Bank–Firm Common Ownership and Corporate Innovation Diffusion: Evidence from Risk-Buffering and Information-Risk Channels

Abstract

Against the backdrop of China’s innovation-driven development strategy, innovation diffusion is a key stage through which firm-level innovation outcomes generate broader economic value. However, this process is often constrained by financing pressure, information asymmetry, and uncertainty in external evaluation. This study examines whether and how bank–firm common ownership, as an ownership-based financial linkage between banks and firms, affects corporate innovation diffusion. Using data on Chinese A-share non-financial listed companies from 2010 to 2023, this paper finds that bank–firm common ownership significantly promotes corporate innovation diffusion. The results remain robust after alternative variable measurements, a higher identification threshold for bank–firm common ownership, lagged explanatory variables, instrumental-variable estimation and propensity score matching. Further mechanism tests show that bank–firm common ownership promotes innovation diffusion mainly through two risk-related channels: liquidity-risk buffering and information-risk reduction. First, it improves firms’ access to commercial credit financing, thereby strengthening their liquidity-risk buffering capacity and helping them withstand financing pressure during the innovation diffusion process. Second, it improves firms’ information disclosure, thereby reducing information asymmetry and external evaluation uncertainty surrounding innovation activities. Further analysis shows that the positive effect of bank–firm common ownership on innovation diffusion is more pronounced among state-owned enterprises and firms with stronger market positions. This study enriches the literature on financial linkages and corporate innovation diffusion, and provides evidence on how bank–firm ownership ties can support innovation diffusion through liquidity-risk buffering and information-risk reduction.

1. Introduction

Innovation diffusion is a crucial channel through which innovation generates economic and social value. Relative to innovation input and innovation output, innovation diffusion places greater emphasis on the process by which new knowledge, technologies, and solutions are disseminated, absorbed, and reused across a broader range of actors (Rogers 2003). Only when innovative outcomes transcend the boundaries of a single firm and are recognized, imitated, adapted, and further developed by other entities can innovation truly translate into industrial upgrading, productivity growth, and improvements in social welfare. In particular, as China accelerates the construction of a science and technology powerhouse and advances high-level technological self-reliance and self-strengthening, it is undoubtedly important whether individual firms innovate; equally important, however, is whether the resulting innovations can be effectively diffused and whether their externalities can be continuously amplified. This bears directly on the overall efficiency of the innovation system and the realization of the innovation-driven development strategy. Innovation diffusion is therefore not a peripheral stage of the innovation process, but a key determinant of whether innovative outputs can be effectively converted into economic and social value.

Yet innovation diffusion is not simply a natural consequence of innovation output. Much of the knowledge embedded in innovation is highly tacit and cannot be fully codified or transmitted at low cost through standardized texts (Agrawal and Goldfarb 2008). Its diffusion thus depends heavily on sustained information exchange, organizational interaction, and collaborative learning among actors. In practice, however, communication costs arising from geographic distance (Cai et al. 2022), transaction barriers created by institutional differences (Donnelly et al. 2024), and pervasive information frictions (Cullen et al. 2025) all undermine the completeness and timeliness of information transmission, thereby increasing the cost of innovation diffusion and reducing its efficiency. This is especially true for high-risk, long-horizon innovation activities, where the stability of external financial support and the smoothness of inter-firm connection mechanisms substantially shape both the quality and extent of diffusion.

In recent years, a growing body of research has begun to examine how external financial conditions affect innovation diffusion from the perspectives of financial development, financial networks, and credit supply. This literature has focused primarily on dimensions such as credit scale and the geographic expansion of bank branches (Comin and Nanda 2019). Although some studies have shown that financial networks can promote regional innovation diffusion through investment channels (Wang and Zhang 2025), the literature as a whole has paid more attention to regional financial linkages and credit relations, while giving insufficient attention to equity linkages as a potentially more deeply embedded micro-level mechanism with governance implications. In particular, under bank–firm common ownership, common shareholders form an important bridge between banks and firms and may strengthen information exchange and resource linkages between the two. Against this backdrop, whether bank–firm common ownership promotes the flow of knowledge and technology across firms, thereby affecting innovation diffusion, constitutes the central research question of this paper.

Using a sample of non-financial A-share listed firms over the period 2010–2023, this paper finds that bank–firm common ownership exerts a significant effect on innovation diffusion. The results remain robust after alternative explanatory and dependent variables, a higher identification threshold for bank–firm common ownership, lagged explanatory variables, instrumental-variable estimation, and propensity score matching. Further analysis shows that commercial credit financing and information disclosure constitute two important mechanisms through which bank–firm common ownership promotes innovation diffusion. Specifically, commercial credit financing enhances firms’ financial buffering capacity and their ability to withstand financing pressure during the diffusion process, while information disclosure reduces information asymmetry and external evaluation uncertainty surrounding innovation activities. Heterogeneity analysis further indicates that the effect is more pronounced among state-owned enterprises and firms with stronger market positions.

The marginal contributions of this paper are threefold. First, it extends the literature on the determinants of innovation diffusion. Existing studies have mainly examined the conditions under which innovation diffusion arises from the perspectives of geographic proximity, institutional environment, financial development, and social trust, while paying relatively little attention to micro-level ownership structures, especially bank–firm common ownership. By incorporating bank–firm common ownership into the analytical framework of innovation diffusion, this paper reveals that common shareholders, as an important bridge linking banks and firms, affect the inter-firm flow of innovative factors by strengthening information connections and resource linkages. In doing so, it provides new empirical evidence on the micro-foundations of innovation diffusion.

Second, this paper deepens our understanding of the economic consequences of bank–firm common ownership. Prior research on bank–firm common ownership has focused largely on internal firm-level effects, such as improvements in firms’ own innovation performance, while paying insufficient attention to whether it generates cross-firm and cross-entity spillovers. This paper shows that bank–firm common ownership affects not only firms’ internal innovation activities but also has a significant innovation-diffusion effect. This shift from firms’ own innovation output to the external diffusion of innovation outcomes helps broaden the analytical scope of bank–firm common ownership and highlights its potential role in cross-firm knowledge spillovers.

