Paradox or Synergy Between Digital Capability and Corporate Social Responsibility to Achieve Ambidextrous Innovation in Chinese Firms

Abstract

This paper provides a new and significant conceptual framework to enhance understanding of how digital capability and corporate social responsibility (CSR) complement each other in achieving the trade-off of ambidextrous innovation. Building on resource orchestration theory, we propose that opportunity recognition can serve as a mediating bridge to convey the positive impact of digital capability on ambidextrous innovation. Furthermore, these effects are likely to be especially pronounced among enterprises with a higher level of CSR implementation according to the reciprocity principle of social capital theory. We conducted a questionnaire-based survey among executives from 225 non-listed companies and a longitudinal panel study of 1897 listed companies from 2009 to 2022. The results support our hypotheses, showing that CSR implementation strengthens the active indirect effect of digital capability on ambidextrous innovation through accurate opportunity recognition. This paper enriches the research on the positive consequences of digital capabilities, introduces opportunity recognition into resource orchestration theory from the perspective of intangible assets, verifies the mediating role of opportunity recognition between digital capabilities and ambidextrous innovation, and sheds light on how an organization’s CSR strategy and digital capabilities are complementary. CSR can catalyze the positive impact of an enterprise’s digital capability on opportunity recognition and ambidextrous innovation. We advise enterprises on sustainable development, emphasizing the importance of fulfilling their CSR strategies while enhancing their digital capabilities.

1. Introduction

As industries transform rapidly, leveraging digital technology becomes crucial for organizations to promote sustainable growth and competitiveness [1]. Ambidextrous innovation, the capacity to pursue exploitative and exploratory innovation simultaneously, is vital for adapting to complex external environments, thereby strengthening competitive strengths and sustainability [2]. Many scholars believe that digital capability has a positive effect on ambidextrous innovation [3].

However, recent skepticism challenges this relationship [4]. Critics argue that digital tools can hinder innovation through over-reliance, potentially depleting firms’ ability to engage in ambidextrous innovation [4,5]. Investing in digital capabilities may yield positive results over time [6], despite the high costs and the need for specialized expertise. This leads to a question about the efficiency of digital capabilities for all enterprises.

To date, there is no consensus on how digital capability influences ambidextrous innovation. While many believe that digital capabilities enhance innovative capacities [7], others warn of the risks of diminished ambidextrous innovation from excessive dependence on technologies [4,6]. Notably, more studies are needed to mitigate the adverse effects of indiscriminate digitalization.

Resource orchestration theory suggests that digital capabilities facilitate ambidextrous innovation by identifying opportunities for growth and development. This involves seeking creative ideas for product development and identifying external market opportunities [8]. Companies with advanced digital capabilities are more sensitive to industry trends and market updates [7]. They leverage such opportunities to achieve a balance between exploratory and exploitative innovation [2]. Thus, we aim to explore whether opportunity recognition mediates the link between digital capability and ambidextrous innovation, arguing that this mechanism mediates the positive effect of digital capability on ambidextrous innovation.

While opportunity recognition can be identified as a facilitator for innovative outcomes, it remains unclear how digital capability affects opportunity recognition for ambidextrous innovation across differing degrees of CSR implementation. Previous literature has suggested that CSR and digital capability are both cost-oriented strategies; they may be detrimental to business development when a company’s financial situation and resources are not strong [9,10]. But we believe that the relationship between the two is not contradictory. According to social capital theory, CSR goes beyond legal and financial obligations, addressing stakeholders’ needs to realize societal welfare [11,12]. The level of CSR implementation may interact differently with digital capability. Specifically, a firm’s responses to market opportunities is inherently tied to how thoroughly it considers the stakeholders’ needs [13]. Implementing CSR provides companies with perspectives on stakeholder needs, which can facilitate the relationship between digital capability and ambidextrous innovation [14]. Drawing on the reciprocity principle, a company’s attention to stakeholder preferences cultivates a more nuanced awareness of industrial dynamics and business opportunities [15]. Therefore, a company with more extensive CSR implementation will likely leverage its digital capability more effectively to achieve ambidextrous innovation through recognizing opportunities.

Building on the foregoing analysis, we investigate two research questions (RQs):

RQ1. Does digital capability have a positive impact on ambidextrous innovation through opportunity recognition?

RQ2. Can CSR be leveraged as fuel to amplify the positive impact of digital capability on opportunity recognition and ambidextrous innovation?

This paper employs two studies to verify the research question, because:

• With the rapid development of digitalization in China, all industries are facing opportunities and challenges brought by digital transformation. If only listed or unlisted enterprises are studied, research conclusions and management implications will lack relative universal value.
• The sample of study 1 in this paper consists of non-listed enterprises in China, and the questionnaire survey with multi-point measurement is adopted. However, due to the limitations of data availability and the characteristics of the research methods, the data in Study 1 are only cross-sectional, making it difficult to objectively and quantitatively observe the changes in the indicators of enterprises over a period of time. In Study 2, we conducted a longitudinal panel study of 1897 listed companies in China from 2009 to 2022. If the two can obtain the same result, it proves that our conclusion is relatively reliable and universal.

We offer several important contributions to theoretical research. First, we address the concerns of previous scholars about enterprises abusing digital capabilities [4,6], enriching the understanding of the positive consequences of digital capabilities, and identify the circumstances under which the benefits of digital capabilities can be more efficiently realized. This study utilizes two studies to explore how firms can utilize digital capabilities more effectively to achieve the equilibrium of ambidextrous innovation.

Second, we explore opportunity recognition as a mechanism that channels digital capability on ambidextrous innovation from an “intangible assets” perspective. Most previous literature has taken perspectives related to “tangible assets”, such as easing financing restrictions and optimizing organizational processes as bridges to explore the positive role of digital capability on ambidextrous innovation [16,17]. Instead, we consider that recognized opportunities are precious resources for companies’ innovation and sustainable development based on resource orchestration theory. Given the data availability and literature review, we leverage the R&D investment ratio and the frequency of mergers and acquisitions to measure opportunity recognition in study 2.

