Effects of Digital Transformation on Dynamic Capital Structure Adjustment: Evidence from China

Abstract

Over the past two decades, there has been a surge in digital innovation. China is the world’s second-largest digital economy entity and the national strategy to build a digital China is in full swing. Chinese enterprises have received great support while pursuing digital transformation. This study aims to explore the impact of digital transformation on firm dynamic capital structure adjustment by taking evidence from Chinese listed firms. Based on the data of 3855 Chinese A-share listed firm-year observations during 2011–2021, this study employed a panel data fixed effects model to ascertain the association between the proposed variables. The results indicate that digital transformation has a significant positive effect on accelerating capital structure adjustment speed. Furthermore, we show that digital transformation promotes dynamic capital structure adjustment through financial flexibility. The impact of digital transformation on dynamic capital structure adjustment is characterized by heterogeneity in the ownership type, asset size, and credit cycle. In addition, reduced financing constraints and growing economic policy uncertainty will moderate the relationship between digital transformation and dynamic capital structure adjustment. Our findings provide insights into how digital transformation is driving companies to accelerate their capital structure adjustment speed to the optimal, and supplements the existing literature on the determinants of dynamic capital structure adjustment.

1. Introduction

The global economy is in the era of digital transformation (hereafter referred to as DT) [1]. Three technological trends, including quantum technologies, exponential intelligence, and ambient computing, reported by Deloitte are all related to digital technologies [2]. According to a survey reported in the Harvard Business Review, DT risk is one of leadership’s concerns [3]. DT has changed the way an organization operates. DT reflects the transformation of the traditional organizing logic to a new digital organizing logic [4]. The financial impact of DT on companies and the unprecedented opportunities it provides has attracted widespread attention worldwide. On the one hand, DT impacts the whole organization, including both internal and external users, especially with regard to business models, operational processes and user experience, which link the entire production, consumption and service chain [5]. On the other hand, DT has significantly broadened access to information, increased the conversion rate of information, reduced information asymmetry, improved corporate governance ability and increased firm performance [6]. However, the benefits of DT inevitably entail significant costs, both financial and labor. Where these costs come from (e.g., financing or not) and where they go (e.g., digital equipment purchases) relates to the firm’s capital structure. Hence, research questions include whether and how a firm’s DT affects its capital structure.

The fact that China places a high priority on the growth of the digital economy is why we are looking at the impact of DT on dynamic capital structure adjustment in a Chinese context. According to a report by the China Academy of Information and Communications Technology on the development of China’s digital economy in 2021, the value added by the digital economy in the 47 countries studied was USD 38.1 trillion, representing 45.0% of GDP. On the scale of the digital economy, China ranked second, with USD 7.1 trillion, after the U.S. These economic data outcomes demonstrate that the global digital economy is growing rapidly worldwide, while China is an important part of the world’s digital economy and is a representative of the digital economy in developing countries. Moreover, it is important to note that policies related to DT have been included frequently in the Chinese government’s key national plans in recent years. For example, in accordance with the report of the latest 20th National Congress of the Communist Party of China in 2022, building a “digital China” is vital. Therefore, a series of policies/initiatives have been introduced in China to support enterprises in pursuing DT. For instance, purposes include boosting research & development and innovation investment, consolidating digital infrastructure, as well as encouraging the convergence of the digital and real economies, etc. In consequence, studying how the DT has affected firm’s capital structure adjustment in the Chinese context is important from both a theoretical and practical standpoint.

The objective of our paper was to examine the impact of a firm’s DT on its capital structure adjustment. Using the data of 3855 Chinese A-share listed firm-year observations from the Shanghai and Shenzhen stock exchanges during 2011–2021, the study employed ordinary least square and panel data fixed effects techniques to ascertain the association between the proposed variables. It was found that DT can speed up the pace of adjustment of firms’ capital structure, and DT can influence dynamic capital structure adjustment through financial flexibility. DT increases the adjustment speed by influencing debt financing or equity financing. In addition, the study’s heterogeneity tests indicated that the impact of DT on dynamic capital structure adjustment varied across the ownership nature, asset size, and credit cycles of firms. Furthermore, it was concluded that reduced financing constraints and growing economic policy uncertainty moderates the relationship between DT and dynamic capital structure adjustment.

This paper contributes to earlier studies in the following respects: First, this study investigated the relationship between DT and dynamic capital structure adjustment. The empirical results demonstrated that DT has a significant effect on the dynamic adjustment of capital structure, and complement current literature concerning the factors that influence dynamic capital structure adjustment. A new theoretical instrument and basis for optimizing the capital structure of firms is presented. Second, to reveal the relationship between DT and dynamic capital structure adjustment more systematically, this paper further explores specific ways in which DT affects firms’ adjustment speed of capital structure and the mediating channel through which DT affects capital structure adjustment. Third, this paper investigates the heterogeneous characteristics of the three dimensions of the nature of ownership, asset size and credit cycle, and the moderating role of financing constraints and economic policy uncertainty. The results provide positive insights into the forms through which companies can adjust their capital structure faster, the channels through which they can accelerate their adjustment speed, how firm management can avoid a one-size-fits-all policy, and how they can adjust their capital structure in different contexts. What is more, from a debt perspective, it provides microscopic advice on which forms of financing should be used for firms aiming to achieve an optimal capital structure, as well as theoretical implications for how enterprises can leverage digital technology to protect themselves against financial risks, which has huge theoretical implications under the conditions of a transition economy. Finally, as previously mentioned, China is the world’s second-largest digital economic entity—as a representative of the developing world, China’s experience in implementing a digital transformation strategy can provide lessons for other countries, especially developing countries.

The rest of this paper is structured as follows: In Section 2, we review the current research status. The theoretical analysis and the development of the hypotheses are discussed in Section 3. Section 4 presents the data and methodology. Section 5 presents the empirical results. Section 6 investigates further studies. Section 7 consists of the discussion and conclusion.

2. Literature Review

2.1. Research on Digital Transformation

DT is characterized as an ongoing process of using new digital technologies, for instance, mobile technology, social media, cloud technology, artificial intelligence, blockchain, the Internet of Things, and big data analytics, to enable value co-creation, and to achieve business model innovation [7,8,9]. Taking Airbnb or Uber as examples, they are classic DT examples illustrating business model innovation [10]. Although some scholars speak of DT as comprising several stages, e.g., “three phases of DT are digitization, digitalization, and DT” [9], and some authors believe DT is a firm-specific and multi-dimensional concept, there is a consensus in most of the literature that DT is the ultimate stage and is the most prevalent [9,10]. In this article, our viewpoint is that DT is a firm-wide phenomenon that impacts the whole corporation and its business practices, resulting from the embracing of digital technologies [11,12].

A considerable amount of literature has highlighted that the use of digital technologies affects firms’ operation. Digital technologies encourage the creation of digital products and services, thus driving industry convergence and generating new industry sectors [13,14]. Moreover, DT enables enterprises to improve their operations and capabilities and increase the creation, delivery and capture of value [15], and enhances firms’ (innovation) performance [16,17]. Additionally, Rupeika-Apoga, Petrovska [18] investigated how DT positively affected SMEs’ revenues and business models. Bai, Quayson [19] argued that the leveraging of digital resources and capabilities by micro and small enterprises helped them survive during COVID-19 and to achieve sustainable development post-pandemic. Moreover, many published articles have reported that the use of digital technologies during COVID-19 increased organizational resilience and elasticity and provided important support for global economic recovery [20]. Furthermore, people have needed to change the way they work. Thanks to digital technologies, working remotely has become the norm [21,22]. Although DT has increasingly altered firms’ economic activities, the influence of DT on dynamic capital structure adjustment remains an underexplored topic.

