How Does TMT Heterogeneity Affect Firm Digital Innovation Resilience?

Abstract

In the digital economy era characterized by heightened uncertainty, strengthening internal governance to bolster firm adaptability and sustain digital innovation resilience has become crucial. As a key strategic resource, top management team (TMT) heterogeneity holds significant theoretical and practical value for enhancing firms’ digital innovation resilience. Using a sample of Chinese manufacturing listed firms, this study examines how TMT heterogeneity affects digital innovation resilience and the underlying mechanisms. The findings indicate that: (1) Greater TMT heterogeneity strengthens firms’ digital innovation resilience. (2) This effect operates primarily through alleviating financing constraints and improving investment efficiency. (3) The impact varies across firm types: it is stronger for small and medium-sized firms than for large firms; more pronounced in state-owned firms than in non-state-owned firms; more significant in low-monopoly firms than in high-monopoly firms; and notably greater for firms in eastern China than for those in central and western regions. (4) Government subsidies can strengthen the positive impact of TMT heterogeneity on firm digital innovation resilience. This study provides theoretical insights and practical guidance for enterprises to build effective TMTs, alleviate financing constraints, and improve investment efficiency, and for the government to provide subsidies, with the ultimate aim of fostering digital innovation resilience.

1. Introduction

The deepening of the digital economy has fundamentally reshaped the process of technological innovation within firms, altering the endowment of innovative factors and the availability of information and innovation resources. Traditional innovation models are increasingly inadequate to meet the demands of modern society and rapid economic development [1]. At the same time, the competitive landscape is shifting from a relatively stable “red ocean” to a complex, volatile, and uncertain “digital deep sea.” In this context, digital innovation has become a critical pillar for building core competitiveness [2]. However, these technological opportunities are accompanied by significant challenges. The emergence of disruptive technologies, the rapid iteration of market demands, and fluctuations in global supply chains expose firms’ digital innovation activities to considerable internal, external shocks and disruptions [3]. Consequently, a key question arises: in the face of internal and external shocks, can firms maintain the stability of their digital innovation processes, demonstrate recovery capacity, and even evolve toward higher levels of shock resistance? This ability—to achieve digital innovation resilience through self-learning, adaptation, and resistance—is crucial not only for the survival and development of firms in the digital age but also for promoting the healthy, stable, and sustainable development of the national economy and shaping new competitive advantages in the global arena.

Digital innovation is ultimately driven by human agency. A firm’s capacity for digital innovation—including its willingness to innovate, responsiveness to complex environments, and overall resilience—is fundamentally shaped by human decisions. In today’s intricate and competitive landscape, innovation can no longer depend on the insight of individual managers. Instead, the top management team (TMT) has become the central locus of strategic control, collectively shaping firm strategy and guiding the firm’s digital innovation trajectory [4]. TMT members vary in attributes such as career background, tenure, gender, and expertise [5], meaning that a team’s impact on digital innovation resilience depends significantly on its heterogeneity. When members collaborate effectively, a heterogeneous TMT can provide richer experience, diverse cognitive perspectives, and more robust decision-making [6]. The interaction of differing values, logic, and behavioral styles may strengthen the team’s overall risk resilience [7], helping the firm better identify risks, seize opportunities, and make innovative yet balanced strategic choices—thereby enhancing investment efficiency [8]. Moreover, the network resources embedded in a diverse TMT can improve access to external financing, alleviate financial constraints, and help the firm withstand shocks while sustaining innovation [9]. Conversely, heterogeneity in attributes such as age or tenure may foster divergent beliefs and erode trust, leading to factional conflict. This can raise the cost of innovation-related decision-making and coordination [10], result in missed opportunities, and lower investment efficiency [11]. Communication barriers within the TMT may also signal high risk to external investors, undermining their trust and willingness to provide financing [12], which tightens financial constraints and weakens digital innovation resilience. Furthermore, digital innovation carries not only the inherent high risk, investment, and uncertainty of traditional innovation, but also new features such as rapid iteration, high data intensity, and strong ecosystem dependence. As a result, it requires even more stable financial support than traditional innovation [13]. Therefore, when TMT heterogeneity influences firm digital innovation resilience, direct financial support from the government—such as subsidies [14]—may be necessary to positively foster such resilience. Consequently, examining how TMT heterogeneity influences digital innovation resilience, how firms can harness it to build effective leadership, and how government subsidies moderate this relationship represents a timely and important research imperative.

While existing literature offers valuable insights, several research gaps remain. First, most studies focus on the impact of individual TMT demographic attributes—such as gender, age, education, functional background, tenure, and risk preference—or their combined effects on firms’ innovation inputs or outputs, yet findings remain inconsistent [4,5,7,8,15,16,17,18,19]. Second, although scholars have examined links between TMT cognitive foundations, social networks, internal stability, diversity, education, overseas experience, or professional background and digital innovation, conclusions are similarly mixed [13,20,21,22,23,24,25]. Furthermore, a limited yet growing body of re-search examines how TMT characteristics influence organizational or innovation resilience, generally suggesting that certain TMT attributes can enhance such resilience [14,26,27]. Third, most studies focus on the direct impact of government subsidies on firm innovation, yet their findings remain inconsistent [28,29,30,31]. Despite these contributions, three critical research gaps remain salient. First, prior studies have not adequately addressed how TMT heterogeneity specifically shapes digital innovation resilience in the context of the digital economy. Second, while TMT heterogeneity ultimately operates through firm-level resource allocation—with investment behavior being a core manifestation—the potential mediating role of investment efficiency (reflecting the effectiveness of capital allocation) remains unexplored. Similarly, financing constraints, a key inhibitor of digital innovation activities and thus of innovation resilience, may also serve as a critical mediating mechanism. Yet, existing literature has not systematically examined the internal pathways through which TMT heterogeneity affects digital innovation resilience via these two channels. Third, the impact of government subsidies on the relationship between TMT heterogeneity and firm digital innovation resilience has not been examined in existing research.

Based on the above analysis and drawing on upper echelons theory, organizational resilience theory and government intervention theory, this study proposes a theoretical model with eight hypotheses to examine how TMT heterogeneity influences digital innovation resilience and the underlying mechanisms involved, and the moderating role of government subsidies in the relationship between TMT heterogeneity and firm digital innovation resilience. Specifically, it addresses three research questions: (1) How does TMT heterogeneity affect a firm’s digital innovation resilience? (2) Does TMT heterogeneity influence digital innovation resilience through the mediating channels of investment efficiency and financing constraints? (3) What role do government subsidies play between TMT heterogeneity and firm digital innovation resilience? To examine the proposed theoretical model and address the research questions, this study employs a sample of Chinese manufacturing listed firms and conducts multiple regression analyses. Robustness checks and endogeneity tests are performed to ensure the reliability of the findings.

This paper makes contributions primarily in the following aspects. First, research content contribution. Prior research has seldom addressed the issue of firm digital innovation resilience, nor has it considered the impact of TMT heterogeneity on such resilience [4,5,7,8,13,15,16,17,18,19,20,21,22,23,24,25]. This study reveals the positive effect of TMT heterogeneity on firm digital innovation resilience; it helps enrich research on firm digital innovation resilience within the upper echelons theory and it assists firms in maintaining the stability of digital innovation and enhancing their digital innovation resilience. Second, mechanism research contribution. Existing studies have mainly focused on the direct impact of TMT characteristics on organizational or innovation resilience [14,26,27], without examining the role played by investment efficiency and financing constraints in the relationship between TMT heterogeneity and firm digital innovation resilience. This paper uncovers the mediating role of investment efficiency and financing constraints between TMT heterogeneity and firm digital innovation resilience, it contributes to expanding the application of organizational resilience theory in the field of firm innovation management and it helps firms improve investment efficiency, reduce financing constraints, and thereby further strengthen their digital innovation resilience by building a well-composed TMT. Third, research perspective contribution. While existing research has primarily examined how government subsidies influence firm innovation [28,29,30,31], scant attention has been paid to their moderating role in the relationship between TMT heterogeneity and firm digital innovation resilience. This study reveals that government subsidies can positively moderate the enhancing effect of TMT heterogeneity on digital innovation resilience. This finding helps extend the application of government intervention theory to the field of firm innovation management. It provides insights for the government to strengthen, through subsidies, the positive impact of TMT heterogeneity on digital innovation resilience, thereby further augmenting such resilience.