Third, this paper explores the complementary roles of CSR and digital capability from the lens of the reciprocal relationship between firms and stakeholders. We address some scholars’ concerns about the paradoxical roles of CSR and digital capability [18]. This highlights the crucial role of the interaction between CSR and digital capabilities in fostering sustainable ambidextrous innovation [19].

2. Literature Review and Hypotheses Development

2.1. Digital Capability and Opportunity Recognition

Based on resource orchestration theory, which emphasizes the importance of utilizing resources effectively, companies with advanced digital capabilities gain a competitive advantage through broader and deeper information search capabilities, thereby fostering strategic decision-making [20]. By leveraging these advanced digital capabilities to analyze and process extensive data, including customer preferences and competitor strategies [21], firms gain a more in-depth understanding of market dynamics. These capabilities enable organizations to derive actionable intelligence from complex datasets, identifying business opportunities that would otherwise remain hidden [22].

In contrast, companies with limited digital capabilities often struggle to identify opportunities [23]. These firms typically face challenges in acquiring relevant market knowledge and accurately interpreting information, which hinders their ability to identify opportunities within the business environment [24]. This suggests that companies with less digital capability are less accurate in recognizing unique opportunities than their peers. We anticipate that digital capability is positively correlated with the recognition of opportunities.

Hypothesis 1: 

Digital capability is positively related to opportunity recognition.

2.2. Opportunity Recognition and Ambidextrous Innovation

According to resource orchestration theory, opportunities identified by enterprises can be regarded as intangible resources. Ambidextrous innovation has been universally regarded as a necessity for sustainability [2]. However, pursuing ambidextrous innovation is often challenging [15]. Research reveals that a primary reason for failure in ambidextrous innovation is that companies prioritize exploitative over exploratory innovation [5,25]. This is because exploitative innovation leverages existing knowledge and resources to achieve immediate gains, requiring low input costs and yielding substantial short-term benefits [25,26]. Conversely, exploratory innovation entails high input costs with long-term benefits, resulting in small short-term gains [27]. Consequently, companies cautiously pursue exploratory innovation.

Other studies attribute failures in ambidextrous innovation to decision-makers’ traits, such as being overly powerful, myopic, and risk-averse [28]. We argue that these failures may stem from a company’s inability to identify opportunities, which can create anxiety and resistance in exploratory innovation decisions [2]. Thus, the primary causes of ambidextrous innovation failure are a lack of trust and security in exploratory strategies and the inability to identify suitable opportunities for exploration.

Hypothesis 2: 

Opportunity recognition is positively related to ambidextrous innovation.

2.3. The Mediating Role of Opportunity Recognition

Resource management encompasses structuring, bundling, and leveraging firms’ resources to create core competitiveness [28]. Structuring involves acquiring, accumulating, and divesting resources to build the firm’s resource portfolio. Bundling integrates these resources to form capabilities while leveraging these capabilities to capitalize on specific opportunities [29]. Central to resource orchestration theory is how firms effectively utilize their resources.

In this context, a company’s digital capability serves as an external probing mechanism, enabling it to access various resources and information. We suggest that digital capability can be viewed as a form of “structuring” in resource orchestration theory. Opportunity recognition can effectively integrate information from the business environment with a firm’s existing resources and capabilities. This can be considered as the second stage, “bundling”. Companies equipped with the ability to acquire external resources and assess their existing resources can allocate resources more efficiently, thereby promoting ambidextrous innovation. This leads us to the third stage, leveraging resources, to achieve the goal of sustainable development for enterprises.

Hypothesis 3: 

The positive relationship between digital capability and ambidextrous innovation is mediated by the recognition of opportunities.

2.4. The Moderating Role of CSR Implementation on the Digital Capability–Opportunity Recognition Relationship

Some scholars have begun exploring the role of CSR strategies in the digital era [30]. However, there is a scarcity of studies examining whether CSR implementation accelerates the positive effect of digital capability on opportunity recognition. Building on the IT paradox, superior digital capabilities can lead enterprises to experience information overload [31]. Many organizations struggle to discern which opportunities align with their current development stage, which can potentially trigger the negative aspects of digital capability [4]. We posit that CSR implementation can mitigate these negative impacts.

On the one hand, digital capabilities emphasize the adjustment of an enterprise’s strategic priorities and the upgrading and transformation of management processes, with the aim of enabling the enterprise to adapt to the current constantly evolving external environment and enhance its sustainable competitiveness [1]. Similarly, the implementation of CSR emphasizes that while enterprises pursue economic benefits, they should actively consider the impact of their strategic activities on stakeholders [11]. In other words, enterprises with a high degree of CSR implementation not only focus on profitability when making decisions but also pay close attention to the dynamics of stakeholders (such as consumer preferences, market segment trends, etc.). The fundamental purpose of enterprises performing all these actions is to enable them to seek opportunities conducive to innovation and sustainable development [12], which coincides with the essence of digital capabilities. From this perspective, CSR can help enterprises maximize the positive impact of digital capabilities on opportunity recognition [14].

On the other hand, the essence of CSR is to elaborate on the relationship between enterprises and stakeholders [11]. According to social capital theory, firms with high levels of CSR implementation are better equipped to assess their current situation, understand market dynamics, and align with stakeholder needs [32]. Given the reciprocal principle, implementing CSR is not only an ethical obligation as part of corporate citizenship [10], but more importantly, doing so helps firms establish effective strategies that differentiate themselves from other competitors [33]. Companies that thoughtfully execute CSR strategies experience tangible benefits attributed to such “altruistic behavior” [34]. Therefore, the relationship between CSR and digital capabilities is complementary. Suppose an enterprise wants to identify appropriate opportunities through high-level digital capabilities. In that case, it needs high-level CSR to help the enterprise screen information, position the market, clarify the supply and demand relationship, identify innovation opportunities, and avoid falling into a decision-making predicament of information overload. Otherwise, digital capabilities may be misused.