2.2. Research on Dynamic Capital Structure Adjustment

“Capital structure is represented by long- and short-term debt, total debt, and liquidity” [23]. According to the Modigliani–Miller theorem [24], without market imperfections, firms’ capital structure should not influence corporate value. However, in reality, capital structure and corporate value have a clear correlation because of market frictions, such as bankruptcy cost, information asymmetry, agency cost, etc. This indicates that there is an optimal capital structure for maximizing firm value, or, in other words, a balance between equity and debt that reduces the cost of capital and increases the value of the firm. The value of the optimal capital structure is also highly controversial among different scholars. There are two main views: (1) An optimal capital structure is a range over which a firm allows its debt ratio to vary, rather than a fixed value [25]; (2) the optimal capital structure is a specific value [26]. This article adopts the view that the optimal capital structure refers to the point at which the agency costs of equity and debt financing are minimized. That is, a firm achieves an optimal capital structure when the reduction in the agency cost of equity due to debt financing is exactly equal to the agency cost of debt [26]. The optimal capital structure is the criterion by which enterprises aim to adjust their capital structure when it deviates from the target level [25]. The target level is the capital structure that is closer to the optimal capital structure. Harford, Klasa [27] provide evidence that enterprises with target capital structures try to not deviate from the optimal capital structure. Accelerating the adjustment speed to an optimal capital structure is the essence of dynamic capital structure adjustment. Therefore, helping firms find the optimal financing decision through dynamic capital structure adjustment is essential [28]. Following Modigliani and Mille [24], subsequent scholars developed a series of theories to provide rational explanations for the adjustments of the level of corporate debt, arguing the importance of optimal capital structure for firms. For example, trade-off theory [29], pecking order theory [30,31], market timing theory [32], agency theory [26], signaling theory [33], free cash flow theory, and so on [34]. A more obvious practical example in China was provided at the 11th meeting of the CPC Central Leading Group for Financial and Economic Affairs in 2015. President Xi first introduced the idea of “supply-side structural reform”, indicating that one of the key reform policies of “supply-side reform” is “deleverage”. “Deleverage” refers to reducing the leverage ratio to a threshold that is effective in preventing financial risk rather than reducing the debt ratio to zero. Specifically, on the one hand, those enterprises that are technologically backward, inefficient, lack marketability and operate at a long-term loss should not be maintained by increasing debt funding but should be deleveraged through bankruptcy or mergers and acquisitions. Moreover, efforts should be made to reduce or even remove the leverage of those enterprises that are environmentally unfriendly and have been ineffective for a long time to encourage the growth of a green economy. On the other hand, for those firms with advanced technology, better sales and higher profitability, not only should the supply of debt funding not be reduced, but their borrowing needs should also be met in time. The essence of these two aspects of deleveraging is dynamic capital structure adjustment. Hence, it is essential to understand the logic or determinants of why enterprises are required to dynamically adjust their capital structure.

The dynamic capital structure adjustment is influenced by various factors, internally (firm-specific variables) and externally (macroeconomic conditions). Externally, a number of scholars have empirically determined that macroeconomic conditions (e.g., environmental policy, market price, reforms in legislation, volatile seasonal demand, etc.) influence the dynamics of capital structure, and conclude that enterprises change their capital structure more quickly under favorable macroeconomic situations than in unfavorable ones [35,36,37,38]. Internally, corporate governance capacity affects the capital structure adjustment speed. Corporate governance capacity includes the education level and work experience of firms’ CEOs, information accessibility, foreign ownership, financial constraints [39], and so on. For firms with strong governance capacities, the adverse impact of economic uncertainty on the speed of capital structure adjustment is substantially smaller [40,41,42,43]. This study adds to the body of knowledge by demonstrating how enterprises’ DT influences capital structure adjustment.

It remains unclear how DT affects the dynamic capital structure adjustment. However, the adoption of digital technologies by firms may hasten the adjustment speed. Firstly, DT policies can entice enterprises to invest in R&D to pursue transformation of the corporate structure as DT has been empirically confirmed by Tian, Li [44] to have a positive impact on corporate risk-taking. The enhancement of digitalization could boost operating flexibility and financing availability. Second, DT may accelerate capital structure changes by improving stock liquidity or upgrading the bond market. The essence of this is that information asymmetry may be mitigated via DT [45,46].

On the other side, DT may decrease the adjustment speed toward the optimal capital structure. First, DT may not be suitable for all enterprises. Zhai, Yang [16] found that, for the early stages of a product (immature stage), investment in R&D for digital technologies may not be very effective as it will lead to waste of the cost of capital. In addition, as stated by Gebauer, Fleisch [47], DT may leave companies in a digital paradox trap. Companies may invest in digital transformation strategies, but seldom seem to achieve a corresponding boost in revenue, which will lead to a negative financial effect. Digital transformation strategy here refers to the new organizing logic in transforming a company-wide business model through digital technologies rather than just the adoption of digital technologies. Specifically, a digital transformation strategy means creating value (business model innovation) and the capability of adopting/exploring/exploiting new digital technologies. It is an integral part of the corporate strategy formula [48,49]. These uncertainties will lead to the failure of the DT of firms. What is more, the support of policies to promote DT can lead to many companies borrowing externally for short-term debt but for long-term usage, as they are unable to pay back in time. This is because the implementation of digital technology investments is a process that pays off in the long term [50].

3. Theoretical Analysis and Hypotheses Development

According to trade-off theory, there is an optimal capital structure that maximizes the firm value. However, due to factors such as market frictions, firms have adjustment costs. Therefore, firms will only adjust their capital structure upwards or downwards to achieve a target capital structure if the benefits of adjustment exceed the costs of adjustment. Adjustment costs and adjustment benefits affect the firm’s capital structure adjustment speed [51,52].

From one point of view, DT may accelerate the capital structure adjustment speed. Firstly, DT reduces the degree of information asymmetry within and outside the firm and optimizes the information interaction environment. According to asymmetry information theory [53], in market economic activities, there is an information difference known by the internal and external parties (firms and investors). From the internal side, DT enhances enterprises’ information accessibility, increases their access to market information, and the application of digital technology accelerates the enterprises’ ability to compile and process external information. Externally, as DT drives companies to improve their information disclosure mechanisms, the relative openness and transparency of corporate information increase investors’ understanding of corporate financial information, especially for financial institutions to review corporate credit and qualifications and consider whether to grant financing. Therefore, DT will reduce the information gap both internally and externally, thus reducing the financing cost of enterprises, which, in turn, reduces the cost of adjusting the capital structure and accelerates the adjustment speed. In addition, DT changes the business model and promotes the high-quality development of firms. DT has driven business model innovation, with a large number of business entities shifting their business to both online and offline. Pursuing the aim of a “Digital China”, many enterprises have received financial support for their transformation more easily than before, especially high-tech enterprises. For companies that have successfully achieved DT, DT reduces their adjustment costs and increases their adjustment benefits. Moreover, DT helps the management team to strengthen its risk assessment and crisis response capabilities, which, in turn, affects the adjustment speed. With the adoption of digital technologies and the rapid development speed of the digital economy, companies’ business strategies need to be adjusted in time. Enhanced access to information by the corporate management team, on the other hand, can mitigate agency problems, whilst also improving their decision-making capabilities [31,54].

