The U-Shaped Effect of Non-CEO Executives’ Internal Governance on Corporate Innovation Investment: Evidence from China

Abstract

Against the backdrop of the increasingly salient constraints of resource scarcity and environmental pressures on global economic development, sustainable innovation emerges as an imperative strategic pathway for corporations to secure a competitive edge in the international marketplace. Corporate innovation capability serves as the critical factor for both the advancement of sustainable innovation and the maintenance of the corporate competitive edge. While the extant literature has extensively explored how internal and external governance mechanism forces shape corporate investment decision-making, the critical role of non-CEO executives in the process of corporate innovation investment decision-making remains conspicuously underexplored. This study examines the effect of bottom–up governance mechanisms within executive teams on corporate innovation investment from the perspective of non-CEO executive independence. We used a sample of A-listed companies on the Shanghai and Shenzhen stock exchanges from 2007 to 2021 for empirical tests. We found a U-shaped relation between non-CEO executive independence and corporate innovation investment, and this finding still held after addressing endogeneity issues and conducting a series of robustness tests. Mechanism analysis revealed that both non-CEO executives’ decision horizon and firm agency costs positively moderate this U-shaped relationship. This U-shaped effect is pronounced in firms with lower CEO power, lower levels of corporate governance, and non-state-owned firms. Our findings provide an important basis for clarifying the internal governance mechanism of the executive teams while offering new insights for optimizing the allocation of corporate resources and promoting corporate innovation from the perspective of improving corporate governance.

1. Introduction

In the current era of information and knowledge-based economy, science and technology have emerged as pivotal competitive elements in enhancing national competitiveness, with innovation serving as the primary driver of modern economic growth. According to a report by China’s 19th National Congress, the Chinese economy is undergoing a critical phase characterized by economic structure optimizing and growth momentum transformation, wherein innovation serves as the first driving force to lead development. In recent years, within the context of economic globalization, there has been increased pressure for a gradual economic downturn in China and internationally. The Chinese economy is currently experiencing a significant transitional phase marked by industrial restructuring and upgrading, necessitating in-depth analysis of economic globalization development trends and the identification of novel economic growth drivers. As the primary catalyst for leading high-quality development, innovation plays a crucial role in driving scientific and technological advancement, serving as a vital strategic pillar for China’s pursuit of sustainable economic development. Furthermore, enterprises constitute the principal entities of innovation, with the enhancement of national independent innovation capacity fundamentally dependent on the advancement of corporate innovation [1]. Corporate innovation not only enhances enterprise sustainable competitive capacity and elevates corporate value but also fosters economic sustainable development while stimulating economic growth [2,3]. Consequently, the question of how to advance corporate innovation capabilities constitutes a critical determinant in driving corporate sustainable innovation and sustaining competitive advantages.

In recent years, issues related to corporate innovation have garnered significant attention from both governmental entities and academic communities. Existing studies have extensively examined various factors affecting corporate innovation, mainly encompassing executive characteristics [4], corporate governance [5,6], economic policy uncertainty [7,8], and macro-industrial policies [9,10]. As the leader of corporate investment decisions, the behavior of the executive team significantly influences corporate innovation decisions. The corporate executive team comprises both CEO and non-CEO executives. While the CEO, as the principal decision-maker and leader, plays a pivotal role in corporate strategy, non-CEO executives serve dual functions as both decision contributors and implementation facilitators [11]. In practice, strategic decision-making is not solely the CEO’s prerogative, as non-CEO executives play critical roles of coordinating strategy formulation and corporate decision-making oversight. Consequently, the impact of non-CEO executives on corporate investment decisions should not be ignored [12,13].

Existing studies have identified that non-CEO executives may exercise supervisory and constraining functions over CEO behavior, reflecting a bottom–up governance mechanism within executive teams, termed executive team internal governance [14,15]. However, existing research has predominantly investigated the determinants of corporate innovation decisions from the perspective of formal corporate governance mechanisms [5,6]. While some studies have acknowledged informal governance effects within executive teams, scant attention has been paid to how the unique relationships between non-CEO executives and CEOs influence corporate innovation decision-making. Furthermore, existing research on corporate innovation decision-making predominantly focuses on the individual traits of the CEO [16] or examines the issue from the perspective of the executive team as a whole [17]. Although existing research has explored the influence of subordinate executives’ decision-making horizons and relative compensation on corporate innovation [18,19], it has overlooked the special connection between non-CEO executives and the CEOs. Consequently, from the perspective of non-CEO executive independence, this study explores two pivotal questions: (1) Does the internal governance effect of non-CEO executives enhance corporate innovation investment? (2) If so, what is the underlying mechanism? These are the primary questions addressed in this study.

To address these questions, this paper empirically examines the relationship between the internal governance effect of non-CEO executives and corporate innovation investment, with a focus on the non-CEO executive independence as the entry point. The primary contributions of this study can be delineated as follows: First, from the perspective of non-CEO executive independence, this research broadens the theoretical framework on the influencing factors of corporate innovation in terms of corporate managers. The extant literature predominantly examines either core managers’ (CEOs or chairpersons) personal characteristics [16] or the executive team as a whole [17] on corporate innovation decision-making, with scant attention paid to the potential internal governance effects of non-CEO executives within the top management team in innovation decision-making. This study attempts to examine the influence of non-CEO executive independence on corporate innovation investment from the perspective of bottom–up informal governance mechanisms within the executive team. Second, this study takes into account the non-linear relationship between non-CEO executive independence and corporate innovation investment, offering a novel perspective on the role positioning of non-CEO executives in corporate innovation decision-making. Moreover, it reveals that the decision-making horizon of non-CEO executives and corporate agency costs also modulate the relationship between non-CEO executive independence and corporate innovation investment, thereby extending the existing analytical framework. Finally, existing studies have overlooked the contingent impact of internal and external corporate environments on managerial behavior. This study further explores the differences in the effect of CEO power, corporate governance level, and ownership structure on the aforementioned relationship, thereby expanding the research context in which non-CEO executive independence influences corporate innovation investment. This has enlightening implications for understanding the internal governance effects of non-CEO executives across diverse contexts.

2. Literature Review

At present, there is a dearth of research on the connection between internal governance and corporate innovation investment. Therefore, this paper categorizes and analyzes existing research across the following:

Firstly, innovation serves as a critical determinant for the survival of enterprises in the fiercely competitive market environment and plays a crucial role in sustainable development. Corporate innovation has garnered significant attention from scholars both domestically and internationally. Scholars generally demonstrate that corporate innovation enhances competitive advantage and facilitate value growth. Generally, an increase in innovation investment can bolster a company’s competitive edge in developing new products and technologies, thereby strengthening its core competitiveness and ultimately enhancing corporate value [2,3]. As the primary decision-makers in corporate investment, managers play a critical role in the outcome of corporate innovation activities. Given that corporate innovation is characterized by high risk, high cost, and lengthy R&D cycles, it is necessary that managers give greater personal cost in innovation progress. The progression of innovation activities and innovation investment decisions fundamentally depend on managers’ trade-offs between the costs they incur and the benefits they anticipate [20]. Scholars have extensively explored the determinants of corporate innovation decision-making from the perspective of managerial traits, positing that demographic background traits such as age, tenure, and experience [2,16], as well as psychological traits like overconfidence [21,22,23], significantly influence their strategic choices and, consequently, corporate innovation decision-making. Additionally, existing research has extensively examined the effect of corporate governance [6], executive power [3], and employee motivation [24] on corporate innovation decision-making.

Secondly, existing research related to corporate innovation involves corporate governance, primarily discussing formal corporate governance mechanisms. These mechanisms encompass the use of shareholder meetings, supervisory boards, internal controls, and boards of directors to incentivize and monitor managers in serving the interests of the company’s owners, thereby enhancing the firm’s innovation capabilities and promoting corporate sustainable development [25,26,27,28]. However, in studies concerning internal corporate governance, scholars have primarily focused on the CEO as an individual [16] or the executive team as a whole [2]. These studies have yet to consider the fact that the executive team is an assemblage of executives with diverse career aspirations and varying interests. The heterogeneous preferences of these executives may lead to the creation of internal checks and balances within the executive team [29,30]. Existing research has shown that, based on principal–agent theory, CEOs may exploit their informational advantages to engage in opportunistic behaviors [31], such as earnings management for personal gain, which can undermine corporate performance and reduce corporate investment efficiency, thereby harming the overall interests of the company [32].