The remainder of this paper is structured as follows. Section 2 presents the theoretical analysis and research hypotheses. Section 3 introduces the research design, including sample selection and data sources, key variables, and model specification. Section 4 reports the empirical results, covering the impact of TMT heterogeneity on firm digital innovation resilience, robustness and endogeneity tests, heterogeneity analysis, and mechanism analysis. Section 5 discusses the differences between the findings of this study and prior research. Section 6 summarizes the conclusions and limitations, and suggests directions for future research.

2. Theoretical Analysis and Research Hypotheses

2.1. The Impact of TMT Heterogeneity on Digital Innovation Resilience in Firms

Upper echelons theory posits that the TMT is the group of senior executives vested with the authority for firm operational and strategic decision-making, responsible for coordinating internal and external resources and formulating the firm’s future development strategy [7]. The heterogeneity in the TMT’s demographic backgrounds influences its cognitive base and values, thereby shaping the firm’s overall strategic choices [4]. The decision to undertake digital innovation, in particular, is a collective outcome reached by the TMT through information integration, resolution of cognitive conflicts, and alignment of interests, based on the firm’s internal and external context. The strength of its digital innovation intention is closely linked to heterogeneous characteristics such as managers’ gender, age, education, professional background, overseas experience, and financial expertise [20].

Consequently, when confronted with internal and external shocks—such as changes in firm governance, managerial competence, or market and institutional conditions—the ability of a firm to maintain stable digital innovation through self-learning, adaptation, and resistance, thereby possessing the resilience to recover or even evolve to a higher level of shock-absorption capacity [26] and maintain sound digital innovation resilience, is influenced by TMT heterogeneity. If heterogeneous TMT members achieve effective communication enabling knowledge, information, and resource sharing, they can fully leverage the diverse experiences, technical skills, and market insights within the team. This provides more comprehensive information and knowledge for innovation decision-making [32], which not only reduces firm short-sighted opportunism and enhances the overall risk-taking capacity of the TMT [7] but also enables the firm to better identify risks, more accurately recognize opportunities in the digital innovation process, and stimulate vitality in digital innovation [13]. Especially when a firm faces digital innovation failure and subsequent recovery due to complex internal and external shocks, the diverse knowledge, skills, and experience of TMT members facilitate consensus-building in subsequent innovation decisions and offer varied, non-standardized solutions to complex problems [25]. This enhances the firm’s capacity to effectively resist and recover from various shocks [27], maintains the stability of its digital innovation, and ultimately strengthens its digital innovation resilience.

Conversely, if demographic heterogeneity leads to irreconcilable differences in perspectives among TMT members, it may result in a lack of cohesion and a crisis of trust. Members may become more focused on protecting their own subgroup interests, leading to fragmentation and factional opposition [33]. This impedes the free flow of critical digital innovation information and resources, increases the cost of decision-making and coordination, severely weakens the firm’s overall adaptability and collaborative capacity, and makes it difficult for the firm to seize digital innovation opportunities promptly in response to environmental changes [23]. Consequently, when subjected to complex shocks, the firm lacks the capacity to resist and recover, struggles to maintain stable digital innovation, and thus experiences weakened digital innovation resilience.

Based on the preceding analysis, this paper proposes the following competing hypotheses:

H1a: 

TMT heterogeneity enhances firm digital innovation resilience.

H1b: 

TMT heterogeneity undermines firm digital innovation resilience.

2.2. The Intermediary Effect of Financing Constraints

Financing constraints constitute a key limiting factor in a firm’s ability to maintain stable digital innovation [34] and significantly influence whether the firm can develop sufficient resilience to withstand and recover from complex internal and external shocks—that is, to possess adequate digital innovation resilience. The collective cognitive base, social relationships, and network resources of the TMT directly affect the efficiency and effectiveness with which a firm accesses and utilizes external financing channels, thereby shaping its financing constraints [9] and, in turn, influencing its digital innovation resilience.

From a positive perspective, heterogeneity in TMT members’ educational, professional, financial, and overseas backgrounds can generate an information effect for the firm, broadening its external information sources and channels [35]. This improves the timeliness and relevance of information acquisition, effectively mitigating information asymmetry both inside and outside the firm [15]. By providing a diversified cognitive foundation and information channels for strategic decision-making, it helps reduce financing difficulties [16], increase loan scale, lower financing costs, and alleviate financing constraints [17]. Simultaneously, TMT diversity can curb managerial short-term opportunism aimed at pursuing immediate performance growth [19], enhancing the firm’s ability to accurately identify the value of R&D talent, funding, and innovation projects, thereby strengthening innovation capability and competitive advantage [24], and send positive signals to the market to attract external investor attention and funding support, further easing financing constraints [36]. Together, these mechanisms provide sufficient financial assurance for digital innovation activities, enabling the firm to develop the capacity to resist and recover from complex shocks and ultimately enhancing its digital innovation resilience.

From a negative perspective, heterogeneous TMT members may also categorize themselves and others based on demographic traits such as age, gender, and education, showing preference for similar “in-group” members. This tendency often leads to subgroup formation and identity-based categorization within heterogeneous teams, increasing team conflict and communication costs, and resulting in a lack of cohesion or shared vision among TMT members [33]. Such dynamics can also erode external investors’ trust in the TMT, reducing their willingness to provide funding and exacerbating the firm’s financing constraints [18]. As a result, the firm may lack adequate financial support during its digital innovation process, struggle to respond effectively to complex internal and external shocks, find it difficult to sustain stable digital innovation, and ultimately see its digital innovation resilience diminished.

Based on the preceding analysis, this paper proposes the following competing hypotheses:

H2a: 

TMT heterogeneity alleviates firm financing constraints, thereby enhancing digital innovation resilience.

H2b: 

TMT heterogeneity exacerbates firm financing constraints, thereby undermining digital innovation resilience.

2.3. The Intermediary Effect of Investment Efficiency

Existing research indicates that firm strategic decisions are largely explained by the heterogeneous background characteristics of the TMT, as these characteristics reflect its overall cognitive base, values, and insights [22]. As one of the most important strategic decisions, investment efficiency is also profoundly influenced by the background features of a heterogeneous TMT [11]. This influence manifests primarily in the processes of identifying, evaluating, and selecting investment opportunities, ultimately leading to differences in firm investment efficiency [37]. Investment efficiency, as a core reflection of a firm’s resource allocation capability, serves as the material foundation and practical guarantee for digital innovation resilience. When facing complex internal and external shocks, the level of investment efficiency directly determines the survival and adaptability of a firm’s digital innovation system [3].

From a positive perspective, if heterogeneous TMT members—varying in gender, age, education, professional background, overseas experience, and financial expertise—can engage in communication, cooperation, and knowledge sharing, they can fully leverage their respective experiences, technical skills, and market sensing abilities. This helps effectively mitigate the tendency of the team as a whole to be either overly risk-taking or overly conservative in the investment process [4], enhancing the firm’s sensitivity to environmental changes such as market trends, technological shifts, and policy adjustments [38], and promoting more rational investment decisions at the TMT level. Consequently, it restrains overinvestment, reduces blind investment, lowers the probability of innovation project failure, effectively improves the quality of investment decisions [6], and enhances investment efficiency [10]. This provides solid support for digital innovation activities, enabling the firm to better respond to complex internal and external shocks, develop strong resistance and recovery capacities, maintain the stability of digital innovation, and thereby promote digital innovation resilience.

Conversely, if significant differences in professional goals, values, and cognitive models among heterogeneous TMT members lead to interpersonal conflicts and communication barriers, misunderstandings, biases, and distrust can arise within the team. This severely undermines team cohesion, resulting in delayed investment decisions due to an inability to reach consensus and missed opportunities for innovative investments [12]. Team members may also prioritize their own interests over the collective goals of the team or the firm [11], aggravating overinvestment and wasting R&D resources, thereby intensifying inefficient investment behavior [10]. As a result, the firm lacks effective resource guarantees during its digital innovation process, struggles to cope with complex internal and external shocks, and fails to develop the capacity to resist and recover from disruptions. This inability to maintain stable digital innovation ultimately weakens the firm’s digital innovation resilience.