On the other side, DT may decrease the adjustment speed of the capital structure. First, enterprises may fail in their process of DT. As stated in the report “DT Index Study of Chinese Enterprises” released by Accenture, only 16% of firms could obtain benefits from digitalization [55]. Achieving DT requires companies to invest a lot of money, including for the purchase of digital equipment (hardware and software), to cover management and learning costs (digital technology training), and so on. As DT is often a long-term input-output process, companies may apply for short-term debt but for long-term usage because the cost of investment does not pay off in a short time. In addition, the results of investments in DT may not be as high as expected as there is a risk of failure. A failed transformation can lead to companies not being able to repay their debt or even their going bankrupt, in other words, the adjustment cost could be high, which, in turn, can lead to decrease in adjustment speed. Moreover, financial institutions may be more inclined to invest in high-tech companies, while some traditional industries may not be able to borrow money to carry out DT. Furthermore, as DT is currently in full swing, policy support can lead to banks and other financing agencies easing the financing conditions for companies pursuing DT. Excessive debt lending by banks and other financial institutions will result in high leverage in society. Furthermore, the rapid implementation of DT increases management costs and learning costs, but, on the other hand, there may be a lack of digital leadership among managers and a lack of relevant digital experience among corporate employees, resulting in inability of the corporate management team and employees to keep up with the needs of the DT development. This can result in the benefits of digital benefits not exceeding the cost of the digital inputs. As a result, this leads to an increase in the adjustment costs, thus slowing down the adjustment speed [50,56].

The above theoretical analysis makes no claim to be exhaustive. However, it represents the most suitable hypotheses from our investigation. Therefore, based on the above analysis, the following competing hypotheses are proposed:

Ha. 

DT will accelerate the adjustment speed of capital structure.

Hb. 

DT will decrease the adjustment speed of capital structure.

4. Data and Methodology

4.1. Model

Referring to Byoun’s [57] study, we assess the effect of DT on the optimization of enterprise capital structure using the standard partial adjustment model.

$$Le{v}_{i,t}-Le{v}_{i,t-1}=\lambda \left(Le{v}_{i,t}^{*}-Le{v}_{i,t-1}\right)+{\eta }_t+{\mu }_i+{\nu }_{i,t}$$

(1)

where $i,t$ represent the enterprise and year, respectively, ${Lev}_{i,t}$ is the actual capital structure of the enterprise for the year, Lev*_i,t is the target capital structure of the enterprise for the year, ${\eta }_t$ is the time fixed effect, ${\mu }_i$ is the firm fixed effect, and ${\epsilon }_{i,t}$ is the random disturbance term. We record the difference in actual capital structure between the current year and the previous year as $\Delta Le{v}_{i,t}=Le{v}_{i,t}-Le{v}_{i,t-1}$ (hereafter referred to as capital structure adjustment), and the difference between the current year’s projected capital structure and the prior year’s actual capital structure as $De{v}_{i,t}=Le{v}_{i,t}^{*}-Le{v}_{i,t-1}$.

The coefficient we focus on is $\lambda$, which is the proportion of the enterprise capital structure’s actual adjustment to the desired adjustment in the current year, so it reflects the capital structure adjustment speed. If $\lambda$ is less than 0, it shows that the enterprise capital structure has been negatively adjusted, while if $\lambda$ is greater than 0, it indicates positive adjustment, and the greater the value, the faster the adjustment speed.

Since firms’ target capital structure cannot be directly observed, we refer to Byoun et al. [57] and adjust the firm target capital structure using the enterprise characteristic factors connected to the capital structure. The relevant regression Equation is as follows:

$${Lev}_{i,t}^{*}={\sum }_k{\beta }_k{{\rm X}}_{k,i,t-1}+{\eta }_t+{\mu }_i+{\nu }_{i,t}$$

(2)

where $X_{i,t}$ is the characteristic variable of the enterprise, according to the research of Huang and Gong [58]. We chose the following variables as characteristic variables: mortgage ability (FA), enterprise size (Size), profitability (EBIT), non-debt tax shield (DEP), growth opportunity (MB), and industry capital structure (Med_Lev). The characteristic variable construction method is described below.

Substituting Equation (2) into Equation (1), we obtain:

$${Lev}_{i,t}=(1-\lambda ){Lev}_{i,t-1}+{\sum }_k{\lambda \beta }_k{{\rm X}}_{k,i,t}+{\eta }_t+{\mu }_i+{\nu }_{i,t}$$

(3)

Equation (3) aims at obtaining the estimated coefficients $1-\hat{\lambda }$ and $\widehat{\lambda {\beta }_k}$. The target capital structure can be obtained by substituting the estimated coefficient into Equation (2). Finally, we put $Le{v}_{i,t}^{*}$ into Equation (1) and add the capital structure deviation and DT interaction term to investigate the impact of DT on the dynamic capital structure adjustment. The model is set as follows:

$$Le{v}_{i,t}=\left(1-\lambda \right)Le{v}_{i,t-1}+{\displaystyle {\sum }_k\lambda {\beta }_k{X}_{k,i,t}}+{\eta }_t+{\mu }_i+{\nu }_{i,t}$$

(4)

We employ a dual fixed effect model (FE) to regress in order to exclude the impact of individual characteristics and temporal factors on the dynamic capital structure adjustment. The coefficient we focus on in this model is ${\lambda }_1$. If ${\lambda }_1$ is greater than 0, it means that DT can accelerate the adjustment speed. If ${\lambda }_1$ is less than 0, it means that DT will reduce the adjustment speed. If ${\lambda }_1$ is equal to 0 or not significant, it signifies that the adjustment speed is not clearly impacted by DT. In addition, group tests were also conducted on enterprises with different adjustment types. We documented the companies whose goal capital structure was larger than the actual capital structure for the prior year ($De{v}_{i,t}>0$) as a downward capital structure deviation, and the enterprises whose target capital structures were lower than their actual capital structures from the prior year ($De{v}_{i,t}<0$) as an upward capital structure deviation.

4.2. Data and Variables

We used the annual data of A-share non-financial listed enterprises in China from 2011 to 2021 as samples; ST and ST* enterprises were excluded. Finally, 29,292 observed values of 3855 enterprises were retained. To avoid the impact of outliers on the regression findings, we conducted a 1% tail reduction on the data. The data sources for this article were from the Wind database, the CSMAR database, and the annual reports of publicly traded corporations.