Lastly, due to principal–agent and information asymmetry problems, driven by self-interested motives such as earnings management, CEOs are more likely to exploit their authority to implement opportunistic behaviors in corporate investment decisions, which is not conducive to the long-term sustainable development and long-term interests of enterprises [33]. Non-CEO executives constitute a substantial portion of the executive team, with varying power distributions, career aspirations, and personal interests across different positions. Furthermore, corporate performance and future development are intrinsically linked to non-CEO executives’ personal interests. Building on this foundation, existing research has begun to recognize the special association between CEOs and non-CEO executives. On the one hand, as primary strategic decision-makers, CEOs play a pivotal role over corporate investment decision-making. Non-CEO executives, who serve dual roles as both decision-makers and implementers, are essential for the coordination and execution of CEO decision-making. When CEO decisions potentially jeopardize long-term corporate interests, non-CEO executives may intervene through direct (including the refusal to execute decisions) or indirect (such as passive resistance, non-cooperation, or advisory roles) means, thereby mitigating CEO self-interested behaviors, and leveraging their informational advantages to monitor and constrain CEO behavior [34]. On the other hand, within the executive team, non-CEO executives have career aspirations for CEO succession through internal promotion. Motivated by personal reputation and career advancement, non-CEO executives may place greater emphasis on corporate performance and long-term development. Consequently, when CEOs engage in opportunistic behaviors or make flawed decisions, non-CEO executives are more likely to intensify their supervisory efforts, thereby exerting their internal governance effects within the executive team [14].

In practice, the internal governance effect exerted by non-CEO executives on the CEO is not always present, as it hinges on their willingness to supervise the CEO. Non-CEO executives are not entirely independent of the CEO, as the implementation of CEO decisions necessitates their coordination and execution. Simultaneously, non-CEO executives are guided and motivated by the CEO. Existing research indicates that non-CEO executives’ supervisory willingness is closely correlated with their degree of independence from the CEO [14]. The CEO has the authority to directly participate in the hiring or dismissal process of non-CEO executives. Executives promoted or appointed through the CEO’s influence typically demonstrate lower independence levels. Such executives may, out of gratitude towards the CEO, reduce their willingness to monitor the CEO’s self-serving behaviors, and may even support or facilitate the implementation of flawed CEO decisions. Conversely, non-CEO executives appointed prior to the CEO’s tenure generally maintain higher independence. The appointment or promotion of such executives is not influenced by the incumbent CEO, and they consequently exhibit stronger tendencies to monitor the CEO’s self-serving behaviors, which enhances the internal governance effects within the executive team [14]. Research demonstrates that non-CEO executive independence, through internal governance effects, constrains CEO behavior and influences corporate investment decisions, evidenced by reduced earnings management [35], optimized capital structures [14], moderated corporate financialization [36], enhanced risk-taking capacity, and improved corporate investment efficiency [30], ultimately contributing to the enhancement of corporate value. Additionally, other scholars have used the decision-making horizon and relative compensation of subordinate executives to measure their motivation and ability to counterbalance the CEO, revealing positive associations between these factors and corporate innovation [37].

3. Theoretical Analysis and Hypothesis Development

3.1. Non-CEO Executive Independence and Corporate Innovation Investment

The individual heterogeneity and irrational biases of different managers significantly influence corporate innovation decisions, thereby affecting corporate innovation investment [4,21]. Managers may exploit information asymmetry to engage in opportunistic behaviors, while corporate governance mechanisms can incentivize and monitor managers to make decisions aligned with the interests of the company’s owners. The willingness of non-CEO executives to supervise or constrain CEO behavior is closely related to their independence [14]. Those executives who are promoted or appointed through the CEO’s influence often reduce their supervisory willingness out of a sense of obligation to the CEO. In contrast, non-CEO executives who assumed their positions prior to the CEO’s tenure are not subject to the CEO’s constraints and exhibit relatively stronger independence, thereby facilitating the executive team’s internal governance effectiveness. Such dual effects coexist within executive teams [36]. Consequently, as the independence of non-CEO executives increases, it may enhance their willingness to supervise the CEO, effectively exerting the internal governance effects of the executive team and thereby promoting corporate innovation investment. Conversely, it may also lead to a reduction in the supervisory willingness of non-CEO executives, who may collude with the CEO out of self-interest, thereby inhibiting corporate innovation investment. Thus, we consider that the ultimate effect of non-CEO executive independence on corporate innovation investment depends on the relative strength of these opposing effects, potentially exerting asymmetric influences on corporate innovation investment. Threshold theory suggests that the minority faction within an organization must reach a critical scale to exert substantial influence on decision-making. Therefore, the executive team’s internal governance effect surpasses collusion effects only when a threshold level of CEO-independent non-CEO executives is achieved [36]. When the number of non-CEO executives falls below this threshold, they tend to align more closely with the CEO’s behavior, resulting in collusion effects dominating over governance effects, thereby impeding corporate innovation investment. When the number of non-CEO executives exceeds this threshold, the heightened independence of non-CEO executives facilitates the activation of internal governance mechanisms, which diminish or neutralize collusion effects, thereby promoting innovation investment.

Specifically, the impact of non-CEO executive independence on corporate innovation investment may demonstrate a threshold effect, exhibiting a non-linear relationship. When non-CEO executive independence falls below the threshold, it significantly inhibits the effect on corporate innovation investment. Lower non-CEO executive independence indicates that pre-CEO-tenure non-CEO executives constitute a minority, resulting in most executives supporting CEO decisions due to obligation to the CEO. Consequently, independent executives struggle to effectively supervise and may opt to follow majority support for CEO behaviors [34]. As the independence of non-CEO executives increases but remains below the threshold, while the number of independent executives grows, their supervisory capacity remains limited, potentially fostering collusion with the CEO. This enhanced collusion effect amplifies CEO autonomy and self-serving behaviors, ultimately reducing innovation investment. Conversely, when the independence of non-CEO executives exceeds the threshold, CEO-unconstrained executives form the majority, enabling them to provide diverse, objective, and neutral perspectives that play an active supervisory role in corporate innovation decisions. This effective supervision constrains CEO myopia, aligns CEO behavior with the firm’s long-term value creation, and fosters innovation activities. In summary, this study posits that below the threshold, collusion effects of non-CEO executives dominate their internal governance effects, significantly inhibiting innovation investment. Beyond the threshold, internal governance effects surpass collusion effects, significantly promoting innovation investment.

Based on the above discussion, we hypothesize the following:

H1: 

Non-CEO executive independence exhibits a significant U-shaped relationship with corporate innovation investment.

3.2. The Moderating Effect of Non-CEO Executives’ Decision-Making Horizon

The efficacy of internal governance within the executive team is also contingent upon non-CEO executives’ monitoring motivation toward the CEO, which is significantly influenced by their decision horizons. Non-CEO executives who are older or nearing the end of their tenure tend to exhibit shorter decision horizons. When such executives perceive limited promotion opportunities despite their efforts, their willingness to engage in high-risk corporate innovation activities diminishes, concurrently reducing their motivation to monitor CEO behavior. Conversely, younger non-CEO executives or those with shorter tenures are more likely to have opportunities for promotion to CEO positions. These executives typically have longer decision horizons and, driven by promotion prospects, are more inclined to actively participate in corporate innovation activities that benefit the firm’s long-term development, thereby intensifying their motivation to monitor CEO behavior [29]. In summary, considering the effect of decision horizons on the non-CEO executives’ supervisory motivation and the corporate innovation decisions, this study posits that decision horizons may moderate the relationship between non-CEO executive independence and corporate innovation investment.