When confronted with issues such as information overload brought by high-level digital capabilities, a company’s high-level CSR fulfillment will become a “probe” that delves into a vast amount of information, enabling the company to identify business opportunities efficiently that are conducive to innovation and sustainable development. At the same time, it is essential to clearly understand its strategic positioning [10]. CSR implementation can help enterprises identify stakeholder preferences and popular market trends [35]. This insight enables organizations to filter out redundant information and resources generated by prominent digital capabilities, selecting those that are most suitable for the current stage of the organization’s development, thereby fostering ambidextrous innovations [36]. Specifically, companies with more voluntary CSR implementation and digital capability can find opportunities that better cater to target consumers’ preferences. In summary, CSR initiatives enhance the positive effect of digital capability on opportunity recognition by helping companies navigate complex market conditions and align with stakeholder needs.

In summary, CSR initiatives enhance the positive effect of digital capability on opportunity recognition by helping companies navigate complex market conditions and align with stakeholder needs.

Hypothesis 4: 

CSR Implementation moderates the relationship between digital capability and opportunity recognition, such that the relationship is more potent when CSR implementation is high rather than low.

2.5. The Moderating Role of CSR Implementation on the Opportunity Recognition–Ambidextrous Innovation Relationship

According to the principle of reciprocity in social capital theory, firms with high CSR implementation levels can significantly facilitate the stages of identifying opportunities and transforming them into ambidextrous innovations [37]. First, according to the reciprocal principle, enterprises with strong CSR are more likely to receive government support such as special subsidies, preferential policies for tax reduction, and social financing while also contributing to society [38]. This ensures a mutually beneficial relationship between enterprises and society [39]. Enterprises will utilize these funds and technical support to “identify business opportunities and transform them into ambidextrous innovation” to create a virtuous circle of reciprocal relationships [40].

Second, firms that are more involved in CSR activities are more likely to focus on cross-field knowledge, such as industry alliances and industry-university-research collaborations, which will shorten the conversion cycle of ambidextrous innovation [41]. Furthermore, internally, companies that fulfill CSR pay more attention to their employees’ career development and well-being [42]. Based on the principle of reciprocity between the company and its employees, the company will increase employee identification with the organization’s strategy through CSR concern for its employees [43] and promote the efficiency of cross-functional collaboration within the company, which will fuel the process of transforming business opportunities into ambidextrous innovations.

Hypothesis 5: 

CSR implementation moderates the relationship between opportunity recognition and ambidextrous innovation, such that the relationship is more potent when CSR implementation is high rather than low.

To illustrate our proposed conceptual model, please refer to Figure 1.

3. Study 1

3.1. Participants and Procedures for Study 1

To ensure the questionnaire’s validity, we first conducted a pilot test with 30 voluntary DBA students to check the appropriateness of the survey items. We drew our sample from the 2023 Guangdong Top 500 Enterprises Ranking, a representative list of firms in the Pearl River Delta (PRD) region in China. Of the 500 companies contacted, 360 agreed to participate. Research assistants then distributed electronic questionnaire links and QR codes via WeChat, requesting each company to designate one manager to complete the survey. Each firm submitted a three-wave questionnaire set (from May 2024 to August 2024), ensuring that our study operates at the organizational level [44]. The three separate times adopted in Study 1 help to reduce the common method bias [2,45].

To ensure privacy and credibility, we asked participants to write down a unique five-digit phone number (e.g., 91962) as an identification code to match the questionnaires across the three waves (1 month per wave). The response rates for the three questionnaires were 95.56%, 81.69%, and 83.27%. After excluding mismatched questionnaires, the final sample size was 225.

3.2. Measures

The measurement instrument used in this study was a 5-point scale, ranging from 1 (strongly agree) to 5 (strongly disagree). Since the original scales were in English, we followed a translation and back-translation procedure [46].

3.2.1. Digital Capability

We utilized the digital capability scale developed and validated by Zhou and Wu (2010) and examined by Pratono (2024) [5,47]. The scale demonstrated good reliability (Cronbach’s α = 0.88).

3.2.2. Ambidextrous Innovation

We used the scale developed by He and Wong (2004) [23] to measure ambidextrous innovation, which includes two dimensions: exploitative innovation (four items, Cronbach’s α = 0.87) and exploratory innovation (four-item scale, Cronbach’s α = 0.86). To visualize the balance of ambidextrous innovation, we treat it as a single variable in hypothesis testing [48,49]. We follow Tang et al. (2021) in representing it using the additive index of the two dimensions [2].

3.2.3. Opportunity Recognition

Opportunity recognition was measured using a five-item scale from Kuckertz et al. (2017) [8] (Cronbach’s α = 0.88).

3.2.4. CSR Implementation

We followed Tian et al. (2015) in measuring CSR implementation using a 12-item scale (Cronbach’s α = 0.89) [11].

3.2.5. Control Variables

Control variables were selected based on a review of prior studies to minimize spurious predictions of explanatory variables. We controlled for firm size, industry, ownership, and other factors.

3.3. Results

Table 1. Study 1: Descriptive statistics and correlations.

Variables	Mean	SD	1	2	3	4	5	6	7	8	9	10	11	12	13
1. Gender	0.48	0.50	1
2. Age	42.91	5.85	−0.23 **	1
3. Education	2.41	0.650	−0.14 *	0.05	1
4. Work experience	9.10	6.64	−0.06	0.35 **	−0.26 **	1
5. Position	1.21	0.41	−0.15 *	−0.08	0.15 *	0.06	1
6. Industry	4.04	2.09	0.01	0.17 *	−0.07	0.03	−0.11	1
7. Size	830.71	880.45	−0.13 *	0.03	0.03	0.09	−0.12	−0.03	1
8. Firm age	26.51	17.73	−0.03	0.16 *	0.01	0.28 **	−0.22 **	−0.05	0.36 **	1
9. Ownership	1.72	0.84	0.04	−0.18 **	0.07	−0.03	0.25 **	−0.41 **	0.02	−0.07	1
10. Digital capability	3.17	0.63	−0.07	0.02	−0.06	−0.01	0.09	−0.00	0.09	−0.03	−0.04	0.77
11. CSR implementation	3.49	0.68	−0.16 *	0.05	−0.04	−0.00	0.07	0.02	0.02	0.11	0.04	0.15 *	0.89
12. Opportunity recognition	3.84	0.72	−0.05	0.15 *	−0.12	0.12	−0.01	0.12	−0.01	−0.05	−0.00	0.44 **	0.21 **	0.88
13. Ambidextrous innovation	7.96	1.55	−0.19 **	0.08	−0.10	0.02	0.16 *	−0.04	0.01	−0.01	0.02	0.43 **	0.25 **	0.58 **	0.89