Definition and specific description of the main variables were displayed in Table 1. The main variables’ construction methods were as follows:

• Enterprise capital structure. The ratio of total liabilities to total assets was employed to measure corporate capital structure (Lev).
• Digital transformation. A proxy of DT using text analysis was built by crawling DT-related keywords in annual reports. Step one: according to the research of Zhao et al. [59], we selected 99 DT-related keywords in four dimensions (the specific keywords are supplied in Appendix A). Step two: we utilized Python to search for DT-related keywords in the annual reports and counted the occurrence frequency of each keyword. Step three: we summed the occurrence frequency of each keyword and used logarithm processing to obtain a measurement index (DT) of enterprise DT. We also used the entropy method to determine the weight of the keywords, and then synthesized an enterprise DT index (DT_Score) to test robustness. In addition, in order to avoid subjective keyword selection, we also crawled enterprise DT data according to the keywords used by Wu et al. [6] so as to synthesize an enterprise DT index (DT2) to test robustness.
• Characteristic variables. According to Huang and Gong [58], the following variables were selected as characteristic variables: mortgage ability (FA), measured by the ratio of fixed assets to total assets of an enterprise; enterprise size (Size), measured by the logarithmic form of the total enterprise assets; profitability (EBIT), measured by the ratio of EBIT to total assets; non-debt tax shield (DEP), measured by the ratio of depreciation of fixed assets to total assets; growth opportunity (MB), measured by the ratio of the sum of the stock market value and the value liabilities to the total assets; and industry capital structure (Med_Lev), measured by the median capital structure of the industry.

Table 1. Definition and description of variables.

Variable Name	Variable Symbol	Variable Specification
Capital structure	Lev	total liabilities/total assets of the enterprise
Target capital structure	Lev*	refer to Equation (2) for calculation
Capital structure adjustment	ΔLev	the difference in actual capital structure between the current year and the previous year
Capital structure deviation	Dev	the difference between the target capital structure of the current year and the actual capital structure of the previous year
DT	DT	logarithmic form of the sum of occurrence frequency of DT-related keywords in the annual report
Mortgage ability	FA	fixed assets/total assets
Enterprise size	Size	logarithmic form of total enterprise assets
Profitability	EBIT	EBIT/total assets
Non-debt tax shield	DEP	EBIT/total assets
Growth opportunity	MB	(stockmarket value + liability value)/total assets
Industry capital structure	Med_Lev	the median capital structure of the industry

Table 1. Definition and description of variables.

Variable Name	Variable Symbol	Variable Specification
Capital structure	Lev	total liabilities/total assets of the enterprise
Target capital structure	Lev*	refer to Equation (2) for calculation
Capital structure adjustment	ΔLev	the difference in actual capital structure between the current year and the previous year
Capital structure deviation	Dev	the difference between the target capital structure of the current year and the actual capital structure of the previous year
DT	DT	logarithmic form of the sum of occurrence frequency of DT-related keywords in the annual report
Mortgage ability	FA	fixed assets/total assets
Enterprise size	Size	logarithmic form of total enterprise assets
Profitability	EBIT	EBIT/total assets
Non-debt tax shield	DEP	EBIT/total assets
Growth opportunity	MB	(stockmarket value + liability value)/total assets
Industry capital structure	Med_Lev	the median capital structure of the industry

4.3. Descriptive Statistics

Table 2. Descriptive statistics of capital structure.

	Mean	Max	Min	Sd	Average
Panel A: full sample
Lev*i,t	0.4306	0.7783	0.0954	0.0895	0.4196
ΔLevi,t	−0.0078	0.8557	−0.8743	0.0979	−0.0008
Devi,t	0.0100	0.5518	−0.8028	0.1762	0.0193
Panel B: downward deviation
Lev*i,t	0.4095	0.7624	0.1663	0.0834	0.3979
ΔLevi,t	−0.0074	0.8557	−0.8743	0.0995	−0.0015
Devi,t	0.1415	0.5518	0.0001	0.0927	0.1311
Panel C: upward deviation
Lev*i,t	0.4555	0.7783	0.0954	0.0900	0.4481
ΔLevi,t	−0.0085	0.5402	−0.8743	0.0942	0.0000
Devi,t	−0.1446	−0.0000	−0.8028	0.1151	−0.1195

reports descriptive statistics for the enterprise capital structure, among which Panel A reports descriptive statistics of the full sample, Panel B reports descriptive statistics of capital structure downward deviation enterprises, and Panel C reports descriptive statistics of capital structure upward deviation enterprises. It can be seen that, in the full sample, the maximum value, minimum value and standard deviation of Lev* were 0.7783, 0.0954 and 0.0895. The maximum value, minimum value, and standard deviation of ΔLev were 0.8557, −0.8743 and 0.0979. The maximum value, minimum value, and standard deviation of Dev were 0.5518, −0.8028 and 0.1762. These results indicate that there were great differences in capital structure characteristics among the samples. By comparing Panel B and Panel C, it can be seen that the mean values of Lev*, ΔLev and Dev in Panel B were 0.4095, −0.0074 and 0.1415, respectively. Their absolute values were all smaller than those in Panel B, indicating that there was a considerable difference between capital structure downward deviation enterprises and capital structure upward deviation enterprises. Therefore, it is necessary to perform an analysis based on the different types of capital structure adjustment.

Table 3. Descriptive statistics of other major variables.

	Mean	Max	Min	Sd	Average
DTi,t	2.8848	5.8201	0.0000	1.2522	2.8332
FAi,t−1	0.2108	0.6918	0.0024	0.1594	0.1805
Sizei,t−1	21.4992	25.9501	18.1392	1.5422	21.3958
EBITi,t−1	0.0712	0.3253	−0.1516	0.0760	0.0598
DEPi,t−1	0.0202	0.0697	0.0005	0.0148	0.0169
MBi,t−1	2.6723	12.6003	0.8462	2.0564	2.0006
Med_Levi,t−1	0.4211	0.6940	0.3032	0.0792	0.4154

reports descriptive statistics for the other variables. The maximum value of DT was 5.8201, the minimum value was 0, the average value was 2.8848, and the standard deviation was 1.2522, indicating that there were significant disparities in the sample firms’ levels of DT. The statistical results for the other major variables were in line with expectations.

5. Empirical Results

5.1. Baseline Results

Column (1) in

Table 4. Baseline results.

	(1)	(2)	(3)
	ΔLevi,t	ΔLevi,t	ΔLevi,t
	Full Sample	Downward Deviation	Upward Deviation
Devi,t	0.2738 ***	0.2572 ***	0.4030 ***
	(17.8869)	(10.9819)	(13.4740)
DTi,t × Devi,t	0.0407 ***	0.0317 ***	0.0466 ***
	(8.6816)	(4.7405)	(5.2913)
Cons	0.0095 ***	−0.0379 ***	0.0913 ***
	(4.5919)	(−10.9930)	(21.1875)
Year_effect	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes
R2	0.1950	0.0901	0.1828
N	29,292	15,839	13,453

Note: *** p < 0.01.

illustrates the baseline regression results. The coefficient of Dev_i,t was 0.2738, which was significant at the significance level of 1%, indicating that the capital structure of the enterprises was improving. The coefficient of DTi,t × Dev_i,t was 0.0407, which was significant at the significance level of 1%, indicating that DT improved the adjustment speed. Columns (2) and (3) show the sub-sample regression results of enterprises with downward deviation of capital structure and enterprises with upward deviation of capital structure, respectively. The coefficients of Dev_i,t were 0.2572 and 0.4030, respectively, and the coefficients of DT_i,t × Dev_i,t were 0.0317 and 0.0466, respectively, which were all significant at the significance level of 1%. The regression results show that both companies with upward deviations of capital structure and those with downward deviations were in a state of positive capital structure adjustment situation, which means DT improved the adjustment speed for both types. However, by comparing the two sets of regression coefficients, we found that firms’ adjustment speed was faster with an upward deviation of capital structure, which may be related to China’s “deleveraging” policy in recent years. At the same time, DT had a stronger influence on the adjustment speed for firms with an upward deviation.