On the left side of the U-shaped curve, where the collusion effect between non-CEO executives and the CEO outweighs the internal governance effect, non-CEO executives with longer decision horizons exhibit greater supervisory motivation. In this context, despite heightened supervisory motivation, these executives may still collude with the CEO for self-interest, thereby intensifying the inhibitory effect of non-CEO executive independence on corporate innovation investment. On the right side of the U-shaped curve, when non-CEO executives have longer decision horizons, their myopic tendencies diminish, fostering greater focus on long-term benefits that enhance corporate value and enhancing innovation investment motivation [18]. In this context, non-CEO executives’ internal governance effects strengthen the promotion of corporate innovation investment. Therefore, this study concludes that longer decision horizons accentuate the U-shaped relationship between non-CEO executive independence and corporate innovation investment.

Based on the above discussion, we hypothesize the following:

H2: 

The decision horizon positively moderates the relationship between non-CEO executive independence and corporate innovation investment.

3.3. The Moderating Effect of Corporate Agency Costs

Furthermore, due to the principal–agent problem, conflicts of interest arise between the CEO and shareholders, potentially leading to CEO opportunistic behaviors driven by self-interest. The conflict of interest stemming from agency problems is one of the significant challenges faced by firms, substantially impacting long-term corporate value. As Type I agency problems intensify, exacerbating CEO–shareholder interest conflicts, CEOs become more inclined to make decisions diverging from shareholder value maximization, thereby amplifying the significance of non-CEO executives’ internal governance role in corporate innovation activities. Existing research has also shown that firms with higher agency costs often face severe Type I agency problems [14,38]. In summary, this study posits that agency costs significantly influence the relationship between non-CEO executive independence and corporate innovation investment. On the left side of the U-shaped curve, higher agency costs intensify a CEO’s motivation for self-serving behavior, thereby strengthening the collusion effect between non-CEO executives and the CEO. Consequently, when non-CEO executive independence remains below a certain threshold, the inhibitory effect of non-CEO executive independence on corporate innovation investment becomes more pronounced with rising agency costs. On the right side of the U-shaped curve, non-CEO executives can effectively exert internal governance effects within the executive team. As agency costs increase, the CEO myopic behavior becomes more severe. In this context, as non-CEO executive independence further increases, its positive impact on promoting corporate innovation investment becomes stronger. Based on the above analysis, this study concludes that higher agency costs accentuate the U-shaped relationship between non-CEO executive independence and corporate innovation investment.

Based on the above discussion, we hypothesize the following:

H3: 

Agency costs positively moderate the relationship between non-CEO executive independence and corporate innovation investment.

The conceptual framework is shown in Figure 1.

4. Research Design

4.1. Sample Selection

This study selected China’s A-share listed firms on the Shanghai and Shenzhen stock exchanges from 2007 to 2021 as the research sample. The sample selection processed adheres to the following criteria: (1) exclusion of ST-, *ST-, and PT-listed companies; (2) removal of financial and insurance companies; (3) elimination of companies with less than one year of listing; (4) exclusion of companies experiencing CEO turnover during the year; and (5) removal of observations with missing data for key variables. The final initial usable sample consisted of 19,006 observations. The primary sources included the China Research Data Service Platform (CNRDS), the China Stock Market and Accounting Research (CSMAR), and the Wind databases (Wind). Finally, to avoid the adverse effect of bounding values on the empirical results, we winsorized the continuous variables at the 1% and 99% levels. Additionally, firm-level clustered standard errors were employed in all regression analyses.

4.2. Variable Measurement

4.2.1. Corporate Innovation Investment

Existing research predominantly employs the ratio of R&D investment to measure corporate innovation investment. Following the methodology of Xie et al. [39], this study employs the R&D intensity ratio, calculated as R&D input divided by main business revenue, to measure corporate innovation investment. This metric offers several of the following advantages: (1) it objectively reflects the extent to which a firm creates new products and knowledge through increased innovation investment [40]; (2) it is highly replicable and stable, making it widely used by researchers; and (3) innovation input provides a more direct reflection of a firm’s current investment decisions and resource allocation compared to innovation output. R&D input directly captures capital financial allocation to innovation activities, and since there is typically a lag between R&D input and patent applications, R&D input serves as a more representative indicator of a firm’s innovation decisions. Therefore, this study adopts the R&D-to-sales ratio as the primary proxy variable for corporate innovation investment.

4.2.2. Non-CEO Executive Independence

To accurately measure executive team internal governance, this study employs the proportion of non-CEO executives appointed prior to the current CEO’s tenure as our proxy variable for non-CEO executive independence. This choice is grounded in two critical considerations. On the one hand, given the CEO’s influence over the appointment processes of other executives, post-appointment CEOs may appoint executives aligned with their decision-making objectives, fostering collaborative dynamics. CEO-appointed executives frequently feel a sense of gratitude and are more inclined to support the CEO’s decisions, exhibiting weaker motivation to monitor or constrain the CEO’s self-serving behaviors. Correspondingly, the lower level of non-CEO executive independence corresponds to a weaker effect of executive team internal governance. On the other hand, executives appointed prior to the CEO’s tenure remain insulated from such influence, enabling autonomous decision-making. These individuals demonstrate heightened motivation in monitoring and constraining the CEO’s self-serving behaviors [17]. Correspondingly, higher levels of independence among these non-CEO executives consequently strengthen executive team inter-governance.

To summarize the above, building upon the methodologies of Landier, Sraer and Thesmar [15]; Khanna et al. [41]; and Landier, Sauvagnat, Sraer and Thesmar [14], this study measures non-CEO executive independence as the proportion of non-CEO executives appointed prior to the current CEO’s tenure, calculated as Intgov = (Number of non-CEO executives appointed before the CEO and ranked in the top four for compensation)/(Total number of non-CEO executives ranked in the top four in compensation). To ensure comparability, observations with fewer than four non-CEO executives were excluded from the primary analysis. Non-CEO executives refer to members of the executive team other than the CEO, typically including positions such as deputy general managers, operations directors, sales directors, and financial directors, among others, as determined by their inclusion in the firm’s executive team. Consequently, higher Intgov values indicate stronger non-CEO executive independence relative to the CEO, reflecting higher level of executive team internal governance. Notably, observations with fewer than four non-CEO executives constitute approximately 20% of the total sample, which are subsequently incorporated in robustness checks.

4.2.3. Control Variables

Building upon the methodologies of Yuan and Wen [42], we consider the following control variables: firm size (Size), leverage ratio (Lev), corporate growth opportunities (Tq), profitability (Roa), cash holdings (Cash), fixed asset ratio (PPE), firm age (Age), state ownership (State), ownership concentration (Share), CEO duality (Dual), board size (Board), board independence (Boainpd), managerial ownership (Hold), institutional ownership (Inshare), analyst coverage (Analyst), CEO tenure (Tenure), and internal CEO promotion (Intpromoted). Furthermore, the analysis incorporates both year and industry fixed effects.

All variables in this paper are summarized in

Table 1. Definitions and descriptions of variables.

Type	Name	Symbol	Definition
Dependent variable	Corporate innovation investment	Rd	R&D input/main business revenue, (%)
Independent variable	Non-CEO executive Independence	Intgov	(Number of non-CEO executives appointed before the CEO and ranked in the top four for compensation)/(Total number of non-CEO executives ranked in the top four in compensation)
Control variable	Firm size	Size	Ln (total assets of the enterprise)
	Leverage ratio	Lev	Total liabilities/total assets
	corporate growth opportunities	Tq	Tobin Q
	Profitability	Roa	Net profit/total assets
	Cash holdings	Cash	Ln (cash holding)
	Fixed asset ratio	PPE	Fixed asset/total assets
	Firm age	Age	Ln (firm’s actual age since listing + 1), (years)
	State ownership	State	State-owned enterprises are assigned a value of 1, or 0 otherwise
	Ownership concentration	Share	Percentage of shareholding of the top ten largest shareholder, (%)
	CEO duality	Dual	Assigned a value of 1 if the chairman and general manager are the same person, or 0 otherwise
	Board size	Board	Ln (total number of board members + 1)
	Board independence	Boainpd	Number of independent directors/total number of board members, (%)
	Managerial ownership	Hold	The ratio of shares held by directors, supervisors, and executives to the total shares outstanding, (%)
	Institutional ownership	Inshare	The percentage of shares held by institutional investors
	Analyst coverage	Analyst	Ln (number of analysts following the firm in the current year + 1)
	CEO tenure	Tenure	Ln (CEO’s tenure + 1), (year)
	Internal CEO promotion	Intpromoted	A dummy variable equal to 1 if the CEO was promoted internally, and 0 otherwise

.

