involve external financial institutions in establishing a trustworthy financial system. The financial support from external financial institutions may substitute for the TC provided by enterprises. In addition, the financing environment improved for SMEs will reduce the required TC provision.

4.2. Robustness Tests

To account for the potential bias caused by the measurement methods of variables and models, we conduct a series of robustness checks to validate our results using various approaches, including the following: (1) alternative measures of TC; (2) alternative measures of SCD; (3) alternative measures of TMT_POWER; (4) alternative methods of models; (5) alternative methods of sample selection; and (6) the IV method.

4.2.1. Alternative Measures of TC

We added three alternative measures in model (4), which are discussed in

Table 5. The robustness tests of alternative variables.

Variables	TC2	TC3	TC_YN	TC
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
SCD	0.112 ***	0.153 **	1.164 ***			0.288 ***	0.172 ***
	(0.015)	(0.064)	(0.150)			(0.046)	(0.031)
SCD_IND1				0.003 ***
				(0.001)
SCD_IND2					0.080 ***
					(0.011)
SCD * TMT_POWER1						−0.113 **
						(0.050)
TMT_POWER1						−0.044 ***
						(0.005)
SCD * TMT_POWER2							0.109 **
							(0.048)
TMT_POWER2							0.012 ***
							(0.004)
Control variables	YES	YES	YES	YES	YES	YES	YES
Constant	0.094	0.923 *	0.679 ***	0.486 **	2.052	0.702 ***	0.736 ***
	(0.125)	(0.544)	(0.213)	(0.217)	(1.381)	(0.221)	(0.225)
YEAR/Prov/IND	YES	YES	YES	YES	YES	YES	YES
R2	0.277	0.246	0.271	0.273	0.052	0.278	0.274
N	12,396	12,396	17,243	17,239	17,239	17,248	17,049

Note: *** P < 0.01; ** P < 0.05; * P < 0.1. Standard errors in parentheses.

. We define TC2 as the ratio of the sum of notes receivable, accounts receivable and prepaid accounts to total assets [17]; and we define TC3 as the ratio of the sum of notes receivable, accounts receivable and prepaid accounts to operating costs [19]. The results in columns (1) and (2) show that DSC has a positive and significant correlation with both TC2 and TC3. Moreover, we create TC_YN, which takes the value of 1 for enterprises with TC above the industry median and 0 otherwise. In Probit regression of column (3), we find that DSC has a positive and significant correlation with TC_YN. These findings are consistent with the baseline regression results.

4.2.2. Alternative Measures of DSC

To check the robustness, following Xiao et al. [65], we also employed two alternative measures for DSC: DSC_IND1 and DSC_IND2. To construct these two measures, we calculated the industry mean (DSC_mean), maximum (DSC_max) and minimum (DSC_min) based on yearly data. DSC_IND1 is obtained by adjusting DSC by DSC_mean, and DSC_IND2 is obtained by adjusting DSC by the value of DSC_max minus DSC_min, as shown in model (6) and model (7). In Table 5, columns (4) and (5), we find that TC has a positive and significant correlation with both DSC_IND1 and DSC_IND2, which is consistent with the baseline regression results.

DSC_IND1_i,j = DSC_i,j/DSC_mean

(6)

DSC_IND2_i,j = (DSC_i,j − DSC_min)/(DSC_max − DSC_min)

(7)

4.2.3. Alternative Measures of TMT_POWER

We will test whether the moderating effect is stable. We argue that TMT power is affected by whether TMT members hold positions in shareholder units or whether the CEO holds dual roles (CEO and chairman) [56,57]. The first one implies that shareholder units have more influence over enterprise-decision-making behavior, which constrains TMT power [40]. The second one implies that TMT has more authority over the board agenda [64]. We created TMT_POWER1, which takes the value of 1 for TMT holding positions in shareholder units and 0 otherwise. We also created TMT_POWER2, which takes the value of 1 for TMT holding dual roles and 0 otherwise. In column (6) of Table 5, the interaction term between DSC and TMT_POWER1 has a negative and significant coefficient, and so does TMT_POWER1. In column (7), the interaction term between DSC and TMT_POWER2 has a positive and significant coefficient, and so does TMT_POWER2. These findings are consistent with the baseline regression results.