Based on the preceding analysis, this paper proposes the following competing hypotheses:

H3a: 

TMT heterogeneity improves firm investment efficiency, thereby enhancing digital innovation resilience.

H3b: 

TMT heterogeneity reduces firm investment efficiency, thereby undermining digital innovation resilience.

2.4. The Moderating Effect of Government Subsidies

Digital innovation not only inherits the inherent characteristics of traditional innovation—such as high risk, substantial investment, and significant uncertainty [39]—but also exhibits new attributes, including rapid iteration, high data-factor intensity, and strong ecosystem dependence. These features make digital innovation even more dependent on stable financial support than traditional innovation. Consequently, when TMT heterogeneity influences corporate digital innovation resilience, the guidance and support of an effective government through sound policies become essential to foster stronger resilience. Government intervention theory posits that subsidies serve as a key mechanism to address market failures and innovation deficits [28]. Thus, subsidies can be a lever for governments to exert influence in the relationship between TMT heterogeneity and digital innovation resilience.

Proponents argue that subsidies provide direct financial support, lowering innovation risks [29]. Moreover, they signal a firm’s technological strength, financial health, and strong government ties, helping attract external investors, ease financing constraints, and gain innovation opportunities and social capital [30]. In this view, subsidies play a positive, constructive role in enhancing the effect of TMT heterogeneity on resilience.

Critics, however, contend that subsidies can undermine firms’ intrinsic motivation for innovation, fostering a path dependency where funds are used for cash flow rather than R&D [40]. Information asymmetry may also trigger adverse-selection behaviors: firms might manipulate innovation metrics, engage in strategic or “subsidy-seeking” innovation, or pursue political connections for preferential support [34,41]. From this perspective, subsidies can have a negative or counterproductive effect when TMT heterogeneity influences digital innovation resilience.

Based on the preceding analysis, this paper proposes the following competing hypotheses:

H4a: 

Government subsidies have a positive moderating effect on the heterogeneity of TMT and the resilience of digital innovation in firms.

H4b: 

Government subsidies have a negative moderating effect on the heterogeneity of TMT and the resilience of digital innovation in firms.

3. Research Design

3.1. Sample Selection and Data Sources

Given variations in the listing timelines of listed firms, and to mitigate “survivorship bias” that may arise from pursuing a balanced panel, this study retains all available observations to the greatest extent possible. Following China’s Guidelines for Industry Classification of Listed Companies (2012 Revised Edition), manufacturing listed firms are selected as the research sample. An unbalanced panel dataset of manufacturing listed firms from 2015 to 2024 is constructed, and the data are processed as follows:

First, listed firms designated as ST, PT, *ST during the research period, as well as those with abnormal relevant data, are excluded.

Second, by integrating the CSMAR and CNRDS databases, listed firms’ annual reports, and the patent information database of the China National Intellectual Property Administration, data on the number of digital technology patent applications are manually collected based on listed firms’ stock codes.

Third, data on the characteristics of firm executives are primarily sourced from the CSMAR and WIND databases, supplemented by searches and additions from Sina Finance and firm annual reports to address missing executive characteristic data.

The remaining data were sourced from databases such as CSMAR and WIND, ultimately yielding 23,804 valid research samples from 3614 non-overlapping and distinct groups of listed enterprises. Among these, 1190 enterprises were listed continuously from 2015 to 2024, 1701 were listed continuously from 2015 to 2020, and 723 were listed continuously from 2021 to 2024.

3.2. Variable Definition

(1) Firm digital innovation resilience ($\mathit{Digit}$). This paper posits that digital innovation resilience refers to a firm’s ability, when confronted with internal and external environmental shocks—such as changes in firm governance structure, managerial competence, and market or institutional conditions—to maintain the stability of its digital innovation, recover, and even evolve to a higher level of shock resistance through self-learning, adaptation, and resistance. This is primarily reflected in the relative stability and recoverability of a firm’s digital innovation patent applications [14,42,43,44]. Prior research has primarily adopted two approaches to measure resilience: the composite indicator method and the core variable method. The composite indicator method involves dividing resilience into multiple dimensions, selecting corresponding indicators, and constructing an evaluation index system. Scholars often select indicators from dimensions such as resistance and recovery capability, adaptation and adjustment capability, and transformation and development capability to build a resilience evaluation index system, typically measured using the entropy weight method [44]. In contrast, the core variable method measures resilience by selecting key variables that reflect the degree of change in operational or market indicators most affected by shocks [14,44]. Since the core variable method measures resilience by selecting the most shock-sensitive core variables, it has been widely adopted by most scholars. Furthermore, the number of digital technology patents filed by a firm is closely tied to the R&D environment of the given year and is sensitive to internal and external shocks. This sensitivity captures the firm’s susceptibility to disruptions (lack of absorption) in its digital innovation patent applications and the speed at which it recovers to resume filing them. Therefore, drawing on the approach of Martin [42] and Luo et al. [43], this paper also employs the core variable method—specifically, the sensitivity-based core variable approach—using the number of digital technology patent applications volatility as the key proxy indicator. Digital innovation resilience is measured by changes in the volume of digital technology patent applications. The specific calculation method is as follows:

$${\mathit{Digit}}_i^t=(\Delta Y_i-\Delta E)/|\Delta E|$$

(1)

$$\Delta Y_i=Y_i^t-Y_i^{t-1}$$

(2)

$$\Delta E=\left(\right(Y_{\mathrm{r}}^t-Y_{\mathrm{r}}^{t-1})/Y_{\mathrm{r}}^{t-1})Y_i^{t-1}$$

(3)

where ${\mathit{Digit}}_i^t$ represents the relative digital innovation resilience of firm i in year t; $\Delta Y_i$ denotes the actual change in the number of digital technology patent applications filed by firm i from year t − 1 to year t; $\Delta E$ indicates the predicted number of digital technology patent applications for firm i from year t − 1 to year t, estimated based on the overall change in digital technology patent applications in the province where the firm is located; $Y_i^t$ and $Y_i^{t-1}$ represent the number of digital technology patent applications filed by firm i in year t and year t − 1, respectively; and $Y_{\mathrm{r}}^t$ and $Y_{\mathrm{r}}^{t-1}$ represent the total number of digital technology patent applications in the province where the firm is located in year t and year t − 1, respectively.

The resulting value for corporate digital innovation resilience ($\mathit{Digit}$) is a relative indicator with clear economic significance. The magnitude and sign of its value directly reflect a firm’s differentiated response to common shocks at the provincial level. The core theoretical premise of this measurement design is to treat the annual fluctuation in the total volume of provincial digital innovation patents as an effective “proxy variable for common shocks” [14,31,36,42,43,44]. This regional trend captures the general pressure exerted on most firms within the province by factors such as macro policies, economic cycles, and regional market turbulence. Therefore, a firm’s output deviation from this benchmark does not measure its unconditional, absolute innovation “performance” level, but rather characterizes the “conditional robustness” of its innovation system when responding to widespread external pressure—that is, under the same circumstances, some firms exhibit stability (high resilience), while others experience significant volatility (low resilience). This method (the core variable method) precisely infers and measures the firm’s intrinsic, difficult-to-directly observe resilience capability by observing the sensitive response trajectory of a core output indicator to common shocks. To enable comparative analysis across all firms, the calculation results are centered, and Model (4) is constructed as follows:

$$R_i=({\mathit{Digit}}_i^t-{\displaystyle \sum _i^{\mathrm{n}}{\mathit{Digit}}_i^t/n})$$

(4)

Among them, $R_i$ can be directly used to compare the digital innovation resilience of all firms.

When $R_i>0$, it indicates that the digital innovation resilience of firm i exceeds the average of the other firms. A larger value signifies better performance of the firm’s digital innovation resilience relative to the overall provincial context.

When $R_i<0$, it indicates that the digital innovation resilience of firm i falls below the average of the other firms. A smaller value reflects poorer performance of the firm’s digital innovation resilience within the overall provincial context.