4.3. Model Construction

To examine the U-shaped relationship between non-CEO executive independence and corporate innovation investment, this study employs the following baseline regression model, adopting the non-linear relationship testing approach developed by Wang et al. [43]:

$${Rd}_{i,t}={\alpha }_0+{\alpha }_1{Intgov}_{i,t}+{\alpha }_2{Intgov2}_{i,t}+\sum Controls+\sum Year+\sum Ind+{\epsilon }_{i,t}$$

(1)

where i and t represent enterprise and year, respectively; Rd represents corporate innovation investment, with higher values indicating greater corporate commitment to innovation; Intgov represents non-CEO executive independence, with Intgov2 representing its quadratic term of Intgov; Controls includes all control variables; Year represents year fixed effects; Ind represents industry fixed effects; ${\alpha }_2$ is the coefficient to be estimated, which is expected to be significantly positive; and $\epsilon$ is the error term.

To examine the moderating effect of non-CEO executives’ decision-making horizon and corporate agency costs on the relationship between non-CEO executive independence and corporate innovation investment (H2, H3), we use the following two methods:

Method 1 involves introducing interaction terms of independent variables and moderating variables to model (1), yielding the model (2).

$${Rd}_{i,t}={\beta }_0+{\beta }_1{Intgov}_{i,t}+{{\beta }_2{Intgov}_{i,t}\times M_{i,t}+\beta }_3{Intgov2}_{i,t}+{\beta }_4{Intgov2}_{i,t}\times M_{i,t}+{\beta }_5{M}_{i,t}+\sum Controls+\sum Year+\sum Ind+{\epsilon }_{i,t}$$

(2)

where the meaning of the variable symbol is the same as in model (1). $M_{i,t}$ indicates the moderating variables (non-CEO executives’ decision-making horizon, corporate agency costs). ${\beta }_4$ is the crucial parameter in model (2). If ${\beta }_4$ is significantly positive, it suggests the existence of a positive moderating effect.

Method 2 involves constructing a grouped regression model (3) based on model (1).

$${Rd}_{i,t}={\gamma }_0+{\gamma }_1{Intgov}_{i,t}+{\gamma }_2{Intgov2}_{i,t}+\sum Controls+\sum Year+\sum Ind+{\epsilon }_{i,t} \left(M>M_T or M\le M_T\right)$$

(3)

The meaning of the variable symbol is the same as in model (1). Where $M_T$ is the grouping threshold of the moderating variable M, set to the year–industry median of M.

5. Empirical Analyses

5.1. Basic Statistics

Table 2. Descriptive statistical analysis.

Variable	N	Mean	Std	Min	Max	Median
Rd	19,006	4.571	4.599	0.003	26.395	3.581
Intgov	19,006	0.605	0.335	0	1	0.666
Size	19,006	22.223	1.243	20.064	26.179	22.036
Lev	19,006	0.413	0.194	0.054	0.860	0.409
Tq	19,006	2.051	1.232	0.861	7.926	1.659
Roa	19,006	0.040	0.057	−0.203	0.200	0.038
Cash	19,006	20.373	1.280	17.629	24.195	20.260
PPE	19,006	0.921	0.087	0.538	0.999	0.951
Age	19,006	2.066	0.769	0.693	3.295	2.197
State	19,006	0.351	0.477	0	1	0
Share	19,006	58.542	14.723	24.231	90.272	59.247
Dual	19,006	0.289	0.453	0	1	0
Board	19,006	2.250	0.166	1.791	2.708	2.302
Boainpd	19,006	37.367	5.222	33.330	57.140	33.330
Hold	19,006	14.736	20.010	0	68.073	1.584
Inshare	19,006	0.441	0.255	0.001	0.934	0.457
Analyst	19,006	1.520	1.185	0	3.806	1.609
Tenure	19,006	0.063	0.320	0	2.079	0
Intpromoted	19,006	0.805	0.396	0	1	1

presents the descriptive statistics of the main variables for the full sample. The mean of Rd is about 4.571, with a median of 3.581, a minimum value of 0.003, and a maximum value of 26.395. The standard deviation of Rd is 4.599, indicating significant variation in R&D investment across firms and a relatively dispersed distribution. The mean value of Intgov is 0.605, suggesting that approximately 60% of non-CEO executives, who are appointed before the current CEO, are independent from the CEO. The standard deviation of Intgov is 0.335, and the median is 0.666, reflecting considerable variation in non-CEO executive independence across firms. The mean value of State is 0.351, indicating that state-owned enterprises account for 35.1% of the selected sample. The statistical values of the remaining control variables are consistent with the realities of listed companies in China, showing no significant anomalies.

Table 3. Correlation analysis of key variables.

Variable	Rd	Intgov	Size	Lev	Tq	Roa	Cash	PPE
Rd	1
Intgov	−0.02 ***	1
Size	−0.26 ***	−0.02 ***	1
Lev	−0.32 ***	−0.02 ***	0.52 ***	1
Tq	0.28 ***	−0.04 ***	−0.32 ***	−0.30 ***	1
Roa	0.01	0.01	−0.03 ***	−0.38 ***	0.26 ***	1
Cash	−0.12 ***	−0.01	0.84 ***	0.29 ***	−0.21 ***	0.11 ***	1
PPE	−0.08 ***	0.02 ***	−0.02 ***	0.06 ***	−0.01	0.04 ***	0.05 ***	1
Age	−0.22 ***	−0.08 ***	0.47 ***	0.36 ***	−0.09 ***	−0.19 ***	0.33 ***	−0.02 ***
State	−0.23 ***	0.06 ***	0.37 ***	0.31 ***	−0.14 ***	−0.11 ***	0.29 ***	0.11 ***
Share	−0.04 ***	0.09 ***	0.09 ***	−0.11 ***	−0.04 ***	0.22 ***	0.16 ***	0.02 ***
Dual	0.16 ***	−0.11 ***	−0.19 ***	−0.16 ***	0.08 ***	0.04 ***	−0.14 ***	−0.03 ***
Board	−0.14 ***	0.02 ***	0.24 ***	0.15 ***	−0.10 ***	0.003	0.19 ***	0.0333 *
Boainpd	0.05 ***	−0.01 **	0.01 **	−0.0002	0.02 ***	−0.01	0.02 ***	−0.004
Hold	0.24 ***	−0.01	−0.37 ***	−0.32 ***	0.05 ***	0.14 ***	−0.26 ***	−0.03 ***
Inshare	−0.22 ***	0.04 ***	0.41 ***	0.21 ***	0.01	0.11 ***	0.38 ***	0.06 ***
Analyst	0.03 ***	−0.04 ***	0.33 ***	−0.02 ***	0.18 ***	0.40 ***	0.36 ***	−0.04 ***
Tenure	−0.08 ***	−0.07 ***	0.11 ***	0.09 ***	−0.03 ***	0.0003	0.09 ***	0.02 ***
Intpromoted	0.12 ***	−0.08 ***	−0.15 ***	−0.13 ***	0.05 ***	0.08 ***	−0.10 ***	0.01

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

presents the correlation analysis results for some of the main variables in the full sample. The correlation coefficient between Rd and Intgov is −0.02, which is significant at the 1% level, preliminarily indicating a significant negative correlation between non-CEO executive independence and corporate innovation investment. However, the specific nature of this relationship requires further empirical investigation. Most control variables demonstrate statistically significant correlations with Rd, with the absolute values of the correlation coefficients being below 0.5, falling within a reasonable range. Additionally, variance inflation factors (VIFs) remain well below 10, indicating no substantial multicollinearity concerns in the model.