Third, this paper contributes to the literature on financial linkages and innovation under uncertainty by highlighting the role of risk-related conditions in the diffusion stage of innovation. Existing studies have mainly examined how financial linkages affect innovation through credit supply, financing costs, and capital allocation, while paying less attention to whether such linkages help firms maintain financial stability and reduce information frictions after innovation outcomes are generated. This paper shows that bank–firm common ownership facilitates innovation diffusion through two channels: commercial credit financing, which strengthens firms’ capacity to withstand financing pressure, and information disclosure, which reduces information asymmetry and external evaluation uncertainty. These findings provide new evidence on how ownership-based financial linkages support the diffusion of innovation outcomes under uncertainty.

2. Literature Review and Hypothesis Development

Innovation diffusion concerns whether the innovative outputs of one firm can further shape the innovation behavior of other firms. The effective realization of innovation diffusion depends not only on the technological capability of the focal innovating firm itself, but also on whether innovation-related information can be observed in a timely manner by other firms, whether relevant experience can be effectively transmitted (Polidoro and Jacobs 2024), and whether interfirm linkages exist that facilitate the dissemination of innovative outcomes (You et al. 2024; Celik et al. 2022). In the absence of stable connections among firms, even when some firms have already generated high-quality innovations, the impact of such innovations may remain confined within firm boundaries and be difficult to convert into broader diffusion.

Bank–firm common ownership may provide favorable conditions for innovation diffusion by allowing common shareholders to serve as bridges between banks, firms, and other market participants. Through these ownership-based linkages, common shareholders can help reduce informational segmentation among different entities and strengthen external attention to the focal firm’s technological orientation, innovation quality, and market prospects. As a result, the focal firm’s innovation outcomes become more visible and credible to peer firms, suppliers, customers, investors, and other potential adopters.

In addition, the relational network created by bank–firm common ownership may improve the external connectivity of the focal firm. Stronger connectivity can make the focal firm’s technological orientation, patent outputs, and innovation-related information more visible to external firms and other market participants. When external actors can more readily identify and evaluate these innovation outcomes, the focal firm’s patented technologies are more likely to attract attention and be cited in subsequent technological search and development. Therefore, bank–firm common ownership may not only affect the focal firm’s own innovation activities, but also improve the external visibility and citation-based diffusion of its innovation outcomes. Based on the above analysis, this paper proposes the following hypothesis:

Hypothesis 1.

Bank–firm common ownership promotes corporate innovation diffusion.

Commercial credit is not only an important source of short-term financing for firms, but also a manifestation of transactional relationships and mutual trust among firms. Compared with bank credit, commercial credit relies more heavily on information obtained from actual transactions, expectations of contract performance, and sustained cooperative relationships. Accordingly, improvements in a firm’s access to commercial credit depend not only on its financing needs but also on whether trading partners are willing to provide financing arrangements such as accounts payable extensions based on their assessment of the firm’s operating conditions, debt-servicing capacity, and prospects for continued cooperation. In the context of innovation diffusion, commercial credit matters not only because it alleviates firms’ liquidity pressure (Fabbri and Menichini 2010) but also because it sustains transactional relationships and thereby tightens linkages between upstream and downstream firms. Greater stability in trading relationships strengthens interfirm information exchange, technological interaction, and market feedback, thereby improving the conditions under which innovation outcomes can spread along the supply chain.

Bank–firm common ownership may first help mitigate information asymmetry among firms. A major constraint on the provision of commercial credit is that trading partners often find it difficult to assess firm risk accurately and in a timely manner (Ersahin et al. 2024). The capital linkages created by bank–firm common ownership may increase the observability of firms’ operating information and financial conditions, reduce counterparty uncertainty, and thereby strengthen the willingness of trading partners to extend commercial credit (Hu et al. 2026). Second, bank–firm common ownership may enhance transactional trust. Commercial credit is essentially an intertemporal transfer of credit based on future payment commitments, and its formation and maintenance depend critically on confidence in a firm’s repayment capacity and ability to sustain operations (Luo et al. 2023). When common shareholders simultaneously hold equity in both banks and firms, they may transmit a credible endorsement signal to external markets, thereby increasing suppliers’ willingness to maintain long-term credit arrangements. Finally, bank–firm common ownership may stabilize firms’ financing expectations. When external financing is unstable, trading partners often tighten credit supply out of concern over future repayment risk. By improving firms’ access to bank loans, bank–firm common ownership strengthens firms’ ability to sustain normal trading relationships (Hu et al. 2026), which in turn translates into more favorable commercial credit terms. Once commercial credit conditions improve, firms can obtain more flexible financing support from suppliers and customers, which helps them maintain transaction continuity and withstand liquidity pressure during the innovation diffusion process. In this sense, commercial credit financing is not only a source of external funds but also a risk-buffering mechanism that allows firms to better cope with financing pressure, payment delays, and uncertainty in continued innovation-related investment. More stable commercial credit relationships also strengthen information exchange, technological interaction, and expectations of collaboration among upstream and downstream firms, thereby providing a more favorable foundation for the transmission of knowledge, technology, and innovation experience across firms. Based on the above analysis, this paper proposes the following hypothesis:

Hypothesis 2.

Bank–firm common ownership promotes innovation diffusion by improving firms’ access to commercial credit.

Innovation information disclosure alters whether external actors can see, understand, and are willing to adopt information, thereby influencing both the speed and the scope of innovation diffusion. Compared with ordinary information disclosure, innovation information disclosure typically concerns key information related to a firm’s technological innovation, such as R&D activities, technological trajectories, and application scenarios; in essence, it transforms private information that is otherwise difficult to interpret into public information. Such disclosure lowers the information search costs faced by industry competitors and supply chain partners, enabling external firms to identify more accurately the potential value and applicability of a technology. Chang et al. (2024) argue that the magnitude of technology spillovers depends on firms’ distance and visibility in the technology space. High-quality information disclosure narrows cognitive distance, making firms’ innovation outcomes more likely to attract potential collaborators, stimulate industry-wide knowledge exchange and technological diffusion, and be searched, benchmarked, and utilized in subsequent R&D, thereby expanding the scope of innovation diffusion.