Note: N = 225. Gender (men = 0, women = 1); Education (1 = college education; 2 = undergraduate degree, 3 = Master’s degree, 4 = Doctor’s degree); Industry (1 = Manufacturing industry; 2 = Service industry; 3 = Financial industry; 4 = Internet/Software; 5 = Construction/Real Estate; 6 = Education and training industry; 7 = Health and social work; 8 = Others). Diagonal values are the square roots of the average variance extracted for the key variables. * p < 0.05; ** p < 0.01.

reports the descriptive statistics and correlation analysis. We applied confirmatory factor analysis (CFA) to verify that our measurement model exhibited good discriminant validity, as shown in

Table 2. Study 1: Comparison of measurement models.

Model	Descriptions	χ2	df	TLI	CFI	RMSEA	SRMR
Model 1	Five factors: digital capability, opportunity recognition, CSR implementation, exploitative innovation, and exploratory innovation.	520.82	391	0.96	0.96	0.04	0.05
Model 2	Four facto—exploitative innovation and exploratory innovation—were combined into one factor.	674.81	395	0.91	0.92	0.06	0.05
Model 3	Four factors: digital capability and opportunity recognition were merged into one factor.	882.34	395	0.85	0.86	0.07	0.07
Model 4	Three factors—CSR implementation, digital capability, and opportunity recognition—were merged into one factor.	2029.20	402	0.51	0.54	0.13	0.14
Model 5	Two factors—CSR implementation, digital capability, and opportunity recognition—were merged into one factor. Exploitative innovation and exploratory innovation were also merged into a single factor.	2159.49	404	0.47	0.51	0.14	0.16
Model 6	One factor, comprising digital capability, opportunity recognition, CSR implementation, exploitative innovation, and exploratory innovation, was loaded into a single factor.	2425.92	405	0.40	0.43	0.15	0.15

Note: N = 225. Abbreviations: χ², chi-squared value; df, degree of freedom; TLI, Tucker–Lewis Index; CFI, comparative fit index; RMSEA, root mean square error of approximation; SRMR, standardized root mean square residual.

. Compared with other models in Table 2, the goodness of fit for the five-factor basic model (χ² = 520.82, df = 391, TLI = 0.96, CFI = 0.96, RMSEA = 0.04, SRMR = 0.05) is ideal. For example, the goodness of fit for the four-factor model, where ambidextrous innovation is combined into one dimension, is significantly worse (Δχ² = 153.99, p < 0.001) than that for the five-factor basic model. The results prove that each construct in Study 1 has a high degree of differentiation and independence among variables.

Table 3. Study 1: Construct validity and reliability.

First-Order Construct	Composite Reliability (CR)	Root of Average Variance Extracted (AVE)	α
Digital capability	0.88	0.77	0.88
Opportunity recognition	0.88	0.78	0.88
Ambidextrous innovation	0.93	0.78	0.89
Exploitative innovation	0.87	0.79	0.87
Exploratory innovation	0.86	0.78	0.86
Corporate social responsibility implementation	0.95	0.79	0.89

Note: N = 225.

reports the square root of the average extracted variance (AVE) and composite reliability (CR). The square root of AVE for each variable is higher than its correlation coefficient (see Table 1), and the combined reliability (CR) is greater than 0.80, which indicates that the discriminant validity is qualified [50].

3.4. Test of Hypothesis

We tested the hypotheses as follows.

Table 4. Study 1: Regression analysis.

Variables	Opportunity Recognition		Ambidextrous Innovation
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)	B (SE)
Constant	1.65 ** (0.49)	1.18 * (0.50)	4.44 ** (1.05)	2.64 ** (0.93)	2.27 * (0.94)	1.84 † (0.98)
Gen	−0.03 (0.09)	−0.01 (0.09)	−0.49 * (0.20)	−0.45 ** (0.17)	−0.43 ** (0.16)	−0.40 * (0.17)
Age	0.01 † (0.01)	0.01 (0.01)	0.02 (0.02)	0.01 (0.02)	0.00 (0.02)	0.01 (0.02)
Education	−0.07 (0.07)	−0.05 (0.07)	−0.33 * (0.15)	−0.25 † (0.13)	−0.22 † (0.13)	−0.19 (0.13)
Work experience	0.01 (0.01)	0.01 (0.01)	−0.02 (0.02)	−0.03 † (0.01)	−0.03 † (0.01)	−0.02 † (0.01)
Position	−0.13 (0.12)	−0.19 † (0.11)	0.46 † (0.25)	0.60 ** (0.22)	0.48 * (0.21)	0.58 ** (0.21)
Industry	0.04 † (0.02)	0.03 (0.02)	−0.02 (0.05)	−0.07 (0.04)	−0.10 * (0.04)	−0.07 (0.04)
Size	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)
Firm year	−0.00 (0.00)	−0.01 † (0.00)	0.01 (0.01)	0.01 (0.01)	0.01 (0.01)	0.01 (0.01)
Ownership	0.09 (0.06)	0.08 (0.06)	0.05 (0.13)	−0.05 (0.11)	−0.06 (0.10)	−0.08 (0.11)
Digital capability	0.51 ** (0.07)	0.50 ** (0.07)	1.01 ** (0.15)	0.45 ** (0.15)	0.55 ** (0.14)	0.32 * (0.15)
Opportunity recognition				1.09 ** (0.13)	0.91 ** (0.13)	1.18 ** (0.14)
CSR Implementation		0.20 ** (0.06)			0.30 * (0.12)	0.23 † (0.12)
INT 1		0.40 ** (0.10)			0.84 ** (0.19)
INT 2						0.38 * (0.16)
Adjust_R Square	0.21 **	0.28 **	0.21 **	0.41 **	0.46 **	0.43 *
R Square	0.25 **	0.32 **	0.24 **	0.44 **	0.49 **	0.46 *

Note: N = 225. Unstandardized coefficients are presented. Standard errors are reported in parentheses. * p < 0.05, ** p < 0.01, † p < 0.1 (two-tailed).

displays the results of the ordinary least squares (OLS) regression analysis. Digital capability is positively correlated with opportunity recognition (β = 0.51, p < 0.001; Table 4, Model 1), and opportunity recognition is significantly positively correlated with ambidextrous innovation (β = 1.09, p < 0.001; Table 4, Model 4), which supports Hypotheses 1 and 2.