5.2. Hypothesis Testing

5.2.1. Baseline Results: Impact of Non-CEO Executive Independence on Corporate Innovation Investment

This section presents empirical tests of H1 using Model (1), with results detailed in

Table 4. Baseline results.

Variable	(1)	(2)	(3)	(4)	(5)
Intgov		−0.045	−0.016	−1.378 ***	−1.278 ***
		(−0.51)	(−0.20)	(−4.09)	(−4.13)
Intgov2				1.219 ***	1.153 ***
				(4.11)	(4.26)
Size	−0.751 ***	−0.751 ***	−0.573 ***	−0.748 ***	−0.572 ***
	(−13.01)	(−13.01)	(−10.41)	(−12.95)	(−10.39)
Lev	−5.318 ***	−5.320 ***	−4.298 ***	−5.337 ***	−4.311 ***
	(−25.31)	(−25.30)	(−21.20)	(−25.37)	(−21.26)
Tq	0.842 ***	0.841 ***	0.672 ***	0.843 ***	0.674 ***
	(21.46)	(21.43)	(17.39)	(21.50)	(17.46)
Roa	−16.750 ***	−16.748 ***	−15.557 ***	−16.750 ***	−15.552 ***
	(−20.29)	(−20.29)	(−19.75)	(−20.32)	(−19.77)
Cash	0.995 ***	0.995 ***	0.559 ***	0.993 ***	0.558 ***
	(21.05)	(21.06)	(12.82)	(21.03)	(12.82)
PPE	−3.732 ***	−3.730 ***	−2.671 ***	−3.724 ***	−2.668 ***
	(−9.62)	(−9.63)	(−7.23)	(−9.62)	(−7.23)
Age	−0.625 ***	−0.628 ***	−0.702 ***	−0.638 ***	−0.710 ***
	(−10.39)	(−10.36)	(−12.16)	(−10.52)	(−12.33)
State	−0.162 **	−0.159 **	0.172 **	−0.152 *	0.180 **
	(−2.07)	(−2.03)	(2.32)	(−1.95)	(2.43)
Share	−0.034 ***	−0.034 ***	−0.029 ***	−0.034 ***	−0.029 ***
	(−9.19)	(−9.19)	(−8.89)	(−9.14)	(−8.88)
Dual	0.434 ***	0.430 ***	0.297 ***	0.436 ***	0.303 ***
	(5.95)	(5.91)	(4.47)	(5.99)	(4.56)
Board	−0.595 ***	−0.595 ***	0.024	−0.599 ***	0.026
	(−2.72)	(−2.72)	(0.12)	(−2.73)	(0.13)
Boainpd	0.018 ***	0.018 ***	0.021 ***	0.017 ***	0.021 ***
	(2.73)	(2.73)	(3.56)	(2.64)	(3.48)
Hold	0.024 ***	0.024 ***	0.024 ***	0.024 ***	0.023 ***
	(7.32)	(7.30)	(8.12)	(7.36)	(8.23)
Inshare	−0.354	−0.355	0.797 ***	−0.358	0.756 ***
	(−1.23)	(−1.23)	(3.12)	(−1.24)	(3.07)
Analyst	0.203 ***	0.203 ***	0.365 ***	0.204 ***	0.366 ***
	(6.29)	(6.26)	(11.48)	(6.31)	(11.56)
Tenure	−0.190 ***	−0.192 ***	0.059	−0.187 ***	0.053
	(−2.66)	(−2.68)	(0.86)	(−2.60)	(0.85)
Intpromoted	0.365 ***	0.361 ***	0.196 ***	0.369 ***	0.202 ***
	(5.14)	(5.04)	(2.99)	(5.14)	(3.10)
_cons	8.673 ***	8.709 ***	5.778 ***	8.918 ***	6.004 ***
	(8.76)	(8.77)	(5.69)	(8.98)	(5.95)
Year/Ind	NO	NO	YES	NO	YES
N	19006	19006	19006	19006	19006
Adj.R2	0.243	0.243	0.378	0.244	0.379

Note: z statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01.

. Column (1) presents the baseline model results including only control variables, demonstrating that most control variables significantly affect innovation investment. Columns (2) and (3) report results incorporating the Intgov, with and without controlling for year and industry fixed effects, respectively. The results show the non-significant coefficients of Intgov, suggesting the absence of a linear relationship between non-CEO executive independence and corporate innovation investment. Column (4) introduces the Intgov2 without year and industry controls. The results show that the coefficient of Intgov2 is significantly positive at the 1% level (1.219), while the coefficient of Intgov is significantly negative (−1.378). Column (5) presents results including year and industry fixed effects, which are controlled. The results indicate that the coefficient of Intgov2 remains significantly positive at the 1% level (1.153), and the coefficient of Intgov is significantly negative. These findings support a significant U-shaped relationship between non-CEO executive independence and corporate innovation investment, thereby validating H1.

To further validate whether this non-linear relationship satisfies the conditions of a U-shaped curve, the following steps were taken. First, a U-shaped relationship test was conducted on Column (5). The test results indicate that the model exhibits a significant U-shaped relationship at the 1% statistical level, with an inflection point at 0.554—well within the [0, 1] range of Intgov. This indicates that the relationship between non-CEO executive independence and corporate innovation investment indeed follows a U-shaped curve. Specifically, non-CEO executive independence initially suppresses and subsequently promotes corporate innovation investment. This further confirms that the effect of non-CEO executive independence on corporate innovation investment adheres to a U-shaped relationship. Second, based on Column (5), the value of Intgov at the inflection point was identified, and a new predicted value was generated at this point. A regression discontinuity design (RDD) was then performed. We observe that the regression coefficients at the minimum and maximum values of Intgov are significant, at −0.144 and 0.149, respectively, indicating a significant U-shaped relationship between non-CEO executive independence and corporate innovation investment. In summary, the above analysis provides strong support for H1.

5.2.2. Moderating Effects: Non-CEO Executives’ Decision Horizons

To test H2, following the methods of Acharya, Myers, and Rajan [29], we construct a non-CEO executives’ decision horizon measure (Horizon) based on non-CEO executives’ age and tenure, calculated as ${Horizon}_{i,t}=\left({INAge}_{i,t}-{NAge}_{i,t}\right)+\left({INTenure}_{i,t}-{NTenure}_{i,t}\right)$, where ${INAge}_{i,t}$ represents the average age of the top four highest paid non-CEO executives in the industry of firm i in year t; ${NAge}_{i,t}$ represents the average age of the top four highest paid non-CEO executives in firm i in year t; ${INTenure}_{i,t}$ represents the average tenure of the top four highest paid non-CEO executives in the industry of firm i in year t; and ${NTenure}_{i,t}$ represents the average tenure of the top four highest paid non-CEO executives in firm i in year t. Lower Horizon values correspond to older executives with longer tenures, reflecting a shorter decision-making horizon. Building upon model (2), we incorporate an interaction term between non-CEO executive independence and decision horizon, with results reported in Column (1) of

Table 5. Results of the moderating effects: non-CEO executives’ decision horizons.

	(1)	(2)	(3)
	Interaction Term	Long Decision Horizon	Short Decision Horizon
Intgov	−1.111 ***	−1.560 ***	−0.672
	(−3.52)	(−3.59)	(−1.52)
Intgov × Horizon	−0.131 **
	(−2.01)
Intgov2	1.070 ***	1.539 ***	0.629 *
	(3.87)	(3.90)	(1.68)
Intgov2 × Horizon	0.119 **
	(2.10)
Horizon	0.040 **
	(2.38)
Controls	YES	YES	YES
_cons	5.477 ***	1.817	8.471 ***
	(5.28)	(1.09)	(6.51)
Year/Ind	YES	YES	YES
N	19006	9471	9535
Adj.R2	0.379	0.388	0.369
p-value		0.093

Note: z statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01. The p-values for the between-group coefficient differences are calculated based on the estimation results of the seemingly unrelated regression (SUR) model and are largely consistent with the results obtained from Fisher’s exact test and the Chow test.