5.2. Robustness Checks

To enhance the robustness of the empirical findings, the following robustness tests were applied:

Firstly, we replaced the measurement method of DT. We used the entropy weight method to first synthesize the enterprise DT index. In column (1) to column (3) of

Table 5. Replace the measurement method of DT.

	(1)	(2)	(3)	(4)	(5)	(6)
	Full Sample	Downward Deviation	Upward Deviation	Full Sample	Downward Deviation	Upward Deviation
	ΔLevi,t	ΔLevi,t	ΔLevi,t	ΔLevi,t	ΔLevi,t	ΔLevi,t
Devi,t	0.3593 ***	0.3334 ***	0.4943 ***	0.3347 ***	0.3173 ***	0.4630 ***
	(41.4609)	(24.5134)	(24.0108)	(33.3521)	(21.1376)	(20.7604)
DT_Scorei,t × Devi,t	14.0125 ***	9.2092 ***	16.8870 ***
	(6.2052)	(2.8931)	(3.2976)
DT2i,t × Devi,t				0.0371 ***	0.0225 ***	0.0483 ***
				(8.3396)	(4.1528)	(5.2997)
Cons	0.0104 ***	−0.0394 ***	0.0919 ***	0.0098 ***	−0.0381 ***	0.0884 ***
	(5.0438)	(−11.5614)	(21.4901)	(4.7692)	(−11.0244)	(20.5049)
Year_effect	Yes	Yes	Yes	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.1915	0.0887	0.1792	0.1944	0.0892	0.1832
N	29,292	15,839	13,453	29,292	15,839	13,453

Note: *** p < 0.01.

, the coefficients of DT_Score_i,t × Dev_i,t were all greater than 0 at the significance level of 1%. Then, we constructed an enterprise DT index according to the keywords used in the study of Wu et al. [6]. In columns (4) to (6), the coefficients of DT2_i,t × Dev_i,t were all greater than 0 at the significance level of 1%, which is consistent with the conclusion above.

Second, we replaced the regression sample. In Panel A of

Table 6. Replace the regression sample.

Panel A: Manufacturing Sample
	(1)	(2)	(3)
	Full Sample	Downward Deviation	Upward Deviation
	ΔLevi,t	ΔLevi,t	ΔLevi,t
Devi,t	0.2674 ***	0.2427 ***	0.4341 ***
	(15.5471)	(8.9817)	(10.7244)
DTi,t × Devi,t	0.0446 ***	0.0335 ***	0.0401 ***
	(7.6100)	(4.2746)	(3.0073)
Cons	0.0052 **	−0.0337 ***	0.0918 ***
	(2.0494)	(−8.2196)	(15.6424)
Year_effect	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes
R2	0.1944	0.0891	0.1815
N	18,594	10,952	7642
Panel B: After 2013 (Year of Digital Finance)
	(1)	(2)	(3)
	Full Sample	Downward Deviation	Upward Deviation
	ΔLevi,t	ΔLevi,t	ΔLevi,t
Devi,t	0.3208 ***	0.3071 ***	0.4551 ***
	(17.9701)	(11.2304)	(12.8350)
DTi,t × Devi,t	0.0431 ***	0.0339 ***	0.0459 ***
	(8.2892)	(4.3400)	(4.5886)
Cons	0.0062 ***	−0.0447 ***	0.0954 ***
	(3.6592)	(−13.9767)	(22.1330)
Year_effect	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes
R2	0.2226	0.1105	0.1956
N	25,490	13,869	11,621
Panel C: Before Supply-Side Reform was Proposed
	(1)	(2)	(3)
	Full Sample	Downward Deviation	Upward Deviation
	ΔLevi,t	ΔLevi,t	ΔLevi,t
Devi,t	0.4540 ***	0.4349 ***	0.5431 ***
	(14.8689)	(11.6933)	(11.1620)
DTi,t × Devi,t	0.0566 ***	0.0347 ***	0.0648 ***
	(6.3744)	(3.0184)	(4.6434)
Cons	0.0035 *	−0.0730 ***	0.1145 ***
	(1.7896)	(−14.8558)	(17.3087)
Year_effect	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes
R2	0.2563	0.1594	0.2410
N	10,439	5457	4982

Note: * p < 0.1, ** p < 0.05, *** p < 0.01.

, the sample of manufacturing firms for regression is represented. The coefficients of DT_i,t × Dev_i,t were all greater than 0 at the significance level of 1%. Moreover, after 2013, which is seen as the first year of digital finance in China, the DT of enterprises became more professional and directional, so we used the samples after 2013 for regression analysis. The coefficients of DT_i,t × Dev_i,t in Panel B of

Table 7. 2SLS test.

	(1)	(2)	(3)	(4)	(5)	(6)
	Full Sample		Downward Deviation		Upward Deviation
	DTi,t × Devi,t	ΔLevi,t	DTi,t × Devi,t	ΔLevi,t	DTi,t × Devi,t	ΔLevi,t
Devi,t	1.4262 ***	0.1352 ***	1.8746 ***	0.1785 ***	1.1159 ***	0.2123 ***
	(31.5092)	(5.4629)	(29.3587)	(3.4667)	(12.4454)	(3.4817)
Industry_DTi,t × Devi,t	0.6015 ***		0.5042 ***		0.6592 ***
	(32.9243)		(18.5555)		(14.8757)
DTi,t × Devi,t		0.0953 ***		0.0594 ***		0.1272 ***
		(10.5189)		(3.4129)		(5.2772)
Year_effect	Yes	Yes	Yes	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes	Yes	Yes	Yes
Wald F	7666.699		1419.109		1670.431
KP LM	1645.590		561.965		305.022
R2	0.7694	0.1842	0.7218	0.0880	0.6384	0.1652
N	29,289	28,866	15,836	15,265	13,453	13,088

Note: *** p < 0.01.

were all greater than 0 at the significance level of 1%. Finally, we eliminated the samples after “supply-side reform” was proposed (2015) to exclude the impact of policy intervention. In Panel C of Table 7, the coefficients of DT_i,t × Dev_i,t were all greater than 0 at the significance level of 1%, which is consistent with the conclusion above.

Thirdly, for two-stage regression, we utilized the mean DT_i,t of other enterprises in the same industry as an instrumental variable. The coefficients of Industry_DT_i,t in columns (1), (3) and (5) of Table 7 were significant at the 1% significance level, indicating that instrumental variables had better explanatory force on explanatory variables. In column (2), column (4) and column (6), the coefficients of DT_i,t × Dev_i,t were all greater than 0 at the significance level of 1%, which is consistent with the above conclusion.