4.2.4. Alternative Methods of Modeling

The choice of model also introduce some bias in our research results. Therefore, we employed four alternative methods: first, we lagged the dependent variables (TC) by two periods; second, we used enterprise-fixed effect; third, we used enterprise-level clustered standard errors; fourth, we tested for the possibility of omitted control variables in the model. We included corporate risk taking as an additional control variable. We defined RISK as the standard deviation of industry-adjusted ROE over a three-year window (t − 1, t, t + 1). Finally, since the sample from 2008 to 2010 only constitutes 10% of the total, and our conclusion may be prone to biases across different temporal contexts, we dropped the observations prior to 2011 and reran the regressions. In columns (1) to (5) of

Table 6. The tests of other robustness methods.

Variables	TC_F	TC
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
DSC	0.205 ***	0.085 *	0.216 ***	0.221 ***	0.218 ***	0.176 ***	0.214 ***
	(0.033)	(0.049)	(0.048)	(0.028)	(0.028)	(0.030)	(0.049)
Risk				0.389 ***
				−0.051
Control variables	YES	YES	YES	YES	YES	YES	YES
Constant	0.487 *	−0.652 **	0.652 **	0.706 ***	0.638 **	0.305	0.318
	(0.281)	(0.304)	(0.281)	(0.214)	(0.306)	(0.290)	(0.237)
YEAR/Prov/IND	YES	YES	YES	YES	YES	YES	YES
Firm	NO	YES	NO	NO	NO	NO	NO
R2	0.289	0.052	0.274	0.277	0.285	0.274	0.275
N	12,396	17,248	17,248	17,248	14,882	11,531	14,249

Note: *** P < 0.01; ** P < 0.05; * P < 0.1. Standard errors in parentheses.

, the coefficient of DSC is positive and significant, consistent with the baseline regression results.

4.2.5. Alternative Methods of Sample Selection

Our research may be affected by sample selection bias due to the strategies of different firms. The quality of the information on the annual reports of listed companies may influence the reliability and validity of the data. Moreover, some industries’ listed companies have edges in DSC. To address these two biases of sample selection, we used two methods to exclude some samples: in Table 6, column (6) excludes samples with the rating of annual-reports transparency at “C” and “D”, which is issued by the China Securities Regulatory Commission; column (7) excludes samples from industries such as “computer, communication and other electronic equipment manufacturing”, “software internet” and “scientific research and technical services”. The coefficients above are positive and significant. It is found that the results are consistent with the baseline regression results.

4.2.6. The IV Method

To address the endogenous issues, we employed the two-stage IV method and constructed two variables in

Table 7. The IV tests.

Variables	IV = IV1		IV = IV2
	(1) SCD	(2) TC	(3) SCD	(4) TC
	First Stage	Second Stage	First Stage	Second Stage
IV1	1.353 ***
	(0.013)
IV2			0.623 ***
			(0.048)
SCD		0.118 ***		1.074 ***
		(0.044)		(0.251)
Control variables	YES	YES	YES	YES
Constant	−0.025	0.737 ***	0.118 **	0.551 **
	(0.056)	(0.222)	(0.071)	(0.232)
YEAR/Prov/IND	YES	YES	YES	YES
R2	0.569	0.273	0.301	0.219
N	17,248	17,248	17,248	17,248
Kleibergen–Paap rk LM statistic	252.767		35.987
Kleibergen–Paap rk Wald F statistic	1487.123		38.236

Note: *** P < 0.01; ** P < 0.05. Standard errors in parentheses.

. IV1 is the interaction term between the initial level of DSC and the number of Internet users per 100 people (the data were discontinued in 2016, which can more effectively deal with the issue) in each province from 2004 to 2016 [38]. IV2 is the industry-level means of DSC after excluding the listed companies themselves [65]. In columns (1) and (3), the coefficients of IV1 and IV2 are positive and significant and indicate that IV1 and IV2 are related to DSC. In addition, based on Kleibergen–Paap rk LM statistics and Kleibergen–Paap rk Wald F statistics, there are no weak instruments and under-identification problems in IV1 and IV2. In columns (2) and (4), the coefficients of DSC are positive and significant, consistent with the baseline regression results after accounting for the endogenous issue.