(2) TMT heterogeneity ($\mathit{Manager}$). This paper defines TMT heterogeneity as the diversity across multiple dimensions—including age, gender, education, tenure, professional background, overseas experience, and financial background—within the group of senior executives (such as the chairman, vice chairman, general manager, and deputy general manager of a listed firm) who are primarily responsible for allocating internal and external resources, hold decision-making authority over firm operations and management, and formulate the firm’s future development strategy; they are also tasked with responding to internal and external shocks, thereby shaping the governance of the firm’s digital innovation resilience in the face of such challenges. In existing research, the measurement of TMT heterogeneity has not been standardized, with two main approaches commonly adopted. First, heterogeneity is measured separately across three dimensions: education, tenure, and age. Second, a multidimensional index system is constructed that includes gender, age, education, profession, overseas background, and financial background. For each dimension, individual heterogeneity values are calculated using either the Herfindahl–Hirschman Index or the coefficient of variation. These values are then normalized, and an overall TMT heterogeneity index is derived by averaging the normalized scores [45,46]. A higher index value indicates a greater degree of heterogeneity in categorical variables. This second approach is widely used in the literature. Therefore, this study also adopts the second method to calculate the TMT heterogeneity index, which serves as the proxy variable for TMT heterogeneity. The specific calculation procedure is as follows:

First, heterogeneity in age, gender, education, tenure, professional background, overseas background, and financial background is measured separately.

Specifically, heterogeneity in age and tenure is measured by the coefficient of variation (the ratio of the standard deviation to the mean). Heterogeneity in gender, education, professional background, overseas background, and financial background is calculated using the Herfindahl–Hirschman Index. The calculation method is as follows:

$$H=1-{\displaystyle \sum _{i=1}^{\mathrm{n}}{P}_{\mathrm{ijt}}^2}$$

(5)

In Equation (5), Pijt is the proportion of the i-th member in the TMT of firm j in the t-th year, and the range of this index is [0, 1]. The closer the value is to 1, the higher the heterogeneity of team members in a specific attribute, and the closer the value is to 0, the lower the heterogeneity.

Second, the individual heterogeneity scores obtained above are normalized, and then averaged to derive an overall index of TMT heterogeneity. A higher value of this index represents a greater degree of TMT heterogeneity.

(3) Financing constraints ($\mathit{Sa}$). This paper defines financing constraints as a phenomenon in which significant disparities between the costs of internal and external financing—arising from information asymmetry within and outside the firm, principal–agent problems between managers and shareholders or creditors, and transaction costs such as auditing, underwriting, and legal consulting fees incurred during external financing—restrict the firm’s investment capacity. Existing research primarily employs three methods to measure financing constraints. The first is the KZ index, which integrates multiple financial indicators such as cash flow and leverage into a composite index through ranking or modeling; a higher value indicates stronger constraints [47]. The second is the WW index, constructed via a structural model that includes variables like dividend payout and cash flow [48]. The third is the SA index, which is calculated based on exogenous and less manipulable indicators such as firm size and age [49]. Because the first two approaches incorporate many endogenous variables (e.g., cash flow, financial leverage, Tobin’s Q) that are themselves influenced by the degree of financing constraints, and because the SA index excludes such endogenous variables and reduces the risk of subjective manipulation of core indicators, it offers stronger operability and objectivity. Therefore, to mitigate endogeneity concerns, this study follows Hadlock and Pierce [49] and adopts the SA index to measure firm financing constraints. The SA index takes negative values. The smaller its absolute value (the larger the numerical value), the higher the degree of financial constraints faced by the firm. Conversely, the larger its absolute value (the smaller the numerical value), the lower the degree of financial constraints. The specific calculation formula is as follows:

$$Sa=-0.737size+0.043siz{e}^2-0.04age$$

(6)

Among them, size = ln (total assets of the firm/100,000), total assets unit is yuan, age is the age of the firm, and the index is calculated from this.

(4) Investment efficiency ($\mathit{Resid}$). This paper defines investment efficiency as the degree of deviation between a firm’s actual investment and its optimal level of investment. This deviation reflects the state of the firm’s investment activities: the smaller the deviation, the closer the investment is to the optimal state, and thus the higher the investment efficiency. In existing research, most scholars adopt the Richardson model to measure firm investment inefficiency [50]. This indicator is negative-oriented; a larger absolute value indicates a higher degree of inefficient investment. Therefore, this study follows the same approach to measure firm investment efficiency. The specific calculation formula is as follows:

$$\begin{array}{l}I{\mathit{nv}}_{(i,t)}={\alpha }_0+{\alpha }_1{\mathit{Tobin}}_{(i,t-1)}+{\alpha }_2{\mathit{Debt}}_{(i,t-1)}+{\alpha }_3{\mathit{Cash}}_{(i,t-1)}+{\alpha }_4{\mathit{Age}}_{(i,t-1)}\\ +{\alpha }_5{\mathit{Size}}_{(i,t-1)}+{\alpha }_6{\mathit{Return}}_{(i,t-1)}+\sum Industry+\sum Year+{\mu }_{(i,t)}\end{array}$$

(7)

Among them, $\mathit{Inv}$ denotes firm investment, measured as the ratio of cash flow from investing activities to total assets at the beginning of the period. $Tobin$ represents Tobin’s Q, defined as the ratio of the market value of the firm’s assets to their replacement cost. $Debt$ is the asset-liability ratio, calculated as total liabilities divided by total assets. $Cash$ indicates the firm cash flow utilization rate, measured as the ratio of net cash flow from operating activities to total assets. $\mathit{Age}$ stands for the firm’s listing age, computed as the natural logarithm of (current year − listing year). $\mathit{Size}$ reflects firm size, expressed as the natural logarithm of total assets. $\mathit{Return}$ is the stock return, measured as earnings divided by the initial investment. $\mathit{Industry}$ denotes the industry to which the firm belongs. Based on these definitions, a fixed-effects model is employed for the regression analysis. The absolute value of the residuals obtained from this regression serves as the proxy for firm investment inefficiency. A larger absolute value indicates a greater degree of investment inefficiency, implying lower firm investment efficiency.

(5) Government subsidies ($\mathit{Subsidy}$). This paper defines government subsidies as the monetary or non-monetary assets obtained by enterprises from the government without compensation. Existing studies have employed various methods to measure government subsidies, primarily including: the natural logarithm of the total amount of government subsidies [28], the ratio of total government R&D subsidies to operating revenue [31], and the ratio of the total amount of government subsidies to total assets [39]. Considering data availability and following the approach of Luo and Wang [28], this paper measures government subsidies using the natural logarithm of the amount of subsidies received by a firm in a given year.

(6) Control variables. Based on existing literature, this paper defines the following control variables: firm size ($\mathit{Size}$, measured as the natural logarithm of total assets at year-end), asset–liability ratio ($Debt$, total liabilities divided by total assets), core business profit margin ($Profit$, operating profit divided by operating revenue), cash flow utilization rate ($Cash$, net cash flow from operating activities divided by total assets), proportion of independent directors ($\mathit{Indep}$, number of independent directors divided by board size), shareholding ratio of the largest shareholder ($Top$, shares held by the largest shareholder divided by total shares outstanding), separation ratio between control and ownership rights ($Right$, difference between the actual controller’s voting rights and cash-flow rights in the listed firm), firm growth ($\mathit{Grow}$, growth rate of operating revenue), year ($\mathit{Year}$) and firm individual ($\mathit{Stock}$).

3.3. Model Construction

3.3.1. Benchmark Regression Model

In accordance with the research Hypotheses H1a and H1b, the following benchmark regression model is established to examine the impact of TMT heterogeneity on firm digital innovation resilience:

$$\begin{array}{l}{\mathit{Digt}}_{(i,t)}=\alpha +{\beta }_1{\mathit{Manager}}_{(i,t)}+{\beta }_2{\mathit{Size}}_{(i,t)}+{\beta }_3{\mathit{Debt}}_{(i,t)}+{\beta }_4{\mathit{Profit}}_{(i,t)}+{\beta }_5{\mathit{Cash}}_{(i,t)}\\ +{\beta }_6{\mathit{Indep}}_{(i,t)}+{\beta }_7{\mathit{Top}}_{(i,t)}+{\beta }_8{\mathit{Right}}_{(i,t)}+{\beta }_9{\mathit{Grow}}_{(i,t)}+\mathit{Year}+\mathit{Stock}+{\mu }_{(i,t)}\end{array}$$

(8)

Among them, $Digi{t}_{(i,t)}$ and $Manage{r}_{(i,t)}$ represent the digital innovation resilience and the level of TMT heterogeneity of sample firm i in period t, respectively. The remaining variables are control variables. $\mathit{Year}$ and $\mathit{Stock}$ denote the year and firm individual fixed effects. α is the constant term, β the regression coefficient, and μ the random disturbance term.