. The results show that the coefficient of the interaction term (Intgov2 × Horizon) is significantly positive at the 5% level (0.119), suggesting that non-CEO executives’ decision horizon strengthens the U-shaped relationship between non-CEO executive independence and corporate innovation investment.

Furthermore, building upon Model (3), we partition the full sample into long (above year–industry median of Horizon, coded as 1) and short (below or equal to median of Horizon, coded as 0) decision horizon groups, with results displayed in Columns (2) and (3) of Table 5. The results show that in the long decision horizon group, the coefficient of Intgov2 is significantly positive at the 1% level (1.539), while in the short decision horizon group, the coefficient of Intgov2 is significantly positive at the 10% level (0.629). This suggests that the longer the non-CEO executives’ decision horizons, the stronger the U-shaped relationship between non-CEO executive independence and corporate innovation investment.

In summary, these findings provide robust support for H2.

5.2.3. Moderating Effects: Corporate Agency Costs

To examine H3, we construct an agency cost index (Fee) based on management and selling expenses, calculated as follows:

$${Fee}_{i,t}=\left[{\left(Management expenses\right)}_{i,t}+{\left(Sales expenses\right)}_{i,t}\right]÷{\left(Operating revenue\right)}_{i,t}$$

where i and t represent the firm and year, respectively. Higher Fee values indicate greater agency costs within the firm. Building upon Model (2), we incorporate an interaction term between non-CEO executive independence and corporate agency costs, with results present in Column (1) of

Table 6. Results of the moderating effects: corporate agency costs.

	(1)	(2)	(3)
	Interaction Term	High Agency Cost	Low Agency Cost
Intgov	−1.066 ***	−2.456 ***	−0.025
	(−3.76)	(−4.73)	(−0.09)
Intgov×Fee	−8.939 **
	(−2.38)
Intgov2	0.960 ***	2.036 ***	0.174
	(3.83)	(4.51)	(0.72)
Intgov2 × Fee	6.182 *
	(1.88)
Fee	10.776 ***
	(28.74)
Controls	YES	YES	YES
_cons	0.850	4.859 ***	2.043 **
	(0.86)	(2.66)	(2.12)
Year/Ind	YES	YES	YES
N	19006	9499	9507
Adj.R2	0.440	0.397	0.384
p-values		0.0003

Note: z statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01. The p-values for the between-group coefficient differences are calculated based on the estimation results of the seemingly unrelated regression (SUR) model and are largely consistent with the results obtained from Fisher’s exact test and the Chow test.

. The coefficient of the interaction term (Intgov2 × Fee) is significantly positive at the 10% level (6.182), suggesting that agency costs strengthen the U-shaped relationship between non-CEO executive independence and corporate innovation investment.

Building upon Model (3), we further partition the full sample into high (above year–industry median of Fee, coded as 1) and low (below or equal to median of Fee, coded as 0) agency cost groups, with results displayed in Columns (2) and (3) of Table 6. The results show that in the high agency cost group, the coefficient of Intgov2 is significantly positive at the 1% level (2.036), while in the low agency cost group, the coefficient of Intgov2 is not significant. This suggests that the higher the corporate agency costs, the stronger the U-shaped relationship between non-CEO executive independence and corporate innovation investment.

In summary, these findings provide robust support for H3.

5.3. Robustness Test

To ensure the reliability of the empirical test results, we conduct robustness tests to examine the validity of the impact of non-CEO executive independence on corporate innovation investment. Specifically, the following methods were employed:

Firstly, we use alternative variables of corporate innovation investment. On the one hand, to mitigate the potential impact of earnings management on operating revenue, this study re-measures corporate innovation investment (RD) using the ratio of R&D investment to total assets. The specific calculation formula is as follows: RD = 100 × (R&D investment ÷ Total assets). This alternative variable (RD) is incorporated into Model (1), with results presented in Column (1) of

Table 7. Robustness test results.

	(1)	(2)	(3)	(4)	(5)
Variable	RD	Patent	Rd	Rd	Rd
Intgov	−0.200	−0.365 ***	−0.940 ***	−6.483 ***	−1.092 ***
	(−1.42)	(−3.45)	(−3.44)	(−5.61)	(−2.83)
Intgov2	0.220 *	0.273 ***	0.837 ***	3.349 ***	0.959 ***
	(1.81)	(2.94)	(3.49)	(4.40)	(2.84)
Controls	YES	YES	YES	YES	YES
_cons	3.081 ***	−13.066 ***	5.504 ***	10.414 ***	6.730 ***
	(6.59)	(−34.86)	(5.93)	(9.16)	(5.45)
Year/Ind	YES	YES	YES	YES	YES
N	19006	19006	23809	19006	10170
Adj.R2	0.324	0.372	0.361	0.383	0.372

Note: z statistics in parentheses. * p < 0.1, *** p < 0.01.

. The coefficient of Intgov2 is significantly positive at 0.220, consistent with the previous findings. On the other hand, following the method of Gao, Jiang, and Mekhaimer [11], this study re-measures corporate innovation investment (Patent) using the number of patent applications. The specific calculation formula is as follows: Patent = Ln (Number of invention patents filed in the year + Number of utility model patents filed in the year + 1). The variable Patent is introduced into Model (1) for re-estimation, and the results are reported in Column (2) of Table 7. The coefficient of Intgov2 is significantly positive at 0.273, further corroborating our findings.

Secondly, we use alternative variables of non-CEO executive independence. On the one hand, we retain observations with fewer than four non-CEO executives and re-estimate Model (1), with results presented in Column (3) of Table 7. The coefficient of Intgov2 is significantly positive at 0.837, consistent with our primary findings. On the other hand, following existing research [36], this study re-measures non-CEO executive independence using the proportion of the non-CEO executives ranked in the top four for compensation. The specific calculation formula is as follows: Intgov = (Number of non-CEO executives ranked in the top four for compensation ÷ Total number of non-CEO executives). Higher Intgov values indicate stronger internal governance within the executive team. The new Intgov is substituted into Model (1) for re-estimation, and the results are reported in Column (4) of Table 7. The coefficient of Intgov2 is significantly positive at 3.349, further validating our results.

Lastly, in order to avoid potential effects from the 2018 accounting standard changes, we exclude post–2017 observations on this robustness check. We re-estimate Model (1) using the 2007–2017 subsample, with results presented in Column (5) of Table 7. The coefficient of Intgov2 is significantly positive at the 1% level (0.959), consistent with our primary findings.

5.4. Endogeneity Test

To address the endogeneity issue, this study uses the following methodology: Firstly, although we control for observable factors potentially affecting empirical results, the number of non-CEO executives appointed prior to the CEO’s tenure were not randomly distributed, and their proportion remains subject to CEO influence. Furthermore, the number of non-CEO executives independent of the CEO might still be influenced by other latent confounding factors [34]. Consequently, our findings may be susceptible to self-selection bias in the sample. To mitigate endogeneity issues arising from self-selection bias, this study employs propensity score matching (PSM) to construct paired samples as a robustness check. The samples after matching are significantly more balanced on the core covariates, thereby strengthening the credibility of causal inference. Specifically, the methods are as follows: We classify samples based on whether Intgov values exceed industry-year medians. Samples with Intgov less than or equal to the industry–year median are classified as the low non-CEO executive independence group, while those with Intgov greater than the median are classified as the high non-CEO executive independence group. The following characteristic variables are selected for Logit regression: firm size (Size), leverage ratio (Lev), corporate growth opportunities (Tq), profitability (Roa), cash holdings (Cash), fixed asset ratio (PPE), firm age (Age), state ownership (State), ownership concentration (Share), CEO duality (Dual), board size (Board), board independence (Boainpd), managerial ownership (Hold), institutional ownership (Inshare), analyst coverage (Analyst), CEO tenure (Tenure), and internal CEO promotion (Intpromoted). Based on propensity scores, we implement 1:1 nearest-neighbor matching with replacement. The balance test results show that, after matching, the standard deviations of all matching variables between the low and high non-CEO executive independence groups are less than 10%, indicating good matching quality. The matched sample is then used to re-estimate Model (1), and the results are reported in Column (1) of

Table 8. Results of the endogeneity test.