We used the bootstrapping approach with 10,000 replications and a 95% Confidence Interval (CI) to test Hypothesis 3 [50].

Table 5. Study 1: Mediation test.

Effect	B	CI
Indirect effect	0.52	[0.32, 0.75]
Total effect	1.07	[0.77, 1.36]

Note: N = 225. The indirect effect was significant if the confidence interval did not contain zero. Abbreviation: CI, confidence interval.

shows the significant positive indirect effect of opportunity recognition between digital capability and ambidextrous innovation. The estimation of the indirect effect is 0.52 (95% CI = [0.32, 0.75]), supporting Hypothesis 3. We employed a mean-centered digital capability and CSR implementation to reduce potential collinearity problems [51] and then multiplied them to create a new interactive variable, INT 1. As shown in Table 4, the coefficient of INT 1 is significantly positive at the 1% level, indicating that CSR implementation can substantially strengthen the positive impact of digital capability on opportunity recognition (β = 0.40, p < 0.001, see Model 2), supporting Hypothesis 4. We also mean-centered opportunity recognition and CSR implementation to obtain a new interactive variable INT 2. The coefficient of INT 2 is significantly positive at the 5% level, indicating that CSR implementation can significantly strengthen the positive impact of opportunity recognition on ambidextrous innovation (β = 0.38, p < 0.05; see Model 6), supporting Hypothesis 5. INT 1 represents the interaction term between the independent variable and the moderating variable, and INT 2 represents the interaction term between the mediating variable and the moderating variable. According to the theoretical model presented in this paper, INT 1 has an effect on the mediating effect in the first half, while INT 2 has an effect on the mediating effect in the second half. The two are different.

Furthermore, we applied 10,000 bootstrap replications using the PROCESS Macro [50] and created two simple slope plots. Figure 2a shows that the link between digital capability and opportunity recognition was significantly positive when CSR implementation was high (1 SD above the mean; β = 0.79, p < 0.001) but insignificant when CSR implementation was low (1 SD below the mean; β = 0.18, p = n.s.).

Finally, we used Model 58 from the PROCESS Macro with 10,000 replications to conduct supplementary analysis of moderated mediation.

Table 6. Study 1: Moderated mediation.

Effect	B	95% CI
Mediating effect	0.56	[0.34, 0.82]
High CSR (+1 SD)	1.12	[0.60, 1.69]
Low CSR (−1 SD)	0.14	[−0.09, 0.38]

Note: N = 225. The indirect effect was significant if the confidence interval did not contain zero. Abbreviation: CI, confidence interval.

shows that when CSR implementation is high, the indirect effect is significantly positive (β = 1.12, 95% CI = [0.60, 1.69]) but insignificant when CSR implementation is low (β = 0.14, 95% CI = [−0.09, 0.38]). Moreover, we drew simple slope plots, as displayed in Figure 2b, to show this effect.

4. Study 2

4.1. Participants and Procedures for Study 2

We used databases to obtain longitudinal panel data for listed enterprises in two stock exchanges (Shanghai & Shenzhen) in Study 2. Listed companies face more stringent regulatory requirements than unlisted companies. Thus, if Study 2 reaches the same conclusion, the theoretical model provides credible evidence for Chinese firms.

Due to availability, we selected data from China’s A-share listed companies spanning from 2009 to 2022. Based on experience from previous research, we deployed the following criteria to clean our samples: (1) We excluded listed companies in industries such as finance and insurance. (2) We excluded samples that have been labeled ST, (*) ST and PT. (3) We excluded insolvent companies. (4) We excluded samples with missing values. Consistent with most studies, we used winsorization at the 1% and 99% levels to mitigate the effects of extreme values. Finally, we obtained 11,420 unbalanced panel data points from 1897 companies.

4.2. Measure

4.2.1. Digital Capability

As a crucial capability for the sustainable development of enterprises, digital capabilities are typically reflected in an enterprise’s annual report. The use of words in the annual report reveals the development priorities and capabilities of the enterprise. Therefore, it is feasible to describe digital capabilities by using the frequencies of digital-related words from annual reports. Referring to existing research [52], we applied Python 3.12.1 crawler functions to filter the firm’s annual reports from the China Stock Market & Accounting Research (CSMAR) database to capture the frequency of keywords related to digital capability. We divided the keyword frequency by the total number of words in the annual report as the proxy variable for digital capability. We followed Luo et al. (2024) to select relevant keywords [53].

4.2.2. Ambidextrous Innovation

We measured ambidextrous innovation by analyzing the number of patents from the Chinese Research Data Services (CNRDS) platform, including utility model and invention patents. Since there are no citations in China’s patent data, we referred to previous studies and used the international patent classification (IPC) numbers for patent applications [54]. This study chose a five-year window. If the first four digits of IPC numbers have appeared at least once in the previous five years, the patent is classified as an exploitative innovation. Otherwise, it is considered exploratory innovation. To visualize the equilibrium of ambidextrous innovation more intuitively, we calculated the additive index (the sum of exploratory and exploitative innovation) as the proxy variable of ambidextrous innovation [2].

4.2.3. Opportunity Recognition

Using the CSMAR database, we analyzed the number of mergers and acquisitions (M&A) and the ratio of R&D investment to operating income as proxy variables for opportunity recognition. Companies engaged in M&A or R&D activities identify business opportunities that foster ambidextrous innovation [55,56]. Based on mainstream research, we believe it is appropriate to use this proxy variable for measurement.