5. Additional Analyses

5.1. Exploring of Mechanism Effects

In the theoretical analysis of Section 2, we proposed that DSC improves the sustainable provision of TC by curbing financialization and increasing asset specialization. To test whether the analysis of our hypothesis is equally statistically significant, drawing on the mediation-effect model, we constructed the following models to test the mechanism effect:

Fin/Asset_i,t = α₀ + α₁DSC_i,t + ∑αCVs_i,t + Year + Ind + Prov + ɛ;

(8)

TC_i,t = α₀ + α₁DSC_i,t−1 + α₂Fin/Asset_i,t−1 + ∑αCVs_i,t−1 + Year + Ind + Prov + ɛ.

(9)

In model (8) and model (9), FIN is enterprise financialization and ASSET is enterprise asset specialization. We define FIN as the ratio of financial assets to total assets [66], and ASSET as the ratio of R&D expenditure to operating income [51]. R&D expenditure reflects the future competitive advantage of firms, indicating the direction of core business in the future and with higher operational risks. Simultaneously, innovation in the new era demands more support from collaborators in the network, and also imposes higher requirements for collaborators’ capital demand [50], which conforms to the definition of asset specialization.

5.1.1. Curbing Financialization

Financialization increases the financial risk and affects the sustainable provision of TC. The more the trend of financialization of companies is curbed, the more sustainable the TC provided by companies would be. In this study, we investigated the mechanism between DSC and TC provision by using Fin as a mediating variable. In column (1) of

Table 8. The results of mechanism effects.

Variables	FIN	TC	ASSET	TC
	(1)	(2)	(3)	(4)
DSC	−0.014 *	0.212 ***	0.028 ***	0.174 ***
	(0.008)	(0.027)	(0.006)	(0.028)
FIN		−0.277 ***
		(0.027)
ASSET				0.999 ***
				(0.068)
Control variables	YES	YES	YES	YES
Constant	−0.025	0.737 ***	0.118 **	0.551 **
	(0.056)	(0.222)	(0.071)	(0.232)
YEAR/Prov/IND	YES	YES	YES	YES
R2	0.277	0.271	0.275	0.270
N	17,248	17,248	14,530	14,530

Note: *** P < 0.01; ** P < 0.05; * P < 0.1. Standard errors in parentheses.

, the coefficient for DSC is −0.014 and statistically significant. This indicates that DSC reduces the financialization of enterprises and frees up financial capital. In column (2), the coefficient for DSC is positive and significant, while the coefficient for FIN is negative and significant, suggesting that financialization inhibits the sustainable TC provided by enterprises. These results imply that lowering financialization is a key mechanism through which DSC enhances TC provision. Curbing the effect of financialization via DSC reduces the financial risk of high volatility of corporate financial assets. Then, stable financial metrics on liquidity can not only provide more confidence that firms can provide sustainable TC; they can also improve the transparency of firms’ financial indicators. Moreover, they can bring more bank loans and enable firms to play a better role in reallocating TC resource in supply chain. We believe that it is a necessary path for sustainable TC provision.

5.1.2. Improving Assets Specialization

Asset specialization raises the operational risk of firms and is an important motivation to provide sustainable TC. In this study, we used ASSET as a mediating variable to explore the mechanism between DSC and TC. In column (3) of Table 8, the coefficient for DSC is positive and significant, implying that DSC increases the level of asset specialization for enterprises. In column (4), the coefficient for DSC is positive and significant, as well as the coefficient for ASSET. These results imply that improving asset specialization is another key mechanism through which DSC enhances TC provision by enterprises. Asset specialization is a source of competitive advantage for enterprises in the future and a necessary process of implementation for DSC. In addition, it can reflect the division of the supply chain driven by DSC; that is, with the help of DSC, enterprises can control the supply chain and compete with other supply chains by mastering the special assets in the supply chain via division. Enterprises can redistribute the structure of assets among the supply chain and gain better dynamic capabilities. However, as a corresponding cost, firms need to provide sustainable TC for their partners.

5.2. Heterogeneous Analysis

In the subsequent sections, we implemented heterogeneous analysis and enriched the empirical evidence for the relationship between DSC and sustainable TC provision. From firm-level perspectives such as agency costs and the cooperation of the supply chain and state ownership, we discussed whether the effect of DSC on sustainable TC provision varies, and additionally from industry-level perspectives such as the degree of industrial competition and nationally supported industries.