3.3.2. Mediating Effect Model

In accordance with the research Hypotheses H2a, H2b, H3a, and H3b, and to examine the mechanistic pathways through which TMT heterogeneity influences firm digital innovation resilience by affecting financing constraints and investment efficiency, this study draws on the approach of Zhang et al. [3]. First, the following regression models are constructed to analyze the relationships between TMT heterogeneity and firm financing constraints and investment efficiency, respectively:

$$\begin{array}{l}{\mathit{Sa}}_{(i,t)}=\alpha +{\beta }_1{\mathit{Manager}}_{(i,t)}+{\beta }_2{\mathit{Size}}_{(i,t)}+{\beta }_3{\mathit{Debt}}_{(i,t)}+{\beta }_4{\mathit{Profit}}_{(i,t)}+{\beta }_5{\mathit{Cash}}_{(i,t)}\\ +{\beta }_6{\mathit{Indep}}_{(i,t)}+{\beta }_7{\mathit{Top}}_{(i,t)}+{\beta }_8{\mathit{Right}}_{(i,t)}+{\beta }_9{\mathit{Grow}}_{(i,t)}+Y\mathrm{ear}+S\mathrm{tock}+{\mu }_{(i,t)}\end{array}$$

(9)

$$\begin{array}{l}{\mathit{Resid}}_{(i,t)}=\alpha +{\beta }_1{\mathit{Manager}}_{(i,t)}+{\beta }_2{\mathit{Size}}_{(i,t)}+{\beta }_3{\mathit{Debt}}_{(i,t)}+{\beta }_4{\mathit{Profit}}_{(i,t)}+{\beta }_5{\mathit{Cash}}_{(i,t)}\\ +{\beta }_6{\mathit{Indep}}_{(i,t)}+{\beta }_7{\mathit{Top}}_{(i,t)}+{\beta }_8{\mathit{Right}}_{(i,t)}+{\beta }_9{\mathit{Grow}}_{(i,t)}+\mathit{Year}+\mathit{Stock}+{\mu }_{(i,t)}\end{array}$$

(10)

Subsequently, regression models incorporating financing constraints and investment efficiency as mediating variables are constructed, respectively:

$$\begin{array}{l}{\mathit{Digt}}_{(i,t)}=\alpha +{\beta }_1{\mathit{Manager}}_{(i,t)}+{\beta }_2{\mathit{Sa}}_{(i,t)}+{\beta }_3{\mathit{Size}}_{(i,t)}+{\beta }_4{\mathit{Debt}}_{(i,t)}+{\beta }_5{\mathit{Profit}}_{(i,t)}+{\beta }_6{\mathit{Cash}}_{(i,t)}\\ +{\beta }_7{\mathit{Indep}}_{(i,t)}+{\beta }_8{\mathit{Top}}_{(i,t)}+{\beta }_9{\mathit{Right}}_{(i,t)}+{\beta }_{10}{\mathit{Grow}}_{(i,t)}+\mathit{Year}+\mathit{Stock}+{\mu }_{(i,t)}\end{array}$$

(11)

$$\begin{array}{l}{\mathit{Digt}}_{(i,t)}=\alpha +{\beta }_1{\mathit{Manager}}_{(i,t)}+{\beta }_2{\mathit{Resid}}_{(i,t)}+{\beta }_3{\mathit{Size}}_{(i,t)}+{\beta }_4{\mathit{Debt}}_{(i,t)}+{\beta }_5{\mathit{Profit}}_{(i,t)}+{\beta }_6{\mathit{Cash}}_{(i,t)}\\ +{\beta }_7{\mathit{Indep}}_{(i,t)}+{\beta }_8{\mathit{Top}}_{(i,t)}+{\beta }_9{\mathit{Right}}_{(i,t)}+{\beta }_{10}{\mathit{Grow}}_{(i,t)}+\mathit{Year}+\mathit{Stock}+{\mu }_{(i,t)}\end{array}$$

(12)

Among them, $S{a}_{(i,t)}$ and $Resi{d}_{(i,t)}$ represent the financing constraints and investment efficiency of sample firm i in period t, respectively. Other symbols are the same as previously described.

3.3.3. Moderating Effect Model

In accordance with the research Hypotheses H4a and H4b, and to examine the moderating effect of government subsidies. The following regression models are constructed to analyze the moderating effect of government subsidies:

$$\begin{array}{l}{\mathit{Digt}}_{(i,t)}=\alpha +{\beta }_1{\mathit{Manager}}_{(i,t)}+{\beta }_2{\mathit{Subsidy}}_{(i,t)}+{\beta }_3{\mathit{Manager}*\mathit{Subsidy}}_{(i,t)}+{\beta }_4{\mathit{Size}}_{(i,t)}+{\beta }_5{\mathit{Debt}}_{(i,t)}\\ +{\beta }_6{\mathit{Profit}}_{(i,t)}+{\beta }_7{\mathit{Cash}}_{(i,t)}+{\beta }_8{\mathit{Indep}}_{(i,t)}+{\beta }_9{\mathit{Top}}_{(i,t)}+{\beta }_{10}{\mathit{Right}}_{(i,t)}+{\beta }_{11}{\mathit{Grow}}_{(i,t)}+\mathit{Year}+\mathit{Stock}+{\mu }_{(i,t)}\end{array}$$

(13)

Among them, ${\mathit{Subsidy}}_{(i,t)}$ represent government subsidies received by the sample firm in period t. Other symbols are the same as previously described.

4. Results

4.1. Benchmark Regression

This study employs a stepwise regression approach to test the core relationship between TMT heterogeneity and firm digital innovation resilience. As reported in

Table 1. Benchmark Regression Results.

	Model (1)	Model (2)
	Digit	Digit
Manager	2.955 **	2.781 **
		(1.317)
Size		−0.949 ***
		(0.187)
Debt		−0.049
		(0.711)
Profit		1.791 **
		(0.733)
Cash		−2.534 **
		(1.238)
Indep		−0.019
		(0.022)
Top		−0.016
		(0.013)
Right		0.017
		(0.020)
Grow		1.060 ***
		(0.259)
Stock/Year	Yes	Yes
Cons	−0.836 **	20.819 ***
	(0.404)	(4.131)
Observations	23804	23,804
R-squared	0.257	0.259

Note: * represents significance at the 10% level, ** represents significance at the 5% level, *** represents significance at the 1% level.

, Model (1) includes only firm and year fixed effects and shows that TMT heterogeneity is positively significant at the 5% level (coef. = 2.955). After adding a full set of control variables in Model (2), TMT heterogeneity remains significant at the 5% level (coef. = 2.781), supporting Hypothesis H1a that higher TMT heterogeneity strengthens digital innovation resilience, while H1b is rejected. This result can be explained by the fact that, in complex and uncertain environments, a heterogeneous TMT facilitates knowledge, information, and resource sharing through effective communication. This enables the integration of diverse experiences, technical expertise, and market-sensing capabilities, which in turn provides a broader cognitive foundation for identifying risks and opportunities, understanding complex technologies, and generating non-standard solutions. Consequently, the firm develops stronger resistance and recovery capacities, maintains innovation stability, and achieves greater digital innovation resilience.

4.2. Robustness and Endogeneity Tests

4.2.1. Cluster Firm Level Standard Error

As shown in Model (1) of

Table 2. Robustness test results.

	Cluster Firm Level Standard Error	Add Industry Region Interaction Items	Increase Control Variables
	Model (1)	Model (2)	Model (3)
	Digit	Digit	Digit
Manager	2.781 **	2.781 **	2.873 **
	(1.396)	(1.317)	(1.318)
Vdt		−0.0001
		(0.011)
Board			−0.082
			(0.736)
Equity			0.846 **
			(0.412)
Gdp			1.148
			(0.754)
Control	Yes	Yes	Yes
Stock/Year	Yes	Yes	Yes
Cons	−0.836 **	20.855 ***	8.404
	(0.404)	(4.144)	(9.015)
Observations	23804	23804	23804
R-squared	0.259	0.259	0.259

Note: * represents significance at the 10% level, ** represents significance at the 5% level, *** represents significance at the 1% level.