	(1)	(2) First Stage	(3) Second Stage	(4)
Variable	Rd	IfRd	Rd	F1Rd
Intgov	−1.523 ***		−1.190 ***	−1.623 ***
	(−3.40)		(−3.66)	(−4.41)
Intgov2	1.420 ***		1.071 ***	1.413 ***
	(3.68)		(3.75)	(4.44)
IMR			0.229
			(0.09)
IV		2.078 *
		(1.71)
Controls	YES	YES	YES	YES
_cons	6.029 ***	−2.503 *	6.620 ***	7.673 ***
	(4.23)	(−1.67)	(4.67)	(6.53)
Year/Ind	YES	YES	YES	YES
N	10019	16897	16756	14118
Adj.R2	0.381	0.215	0.379	0.363

Note: z statistics in parentheses. * p < 0.1, *** p < 0.01.

. The coefficient of Intgov2 is significantly positive at 1.420. The matched sample results remain consistent with our hypothesis tests, demonstrating the robustness of the U-shaped relationship between non-CEO executive independence and innovation investment after addressing self-selection bias.

Secondly, due to the presence of missing values in corporate innovation data, which may lead to endogeneity issues caused by sample selection bias, we employ the Heckman two-stage model. The model implements a two-stage regression framework. Stage I specifies the selection equation by incorporating instrumental variables (IV) to estimate the propensity for sample inclusion. Stage II subsequently estimates the structural equation conditional on these predicted probabilities, thereby effectively mitigating selection bias arising from non-random sampling. Specifically, the methods are as follows: We introduce IfRd, a dummy variable indicating whether a firm discloses R&D investment (1 = discloses R&D inputs; 0 = otherwise). Following the methods of existing research [44], this study uses the mean value of non-CEO executive independence of other firms in the same region and year as an instrumental variable (IV). In the first stage, we estimate a Probit model with IV to estimate the probability of whether a firm will disclose R&D investment, generating the Inverse Mills Ratio (IMR), with results presented in Column (2) of Table 8. The second stage incorporates IMR as a control variable in Model (1) for firms disclosing R&D inputs, with results displayed in Column (3) of Table 8. The coefficient of Intgov2 is significantly positive at 1.071, consistent with our primary findings. This indicates that the results are robust.

Lastly, to mitigate potential endogeneity issues arising from reverse causality, this study re-examines Model (1) by lagging all explanatory and control variables by one period. The lagged variables are theoretically posited to exhibit diminished contemporaneous correlation with the current dependent variable, thereby effectively mitigating endogeneity concerns stemming from reverse causality. The results are reported in Column (4) of Table 8. The coefficient of Intgov2 is significantly positive at the 1% level (1.413), indicating that reverse causality does not affect the U-shaped relationship between non-CEO executive independence and corporate innovation investment.

6. Further Analysis

6.1. Heterogeneity Test: CEO Power

Power influences individuals’ psychological cognition, and possessing power can significantly alter individual behavioral patterns. As the primary decision-maker, CEOs with greater power exert stronger influence over corporate strategic decisions. A high-power CEO has considerable discretion in appointing non-CEO executives, thereby weakening the supervisory effect of non-CEO executives over the CEO. Increased CEO power reduces non-CEO executives’ psychological safety, limiting their ability to express dissenting opinions and fostering compliance with CEO decisions. As CEO power grows, their influence over non-CEO executives expands, exacerbating communication asymmetries within the executive team and increasing CEO decision-making arbitrariness [3], potentially diminishing the effect of non-CEO executive independence on innovation investment. Therefore, we hypothesize that the effect of non-CEO executive independence on innovation investment is more pronounced in firms with less powerful CEO.

To examine the heterogeneity of CEO power (CP), we partition the full sample into high (above annual–industry median of CP) and low (below or equal to median of CP) CEO power groups, conducting separate regressions based on Model (1). Following Haynes and Hillman [45] and Finkelstein [46], we measure CEO power across four dimensions—structural power (SP), ownership power (OP), expert power (EP), and reputational power (RP), using eight binary indicators. Specifically, SP is measured by whether the CEO also serves as a board member and whether they are an internal director of the company (1 if yes, 0 otherwise); OP is measured by whether the CEO holds equity in the company and whether institutional investor ownership is below the industry median (1 if yes, 0 otherwise); EP is measured by whether the CEO holds a senior professional title and whether their tenure exceeds the industry median (1 if yes, 0 otherwise); and RP is measured by whether the CEO has a high level of education (1 for a master’s degree or higher, 0 otherwise) and whether the CEO holds external part-time positions (1 if yes, 0 otherwise). We compute a composite CEO power index by averaging these eight binary indicators. Regression results in Columns (1) and (2) of

Table 9. Heterogeneity test results.

	(1)	(2)	(3)	(4)	(5)	(6)
	High CEO Power	Low CEO Power	High Corporate Governance	Low Corporate Governance	SOEs	Non-SOEs
Intgov	−1.473 ***	−1.042 ***	−0.405	−2.029 ***	0.096	−1.946 ***
	(−2.75)	(−2.75)	(−1.00)	(−4.35)	(0.23)	(−4.78)
Intgov2	1.183 **	1.068 ***	0.539	1.665 ***	0.153	1.629 ***
	(2.43)	(3.28)	(1.51)	(4.09)	(0.41)	(4.56)
Controls	YES	YES	YES	YES	YES	YES
_cons	6.524 ***	6.021 ***	4.382 ***	8.582 ***	9.999 ***	2.949 *
	(3.36)	(4.98)	(3.29)	(5.15)	(8.01)	(1.78)
Year/Ind	YES	YES	YES	YES	YES	YES
N	6189	12817	9507	9499	6676	12330
Adj.R2	0.402	0.366	0.358	0.385	0.327	0.369
p-values	0.843		0.036		0.004

Note: z statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01. The p-values for the between-group coefficient differences are calculated based on the estimation results of the seemingly unrelated regression (SUR) model and are largely consistent with the results obtained from Fisher’s exact test and the Chow test.

show significant positive Intgov2 coefficients for both high (p < 0.01) and low (p < 0.05) CEO power groups, with no statistically significant difference in coefficient magnitude. These findings suggest that non-CEO executive independence has a stronger effect on innovation investment in firms with less powerful CEO.

6.2. Heterogeneity Test: Corporate Governance Quality

Effective corporate governance typically aligns managerial decision-making motivations with those of shareholders and, to some extent, constrains management’s self-serving behaviors [47]. The level of corporate governance directly reflects the effectiveness of the firm’s governance structure [48]. The corporate governance structure represents formal institutional arrangements that constitute external governance mechanisms for executive teams. In contrast, non-CEO executive independence serves as an informal internal governance mechanism within the executive team. Consequently, in firms with lower levels of corporate governance, external governance of the executive team is relatively insufficient, making it easier for non-CEO executive independence to exert internal governance effects. Conversely, when corporate governance levels are high, the external governance effects of the executive team are strengthened, potentially weakening the informal governance effects of non-CEO executive independence. Therefore, we hypothesize that non-CEO executive independence has a greater impact on innovation investment in firms with lower levels of corporate governance.