Under bank–firm common ownership, firms may have stronger incentives to improve the disclosure of innovation information. From the perspective of common shareholders, simultaneous equity holdings in both banks and firms shift attention away from the returns of any single firm toward the value maximization of the overall portfolio. Greater innovation information disclosure helps improve the external information environment of the firm and enhances capital market recognition of the firm’s growth prospects and innovative capabilities, thereby increasing stock liquidity and reducing financing costs, which is of positive significance for common shareholders seeking to maximize portfolio returns (Park et al. 2019). For banks, greater disclosure of innovation information enables a more accurate assessment of firms’ innovation capability, growth potential, and risk profile (Saidi and Žaldokas 2021), thereby mitigating information asymmetry and strengthening banks’ willingness to provide credit support. Moreover, because common shareholders simultaneously hold equity stakes in both banks and firms, firms’ disclosure of innovation information not only helps outside investors recognize their development prospects but also conveys signals about innovation capability and growth potential to banks, thus reinforcing banks’ incentives to provide lending support. When bank–firm common ownership promotes firms’ innovation information disclosure, relevant information—such as technological directions, product positioning, R&D activities, and patenting outcomes—can become more visible and interpretable to external firms. Improved disclosure reduces the information costs faced by external actors when assessing the value and applicability of the focal firm’s innovation outcomes. As a result, these innovation outcomes may be more likely to enter other firms’ technological search processes and be cited in subsequent patenting activities. Based on the above analysis, this paper proposes the following hypothesis:

Hypothesis 3.

Bank–firm common ownership promotes innovation diffusion by improving the level of firms’ information disclosure.

3. Research Design

3.1. Sample Selection and Data Sources

This paper uses Chinese A-share listed firms on the Shanghai and Shenzhen stock exchanges over the period 2010–2023 as the research sample, and applies the following screening procedures: (1) financial firms are excluded; (2) observations with missing data are removed; and (3) firms designated as ST, *ST, or PT are excluded. To mitigate the influence of extreme values, all continuous variables are winsorized at the 1% level. The final sample comprises 28,176 firm-year observations. The detailed sample selection procedure is reported in Appendix A,

Table A1. Sample selection procedure.

Sample Selection Procedure	Observations
Initial A-share firm-year observations, 2010–2023	42,978
Less: financial firms	1106
Less: ST, *ST, and PT firms	2487
Less: observations with missing key variables	11,209
Final sample	28,176

. Data on bank–firm common ownership, financial indicators, innovation-related measures, and firms’ bank-loan details are obtained from the CSMAR, RESSET, and CNRDS databases.

3.2. Variable Definitions and Model Specification

3.2.1. Variable Definitions

• Dependent variable

The dependent variable is patent citations. This paper measures corporate innovation diffusion using patent citations. It should be noted that, because patent applications are subject to a relatively long review cycle, innovation diffusion may in practice already occur during the application period. Accordingly, this paper uses the patent application year as the year in which innovation diffusion occurs. In addition, patents can generally be classified into three categories: invention patents, utility model patents, and design patents. In the literature, invention patents are commonly regarded as the patent type with the highest technological content and the greatest degree of innovativeness. Although China’s Patent Examination Guidelines impose background-technology disclosure requirements for both invention and utility model patent applications, the substantive examination applied to utility model patents is substantially less rigorous than that for invention patents (Dai et al. 2024).

Taken together, and following the approach of Howell et al. (2025), this paper focuses on the total number of citations received by firms’ invention patent applications. Specifically, the number of citations is increased by one and then transformed using the natural logarithm to construct the measure of patent citations (LnCit). As a robustness check, this paper also uses the natural logarithm of one plus the average number of citations received by firms’ invention patent applications.

The core explanatory variable is bank–firm common ownership. Following Ojeda (2018) and Liu et al. (2024), this paper constructs an indicator of bank–firm common ownership. Using quarterly-to-annual shareholding data from CSMAR on the shareholding ratios of the top ten tradable shareholders of non-financial listed firms and banks, and combining these data with bank-loan relationship data from RESSET, this paper identifies the set of lending banks associated with each firm in year t. In constructing the measure, the paper first identifies, at the “firm i–bank j–year t” level, the set of common shareholders between the firm and the bank, namely, the investors that simultaneously appear among the top ten tradable shareholders of firm i and bank j. To measure bank–firm common ownership more accurately, nominal shareholder accounts are traced upward during the data-identification stage. Because a single institution often establishes multiple fund products or asset-management plans with similar functions, and these products may be held under the names of different custodian banks while in substance being controlled by the same investment decision-maker, failing to account for such fragmentation may bias the identification of ownership linkages downward. To address this issue, the paper penetrates and cleans securities investment funds, insurance portfolios, trust accounts, and similar accounts, and consolidates their holdings to the parent-company level. This treatment helps mitigate the underestimation of bank–firm common ownership caused by excessively fragmented statistical reporting.