4.2.4. CSR Implementation

To obtain the latest data, we used average annual ESG scores from the Huazheng database to measure CSR implementation [57]. This approach is more objective and professional than other ratings [58].

4.2.5. Control Variables

Based on a review of previous research [2,59], Study 2 selected several control variables to reduce spurious predictions of the explanatory variables. The proxy variables for measuring the control variables were obtained from the CSMAR database. Detailed definitions of the control variables are shown in

Table 7. Study 2: Control variable definition.

Variable	Measurement
Firm size	The natural logarithm of the total assets of the corporation.
ROA	Net profit/Average balance of total assets
Firm age	Ln (year-year of establishment +1)
SOE	A dummy variable on whether the corporation is state-owned or not.
Management share	The number of shares held by the Board of Directors, the supervisory Board, and senior management/and the Total share capital
Intangible assets	Net intangible assets/Total assets
Gender	Percentage of females in management

.

4.3. Results

We used the two-way fixed effects model to validate hypotheses after conducting the Hausman test [53]. We lagged the independent variables, all control variables, and moderators by one year.

Table 8. Study 2: Correlations.

	1	2	3	4	5	6	7	8	9	10	11
1. Size(/10)	1
2. ROA(/10)	−0.05 **	1
3. Firm age(/10)	0.15 **	−0.06 **	1
4. SOE(/10)	0.26 **	−0.15 **	0.08 **	1
5. Management share(/10)	−0.23 **	0.11 **	−0.11 **	−0.48 **	1
6. Intangible assets(/10)	0.01	0.01	−0.05 **	0.06 **	−0.02	1
7. Gender(/10)	−0.17 **	0.04 **	0.10 **	−0.25 **	0.18 **	−0.03 **	1
8. Ambidextrous innovation(/1000)	0.45 **	0.02 *	0.03 **	0.08 **	−0.05 **	−0.02 *	−0.13 **	1
9. Digital capability(*10)	−0.03 **	0.02 †	0.04 **	−0.12 **	0.12 **	−0.04 **	0.12 **	0.10 **	1
10. Opportunity recognition(/10)	−0.19 **	0.12 **	−0.03 **	−0.30 **	0.32 **	−0.06 **	0.07 **	0.16 **	0.28 **	1
11. CSR implementation	0.28 **	0.10 **	0.04 **	0.10 **	−0.03 **	−0.03 **	−0.00	0.16 **	0.04 **	0.04 **	1

Note: N = 11,420. ** p < 0.01, * p < 0.05, † p < 0.1.

shows the descriptive statistics and correlations for all variables. The correlation coefficients with other variables are all less than 0.50, and the mean variance inflation factors (VIFs) of the variables are below the threshold value of 10 (mean VIF = 1.18). This indicates that the multicollinearity problem in this model is not significant [60].

Table 9. Study 2: Results of OLS regression analyses.

Variables	Opportunity Recognition		Ambidextrous Innovation
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Digital capability	0.46 ** (0.15)	0.39 ** (0.14)	0.34 ** (0.13)	0.30 * (0.13)	0.24 * (0.10)	0.21 * (0.10)		0.28 * (0.13)
Opportunity recognition				0.09 ** (0.02)		0.09 ** (0.01)	0.09 ** (0.01)	0.09 ** (0.01)
CSR implementation		0.00 (0.00)			0.00 (0.00)	0.00 (0.00)	0.01 ** (0.00)	0.01 ** (0.00)
INT 1		0.34 * (0.16)			0.49 ** (0.11)	0.46 ** (0.12)
INT 2							0.01 ** (0.00)	0.01 ** (0.00)
Size	−0.13 (0.09)	−0.14 (0.09)	0.49 ** (0.10)	0.50 ** (0.10)	0.48 ** (0.10)	0.49 ** (0.10)	0.51 ** (0.10)	0.49 ** (0.10)
ROA	0.10 (0.75)	0.11 (0.75)	0.04 (0.62)	0.03 (0.62)	0.05 (0.62)	0.04 (0.62)	0.02 (0.62)	0.00 (0.62)
Firm age	0.79 (0.68)	0.80 (0.68)	0.38 (0.59)	0.30 (0.58)	0.39 (0.59)	0.32 (0.58)	0.23 (0.58)	0.28 (0.58)
SOE	0.05 (0.20)	0.04 (0.20)	0.05 (0.15)	0.05 (0.15)	0.04 (0.14)	0.03 (0.14)	0.02 (0.15)	0.03 (0.14)
Management share	0.01 * (0.01)	0.01 * (0.01)	0.01 * (0.00)	0.01 * (0.00)	0.01 * (0.00)	0.01 * (0.00)	0.01 † (0.00)	0.01 † (0.00)
Intangible assets	0.11 (0.97)	0.14 (0.97)	0.72 (0.75)	0.71 (0.74)	0.77 (0.75)	0.76 (0.74)	0.69 (0.74)	0.73 (0.74)
Gender	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)	−0.00 (0.00)
Constant	0.19 (0.28)	0.22 (0.28)	−1.15 ** (0.26)	−1.17 ** (0.25)	−1.12 ** (0.25)	−1.14 ** (0.25)	−1.15 ** (0.25)	−1.13 ** (0.25)
Firm FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	11,420	11,420	11,420	11,420	11,420	11,420	11,420	11,420
Number of code	1897	1897	1897	1897	1897	1897	1897	1897
R-squared	0.18	0.18	0.18	0.19	0.19	0.20	0.20	0.20

SEs are clustered at the firm level, and SEs are reported in parentheses. ** p < 0.01, * p < 0.05, † p < 0.1.

exhibits the results of the OLS regression. Digital capability is positively linked to opportunity recognition (β = 0.46, p < 0.001; Model 1), thereby supporting Hypothesis 1. Opportunity recognition is positively related to ambidextrous innovation (β = 0.09, p < 0.001; Model 4), supporting Hypothesis 2. Furthermore,

Table 10. Study 2: Mediation test with 10,000 replications.