6. Further Study

In the above, we have demonstrated that DT speeds capital structure adjustment. To systematically reveal DT’s influence on dynamic capital structure adjustment, we carried out the following extended studies: First, we considered whether DT speeds capital structure adjustment by influencing debt financing or equity financing. Secondly, we further examined the transmission channels through which DT affects capital structure adjustment. Thirdly, we considered the heterogeneous characteristics of DT affecting the capital structure adjustment in three dimensions: ownership type, asset size and credit cycle. Finally, we analyzed the moderating roles of reduced financing constraints and growing economic policy uncertainty on the relationship between DT and capital structure adjustment speed.

6.1. Analysis of Adjustment Method

For enterprises with downward deviation of capital structure, it can be adjusted upward by increasing the debt scale and reducing the equity scale. For enterprises with upward deviation of capital structure, it can be adjusted downward by reducing the debt scale and increasing the equity scale. We constructed the regression model using Equation (5) to test the method by which DT affects the adjustment speed.

$$D_{i,t}={\vartheta }_0+{\vartheta }_{1}D{T}_{i,t}+{\displaystyle {\sum }_k{\vartheta }_k{X}_{k,i,t}}+{\eta }_t+{\mu }_i+{\nu }_{i,t}$$

(5)

In Equation (5), $D_{i,t}$ is a dummy variable representing the adjustment method of capital structure. It includes the following: whether the enterprise increases the debt scale (Idebt_i,t)—if the ratio of newly borrowed cash of the enterprise to the total assets exceeds 5%, Idebt_i,t is denoted as 1, otherwise it is denoted as 0; whether the enterprise reduces the scale of equity (Dequity_i,t)—if the ratio of cash paid by means of dividend distribution, profit or interest repayment to the total assets exceeds 5%, Dequity_i,t is denoted as 1, otherwise it is denoted as 0; whether the enterprise reduces the debt scale (Ddebt_i,t)—if the ratio of cash paid by the enterprise to repay debt exceeds 5% of the total assets, Ddebt_i,t is denoted as 1, otherwise it is denoted as 0; whether the enterprise has increased the scale of equity (Iequity_i,t)—if the ratio of cash received by the enterprise’s equity investment to the total assets exceeds 5%, Iequity_i,t is denoted as 1, otherwise it is denoted as 0. Since the explained variables are dummy variables, we used the Logit model for regression.

The coefficients of DT_{i, t} shown in columns (1) and (4) of

Table 8. Adjustment method test results.

	(1)	(2)	(3)	(4)
	Downward Deviation		Upward Deviation
	Idebti,t	Dequityi,t	Ddebti,t	Iequityi,t
DTi,t	0.0183 **	−0.0022	0.0072	0.0372 ***
	(2.2851)	(−0.4283)	(1.6192)	(3.4006)
FAi,t−1	0.2229 ***	−0.1092 *	0.1699 ***	0.2438 *
	(2.7523)	(−1.9477)	(3.0905)	(1.9469)
Sizei,t−1	0.0508 ***	−0.0605 ***	0.0246 **	−0.1313 ***
	(3.2240)	(−5.7616)	(2.4939)	(−7.9252)
EBITi,t−1	−0.2111 *	1.3368 ***	−0.0090	0.3325 *
	(−1.6792)	(13.7771)	(−0.1179)	(1.8932)
DEPi,t−1	1.6074 *	0.7882	0.3011	2.8313 **
	(1.6990)	(1.2951)	(0.6337)	(2.2637)
MBi,t−1	0.0037	0.0057 **	−0.0027	0.0270 ***
	(0.8947)	(2.3328)	(−0.7272)	(3.9437)
Med_Levi,t−1	−0.3809	0.0221	−0.2057	0.3515
	(−1.0798)	(0.0896)	(−1.2317)	(1.0429)
Cons	−0.3343	1.3194 ***	0.4105 *	2.7074 ***
	(−0.8858)	(5.4229)	(1.7485)	(6.9117)
Year_effect	Yes	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes	Yes
R2	0.0096	0.0492	0.0089	0.0758
N	11,914	15,519	12,802	7460

Note: * p < 0.1, ** p < 0.05, *** p < 0.01.

were 0.0183 and 0.0372, respectively, and were significant at the 5% and 1% significance levels, respectively. The coefficients of DT_{i, t} shown in column (2) and column (3) were −0.0022 and 0.0072, but they were not statistically significant. The above regression results show that DT mainly improved firms’ adjustment speed with downward deviation of capital structure by increasing the scale of debt financing; and improved firms’ adjustment speed with upward deviation of capital structure by increasing the scale of equity financing.

6.2. Intermediary Channel Test

We constructed the regression model represented by Equations (6)–(8) to test the intermediary channel through which DT affects the speed of capital structure adjustment.

$$\Delta Le{v}_{i,t}=\left({\lambda }_0+{\lambda }_{1}D{T}_{i,t}\right)\times De{v}_{i,t}+{\eta }_t+{\mu }_i+{\nu }_{i,t}$$

(6)

$$F{F}_{i,t}={\delta }_0+{\delta }_{1}D{T}_{i,t}+{\displaystyle {\sum }_k{\delta }_k{X}_{k,i,t-1}}+{\eta }_t+{\mu }_i+{\nu }_{i,t}$$

(7)

$$\Delta Le{v}_{i,t}=\left({\gamma }_0+{\gamma }_{1}D{T}_{i,t}+{\gamma }_{2}F{F}_{i,t}\right)\times De{v}_{i,t}+{\eta }_t+{\mu }_i+{\nu }_{i,t}$$

(8)

Equation (6) is the basic regression test, which has been verified above. Equation (7) is the regression test of DT on financial flexibility (FF). According to Zeng [60], an enterprise’s financial flexibility is equal to its cash flexibility plus its debt flexibility. The cash flexibility is represented by the enterprise cash ratio after deducting the industrial average, and the debt flexibility is represented by the industrial average asset–liability ratio minus the enterprise asset–liability ratio. Equation (8) is the regression adding intermediary variables. The coefficients we focus on in the above model are ${\lambda }_1$, ${\delta }_1$, ${\gamma }_2$.

The coefficients of DT_i,t shown in column (1) and (3) of

Table 9. Intermediary channel test.

	(1)	(2)	(3)	(4)
	Downward Deviation		Upward Deviation
	FFi,t	ΔLevi,t	FFi,t	ΔLevi,t
DTi,t	−0.1945 ***		0.0204 *
	(−5.2155)		(1.9539)
Devi,t		0.3888 ***		0.5737 ***
		(16.6655)		(14.0652)
DTi,t × Devi,t		0.0170 ***		0.0469 ***
		(2.6887)		(5.8360)
FFi,t × Devi,t		−0.0458 ***		0.2258 ***
		(−19.6800)		(6.1318)
FAi,t−1	−2.8421 ***		−0.2505
	(−8.0477)		(−1.5219)
Sizei,t−1	−0.5523 ***		−0.0028
	(−8.3796)		(−0.1343)
EBITi,t−1	1.9844 ***		0.1250
	(3.9368)		(0.5694)
DEPi,t−1	−12.9018 ***		0.6233
	(−3.7047)		(0.3892)
MBi,t−1	−0.0061		0.0777 ***
	(−0.3818)		(7.3293)
Med_Levi,t−1	2.6323 **		0.2470
	(2.4383)		(0.8307)
Cons	12.8702 ***	−0.0363 ***	−0.9056 *	0.0753 ***
	(7.7138)	(−10.8076)	(−1.7278)	(16.2368)
Year_Effect	No	Yes	No	Yes
Firm_Effect	Yes	Yes	Yes	Yes
R2	0.0931	0.1456	0.0360	0.2905
N	15,859	15,859	13,482	13,482