5.2.1. Agency Costs

It was noted in the existing literature that agency cost reflected the operational efficiency of enterprises. In addition, higher agency costs imply that enterprises are forced to face information asymmetry issues in their supply chains. Therefore, we explored whether the effect of DSC on TC provision varies depending on the level of agency costs. We measured agency costs via the ratio of management expenses and sales expenses to operating income [59]. Next, we split them into two groups based on the industry median and code the higher group as AC = 1. In

Table 9. The tests of other robustness methods.

Variables	TC
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
	AC = 1	AC = 0	SCC = 1	SCC = 0	SOE = 1	SOE = 0	IC = 1	IC = 0	NS = 1	NS = 0
DSC	0.292 ***	0.101 **	0.104 **	0.327 ***	0.102 *	0.244 ***	0.059	0.193 ***	0.066	0.257 ***
	(0.038)	(0.039)	(0.045)	(0.038)	(0.058)	(0.031)	(0.058)	(0.031)	(0.048)	(0.327)
Control variables	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Constant	0.672 *	−0.026	0.802 *	0.368	−0.177	1.284 ***	0.529 *	0.261	0.665 **	0.619 **
	(0.355)	(0.267)	(0.451)	(0.356)	(0.277)	(0.373)	(0.290)	(0.316)	(0.333)	(0.291)
YEAR/Prov/IND	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
R2	0.289	0.052	0.274	0.277	0.274	0.275	0.266	0.276	0.238	0.298
N	8315	8789	6189	6658	6612	10,636	8605	8642	7773	9464
Suest test p-value	0.000		0.000		0.029		0.041		0.001

Note: *** P < 0.01; ** P < 0.05; * P < 0.1. Standard errors in parentheses.

, columns (1) and columns (2) report the results for this group. As can be seen, the coefficients of DSC for the two groups indicate that DSC has significant and positive effects in both groups and has a more positive impact in higher-level groups. In addition, the result of the Suest test indicates that the coefficients are statistically different. It means that in the higher-level group, DSC can sufficiently reduce the information asymmetry problems and exert a more prominent effect.

5.2.2. Cooperation of Supply Chain

The existing literature suggested that the degree of supply chain cooperation reflects the trust between the firm and its supply chain partners [17]. In addition, higher-degree cooperation can mitigate the information asymmetry issues in supply chain. Next, we explored whether the effect of DSC on TC provision varies depending on the degree of supply chain cooperation. We measured it via the sum of two ratios, which are the ratio of purchases from the top five suppliers to total purchases and the ratio of sales to the top five customers to total sales [17]. Then, we split them into high and low groups based on the industry median and code the high group as SCC = 1. In Table 9, columns (3) and (4) report that DSC has significant and positive effects in both groups. It is found that DSC is more significant and positive in the lower group. In addition, the Suest test also shows that the difference is statistically significant. It is meant that in the lower-level group, DSC can increase the level of trust and exert a more prominent effect.

5.2.3. State Ownership

Some studies have pointed out that ownership reflects the objectives and philosophy of an operation [67]. In fact, China’s state-owned enterprises (SOEs) have business objectives that prioritize social benefits over economic benefits. To ensure their social service stability, SOEs prefer to focus on supply chain stability and sustainability and are willing to offer more favorable TC terms [67]. On this basis, we examined whether the effect of DSC on TC provision differs due to different ownership. We coded SOEs as SOE = 1. In Table 9, columns (5) and (6) report that DSC has significant and positive effects for both SOEs and non-SOEs. The coefficients of DSC and the Suest test indicate that the positive effect of DSC is stronger in non-SOEs than in SOEs. This means that DSC has a more prominent role in non-SOEs and raises the priority of externalities in non-SOEs.