Based on the identified bank–firm common shareholders, this paper calculates the shareholding ratios of common shareholders in both the firm and the bank, and further derives a measure of the intensity of bank–firm common ownership. Specifically, let $Investor{s}_{it}$ denote the set of the top ten tradable shareholders of firm i in year t, and $Investor{s}_{jt}$ denote the set of the top ten tradable shareholders of bank j in year t. Then the set of bank–firm common shareholders is given by $Investor{s}_{it}\cap Investor{s}_{jt}$. Let $Share{s}_{mit}$ denote the number of shares of firm i held by common shareholder m in year t, and $TotalShare{s}_{it}$ denote the total shares outstanding of firm i in year t. Similarly, let $Share{s}_{mjt}$ denote the number of shares of bank j held by common shareholder m in year t, and $TotalShare{s}_{jt}$ denote the total shares outstanding of bank j in year t. Accordingly, the shareholding ratio of common shareholders in the firm is defined as PERCENT. A larger value of PERCENT indicates a higher degree of bank–firm common ownership for firm i in year t.

$${\mathit{PERSHARE}}_{i,j,t}={\displaystyle \frac{{\sum }_{m\in \left({\mathit{Investors}}_{\mathit{it}}\cap {\mathit{Investors}}_{\mathit{jt}}\right)}{\mathit{Shares}}_{\mathit{mit}}}{{\mathit{TotalShares}}_{\mathit{it}}}}$$

(1)

$$\mathit{BANK}\text{\_}{\mathit{PERSHARE}}_{i,j,t}={\displaystyle \frac{{\sum }_{m\in \left({\mathit{Investors}}_{\mathit{it}}\cap {\mathit{Investors}}_{\mathit{jt}}\right)}{\mathit{Shares}}_{\mathit{mjt}}}{{\mathit{TotalShares}}_{\mathit{jt}}}}$$

(2)

$${\mathit{PERCENT}}_{i,j,t}={\mathit{PERSHARE}}_{i,j,t}\times \mathit{BANK}\text{\_}{\mathit{PERSHARE}}_{i,j,t}$$

(3)

$${\mathit{PERCENT}}_{i,t}=\sum _{j\in \mathit{Banks}\left(i,t\right)}{\mathit{PERCENT}}_{i,j,t}$$

(4)

2.

Control variables

Following the existing literature, this paper controls for the following variables: return on assets (ROA), cash flow ratio (Cashflow), revenue growth (Growth), the shareholding ratio of the top three shareholders (Top3), large-shareholder capital occupation (Occupy), net profit margin (NetProfit), CEO duality (Dual), and whether the firm directly holds bank equity (F_B). Including whether the firm directly holds bank equity as a control variable is important because, as a bank shareholder, a firm may obtain additional informational advantages, relational linkages, and financing convenience through direct equity holdings, which may independently affect its access to bank credit and cost of financing. Without controlling for this factor, the effect of bank–firm common ownership may be confounded with the effect of firms’ direct equity stakes in banks, thereby biasing the regression results. Detailed variable definitions are reported in

Table 1. Variable Definitions and Descriptions.

Type	Variable Name	Variable Symbol	Variable Definition
Dependent Variable	Number of Citations to Invention Patents	LnCit	Ln(1 + the number of citations received by invention patents applied for)
Independent Variable	Bank–firm common ownership	PERCENT	see Equation (4) for the detailed calculation method.
Control Variable	Return on Assets	ROA	Net profit/average total assets balance
	Cash Flow Ratio	Cashflow	Net cash flow generated from operating activities/total assets
	Revenue Growth Rate	Growth	Current-year operating revenue/previous-year operating revenue − 1
	Shareholding Ratio of the Top Three Shareholders	Top3	Number of shares held by the top three shareholders/total number of shares
	Capital Occupation by Large Shareholders	Occupy	Net other receivables/total assets
	Net Profit Margin	NetProfit	Net profit/operating revenue
	CEO Duality	Dual	1 if the chairman of the board and the general manager are the same person; otherwise 0
	Direct Bank Shareholding by the Firm	F_B	1 if the firm directly holds shares in a bank in the current year; otherwise 0

.

3.2.2. Model Specification

This paper specifies the following baseline model:

$$LnCi{t}_{i,t}={\beta }_0+{\beta }_1{PERCENT}_{i,t}+{\gamma Controls}_{i,t}+{\mu }_i+{\lambda }_t+{\epsilon }_{i,t}$$

(5)

where $LnCit$ denotes corporate innovation diffusion; a larger value indicates a higher degree of innovation diffusion. $PERCENT$ represents bank–firm common ownership, measured as the product of the shareholding ratio that common shareholders hold in the listed firm and the shareholding ratio they hold in the corresponding bank. $Controls$ denotes the set of control variables. ${\mu }_i$ and ${\lambda }_t$ represent industry fixed effects and year fixed effects, respectively, and ${\epsilon }_{i,t}$ is the random disturbance term. A significantly positive ${\beta }_1$ indicates that bank–firm common ownership has a significantly positive effect on corporate innovation diffusion; a significantly negative ${\beta }_1$ would indicate the opposite.

3.3. Regression Results and Analysis

3.3.1. Descriptive Statistics

Table 2. Descriptive statistical results of the main variables.

Variables	N	Mean	Max	Std	Min
LnCit	28,176	2.678	10.56	1.719	0
PERCENT	28,176	0.0027	0.732	0.0226	0
Growth	28,176	0.351	17.11	1.043	−0.928
Cashflow	28,176	0.0476	0.267	0.0686	−0.222
ROA	28,176	0.0385	0.222	0.0626	−0.556
NetProfit	28,176	0.0711	0.555	0.184	−2.110
Dual	28,176	0.303	1	0.460	0
Top3	28,176	0.493	0.869	0.153	0.150
Occupy	28,176	0.0133	0.212	0.0216	0.000100
F_B	28,176	0.404	1	0.491	0

presents the descriptive statistics for the main variables. For the dependent variable, innovation diffusion ($\mathrm{L}\mathrm{n}\mathrm{C}\mathrm{i}\mathrm{t}$) has a mean of 2.678. Its standard deviation is 1.719, with a minimum of 0 and a maximum of 10.56. This indicates substantial variation across sample firms in the citation performance of their innovations: some firms’ innovation outcomes exert relatively strong industry influence, whereas others display much weaker diffusion effects, implying that the variable has good discriminating power.