, robustness tests using firm-clustered standard errors confirm a significantly positive effect of TMT heterogeneity on digital innovation resilience, supporting Hypothesis H1a. This result aligns with the benchmark estimates, underscoring the robustness of our main findings.

4.2.2. Add Industry Region Interaction Items

Add industry region interaction items. To further assess robustness, Model (2) in Table 2 incorporates industry × region interaction terms ($\mathit{Vdt}$). The results continue to show a significantly positive effect of TMT heterogeneity on digital innovation resilience, supporting Hypothesis H1a and confirming the consistency and robustness of the benchmark findings.

4.2.3. Increase Control Variables

As shown in Model (3) of Table 2, this paper introduces additional control variables—including board size ($\mathit{board}$, measured as the natural logarithm of the number of directors), firm ownership nature ($\mathit{equity}$, assigned 1 for state-owned firms and 0 otherwise), and regional GDP level ($\mathit{gdp}$, measured as the natural logarithm of GDP in the firm’s location)—to conduct robustness tests. TMT heterogeneity remains significantly positive, confirming Hypothesis H1a. These results align with the benchmark estimates, reinforcing the robustness of our main conclusion.

4.2.4. Endogeneity Test

Although the baseline model controls for relevant variables, endogeneity concerns such as omitted variable bias or reverse causality may persist. To address this, we employ an instrumental variable (IV) approach. Following Zhang et al. [4], we use the one-period lag of TMT heterogeneity ($L.\mathit{managers}$) as the IV and estimate a two-stage least squares (2SLS) model. The second-stage results in

Table 3. Endogeneity test results.

	2SLS		GMM	LIML
	Model (1) Phase One Managers	Model (2) Phase Two Digit	Model (3) Digit	Model (4) Digit
Managers		35.525 ***	35.525 ***	35.525 ***
		(3.176)	(3.176)	(3.176)
L.Managers	0.003 ***
	(0.0001)
Control	Yes	Yes	Yes	Yes
Stock/Year	Yes	Yes	Yes	Yes
Cons	0.386 ***	12.504 ***	12.504 ***	12.504 ***
	(0.020)	(2.145)	(2.145)	(2.145)
Observations	20365	20365	20365	20365
R-squared	0.063	0.075	0.075	0.075
Underidentification test (Kleibergen-Paap rk LM statistic):				443.79
Chi-sq (1) p-value				0.0000
IV redundancy test (LM test of redundancy of specified instruments):				443.79
Chi-sq (1) p-value				0.0000

Note: * represents significance at the 10% level, ** represents significance at the 5% level, *** represents significance at the 1% level.

in Model (2) show that TMT heterogeneity remains significantly positive, supporting Hypothesis H1a and aligning with the benchmark conclusion. Robustness is further verified using a generalized method of moments (GMM) model (Model 3) and a limited information maximum likelihood (LIML) model (Model 4), which also pass instrument identification and redundancy tests. Together, these results confirm that our core finding is robust to endogeneity concerns.

4.3. Heterogeneity Test

4.3.1. Heterogeneity Test of Firm Scale

Firms are classified as large or small-and-medium firms (SMFs) following China’s Standards for the Classification of Large, Medium, Small, and Micro Firms. A heterogeneity test based on firm size reveals distinct effects: As shown in Model (1) and (2) of

Table 4. Heterogeneity test results.

	Heterogeneity of Firm Scale		Heterogeneity of Firm Nature		Heterogeneity of Monopolistic Firm		Heterogeneity of the Region Where the Firm is Located
	Large Firm	Small and Medium-Sized Firm	State-Owned Firm	Non-State-Owned Firm	High Monopoly Firm	Low Monopoly Firm	Eastern Region Firm	Central Region Firm	Western Region Firm
	Model (1)	Model (2)	Model (3)	Model (4)	Model (5)	Model (6)	Model (7)	Model (8)	Model (9)
	Digit	Digit	Digit	Digit	Digit	Digit	Digit	Digit	Digit
Manager	3.003	4.009 ***	9.005 ***	2.387	0.792	3.266 **	2.919 *	2.750	2.093
	(2.913)	(1.507)	(2.756)	(1.553)	(3.126)	(1.524)	(1.595)	(3.077)	(3.569)
Control	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Stock/Year	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Cons	−23.797 *	27.458 ***	2.263	27.422 ***	20.024 *	21.611 ***	19.400 ***	32.217 ***	10.575
	(14.072)	(5.439)	(8.409)	(4.943)	(11.367)	(4.812)	(5.015)	(10.375)	(11.222)
Observations	5950	17854	5519	18285	5913	17891	17289	3794	2721
R-squared	0.336	0.204	0.338	0.235	0.247	0.265	0.248	0.333	0.254

Note: * represents significance at the 10% level, ** represents significance at the 5% level, *** represents significance at the 1% level.

, while TMT heterogeneity does not significantly influence digital innovation resilience in large firms, it exerts a strong positive impact on SMFs (coef. = 4.009, p < 0.01). This suggests that the benefits of TMT heterogeneity are more pronounced for SMFs. Several factors may explain this pattern. SMFs typically exhibit flatter structures, tighter resource constraints, and more centralized decision-making, enabling heterogeneous perspectives to be translated into strategic actions more rapidly and with less organizational inertia than in larger firms. Moreover, in resource-scarce contexts, digital innovation in SMFs heavily depends on the TMT’s ability to engage in resource bricolage—where heterogeneous members contribute diverse technical knowledge, market channels, and non-standard financing methods, directly facilitating innovation breakthroughs under constraints. Additionally, SMF executives tend to focus intensely on survival and growth, allowing a heterogeneous TMT to efficiently direct collective attention toward the most urgent digital opportunities, thereby avoiding the attention fragmentation and strategic drift more common in larger organizations. Consequently, TMT heterogeneity plays a more substantial role in enhancing digital innovation resilience in SMFs than in large firms.

4.3.2. Heterogeneity Test of Firm Nature

Given distinct resource endowments, objectives, and governance models between state-owned (SOFs) and non-state-owned firms (NSOFs), we examine whether the effect of TMT heterogeneity on digital innovation resilience varies by ownership type. The results in Table 4 (Models 3–4) show that TMT heterogeneity does not significantly affect resilience in NSOEs but exerts a strong positive impact on SOFs (coef. = 9.005, p < 0.01), indicating that its promoting effect is substantially stronger in SOFs. This difference may stem from SOFs’ embeddedness in the national innovation system, where digital innovation is shaped by coercive, mimetic, and normative institutional pressures. Within this complex institutional environment, a heterogeneous TMT acts as a critical decoder and buffer—interpreting diverse external demands, integrating conflicting logics, and translating external pressures into coordinated strategic actions. These capabilities fundamentally enhance SOFs’ digital innovation resilience, explaining why TMT heterogeneity plays a more pronounced role in SOFs than in NSOFs.

4.3.3. Heterogeneity Test of Monopolistic Firm

In line with China’s Anti-Monopoly Law and following Weber et al. [51], we use the Lerner index to gauge firm-level market power. A higher Lerner index signifies greater pricing ability and a stronger monopoly position. For heterogeneity analysis, firms are categorized into groups using the upper quartile of the index distribution. A heterogeneity test based on monopoly status reveals that TMT heterogeneity does not significantly affect digital innovation resilience in high-monopoly firms, but exerts a positive and significant impact on low-monopoly firms (coef. = 3.266, p < 0.05). This disparity can be at-tributed to low-monopoly firms’ greater dependence on external resources and higher survival pressure, which amplify the value of TMT heterogeneity. Specifically, heterogeneous executives provide cross-boundary social networks and knowledge that broaden resource-acquisition channels, buffering external uncertainty and building “resource resilience.” Furthermore, diverse information and perspectives supplied by a heterogeneous TMT enhance dynamic capabilities—such as market-sensing and strategic reconfiguration—that underlie digital innovation resilience. Finally, unlike high-monopoly firms, which often disperse attention across existing operations, low-monopoly firms focus intensely on growth; a heterogeneous TMT helps channel scarce attention toward promising digital opportunities, ensuring more targeted innovation investments. Thus, TMT heterogeneity strengthens digital innovation resilience more markedly in low-monopoly firms than in high-monopoly firms.