Note: * p < 0.1, ** p < 0.05, *** p < 0.01.

were −0.1954 and 0.0204, respectively, which were significant at the significance level of 1% and 10%, respectively, indicating that the DT reduced the financial flexibility of enterprises with downward deviation of capital structure, but improved the financial flexibility of enterprises with upward deviation of capital structure. In column (2) and (4), the coefficients of FF_i,t × Dev_i,t were −0.0458 and 0.2258, which were significant at the 1% significance level, indicating that, for enterprises with downward deviation of capital structure, increasing financial flexibility decreased the adjustment speed, but, for enterprises with upward deviation of capital structure, increasing financial flexibility increased the adjustment speed. This is because for enterprises with downward deviation of capital structure, an optimal capital structure means increasing the size of debt, which is contrary to the aim of high financial flexibility. However, for enterprises with upward deviation of capital structure, capital structure optimization means reducing debt size, which is consistent with the aim of high financial flexibility, and, therefore, companies have lower costs and more incentives to adjust capital structure. In summary, a transmission channel of “DT–financial flexibility–dynamic recapitalization” exists.

6.3. Heterogeneity Test

Since state-owned enterprises (hereafter referred to as SOEs) and large-scale firms are the main beneficiaries of credit rationing, we tested the heterogeneity at the level of ownership type and asset size. The test results are shown in Panel A of

Table 10. Heterogeneity test results.

Panel A
	(1)	(2)	(3)	(4)
	Downward Deviation		Upward Deviation
	SOEs	non-SOEs	SOEs	non-SOEs
	ΔLevi,t	ΔLevi,t	ΔLevi,t	ΔLevi,t
Devi,t	0.2344 ***	0.2830 ***	0.4218 ***	0.3911 ***
	(6.7791)	(9.1492)	(9.8957)	(9.5080)
DTi,t × Devi,t	0.0301 ***	0.0297 ***	0.0261 **	0.0598 ***
	(2.8485)	(3.5049)	(2.2391)	(4.7331)
Cons	−0.0257 ***	−0.0499 ***	0.0841 ***	0.0992 ***
	(−5.0233)	(−11.0991)	(15.1153)	(14.8133)
Year_effect	Yes	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes	Yes
R2	0.0855	0.0972	0.1714	0.1952
N	5052	10,787	6095	7358
Panel B
	(1)	(2)	(3)	(4)
	Downward Deviation		Upward Deviation
	Large-Scale Firms	Small-Scale Firms	Large-Scale Firms	Small-Scale Firms
	ΔLevi,t	ΔLevi,t	ΔLevi,t	ΔLevi,t
Devi,t	0.3069 ***	0.4000 ***	0.4755 ***	0.4008 ***
	(10.0609)	(10.0090)	(13.0754)	(7.7925)
DTi,t × Devi,t	0.0278 ***	0.0158	0.0308 ***	0.0624 ***
	(3.1888)	(1.4177)	(3.0454)	(3.7637)
Cons	−0.0334 ***	−0.0735 ***	0.0861 ***	0.1156 ***
	(−7.6633)	(−10.7970)	(19.4768)	(9.1868)
Year_effect	Yes	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes	Yes
R2	0.1201	0.1194	0.2046	0.1964
N	9876	5963	10,291	3162
Panel C
	(1)	(2)	(3)	(4)
	Downward Deviation		Upward Deviation
	Credit Expansion Periods	Credit Contraction Periods	Credit Expansion Periods	Credit Contraction Periods
	ΔLevi,t	ΔLevi,t	ΔLevi,t	ΔLevi,t
Devi,t	0.3358 ***	0.3643 ***	0.5428 ***	0.4741 ***
	(7.3353)	(11.7880)	(10.2469)	(11.0115)
DTi,t × Devi,t	0.0418 ***	0.0097	0.0402 ***	0.0479 ***
	(3.3725)	(1.1282)	(2.7971)	(3.8141)
Cons	−0.0461 ***	−0.0493 ***	0.0914 ***	0.0974 ***
	(−11.7943)	(−11.4836)	(16.5245)	(17.4002)
Year_effect	Yes	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes	Yes
R2	0.1336	0.1141	0.2200	0.2215
N	8753	7086	7176	6277

Note: ** p < 0.05, *** p < 0.01.

in groups based on the type of firm ownership. The regression findings for the group of enterprises with a downward capital structure deviation are displayed in columns (1) and (2). The coefficients of DT_i,t × Dev_i,t were 0.0301 and 0.0297, respectively, and both were significant at the significance level of 1%. The coefficients of the two groups did not differ significantly from one another. Columns (3) and (4) show the regression results for the group of enterprises with upward deviation of capital structure, in which the coefficients of DT_i,t × Dev_i,t were 0.0261 and 0.0598, which were significant at the 5% and 1% significance levels, respectively. The influence of the DT on the dynamic adjustment of non-SOEs’ capital structures was stronger when the capital structure deviated upward, as shown by a comparison of the coefficients of the two groups.

The regression findings are displayed in Panel B of Table 10 based on the asset size grouping, where firms with assets larger than the median are classified as large-scale enterprises, while those with assets smaller than the median are classified as small-scale firms. The coefficient of DT_i,t × Dev_i,t in column (1) was 0.0278, which was significant at the significance level of 1%, while the coefficient of DT_i,t × Dev_i,t in column (2) was not statistically significant, which indicates that, for enterprises with downward deviation of capital structure, the impact of DT on the adjustment speed only exists in large-scale firms. Columns (3) and (4) show the regression results for the group of enterprises with upward deviation of capital structure, where the coefficients of DT_i,t × Dev_i,t were 0.0308 and 0.0624, respectively, and both were significant at the 1% significance level. Comparing the coefficients of the two groups demonstrated that small-scale businesses were significantly impacted by the DT.

The credit cycle may have an impact on business debt financing, which, therefore, affects the corporate capital structure, so we conducted a test of heterogeneity according to the credit cycle. We applied Hodrick Prescott Filter to the growth rate of credit balances of financial institutions and recorded periods with a cycle fluctuation term greater than 0 as credit expansion periods, and periods with a cycle fluctuation term less than 0 as credit contraction periods. Columns (1) and (2) in Table 10 show the regression results for the group of enterprises with downward deviation of capital structure, where the coefficient of DT_i,t × Dev_i,t in column (1) was 0.0418 and significant at the significance level of 1%. The coefficient of DT_i,t × Dev_i,t in column (2) was not statistically significant, indicating that the speed of enterprise capital structure adjustment was affected by DT only during the credit expansion period. Column (3) and column (4) show the regression results for the group of enterprises with upward deviation of capital structure, where the coefficients of DT_i,t × Dev_i,t were 0.0402 and 0.0479, respectively, and both were significant at the 1% significance level. The differences between the coefficients of the two groups were small.