5.2.4. The Degree of Industrial Competition

Industrial characteristics may influence the fact that firms within an industry exhibit different features, and consequently, they may affect the paths through which DSC influences TC [17]. Among these characteristics, the degree of industrial competition is one of the master aspects that deserves attention. Previous studies suggested that the degree of industry competition could influence the decisions of TC [16]. In our study, we verified that the mechanisms by which DSC enhanced TC functioned via increasing operational risk and decreasing financial risk. However, in industries with a higher degree of monopoly, firms enjoy excess returns and stable cash flows, which may limit the extent to which DSC alters their operational and financial risks, and thus may constrain its effect on TC. In contrast, in industries with more intense competition, the situation may be reversed. To examine this hypothesis, we used the degree of industrial concentration to measure the degree of competition of firms in certain industries and investigate whether the impact of SCD on TC varies. We divided the sample into high and low groups based on the annual median of the index of industrial HHI [17,38]. We coded the industry with a higher degree of HHI as IC = 1, which meant that the degree of industrial competition was higher. In columns (7) and (8) of Table 9, the coefficients of DSC are only significantly positive in the low group. In addition, the Suest test indicates that there is a significant difference between the two groups of samples. It implies that in industries with higher competition, DSC can play a more prominent role in improving sustainable TC.

5.2.5. Nationally Supported Industries

The presence or absence of national support also reflects some features of the industry—in particular, whether the industry in which the firm operates has a forward-looking and high-tech nature. Generally speaking, firms in industries that receive national support can achieve higher economic returns and market valuations. More importantly, these firms tend to focus more on R&D, leading to their possession of larger specific assets [68]. Meanwhile, because their products have higher value, their financial assets are lower. Therefore, these firms face higher operational risks but lower financial risks. This is in line with our findings on firm risk after the effect of DSC. At this point, the effect of DSC may not be able to reach its expected outcome in industries that receive national support. For this reason, we further examine whether the impact of DSC on TC differs in nationally supported industries. We divided them into two groups based on whether the five-year plans of governments at all levels designate the industry in which the firm operates as a national-support industry. In addition, we coded the nationally supported industry as NS = 1. In columns (9) and (10) of Table 9, the coefficients of DSC are only significantly positive in non-nationally supported industries. Moreover, the Suest test indicates that there is a significant difference between the two groups of samples. This indicates that in traditional industries, DSC can play a more prominent role, facilitating the economic reform and transformation of traditional industries.

6. Conclusions and Discussion

TC provision is one of the most crucial short-term financing behaviors and strategic decisions in the supply chain. How to enhance TC provision to actively exert credit allocation function and improve collaboration within the supply chain has been a vital theoretical and practical issue. In recent years, the application of digital technology in economic activities and the transformation of economic agents have aroused widespread attention [20]. Existing studies not only explored the mechanism of digitization on TC financing [23,37] and demonstrated that bank credit may substitute TC financing in the context of digitization [38]; they also found that DSC would increase supply chain resilience and performance from multiple aspects [33]. DSC is an essential component of facilitating enterprise cooperation. Therefore, it should be emphasized that enterprises need to undertake the transformation of DSC proactively [21,25], and the government must also pay attention to bridging the digital gap [69]. This implies that DSC may also exert a significant influence on sustainable TC provision, but existing research could not address this issue. Our study complements this key content and connects the two important research streams of DSC and TC in the field of supply chain. Meanwhile, our study also offers some insights into how to advance the DSC transformation of SMEs. SMEs are constrained by the scarcity of resources and capabilities; hence, their transformations are particularly challenging, especially in the absence of sufficient sources of funding [70]. Some scholars have suggested that large enterprises should be encouraged to provide sustainable support for SMEs’ transformation [3]. However, they did not point out specific and feasible means. In fact, sustainable TC offered by large enterprises may be an important path to achieving this. Compared with existing studies, our study identified a positive relationship between DSC and TC provision in large enterprises. Hence, we offer the theoretical foundation that TC provided from large enterprises can be a sustainable source of funding for SMEs’ digital transformation. By developing a digital network of the supply chain involving large, medium and small enterprises, based on the enhancement of credit allocation efficiency, we can promote the comprehensive transformation of DSC and the co-creation in value of the supply chain [5]. This broadens the significance of our study and also indicates many issues worthy of continuous attention in future research on the supply chain. At the same time, unlike existing research on DSC, our study used data from Chinese listed firms from 2008 to 2021 and adopts the methods of text analysis and TF-IDF calculation to construct the indicators of DSC. This compensates for the deficiency of empirical evidence in existing related research and can provide reference for subsequent research on DSC.