4.3.4. Heterogeneity Test of the Region Where the Firm Is Located

Drawing on China’s regional classification outlined in the policy document Guiding Opinions of the Central Committee of the Communist Party of China and the State Council on Promoting the Rise of the Central Region, we group firms into eastern, central, and western regions and test for location-based heterogeneity. The results in Table 4 (Models 7–9) indicate that TMT heterogeneity does not significantly affect digital innovation resilience in the central and western regions but shows a positive effect for firms in the eastern region (coef. = 2.919, p < 0.10). Thus, the contribution of TMT heterogeneity to resilience is markedly stronger in the east. This regional disparity can be attributed to systematic differences in factor endowments, market dynamism, and institutional support. The eastern region’s dense agglomeration of talent, capital, and technology—coupled with active venture capital markets and strong knowledge spillovers—enables heterogeneous TMTs to more effectively access and integrate high-quality resources. Moreover, intense competition, advanced digital demand, and mature industrial chains in the east create abundant innovation opportunities, raising the strategic value of cognitive diversity in identifying and seizing new niches. Supportive governance, efficient policy implementation, and better digital infrastructure further reduce the friction of implementing TMT-driven change. In contrast, the central and western regions exhibit relatively weaker availability of digital innovation factors, more traditional industrial ecosystems, and a lower sense of urgency toward digital transformation. Although policy support is substantial, supporting ecosystems and market entities are still evolving, and firms in these regions often display greater risk aversion due to resource constraints. These conditions collectively dampen the efficiency with which TMT heterogeneity translates into digital innovation resilience. Consequently, the effect of TMT heterogeneity is more pronounced in eastern China than in the central and western regions.

4.4. Mechanism Verification

This paper employs the three-step regression method (following Zhang et al. [3]) to test the mediating roles of financing constraints and investment efficiency. We also apply the PROCESS macro and use bootstrap sampling (1000 repetitions) to verify significance when indirect effects are not clearly determined by the three-step approach [52]. A mediating effect is considered significant if its 95% bias-corrected confidence interval (LLCI to ULCI) does not include zero.

When financing constraints serve as the mediator, the three-step regression results support the mediating role of financing constraints. First, TMT heterogeneity shows a significant positive effect on digital innovation resilience (coef. = 2.781, p < 0.05; Table 1, Model 2). Second, TMT heterogeneity significantly alleviates financing constraints (coef. = –0.025, p < 0.01;

Table 5. Results of Mechanism Analysis.

	Financing Constraints as Intermediary Variables		Investment Efficiency as an Intermediary Variable
	Model (1)	Model (2)	Model (3)	Model (4)
	Sa	Digit	Resid	Digit
Manager	−0.025 ***	3.041 **	−0.029 ***	2.797 **
	(0.007)	(1.315)	(0.008)	(1.318)
Sa		−10.294 ***
		(1.301)
Resid				−0.574
				(1.176)
Control	Yes	Yes	Yes	Yes
Stock/Year	Yes	Yes	Yes	Yes
Cons	3.365 ***	−13.790 **	−0.091 ***	20.902 ***
	(0.022)	(6.015)	(0.025)	(4.131)
Observations	23804	23804	23804	23804
R-squared	0.858	0.261	0.025	0.259

Note: * represents significance at the 10% level, ** represents significance at the 5% level, *** represents significance at the 1% level.

, Model 1). Finally, when both variables are included, TMT heterogeneity remains significant (coef. = 3.041, p < 0.05) and financing constraints exert a strong negative effect (coef. = –10.294, p < 0.01; Table 5, Model 2). The indirect effect is 0.257, accounting for about 8.5% of the total effect, which supports Hypothesis H2a. This mediating mechanism can be attributed to the diverse social networks (e.g., government, banking, venture-capital, and overseas ties) and reputational signals associated with a heterogeneous TMT. Such networks provide varied industry experience and improve the accuracy of evaluating digital-innovation projects, while the personal reputation of TMT members transmits credible signals to external capital markets. Together, these factors ease access to and allocation of scarce financial resources, thereby enhancing the firm’s ability to sustain digital innovation under external shocks.

When investment efficiency serves as the mediator, first, TMT heterogeneity positively affects digital innovation resilience (coef. = 2.781, p < 0.05; Table 1, Model 2). Second, TMT heterogeneity significantly reduces investment inefficiency—the inverse measure of investment efficiency (coef. = –0.029, p < 0.01; Table 5, Model 3)—implying a positive effect on efficiency itself. Third, when both variables are included, TMT heterogeneity remains significant (coef. = 2.797, p < 0.05) while the direct path from investment efficiency to resilience is not significant (Table 5, Model 4). To test the mediating effect of investment efficiency, this study employed the bootstrap method with 1000 resamples to estimate the bias-corrected (BCa) confidence interval. The results indicate a significant indirect effect, as the 95% confidence interval [BootLLCI = 0.001621, BootULCI = 1.996632] excludes zero (p = 0.000). The estimated indirect effect is (−0.029) × (−0.574) = 0.016646, accounting for approximately 0.6% of the total effect, thereby supporting Hypothesis H3a. This result can be explained by the contrast between homogeneous TMTs, which are susceptible to groupthink and may overinvest in misaligned digital projects, and heterogeneous TMTs, which employ diverse information processing, risk assessment, and opportunity-identification capabilities. By more accurately evaluating the feasibility, returns, and risks of digital technologies, a heterogeneous TMT allocates scarce resources more appropriately, makes more contextually grounded investment decisions, and thus builds a sustainable capacity for iterative innovation in uncertain environments.

4.5. Testing the Moderating Effect of Government Subsidies

In this study, government subsidies are incorporated into the benchmark regression model as a moderating variable. As shown in

Table 6. Results of Moderating Analysis.

	Model (1)	Model (2)
	Digit	Digit
Manager	−174.598 ***	−175.915 ***
	(13.470)	(13.470)
Subsidy	−16.425 ***	−16.234 ***
	(1.387)	(1.387)
Manager × Subsidy	63.891 ***	64.315 ***
	(4.825)	(4.825)
Control	—	Yes
Stock/Year	Yes	Yes
Cons	−0.836 **	20.819 ***
	(0.404)	(4.131)
Observations	23804	23,804
R-squared	0.257	0.259

Note: * represents significance at the 10% level, ** represents significance at the 5% level, *** represents significance at the 1% level.

, Model (1) includes only firm and year fixed effects. The interaction term between TMT heterogeneity and government subsidies shows a positive and significant correlation with corporate digital innovation resilience at the 1% level (coef. = 63.891). After adding all control variables in Model (2), the interaction term remains positively significant at the 1% level (coef. = 64.315), supporting Hypothesis H4a (while H4b is not supported). This indicates that government subsidies play a positive moderating role in the relationship between TMT heterogeneity and corporate digital innovation resilience.

This phenomenon can be explained by two main mechanisms. First, government subsidies directly provide essential financial resources, alleviating budget constraints for firm innovation activities. This enables heterogeneous teams to more fully translate their diverse cognitive perspectives into exploratory practices. Second, subsidies act as a risk buffer and a signaling or certification mechanism. They reduce the perceived risk of high-stakes innovative experimentation for the team while simultaneously attracting external collaboration and resource inflows. Together, these effects amplify the positive impact of TMT heterogeneity on digital innovation resilience.

5. Discussion

5.1. Theoretical Implications

This study is motivated by the need to address three interrelated research questions. Drawing on upper echelons theory, organizational resilience theory, and government intervention theory, we first investigate how TMT heterogeneity influences firms’ digital innovation resilience. Second, we examine the underlying mechanisms through which TMT heterogeneity enhances digital innovation resilience—specifically, by alleviating financing constraints and improving investment efficiency. Third, we explore the moderating role of government subsidies in the relationship between TMT heterogeneity and digital innovation resilience.