The descriptive statistics show that PERCENT has a mean of 0.0027 and a standard deviation of 0.0226. The minimum is 0, and the maximum is 0.732. This distribution suggests that bank–firm common ownership is relatively rare overall, although a small number of firms exhibit comparatively high levels of such ownership. The standard deviation exceeds the mean, indicating substantial heterogeneity in bank–firm common ownership across firms. It should also be noted that the relatively low mean of PERCENT is mainly attributable to the way the variable is constructed. Specifically, it is calculated as the product of the shareholding ratio of common shareholders in the firm and the shareholding ratio they hold in the bank; because the product of two ratios is typically small, the sample mean is correspondingly low.

For the control variables, the descriptive statistics show that sample firms exhibit substantial heterogeneity in profitability, growth, ownership concentration, governance structure, cash flow, and direct bank shareholding. These variations provide a suitable basis for examining the relationship between bank–firm common ownership and corporate innovation diffusion in the subsequent regression analysis.

3.3.2. Baseline Regression Analysis

Table 3. Baseline regression results.

	(1)	(2)	(3)
	LnCit	LnCit	LnCit
PERCENT	3.6344 ***	3.4479 ***	4.0585 ***
	(0.4022)	(0.3989)	(0.5528)
Growth		0.0098	−0.0043
		(0.0111)	(0.0193)
Cashflow		0.6999 ***	1.0248 ***
		(0.1684)	(0.1919)
ROA		1.9008 ***	1.6694 ***
		(0.2932)	(0.4886)
NetProfit		−0.5899 ***	−0.5388 **
		(0.0997)	(0.2042)
Dual		−0.2148 ***	−0.2494 ***
		(0.0223)	(0.0605)
Top3		−0.1667 **	−0.1881
		(0.0681)	(0.3347)
Occupy		1.3479 ***	0.5802
		(0.4865)	(0.9036)
F_B		0.0010	−0.0287
		(0.0207)	(0.0764)
Industry	No	No	Yes
Year	No	No	Yes
Observations	28,176	28,176	28,176
Adj. R-sq.	0.0026	0.0090	0.2839

Note: Cluster-robust standard errors are shown in parentheses **, and *** denote statistical significance at the 5%, and 1% levels, respectively.

reports the baseline regression results on the effect of bank–firm common ownership on innovation diffusion. Specifically, column (1) presents the univariate estimation results; column (2) reports the estimates after adding control variables but without fixed effects; and column (3) reports the results after further controlling for both industry fixed effects and year fixed effects on top of the control variables. The results show that the regression coefficients on the core explanatory variable, PERCENT, are 3.6344, 3.4479, and 4.0585, respectively, all of which are positive and statistically significant at the 1% level. These findings indicate that bank–firm common ownership has a significant positive effect on innovation diffusion. In terms of economic magnitude, the standard deviation of PERCENT is 0.0226. Based on the coefficient of 4.0585 in column (3), a one-standard-deviation increase in PERCENT is associated with an increase of 0.0917 in LnCit. Since LnCit is defined as the natural logarithm of one plus invention-patent citations, this corresponds to an approximately 9.6% increase in innovation diffusion. Therefore, although the mean value of PERCENT is small, its estimated effect is economically meaningful.

With respect to the control variables, Cashflow and ROA are positively associated with innovation diffusion, suggesting that firms with stronger liquidity and profitability are better positioned to support technology dissemination and external collaboration. By contrast, NetProfit and Dual are negatively related to innovation diffusion, indicating that higher profit orientation or more concentrated managerial power may be associated with weaker external diffusion of innovation outcomes. Overall, after controlling for industry and year fixed effects, the baseline result remains robust.

3.3.3. Robustness Tests

1.

Replacing the dependent variable

To enhance the credibility of the findings and rule out potential interference from measurement error and subjective choices in the construction of the dependent variable, this paper follows Howell et al. (2025) and replaces the baseline dependent variable with an alternative measure of innovation diffusion, namely the natural logarithm of one plus the average number of citations received by a firm’s invention patent applications (LnCit2). Column (1) of

Table 4. Robustness analysis.

	(1)	(2)	(3)	(4)
	LnCit2	LnCit	LnCit	LnCit
PERCENT	0.0898 **
	(0.0428)
BFCO		0.6535 ***
		(0.1056)
PERCENT 3			0.8319 ***
			(0.1295)
L.PERCENT				3.6268 ***
				(0.6443)
Controls	Yes	Yes	Controls	Yes
Industry	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
Observations	28,176	28,176	28,176	21,325
Adj. R-sq.	0.5313	0.2937	0.2870	0.2939

Note: Cluster-robust standard errors are shown in parentheses. ** and *** denote statistical significance at the 5% and 1% levels, respectively.

reports the results using this alternative dependent variable. The coefficient on LnCit2 is 0.0898 and remains significantly positive at the 5% level. This suggests that the main conclusions of this paper continue to hold under an alternative measurement of the dependent variable.

To address possible measurement error in the core explanatory variable, this paper follows Liu et al. (2024) and replaces PERCENT with a dummy variable indicating the existence of bank–firm common ownership (BFCO). This variable equals 1 if bank–firm common ownership exists and 0 otherwise. Compared with the continuous shareholding-ratio measure, this dummy variable can reduce potential bias arising from extreme values or from variation in the magnitude of shareholding ratios. Column (2) of Table 4 reports the results using this alternative explanatory variable. The coefficient on BFCO is 0.6535 and is significantly positive at the 1% level. This indicates that the main conclusions of this paper remain robust under an alternative measurement of the core explanatory variable.

To further verify the robustness of the baseline regression results, this paper follows Freeman (2025) and resets the identification threshold for bank–firm common ownership by redefining the core explanatory variable as bank–firm common ownership with a shareholding ratio of at least 3% on both sides (PERCENT3). That is, a substantive bank–firm common ownership relationship is identified only when the common shareholder’s ownership stake in both the bank and the firm is no less than 3%. This stricter criterion effectively excludes spurious correlations driven by passive allocation or very small shareholdings, and instead focuses on common shareholders with greater potential influence and information-sharing capacity. Column (3) of Table 4 shows that, when the sample is restricted to cases in which the bank–firm common ownership ratio exceeds 3%, the coefficient on PERCENT3 is 0.8319 and remains significantly positive at the 1% level. This indicates that the findings are not driven by a small number of extremely low-shareholding common ownership links, thereby further strengthening the reliability of the results.