6.4. Moderating Role Test

To further test the moderating role of financing constraints and economic policy uncertainty, we added the interaction term between the moderating variables and DT to the regression model shown in Equation (9).

$$\Delta Le{v}_{i,t}=\left({\theta }_0+{\theta }_{1}D{T}_{i,t}+{\theta }_{2}M+{\theta }_{3}M\times D{T}_{i,t}\right)\times De{v}_{i,t}+{\eta }_t+{\mu }_i+{\nu }_{i,t}$$

(9)

$M$ represents financing constraints and economic policy uncertainty. The financing constraint (SA) is calculated as SA = −0.737 × Size + 0.043 × Size2 − 0.04 × Age, where Size represents the total assets and Age is the age of the enterprise, and a larger SA index represents a smaller financing constraint for the enterprise. As a stand-in for economic policy uncertainty, we utilized Baker et al.’s (2016) measurement of the Chinese economic policy uncertainty index (news index), grounded in South China Morning Post news keywords (EPU).

The coefficient of SA_i,t × DT_i,t × Dev_i,t in column (1) of

Table 11. Moderating role test results.

	(1)	(2)	(3)	(4)
	Downward Deviation	Upward Deviation	Downward Deviation	Upward Deviation
	Devi,t	Devi,t	Devi,t	Devi,t
Devi,t	−0.9169 ***	−0.2480	0.2386 ***	0.3122 ***
	(−2.9403)	(−0.6916)	(8.4592)	(7.6992)
DTi,t × Devi,t	0.2096 **	0.1354	0.0273 ***	0.0668 ***
	(2.5584)	(1.4404)	(3.2408)	(5.3733)
SAi,t × Devi,t	−0.3107 ***	−0.1751 *
	(−3.7915)	(−1.8460)
SAi,t × DTi,t × Devi,t	0.0475 **	0.0248
	(2.2253)	(1.0151)
EPUt × Devi,t			0.0065	0.0307 ***
			(1.3227)	(3.1887)
EPUt × DTi,t × Devi,t			0.0005	−0.0070 **
			(0.3552)	(−2.5707)
Cons	−0.0299 ***	0.0883 ***	−0.0351 ***	0.0869 ***
	(−7.1953)	(17.9114)	(−9.4799)	(19.0806)
Year_effect	Yes	Yes	Yes	Yes
Firm_effect	Yes	Yes	Yes	Yes
R2	0.0927	0.1852	0.0905	0.1853
N	15,839	13,453	15,839	13,453

Note: * p < 0.1, ** p < 0.05, *** p < 0.01.

was 0.0475, which was significant at the significance level of 5%, but the coefficient in column (2) was not statistically significant, indicating that, for enterprises with downward deviation of capital structure, reduction in financing constraints can enhance the effect of DT on the adjustment speed. The coefficient of EPU_t × DT_i,t × Dev_i,t in column (3) was not statistically significant, while the coefficient in column (4) was −0.0070 and was significant at the 5% level of significance, indicating that, for enterprises with upward deviation of capital structure, growing economic policy uncertainty reduced the strength of DT on the adjustment speed.

7. Discussion and Conclusions

Although current research on DT has examined its influence on the whole industry/organization, study of the impact of DT has focused more on firm performance [16,61] and new business models [8]. The DT impact on corporate finance has been written about less often. Some scholars have investigated the effect of DT on stock price crash risk [62] or corporate risk-taking [44], whereas the specific impact on dynamic capital structure adjustment has been less examined. Therefore, compared with previous studies, this paper sought to innovatively investigate the relationship between DT and the dynamic capital structure. Compared with [6], this paper introduces the effects of heterogeneity of ownership type, asset size, and credit cycle in relation to the impact of digital transformation on dynamic capital structure adjustment. The study addressed whether and how DT influences firm’s dynamic capital structure adjustment. Another original feature of the study relates to consideration of the factors that affect the adjustment speed/direction of a firm’s capital structure.

This study examined how dynamic capital structure adjustment has been impacted by DT in China. We discovered that firm DT may lessen information asymmetry and increase a firm’s transparency for the management team and financial investors, and may improve financial flexibility, thereby accelerating the capital structure adjustment speed. Although companies with upward deviations of capital structure and those with downward deviations are currently in a state of positive capital structure adjustment under the influence of DT, firms’ adjustment speed with upward deviation of capital structure is faster. Moreover, DT has a stronger impact on the firms’ capital structure adjustment speed with upward deviation. Regarding the adjustment method, we discovered that DT mainly improved the adjustment speed with downward deviation of capital structure by increasing the scale of debt financing; and improved the adjustment speed of capital structure with upward deviation of capital structure by increasing the scale of equity financing. Concerning the mediation analysis, we note that DT will reduce the financial flexibility of firms with downward deviation of capital structure but improve the financial flexibility of firms with upward deviation of capital structure. For enterprises with downward deviation of capital structure, increasing financial flexibility will decrease the adjustment speed, but for enterprises with upward deviation of capital structure, increasing financial flexibility can increase capital structure adjustment speed. Furthermore, heterogeneity tests have shown that the impact of DT on the dynamic adjustment of capital structure of firms with upward deviation of capital structure is stronger for non-state-owned enterprises compared to state-owned enterprises and is strong for small-scale firms compared to large-scale firms; the impact of DT on the dynamic adjustment of capital structure of enterprises with downward deviation of capital structure only exists in large-scale firms and is only effective in the period of credit expansion. With respect to moderating effects, our research shows that, for enterprises with downward deviation of capital structure, reduced financing constraints can enhance the strength of DT on the adjustment speed; while for enterprises with upward deviation of capital structure, growing economic policy uncertainty will reduce the strength of DT on the adjustment speed.

These results provide the management team and financial investors with fresh perspectives on how DT affects dynamic capital structure adjustment. Company managers should actively respond to the rapid development of digital technology, fully grasp the opportunities in the era of DT, make financial and manpower investments in the DT of enterprises, encourage the deep integration of digital technologies in each channel of enterprise production, operation and customer service, and facilitate the high-quality development of enterprises. At the same time, managers should not rush into DT, but should take into consideration the situation and carry out DT in accordance with their own development needs. In addition, from the perspective of digital financial risks, enterprise managers should take reasonable risk management actions, improve the efficiency of information transmission in the capital market, and avoid the negative impact brought about by DT. Through multi-management initiatives, the adjustment of the corporate capital structure to a target capital structure can be steadily promoted. Moreover, the findings of this paper also offer some illuminating insights for emerging market economies regarding how to optimize capital structure by leveraging DT. Our results also offer some policy implications. Obviously, it is helpful to continue promoting a DT policy, such as ensuring considerable investment in R&D for DT. However, wise investments are necessary. Regarding “supply-side reform”, the government should allocate funds wisely. For those enterprises with good development prospects, such as high-tech enterprises, new energy enterprises, sustainable enterprises, etc., the financing conditions can be relaxed; for those traditional enterprises that have had no income for a long time, they should be appropriately discarded and no longer given substantial financial support.

This paper has some limitations. First, we drew our conclusions from only a single country. It would be worthwhile to perform further research in other developed and developing nations to verify the results. It would be particularly interesting to see if there exist differences in outcomes that may result from the different levels of digitization in developed and developing countries. Second, as listed companies are regulated equity financed, their capital structure does not reflect the full range of domestic companies and a study on this topic using such companies as a sample may be subject to sample bias.