Departing from prior studies that have predominantly focused on either single dimensions of personal characteristic heterogeneity (e.g., gender, age, education level, functional background, tenure, or risk preference) or the isolated effects of various heterogeneous combinations on innovation input or output [4,5,7,8,15,16,17,18,19], and extending beyond research that has examined the relationship between firm digital innovation and heterogeneity in areas such as TMT cognitive base, social networks, internal relational stability, diversity, educational background, overseas experience, or professional background [13,20,21,22,23,24,25], our study demonstrates that TMT heterogeneity can effectively enhance firms’ digital innovation resilience. We argue that when confronting shocks from complex internal and external environments, a heterogeneous TMT facilitates knowledge, information, and resource sharing through effective communication. This enables the full utilization of diverse experiences, technical expertise, and market-sensing capabilities among team members, thereby providing more varied knowledge and problem-solving perspectives for digital innovation. Such diversity helps the firm identify risks and opportunities in the digital innovation process, comprehend complex digital technologies, and generate a greater number of non-standardized solutions. Consequently, a heterogeneous TMT equips the firm with the capacity to effectively resist and recover from environmental shocks, maintain the stability of its digital innovation, and develop stronger digital innovation resilience. This finding contributes to the existing literature by expanding the application scope of upper echelons theory from the perspective of TMT heterogeneity and offers empirical support for firms seeking to build a rationally composed TMT to strengthen their digital innovation resilience.

Furthermore, while existing research has examined how TMT characteristics influence organizational or general innovation resilience—suggesting that diverse TMT attributes can foster such resilience [14,26,27]—the specific mechanism through which TMT heterogeneity affects digital innovation resilience, particularly via investment efficiency and financing constraints, remains underexplored. Addressing this gap, our study identifies and tests a key transmission pathway: TMT heterogeneity enhances digital innovation resilience by alleviating financing constraints and improving investment efficiency. These findings make two core contributions. Theoretically, they extend organizational resilience theory into the digital innovation domain and clarify the micro-foundational roles of strategic resource allocation. Practically, they provide actionable insights for firms to strengthen digital innovation resilience by intentionally building heterogeneous TMTs that can ease financing pressures and allocate capital more effectively.

Finally, although prior research has primarily examined how government subsidies influence corporate innovation, scant attention has been paid to their moderating role in the relationship between TMT heterogeneity and corporate digital innovation resilience [28,29,30,31]. Our study reveals that government subsidies can positively moderate the enhancing effect of TMT heterogeneity on digital innovation resilience. Specifically, government subsidies provide substantial financial support for digital innovation, thereby reinforcing the promoting effect of TMT heterogeneity. This finding extends the application of government intervention theory to the field of corporate innovation management and offers a pathway for governments to use subsidies to strengthen the positive impact of TMT heterogeneity on digital innovation resilience, ultimately further enhancing such resilience.

5.2. Practical Implications

First, firms should recognize TMT heterogeneity as a core strategic asset and deliberately structure leadership teams around such diversity. This study demonstrates that multi-dimensional heterogeneity—spanning age, gender, education, tenure, professional background, overseas experience, and financial expertise—significantly strengthens a firm’s digital innovation resilience. A cognitively and experientially diverse TMT thus serves as a key strategic resource, enabling the firm to navigate digital innovation setbacks and recover more swiftly. Consequently, especially for firms undergoing digital transformation or operating in high-uncertainty contexts, it is critical to advance beyond superficial diversity markers and embed deep-level, task-relevant heterogeneity into talent strategy and succession planning. This entails a shift in perspective: heterogeneity should be treated not as a static human-resource configuration, but as a dynamic capability requiring continuous investment and systematic management. To operationalize this, firms can develop a “strategic competency–team composition” map, identifying the capabilities required for future digital innovation—such as core-technology breakthroughs or business-model renewal—and then recruiting or developing leaders who bring those scarce backgrounds. Concurrently, mechanisms should be established to activate and integrate diverse cognitive perspectives. Examples include strategy workshops that mandate “challenger viewpoints,” and systematic assessments of differences in risk appetite, information processing, and time orientation among TMT members. The aim is to foster a collective intelligence that balances agility with reflection, and short-term adaptability with long-term development. Only through such integrated practices—spanning strategic planning, team design, and cognitive orchestration—can firms convert TMT heterogeneity into a sustained driver of digital innovation resilience.

Second, firms should focus on refining internal governance and strengthening the resource transmission mechanism. This study reveals that TMT heterogeneity enhances digital innovation resilience primarily through alleviating financing constraints and improving investment efficiency—a critical mediating pathway. Therefore, building an effective governance structure to ensure this transmission mechanism functions smoothly should be a core managerial priority. Merely forming a diverse TMT is insufficient; institutionalized processes are necessary to translate the team’s multifaceted knowledge into superior financing strategies, precise investment decisions, and efficient resource allocation. For example, major digital innovation investments could undergo rigorous review by a cross-functional committee, and a rotating “challenger role” could be introduced to systematically surface blind spots and optimize resource plans. Additionally, for core innovation projects, developing scenario-based resource and financing plans (under optimistic, baseline, and pessimistic outlooks) can proactively mobilize the TMT’s external networks, enhancing organizational flexibility and resilience. Firms should also actively empower the TMT to convert individual external ties into organizational strategic resource channels. Incorporating contributions to “strategic relational capital” into executive evaluations can further incentivize such behavior. Through these measures, the cognitive diversity of a heterogeneous TMT can be effectively translated into tangible advantages in resource allocation and investment efficiency, thereby strengthening the firm’s ability to sustain digital innovation under uncertainty.

Third, the government should strive to accurately identify TMT that already possess effective heterogeneity and, through subsidy mechanisms, amplify and safeguard the efficiency with which this team advantage translates into actual innovation resilience. This study reveals that government subsidies can positively moderate the enhancing effect of TMT heterogeneity on corporate digital innovation resilience. Therefore, subsidy policies should shift from “providing broad-based support to enterprises” to “empowering effective teams.” Mechanisms should be designed into the review process to assess the quality of a firm’s TMT heterogeneity—for example, evaluating its relevance to digital innovation tasks and the presence of internal integration processes. Priority should be given to applicants whose team structure is more likely to transform subsidy funds into efficient investment and robust risk-response capabilities. Concurrently, policies may require recipient enterprises to establish supporting governance mechanisms, such as setting up innovation decision-making committees or introducing challenger roles, to ensure the effective integration of diverse cognitive perspectives. This approach allows public funds to act as a catalyst, activating internal strategic resources and strengthening the “heterogeneity → resilience” pathway. Ultimately, this enables more targeted intervention to enhance the digital innovation resilience of the overall economy.

6. Conclusions and Further Avenues for Research

6.1. Conclusions

When firms experience shocks arising from internal and external environmental uncertainty, TMT heterogeneity functions as a critical strategic resource that enables them to sustain stable digital innovation and cultivate robust digital innovation resilience through self-learning, adaptation, and resistance. Using a sample of Chinese manufacturing listed firms from 2015 to 2024, this study investigates the impact of TMT heterogeneity on corporate digital innovation resilience and explores its underlying mechanisms. The main findings are as follows: (1) TMT heterogeneity significantly enhances corporate digital innovation resilience. (2) This enhancement operates through two distinct pathways: alleviating financing constraints and improving investment efficiency. (3) The effect of TMT heterogeneity on digital innovation resilience exhibits notable heterogeneity across different types of firms. Specifically, the effect is stronger for small and medium-sized enterprises compared to large firms, more pronounced in state-owned enterprises than in non-state-owned ones, and more significant in firms with low market monopoly power than in those with high monopoly power. In addition, the effect is substantially stronger for firms located in the eastern region than for those in central and western regions. (4) Government subsidies positively moderate the relationship between TMT heterogeneity and digital innovation resilience, thereby strengthening the enhancing effect. This study offers both theoretical insights and practical guidance for enterprises seeking to build effective TMTs, alleviate financing constraints, and improve investment efficiency, as well as for governments aiming to design subsidy policies that ultimately foster digital innovation resilience.

6.2. Further Avenues for Research

This study examines the impact of TMT heterogeneity on firm digital innovation resilience and reveals the mechanistic pathway through which such heterogeneity enhances digital innovation resilience by alleviating financing constraints and improving investment efficiency. While these findings meaningfully address gaps in the existing literature, the research also has certain limitations. First, in terms of generalizability, the results of this study are confined to manufacturing firms. Future research could extend the investigation to encompass all industries. Second, this paper does not explore the economic outcomes resulting from the influence of TMT heterogeneity on firm digital innovation resilience. Subsequent studies could further analyze the economic consequences arising from TMT heterogeneity’s effect on digital innovation resilience. Third, existing studies are predominantly confined to listed manufacturing firms. Future research could expand the scope of investigation to include non-listed firms.