Effect	Coefficient	Bootstrap std. Err.	z	p	Normal-Based [95% CI]
Indirect Effect	0.39	0.03	12.68	0.00	[0.33, 0.45]
Direct Effect	0.35	0.07	4.97	0.00	[0.21, 0.49]

Note: The Indirect effect was significant if the confidence interval did not contain zero. N = 11,420. Abbreviation: CI, confidence interval.

shows that the indirect effect of opportunity recognition between digital capability and ambidextrous innovation is significant. The estimated indirect effect is 0.39 (95% CI = [0.33, 0.45]; see Table 10), supporting Hypothesis 3.

We standardized digital capability and CSR implementation to reduce potential collinearity problems [51] and multiplied them to create a new interactive term INT 1. In Table 9, the coefficient of INT 1 is significantly positive at the 5% level, indicating that CSR implementation can significantly strengthen the positive impact of digital capability on opportunity recognition (β = 0.34, p < 0.05; see Model 2), thereby supporting Hypothesis 4. We also standardized opportunity recognition and CSR implementation and multiplied them to obtain a new interactive variable INT 2. The coefficient of INT 2 is significantly positive at the 1% level (β = 0.01, p < 0.01, see model 8), indicating that CSR implementation can significantly strengthen the positive impact of opportunity recognition on ambidextrous innovation, supporting hypothesis 5.

4.4. Robustness Analysis

This study lagged the independent variables, moderators, and control variables by one year before performing OLS regression. This study adopts the method of replacing the focus variable to ensure the reliability of the findings. We directly selected the invention patents of listed companies as a substitute variable for ambidextrous innovation to test the model hypotheses. Specifically, if the first four digits of IPC numbers have appeared at least once in the previous five years, the patent is classified as exploitative; otherwise, it is considered an exploratory innovation. Moreover, we accumulated the two as a substitute variable for ambidextrous innovations. As shown in

Table 11. Study 2: Robustness test.

Variables	Opportunity Recognition		Ambidextrous Innovation
	Model 1	Model 2	Model 3	Model 4	Model 5	Mode 6	Mode 7	Mode 8
Digital capability	0.46 ** (0.15)	0.39 ** (0.14)	1.36 * (0.67)	1.31 * (0.66)	1.14 * (0.57)	1.10 † (0.56)		1.28 † (0.66)
Opportunity recognition				0.11 ** (0.04)		0.11 ** (0.03)	0.12 ** (0.04)	0.11 ** (0.03)
CSR implementation		0.00 (0.00)			−0.00 (0.01)	−0.01 (0.01)	0.01 * (0.00)	0.01 † (0.00)
INT 1		0.34 * (0.16)			1.10 * (0.54)	1.06 * (0.53)
INT 2							0.01 ** (0.00)	0.01 ** (0.00)
Size	−0.13 (0.09)	−0.14 (0.09)	0.36 * (0.17)	0.37 * (0.17)	0.34 * (0.17)	0.35 * (0.17)	0.43 * (0.21)	0.36 * (0.17)
ROA	0.10 (0.75)	0.11 (0.75)	−1.26 (1.59)	−1.27 (1.58)	−1.18 (1.56)	−1.20 (1.56)	−1.24 (1.59)	−1.29 (1.59)
Firm age	0.79 (0.68)	0.80 (0.68)	−0.32 (3.37)	−0.41 (3.39)	−0.30 (3.36)	−0.39 (3.37)	−0.64 (3.50)	−0.43 * (3.39)
SOE	0.05 (0.20)	0.04 (0.20)	−0.07 (0.22)	−0.08 (0.22)	−0.09 (0.22)	−0.10 (0.22)	−0.15 (0.22)	−0.10 (0.22)
Management share	0.01 * (0.01)	0.01 * (0.01)	0.01 ** (0.00)	0.01 ** (0.00)	0.01 ** (0.00)	0.01 ** (0.00)	0.01 ** (0.00)	0.01 ** (0.00)
Intangible assets	0.11 (0.97)	0.14 (0.97)	3.02 (2.23)	3.01 (2.22)	3.16 (2.30)	3.15 (2.29)	2.84 (2.17)	3.02 (2.23)
Gender	−0.00 (0.00)	−0.00 (0.00)	−0.01 (0.01)	−0.01 (0.01)	−0.01 (0.01)	−0.01 (0.01)	−0.01 (0.01)	−0.01 (0.01)
Constant	0.19 (0.28)	0.22 (0.28)	−0.69 (0.59)	−0.71 (0.58)	−0.66 (0.57)	−0.68 (0.57)	−0.77 (0.54)	−0.67 (0.57)
Firm FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observations	11,420	11,420	11,420	11,420	11,420	11,420	11,420	11,420
Number of code	1897	1897	1897	1897	1897	1897	1897	1897
R-squared	0.18	0.18	0.04	0.04	0.04	0.05	0.04	0.04

SEs are clustered at the firm level, and SEs are reported in parentheses. ** p < 0.01, * p < 0.05, ^† p < 0.1.

, the data results are consistent with the OLS regression mentioned above, which indicates that the data results of this model are credible. This study employed the following methods to address endogenous issues caused by missing values and dynamic panel data. First, prior to conducting OLS regression, the independent variables, moderators, and control variables were lagged by one year to avoid the impact of reverse causality.

5. Discussion

This paper investigates how enterprises develop a balance of ambidextrous innovation in response to digital technology. Grounded in resource orchestration theory [61], digital capability does not always positively influence ambidextrous innovation; its effectiveness relies on resource utilization. Based on social capital theory, this research validates the necessity of CSR implementation for firms to rationally utilize their digital capabilities and identify business opportunities that fuel the balance of ambidextrous innovation. Through two different samples, we found that when enterprises enhance their CSR efforts, the positive effect of their digital capabilities on opportunity recognition and ambidextrous innovation is amplified. We also examined how enterprises can achieve balance via the mediating mechanism of opportunity recognition. The two studies, respectively, correspond to two different enterprise samples, but they have obtained consistent data results, jointly supporting the hypotheses proposed in this paper. This further indicates that the conclusion we have drawn has specific reference value for both listed and unlisted enterprises within the territory of China. The results encourage enterprises to enhance their digital capabilities while considering CSR implementation to identify more precise and more segmented business opportunities, thereby promoting the balance of ambidextrous innovation. These insights carry significant theoretical and practical implications.