The baseline regression results indicate a significant positive association between bank–firm common ownership and corporate innovation diffusion. However, this finding may be affected by reverse causality: rather than bank–firm common ownership promoting innovation diffusion, firms with stronger innovation potential may be more likely to attract investors who establish common ownership positions. To alleviate this concern, this paper lags the core explanatory variable PERCENT by one period (L.PERCENT) and re-estimates the model. Column (4) of Table 4 reports the results using the lagged explanatory variable. The coefficient on L.PERCENT is 3.6268 and remains significantly positive at the 1% level. This suggests that the main conclusions continue to hold after accounting for the possibility of reverse causality.

3.3.4. Endogeneity Analysis

1.

Instrumental-variable estimation

Although the baseline regression controls for firm characteristics, governance variables, and industry and year fixed effects, the relationship between bank–firm common ownership and corporate innovation diffusion may still suffer from endogeneity. On the one hand, unobservable firm characteristics may affect both the formation of bank–firm common ownership and innovation diffusion. On the other hand, firms with stronger innovation-diffusion capacity may be more likely to attract institutional investors that also hold bank equity, leading to potential reverse causality. To further alleviate these concerns, this paper employs an instrumental-variable approach. Specifically, we use the leave-one-out industry-year average of bank–firm common ownership as the instrument for firm-level bank–firm common ownership. This instrument is relevant because firms in the same industry and year are exposed to similar institutional-investor preferences and bank–firm ownership-network conditions. At the same time, by excluding the focal firm from the calculation, the instrument is less likely to be directly affected by the focal firm’s own innovation diffusion.

Table 5. Endogeneity Tests.

	(1) First Stage: PERCENT	(2) Second Stage: LnCit	(3) PSM—OLS
PERCENT		4.316 ***	0.602 ***
		(1.010)	(0.113)
IV	0.434 ***
	(0.073)
ATT			0.076 ***
Controls	Yes	Yes	Yes
Industry	Yes	Yes	Yes
Year	Yes	Yes	Yes
Observations	28,171	28,171	2719
First-stage F-statistic	35.71
Adj. R-sq.	—	—	0.0181

Note: Cluster-robust standard errors are shown in parentheses. *** denotes statistical significance at the 1% level.

reports the endogeneity test results. Columns (1) and (2) present the instrumental-variable estimation. The first-stage result shows that the instrumental variable is significantly positively associated with firm-level bank–firm common ownership, and the first-stage F-statistic is 35.71, suggesting that the weak-instrument concern is unlikely to be severe. The second-stage result shows that the coefficient on PERCENT remains positive and statistically significant.

2.

Propensity score matching (PSM-OLS)

To further mitigate selection bias arising from observable firm characteristics, this paper employs propensity score matching. Specifically, the existence of bank–firm common ownership (BFCO) is used as the treatment variable, and a Logit model is estimated to obtain firms’ propensity scores based on Growth, Cashflow, ROA, NetProfit, Dual, Top3, Occupy, and F_B. A 1:1 nearest-neighbor matching procedure without replacement is then implemented to construct a matched sample. Based on the matched sample, this paper further performs OLS regression by using LnCit as the dependent variable and PERCENT as the core explanatory variable, while controlling for the same covariates as well as industry and year fixed effects.

Column (3) of Table 5 reports the results. The ATT is 0.076 and is significantly positive at the 1% level, indicating that firms with bank–firm common ownership exhibit higher innovation diffusion than comparable firms without such ownership. The coefficient on PERCENT is 0.602 and remains significant at the 1% level in the matched-sample regression. These results suggest that the baseline conclusion remains robust after controlling for observable differences between treatment and control firms.

4. Mechanism Tests

The preceding results indicate that bank–firm common ownership is positively associated with corporate innovation diffusion. This section examines whether this relationship operates through two risk-related channels. The first is a liquidity-risk buffering channel, captured by commercial credit financing. Innovation diffusion requires continuous transactions and external financing support; therefore, firms with stronger commercial credit financing are better able to withstand short-term financing pressure during the diffusion process. The second is an information-risk reduction channel, captured by information disclosure. By reducing information asymmetry and external evaluation uncertainty, information disclosure improves the visibility and credibility of firms’ innovation outcomes to external stakeholders. The empirical model is specified as follows:

$$M_{it}={\beta }_0+{\beta }_{1}PERCEN{T}_{it}+{\beta }_{2}Control{s}_{it}+FE+{\epsilon }_{it}$$

(6)

where $M_{it}$ denotes the mechanism variable for the firm $i$ in year $t$, and the definitions of the other variables are consistent with those used in the baseline regressions.

4.1. Commercial Credit Financing as a Liquidity-Risk Buffering Channel

Commercial credit financing reflects firms’ ability to obtain financing support from trading partners. In the innovation diffusion process, such financing support can help firms maintain transaction continuity and reduce the disruption caused by short-term financing pressure. Therefore, this paper uses commercial credit financing to capture the liquidity-risk buffering channel through which bank–firm common ownership may affect innovation diffusion.

Following Liu and Wang (2023), this paper measures firms’ commercial credit financing (NetCredit) as the difference between commercial credit obtained and commercial credit extended, scaled by total assets at the end of the period. Specifically, commercial credit obtained is measured by the sum of accounts payable, notes payable, and advances from customers, divided by total assets at the end of the period. Commercial credit supplied is measured by the sum of accounts receivable, notes receivable, and prepayments, divided by total assets at the end of the period. Net trade credit financing (NetCredit), measured as (accounts payable + notes payable + advances from customers − accounts receivable − notes receivable − prepayments) divided by ending total assets.

The results in column (1) of

Table 6. Results of the Mechanism Tests.