To examine the heterogeneity of corporate governance levels (Gov), we partition the sample into high (above annual–industry median of Gov) and low (below or equal to median of Gov) governance groups, conducting separate regressions based on Model (1). Following Zhang and Lu [49], we measure corporate governance levels using eight indicators: the shareholding ratio of the largest shareholder, the combined shareholding ratio of the second to tenth largest shareholder, ownership nature, the shareholding ratio of the institutional investor, CEO duality, the shareholding ratio of the managerial, independent director ratio, and supervisory board meeting frequency. We construct a comprehensive measure of corporate governance levels index through principal component analysis of these eight indicators. The first principal component extracted from the PCA is used as the measure of Gov, with higher values indicating stronger corporate governance. Regression results are presented in Columns (3) and (4) of Table 9. In the low corporate governance group, the coefficient of Intgov2 is significantly positive at the 1% level (1.665), and the coefficient of Intgov is significantly negative. In contrast, in the high corporate governance group, the coefficients of Intgov2 and Intgov are not significant. The coefficient magnitudes differ significantly between groups. In summary, these findings suggest substitutability between formal governance mechanisms within the firm and the informal governance effect of non-CEO executive independence on the CEO, with the impact of non-CEO executive independence on corporate innovation investment being more pronounced in firms with lower levels of corporate governance.

6.3. Heterogeneity Test: Ownership Structure

Ownership nature is a significant factor influencing corporate investment decisions. In state-owned enterprises (SOEs), managerial positions are often subject to appointment arrangements, resulting in unstable executive positions, which weakens the internal supervisory and governance effects of non-CEO executives within the executive team. In contrast, non-state-owned enterprises (non-SOEs) typically exhibit effective oversight and more stable management structures, where executive promotions are closely tied to individual competence, amplifying the influence of individual executive characteristics on corporate investment decisions [17] and enhancing non-CEO executives to exert internal governance effects within the executive team. Furthermore, due to issues such as the absence of owners and insider control, SOEs often grant CEOs greater decision-making authority, including key personnel appointments and dismissals. Additionally, inadequate incentive mechanisms and power–responsibility imbalances in SOEs exacerbate CEO self-serving behaviors, weakening the impact of non-CEO executive independence on corporate innovation investment. Therefore, we hypothesize that the effect of non-CEO executive independence on corporate innovation investment is more pronounced in non-SOEs.

To examine the heterogeneity of ownership nature, we build on the baseline regression model and partition the sample into state-owned (State = 1) and non-state-owned enterprises (State = 0), with regression results presented in Columns (5) and (6) of Table 9. The results show that in the non-SOE group, the coefficient of Intgov2 is significantly positive at the 1% level (1.629), and the coefficient of Intgov is significantly negative. In contrast, in the SOE group, the coefficients of Intgov2 and Intgov are not significant. The coefficient magnitudes differ significantly between ownership types. These findings demonstrate that the impact of non-CEO executive independence on corporate innovation investment is more pronounced in non-SOEs.

7. Discussion

We conducted this study to explore when and how executive team internal governance improves corporate innovation investment. The relationship between our research findings and previous studies is reflected in the following important aspects:

First, we examined the impact of non-CEO executives’ internal governance on corporate innovation investment. Our findings reveal the existence of executive team internal governance effects, confirming the results of Acharya et al. [29]. Acharya et al. [29] support the role of bottom–up governance by non-CEO executives, suggesting that a firm’s management team comprises diverse agents with different career horizons [13], incentive structures [37], and growth opportunities [11]. Given their capacity to influence CEO decision-making, non-CEO executives may push CEOs toward decisions that prioritize long-term corporate development. Our theoretical framework incorporates the special association between CEOs and non-CEO executives, demonstrating that non-CEO executives’ independence influences corporate long-term-oriented decision-making through a bottom–up governance mechanism, thereby validating the executive team internal governance effect. We further find that non-CEO executives influence corporate innovation investment by monitoring and constraining CEO behavior, which confirms the findings of Xie et al. [19]. These results respond to Fama’s [50] call for further research on internal governance mechanisms.

Second, building on threshold theory, we examine the U-shaped relationship between non-CEO executives’ internal governance and corporate innovation investment, confirming that the governance effect of non-CEO executives is not uniformly effective. This finding aligns with Du and Wang [36], suggesting that only when non-CEO executive independence exceeds a critical threshold does internal governance exert a positive effect.

Finally, building on internal governance theory, we incorporate contingency theory to explore how the impact of non-CEO executives’ internal governance on corporate innovation investment is contingent on contextual factors. Our findings demonstrate how non-CEO executives’ internal governance affects corporate innovation investment across varying levels of CEO power, corporate governance quality, and ownership structure. These findings are consistent with existing studies [11,34], which show that the effect of non-CEO executives’ internal governance on corporate investment decisions exhibits differential impacts across distinct internal and external contexts, such as dictatorships, inside directing, CEO power, accounting information quality, and external supervision.

These relationships to prior studies highlight how our study advances the understanding of the underlying mechanisms linking internal governance to corporate innovation investment.

8. Conclusions and Prospects

8.1. Conclusions

This study explores the impact of bottom–up governance mechanisms within the executive team on corporate innovation investment from the perspective of non-CEO executive independence. Using a sample of Chinese A-share listed companies on the Shanghai and Shenzhen (2007–2021), we identify a significant U-shaped relationship between non-CEO executive independence and innovation investment. After addressing endogeneity issues and conducting a series of robustness tests, we confirm that non-CEO executive independence initially suppresses and subsequently promotes corporate innovation investment. The conclusion indicates that the impact of non-CEO executive independence on corporate innovation investment exhibits a threshold effect. When non-CEO executive independence is below the threshold, it inhibits corporate innovation investment; while above the threshold, higher levels of non-CEO executive independence facilitate corporate innovation investment. Furthermore, we find that the non-CEO executives’ decision horizon and corporate agency costs positively moderate this U-shaped relationship. Heterogeneity analysis demonstrates that this U-shaped effect is more pronounced in firms with less CEO power, lower levels of corporate governance, and non-state-ownership.

Based on our conclusions, the managerial implications of this study are as follows: Firstly, improve corporate governance mechanisms and encourage non-CEO executives within the top management team to actively participate in corporate governance. Companies should fully recognize the bottom–up supervisory role of non-CEO executives in strategic decision-making. When designing internal governance mechanisms, firms can implement differentiated incentive measures, focusing on executives who play significant supervisory roles. Secondly, establish a scientific and reasonable governance structure and strengthen the board of directors’ management of non-CEO executive appointments. Appropriately constraining the CEO interference in hiring process can enhance corporate governance levels and effectively promote corporate innovation. From the perspective of the effectiveness of non-CEO executive governance, the positive effect of non-CEO executive independence on corporate innovation investment is not always effective. Only when non-CEO executive independence falls within an appropriate range can independent non-CEO executives exert their supervisory role and drive corporate innovation. Lastly, firms should comprehensively consider contextual factors when building the executive team. The internal governance effects of non-CEO executives on corporate innovation vary under different contexts, such as CEO power, corporate governance levels, and ownership nature. To promote innovation investment, firms should implement context-specific hiring and incentive strategies for non-CEO executives.

8.2. Limitations and Future Research Directions

While this study offers significant theoretical contributions, the following three limitations warrant further investigation:

Firstly, the research sample focuses on listed companies within China’s unique institutional context, which may limit the generalizability of the findings to other institutional settings, particularly emerging economies or developed markets with distinct ownership structures and governance cultures, where the mechanisms through which non-CEO executive independence influences corporate innovation investment may exhibit heterogeneity. Future research could employ cross-context comparative studies to test such variations across institutional settings.

Secondly, although heterogeneity analysis reveals that the U-shaped effect of non-CEO executive independence on innovation investment is more pronounced in non-state-owned enterprises (non-SOEs), the differences in executive team internal governance mechanisms between SOEs and non-SOEs remain underexplored. Future research could integrate institutional theory and resource dependence theory to examine how power sources of non-CEO executives (government-appointed or market-selected professional managers, etc.) under different ownership structures shape their internal governance effects within the executive team.

Finally, the heterogeneity analysis does not fully account for the potential impact of industry-specific variations. Industry characteristics such as innovation intensity, capital structure, and regulatory stringency may reconfigure the relationship between non-CEO executive independence and innovation investment. Future studies could examine this relationship by segmenting industries for cluster analysis and further develop multi-level models incorporating industry features to uncover contingent patterns of internal governance mechanisms in “industry-firm” linkage contexts.