5.1. Theoretical Implications

First, we investigate the positive consequences of digital capabilities for enterprises, responding to scholars’ concerns about the negative impacts of unthinkingly using digital capabilities. Our study reveals the importance of rationally utilizing digital capabilities and broadens the path to achieving a balance in ambidextrous innovation. Prior literature often independently verified the link between digital capability and ambidextrous innovation [62]. Our study, which employed an additive index, provides a more intuitive and comprehensive view of enterprises’ ambidextrous innovation.

Second, we highlight that the positive effects of digital capability are passed to ambidextrous innovation through opportunity recognition, addressing concerns raised by Usai et al. (2021) [4]. Our empirical data support the notion that successful ambidextrous innovation relies on a firm’s ability to identify suitable opportunities for both exploitative and exploratory innovation [6]. Previous works have primarily discussed the effect of digital capability on ambidextrous innovation but have not explored the intrinsic mechanism from the lens of resource orchestration [7]. Thus, we propose a detailed framework illustrating how strong digital capabilities can achieve the balance of ambidextrous innovation through opportunity recognition through resource orchestration theory.

Third, from a reciprocal perspective, we posit that CSR implementation is an indispensable strategic resource in the digital era, thereby expanding the scope of social capital theory. Our research contributes to digital capability by exploring how CSR can strengthen the effectiveness of digital capabilities. While many scholars viewed digital capability as a driver of innovation [63], it can also overwhelm companies with excessive information [64], potentially causing confusion and a loss of strategic positioning [62]. Moreover, some literature suggests that corporations pursuing CSR and digital capabilities simultaneously could hinder innovation due to the cost-oriented attributes of these two strategies [18]. The impact of digital capability on a company can be either favorable or detrimental, contingent on the extent of CSR implementation. Companies that highly prioritize CSR enhance the merits of digital capability for ambidextrous innovation. In contrast, companies with low CSR implementation can impair a company’s independent R&D capability [6], hindering ambidextrous innovation. Our two-sample verification indicates that enterprises can achieve a balance of ambidextrous innovation through effective CSR strategies amid digital challenges.

5.2. Practical Implications

This paper provides empirical findings on how firms can more efficiently utilize digital capabilities to achieve ambidextrous innovation. Our results indicate that firms can identify more business opportunities through digital capabilities, helping organizations achieve a balance of ambidextrous innovation. Therefore, business managers should not unquestioningly believe that pursuing digital capabilities is always conducive to business innovation. Instead, managers should be more sophisticated in utilizing digital capabilities to identify new market opportunities and promote an ambidextrous innovation equilibrium.

Second, according to resource orchestration theory, this paper demonstrates how firms arrange their resources to achieve organizational goals. It suggests that an organization’s decision-makers should focus on its resources and then understand how the firm can utilize them to gain core competitiveness.

Third, this paper utilizes empirical results to demonstrate that the CSR strategy holds irreplaceable strategic significance in the digital era. Fulfilling the CSR strategy is not only an important initiative to improve corporate reputation but also helps enterprises better recognize their market position in the current stage, clarify their stakeholders’ needs, market trends, and social expectations, and serve as a facilitator in applying their digital capabilities to identify opportunities to achieve a balance of ambidextrous innovation.

5.3. Policy Implications

This paper demonstrates the positive impact of digital capabilities and CSR implementation on the opportunity recognition and the realization of ambidextrous innovation, which also has some reference value for government departments in formulating policies. Firstly, it is essential to assist enterprises in enhancing their digital capabilities and strengthening the development of digital infrastructure. Relevant government departments should continue to build high-speed, secure, and inclusive digital infrastructure, provide infrastructure support for enterprises [65], and introduce favorable policies conducive to enterprises’ participation in the digitalization process. This is particularly important for enterprises with scarce resources and funds. Secondly, government departments should pay attention to the publicity of the importance of CSR, emphasize the significance of CSR implementation for enterprise innovation and sustainable development in the digital era, and formulate corresponding favorable policies to encourage enterprises to take the initiative to fulfill social responsibilities. Finally, government departments should focus on the cultivation and introduction of digital talents, introduce corresponding policies, reform the curriculum system of colleges and universities, and cultivate compound talents with both digital skills and a sense of social responsibility.

5.4. Limitations and Future Directions

However, our paper also suffers from some limitations. First, the samples from Study 1 and Study 2 are limited to China, which lacks different cultural contexts. Future research can add samples from Western cultural contexts to improve external validity. Second, given the data availability and mainstream literature, we use the frequency of keywords from the annual reports to measure digital capability. In the future, there can be more diverse approaches to measure digital capabilities. Third, while we focused on the mediating role of opportunity recognition, other variables may also contribute to balancing ambidextrous innovation with digital capability. For example, the choice of a firm’s competitive strategy (differentiated competitive strategy and cost-competitive strategy) may also interpret the effect of digital capability on ambidextrous innovation [59]. Further studies can concentrate on the role of firms’ differentiated competitive and cost-competitive strategies in bridging the impact of digital capability on ambidextrous innovation. Finally, future research can also focus on the moral and ethical issues that enterprises face when they overly pursue digital capabilities, as well as the harm it causes to companies’ sustainable development [66].

6. Conclusions

This study provides empirical evidence confirming that the positive effect of a firm’s digital capabilities on ambidextrous innovation is mediated by opportunity recognition. Most importantly, this effect varies based on the company’s level of CSR implementation. The findings offer practical solutions for enterprises seeking to balance ambidextrous innovation by effectively allocating resources to optimize their digital capabilities. These conclusions are significant for several reasons. Resource orchestration theory underlines that effectively utilizing resources is more crucial than merely owning them; the efficient application of digital capabilities is essential for businesses to advance with changes. CSR implementation promotes sustainable development, which aligns with the reciprocal principle in social capital theory. These findings underscore the need for organizations to prioritize CSR in the digital age.