	(1)	(2)	(3)
	NetCredit	FERROR	Keywords
PERCENT	0.0007 ***	−0.0102 ***	0.0045 **
	(0.0002)	(0.0031)	(0.0016)
Controls	Yes	Yes	Yes
Industry	Yes	Yes	Yes
Year	Yes	Yes	Yes
Observations	28,176	28,176	28,176
Adj. R-sq.	0.1812	0.0231	0.3819

Note: Cluster-robust standard errors are shown in parentheses. ** and *** denote statistical significance at the 5% and 1% levels, respectively.

show that the coefficient on bank–firm common ownership is 0.0007 and is statistically significant at the 1% level, indicating that bank–firm common ownership strengthens firms’ liquidity-risk buffering capacity and helps alleviate financing pressure during the innovation diffusion process. Accordingly, Hypothesis 2 is supported.

4.2. Information Disclosure as an Information-Risk Reduction Channel

Information disclosure may affect innovation diffusion by reducing the uncertainty faced by external stakeholders when evaluating firms’ innovation activities. In the innovation diffusion process, external firms, investors, suppliers, and potential adopters need to identify and assess the technological direction, patenting activities, and application value of the focal firm’s innovation outcomes. Better disclosure can therefore improve the visibility and interpretability of innovation-related information and reduce information asymmetry and external evaluation uncertainty. Accordingly, this paper uses information disclosure to capture the information-risk reduction channel through which bank–firm common ownership may affect innovation diffusion.

This paper examines information disclosure from two dimensions. First, the analyst’s earnings forecast error is used to measure the firm’s general information environment. As important information intermediaries in capital markets, securities analysts process public and private information to assess firm value and identify potential risks (Yang and Han 2025). The accuracy of analysts’ earnings forecasts reflects the extent to which external analysts understand a firm’s future financial condition, which is closely related to the adequacy and interpretability of corporate disclosure (Liu et al. 2024). Therefore, this paper constructs an analyst earnings forecast error indicator (FERROR) based on the deviation between actual earnings per share and analysts’ forecasted earnings per share. A larger FERROR indicates greater forecast error and a less transparent corporate information environment. This paper constructs an analyst earnings forecast quality indicator (FERROR) as follows:

$$FERROR={\displaystyle \frac{\left|AEPS-FEPS\right|}{\left|AEPS\right|}}$$

(7)

where AEPS denotes actual earnings per share, and FEPS denotes the mean analyst forecast of earnings per share. A larger value of this indicator indicates greater analyst forecast error, implying a less transparent corporate information environment.

Second, innovation information disclosure is used to measure the firm’s technology-specific information environment. Innovation diffusion depends not only on the existence of innovation outcomes but also on whether external actors can observe and understand relevant technological information. Drawing on Sun et al. (2025), this paper uses text-mining techniques to construct an indicator of firms’ innovation information disclosure. Specifically, innovation-related keywords are extracted from annual reports, and the keyword dictionary is expanded based on relevant national policy documents and the Word2vec neural-language model. The final dictionary contains 599 innovation-related keywords, including terms such as “patent,” “scientific research,” “process development,” “technological innovation,” “research,” and “invention.” The innovation information disclosure index (Keywords) is measured as the natural logarithm of one plus the frequency of innovation-related keywords in annual reports. A larger value indicates a higher level of innovation-related disclosure.

Columns (2) and (3) of Table 6 report firms’ information disclosure levels as measured by analyst earnings forecast quality and innovation information disclosure, respectively. The results show that the coefficients on bank–firm common ownership are −0.0102 and 0.0045, significant at the 1% and 5% levels, respectively. These findings indicate that bank–firm common ownership helps reduce information asymmetry and external evaluation uncertainty in the innovation diffusion process. Accordingly, Hypothesis 3 is supported.

4.3. Heterogeneity by Firms’ Market Position

Market position is not only a marker of a firm’s existing competitive strength but also reflects its dynamic capability to mobilize supply chain resources, shape industry rules, and promote technological standardization. Innovation diffusion is, in essence, a process through which knowledge flows and is adopted by others. Industry leaders or core firms typically occupy stronger market positions and often sit at pivotal points in the industrial chain, with extensive supplier and customer networks. When such leading firms introduce a new technology or product, other upstream and downstream firms in the supply chain may be induced or guided to adopt its technological standards, thereby generating innovation diffusion (Akcigit et al. 2021). By contrast, although firms with weaker market positions may also possess high-quality individual technologies or products, their location at the lower end of the supply chain and their more limited industry voice constrain their capacity to diffuse innovation.

In addition, innovation diffusion depends heavily on recipients’ recognition of technological value. A strong market position itself serves as a powerful reputational signal. Markets tend to allocate more attention to leading firms, making the innovation outcomes of high-position firms more visible (Jeon et al. 2025) and more likely to be identified and adopted by analysts, investors, and peer firms. By contrast, innovation originating from peripheral firms faces higher search costs and greater credibility barriers, resulting in lower diffusion efficiency than that of industry leaders. By strengthening both information channels and financing channels, bank–firm common ownership may further amplify the innovation advantages of firms with strong market positions. Accordingly, this paper expects the promoting effect of bank–firm common ownership on innovation diffusion to be stronger among firms with higher market positions.

To examine the role of market position, this paper measures a firm’s market position by the ratio of the firm’s annual sales to total annual industry sales. It further constructs a market-position dummy variable (Position). When this ratio is above the industry-sample median, Position equals 1, indicating that the firm has a relatively strong market position; otherwise, it equals 0. On this basis, this paper includes the interaction term between bank–firm common ownership and market position (PERCENT × Position) in the regression analysis. The results in column (1) of

Table 7. Moderation and Heterogeneity Analysis.