Our study also has special contributions in theoretical application. First, our study enriches the literature on DSC and TC. Taking firm risk as the entry point, we answer why DSC can simultaneously enhance competitive advantage and TC provision of firms, which complements scholars’ understanding of the TC competitive hypothesis and buyer-market theory. Second, we applied upper echelon theory to the research of DSC and TC. We confirmed the contribution of TMT power to the relation between DSC and TC provision and enriched the literature on upper echelon theory. In sum, our study addresses the problem of how to encourage excellent enterprises to provide sustainable TC, which has important significance in managerial theory.

Furthermore, our study indicates that DSC of large enterprises can mitigate the unequal distribution of credit resources along the supply chain. Liu et al. argued that offering TC was detrimental to the sustainability of enterprises [71]. However, the rational allocation of credit resources is a vital way to address the supply chain crisis and enhance the supply chain stability [72]. In the context of Industry 4.0, digital technology reduces the information asymmetry in the supply chain and improves collaboration efficiency among enterprises [73]. In the domain of digital coordination between large enterprises and their partners, constrained by the insufficient funds of SMEs, there are certain ceilings and gaps in DSC of large enterprises. Therefore, a reasonable policy of sustainable TC provision can not only assist in improving the stability level of supply chain but also help to safeguard the interests of supply chain partners and augment the positive externality and sustainability [70,74]. Our study offers the following insights for management practice. First, for large enterprises, DSC will entail changes in financial risks and operational risks. They can enhance the flexibility of TC provision and decide credit policies that align with their operational status. Meanwhile, in this process, large enterprises should be mindful of the role of TMT. Second, for SMEs, DSC of large enterprises provides them with sustainable TC, which can not only lower financing costs but also diminish operational uncertainty and gain the opportunities of “free riding”. Third, for government, DSC enables large enterprises to take on more work of credit allocation in the supply chain, which facilitates the improvement of the allocation efficiency of financial resources. Drawing on the empirical evidence from Chinese listed companies, our study offers evidence to support relevant practitioners to optimize their management decisions.

We derived several related conclusions. First, using data from Chinese listed firms from 2008 to 2020, we found that DSC could enhance the TC provision of firms, addressing the question of whether DSC can affect TC provision. Second, for enterprises with higher-power TMT, the effect of DSC on TC provision is more pronounced. Our main conclusions remain valid after the robustness tests. Third, for both upstream and downstream partners, we found that the effects of DSC were simultaneously positive. In addition, among the sub-indicators of DSC, LFD, PFD and IFD can increase TC provision. However, the impact of CFD on TC provision is non-significant. Furthermore, from the perspectives of financial risk and operational risk, we determined the mechanisms of DSC: DSC curbs the financialization and increases the asset specialization of enterprises. Lastly, based on heterogeneous-analysis tests, we found that the effect of DSC was more pronounced for firms with higher agency cost, lower supply chain collaboration and non-state ownership, and it was also more salient in industries with higher competition and without national support.

Our study has the following limitations. First, our sample is panel data of Chinese listed firms excluding financial and real-estate industries. Based on data from other countries and specific industries on DSC, future research can further explore other potential mechanisms of DSC influencing TC provision. Second, we used the TF-IDF algorithm to replace the TF algorithm, which corrects the overestimation of common word weight and the underestimation of key feature word weight in existing research [71]. However, we still confront the inherent drawbacks of text analysis methods [38]; that is, the text of firms’ annual reports cannot fully capture the process of DSC, and there is no consistent disclosure standard for DSC across firms. Currently, the Chinese government is promoting the incorporation of data assets into firm balance sheets, which involves the standardization of data asset value assessment and digital process assessment. We believe that this trend may be a key measure to address the deficiencies of the annual report text. Future research on DSC can include standardized rules for data asset assessment in China as a basis from which to mitigate the lack of standardization in existing research. Third, the TF-IDF algorithm does not incorporate the position information of feature words in annual reports, and the information in the management discussion section of annual reports may be more valuable [75]. Future research can consider the conditions of weight assignment and introduce algorithms that contain annual report position information to refine the DSC indicators we constructed. Fourth, due to the length of our paper, we focused on the effects of DSC within the organizations but neglected to assess whether DSC across organizations can affect their decisions regarding sustainable TC [28]. Finally, there is a severe digital divide across different countries and regions [76], and we do not discuss where the divide affects the impact of DSC. Based on our findings, future studies can examine the effects of DSC on TC decisions more holistically.