	(1)	(2)
	LnCit	LnCit
PERCENT	1.5551 **	3.3200 ***
	(0.5822)	(0.4596)
PERCENT × Position	1.5571 **
	(0.6478)
Position	1.0136 ***
	(0.1436)
PERCENT × SOE		3.9560 ***
		(0.7140)
SOE		0.6247 ***
		(0.2152)
Controls	Yes	Yes
Industry	Yes	Yes
Year	Yes	Yes
Observations	28,176	28,176
Adj. R-sq.	0.3822	0.3159

Note: Cluster-robust standard errors are shown in parentheses. ** and *** denote statistical significance at the 5% and 1% levels, respectively.

show that the coefficient on the interaction term (PERCENT × Position) is 1.5571 and significantly positive at the 5% level. This indicates that the promoting effect of bank–firm common ownership on corporate innovation diffusion is stronger among firms with higher market positions.

4.4. Heterogeneity by Ownership Type

In the Chinese institutional context, it is necessary to further examine the role of ownership type in shaping the effect of bank–firm common ownership on innovation diffusion. State-owned enterprises (SOEs) bear important responsibilities for R&D investment in foundational innovation and for industrial technological upgrading (Wan and Yu 2022). Unlike private firms, which are more inclined toward applied innovation, SOEs, with their stronger capital base and policy support, are more likely to occupy upstream positions in the industrial chain and to focus on foundational innovation characterized by high technological barriers, long horizons, and strong externalities. Compared with applied research, basic research typically has broader application scenarios and stronger knowledge-spillover effects (Akcigit et al. 2021). This implies that, from the supply side of innovation, SOEs hold a large stock of patent technologies with high citation value.

Moreover, ownership type to some extent shapes firms’ willingness to diffuse innovation. Private firms’ innovation decisions generally follow the principle of profit maximization. To maintain a competitive advantage, they often prefer to build technological barriers and prevent knowledge spillovers through patent exclusion. By contrast, as instruments through which the government corrects market failures in R&D, SOEs possess a clear public orientation and political mission, and their objective function is to maximize social welfare. Nevertheless, although SOEs possess basic research and technologies with strong diffusion potential, institutional constraints and relatively weak market sensitivity often place their innovation outcomes in an environment characterized by difficult commercialization and slow diffusion (Cao et al. 2020). Bank–firm common ownership may improve governance and strengthen resource support, thereby enabling SOEs to release more effectively the diffusion potential embedded in their technological stock. By contrast, because private firms already tend to use patent assets more efficiently and because the use and diffusion of their technological achievements are more tightly disciplined by market competition, the marginal effect generated by bank–firm common ownership may be relatively limited. Accordingly, this paper expects the promoting effect of bank–firm common ownership on innovation diffusion to be stronger for SOEs.

To test this conjecture, this paper includes the interaction term between bank–firm common ownership and ownership type (PERCENT × SOE) in the regression analysis. The results in column (2) of Table 7 show that the coefficient on the interaction term (PERCENT × SOE) is 3.9560 and statistically significant at the 1% level. This indicates that the promoting effect of bank–firm common ownership on innovation diffusion is significantly stronger for state-owned enterprises than for private firms.

5. Conclusions

Using a sample of Chinese A-share listed firms in the Shanghai and Shenzhen stock markets from 2010 to 2023, this paper examines the impact of bank–firm common ownership on corporate innovation diffusion and its underlying mechanisms. The empirical results show that bank–firm common ownership has a positive effect on corporate innovation diffusion. This conclusion remains robust after alternative variable measurements, a higher identification threshold for bank–firm common ownership, lagged explanatory variables, instrumental-variable estimation and propensity score matching. Mechanism tests show that bank–firm common ownership promotes innovation diffusion through commercial credit financing and information disclosure. Specifically, commercial credit financing helps firms buffer financing pressure during the innovation diffusion process, while information disclosure reduces information asymmetry and external evaluation uncertainty surrounding innovation activities. Further analysis shows that the effect of bank–firm common ownership on corporate innovation diffusion is more pronounced among firms with stronger market positions and state-owned enterprises.

The findings have both theoretical and practical implications. Theoretically, this paper extends the literature on innovation diffusion by introducing bank–firm common ownership as a micro-level ownership-based financial linkage, and shifts the focus from firms’ internal innovation output to the external diffusion of innovation outcomes. It also contributes to research on financial linkages and innovation under uncertainty by showing that commercial credit financing and information disclosure are important channels through which ownership-based bank–firm linkages support innovation diffusion. Practically, the results suggest that firms may improve the diffusion of innovation outcomes by strengthening stable financial relationships, supply chain credit arrangements, and innovation-related information disclosure. For banks and institutional investors, bank–firm common ownership may provide useful information channels for evaluating firms’ innovation capacity and risk profile. For regulators, the findings suggest that financial ownership networks can support innovation diffusion, while potential governance risks, conflicts of interest, and market-power concerns associated with common ownership should also be monitored.

This study also has several limitations, which provide directions for future research. First, the sample is limited to Chinese A-share listed companies. Therefore, the conclusions may be more applicable to listed firms operating in bank-centered financial systems, markets with active institutional investors, and institutional environments where commercial credit and bank–firm relationships play an important role. Whether the findings apply to small and medium-sized enterprises, unlisted firms, or firms in countries with more market-based financial systems requires further examination. Second, this paper measures innovation diffusion mainly through patent citations. Patent citations capture an important but partial dimension of innovation diffusion. They reflect citation-based knowledge use and technological recognition, but do not fully capture other diffusion channels such as technology licensing, personnel mobility, or informal knowledge flows. Accordingly, our findings should be interpreted as evidence on citation-based innovation diffusion rather than as a complete measure of all forms of innovation diffusion. In addition, the instrumental-variable analysis is subject to the standard limitation that the exclusion restriction cannot be directly tested. Although the leave-one-out industry-year instrument helps mitigate firm-level reverse causality concerns, the IV results should be interpreted as evidence that alleviates, rather than fully eliminates, potential endogeneity. Finally, future research may further examine whether bank–firm common ownership affects innovation diffusion through human-capital allocation, R&D personnel mobility, or technical-talent attraction, which are not directly tested in this paper.