A team is a dynamic organizational structure, and everyone in the team depends to some extent on cooperation with others to complete work tasks (Alavi and McCormick, 2018). The increasing interdependence of human functions encourages the role of collective subjectivity (Bandura, 2000). Collective efficacy is a subjective understanding of team members’ abilities, and it is the common belief of team members that the team can successfully complete a specific task (Chen et al., 2019; Luo and Lin, 2022). Once collective efficacy enters the belief structure of TMT members, it has a significant impact on individual and team entrepreneurial behavior (Srivastava et al., 2006). In this process, collective efficacy reflects the organizational commitment and responsibility of TMT members to the team and influences individual behaviors in the team (Chen et al., 2019). These include individual’s perceptions and attitudes toward entrepreneurship, how much effort they put into performing group tasks, how well they can work with team members to make the right decisions, and how well they can maintain their previous level of effort when collective efforts fail to achieve goals quickly or when they face opposition (Gibson and Earley, 2007; Donohoo et al., 2018). This view also predicts that high levels of entrepreneurial-oriented behavior are based on TMT collective efficacy.

Social cognitive theory emphasizes the role of psychological cognitive states (Bandura, 2001). From the psychological level, collective efficacy is the effective coordination and integration of self-efficacy (Bandura, 1997). Since the CEO has absolute control and dominance over the team resources, most of the strategic decisions related to the survival and development of the company are made by the CEO (Zhong et al., 2022), and the CEO’s decisions are implemented through the TMT members (Li and Jones, 2019). In this case, on the one hand, entrepreneurial-oriented behaviors are decisions and actions made under the CEO’s bounded rationality. Due to the limitations of his prior experience and cognitive level, the decision-making process also requires the participation of TMT members. TMT collective efficacy can minimize the apathy, passivity, and inaction of TMT (Bandura et al., 1999; Luo and Lin, 2022), while improving their decision-making ability (Donohoo et al., 2018). In a TMT with high collective efficacy, the CEO believes in himself and the enterprise’s ability to achieve the desired results (Chen et al., 2019). Thus, they will constantly follow the market trail, remain vigilant of competitors, seize new market opportunities ahead of competitors, and launch innovative products and services (Elms et al., 2022). On the other hand, in an enterprise, if TMT members have a strong sense of collective efficacy as well as collective power and faith (Alavi and McCormick, 2018), then TMT members can substantially impact their entrepreneurial activity. In particular, faith will stimulate the confidence of the CEO to successfully play the entrepreneurial role and complete the entrepreneurial task. Also, to make decisions in favor of entrepreneurial orientation activities.

However, entrepreneurship is not always successful (Tasa and Whyte, 2005; Wynn and Jones, 2019). TMT characterized by high collective efficacy are more likely to recover from failure and are more likely to create a positive work environment and emotional climate (DeRue et al., 2010; Rapp et al., 2021). This is because the positive beliefs brought by TMT collective efficacy can inhibit and alleviate this failure cognition, thus effectively controlling the team’s anxiety and reducing the negative emotions associated with the execution of corporate tasks (Elms et al., 2022). In this case, the executive team tends to view the situation as an opportunity full of positivity and development, rather than complete negativity and danger. CEOs are also more willing to take risks associated with active competitive behavior and are more willing to redo Innovative thinking (Knight et al., 2001), always stick to the task direction, and implement entrepreneurial-oriented decision-making. Based on this, we propose that:

As a psychological perception with strong subjectivity, TMT collective efficacy is mainly generated from the interdependence and interaction of TMT members (Banks et al., 2014; Chen et al., 2019). Research has shown that the most effective interactions occur between the most powerful member of the team (the CEO) and other members of TMT. In the management of the TMT, the CEO should not only pay attention to whether the corporate goals can be achieved, but also whether he or she and other members of the TMT can communicate and support each other (Herdman et al., 2017). Therefore, CEO-TMT interface is particularly important (Lin and Lin, 2019). CEO-TMT interface refers to the dynamic process in which the CEO and other members of TMT connect and interact with each other (Georgakakis et al., 2022). It is the opposite of a state of “either I or he.” Such interaction can only be achieved when the team presents a kind of mutual identification and integration (Qiao et al., 2021).

Due to the limited personal ability and resources of the CEO, it is impossible to comprehensively identify and judge the internal conditions and external environment of the enterprise (Hambrick, 2007). Differences in the characteristics of the CEO and other members of TMT will lead to different values and cognitive behaviors of team members, which will affect the perspective of TMT members on the problem and the way to deal with the problem. CEO-TMT interface is the result of their interaction (Simsek et al., 2018). Although the shared beliefs of team members increase when TMT collective efficacy comes into play, the CEO and other members of TMT are also prone to an “us-them” adversarial situation. At this time, serious communication problems and conflicts often occur within the TMT (Bezrukova et al., 2012), which can weaken the TMT’s collective efficacy and reduce the quality of decision-making.

Prior research underscored that a successful TMT requires effective interaction between members, especially between the CEO and members of the TMT (Bachrach et al., 2022). On the one hand, as the interaction between the CEO and other members of TMT becomes stronger and stronger, the information exchange among members will increase and show a diversified pattern (Herdman et al., 2017), and the emotional conflict between members will gradually decrease (Yang et al., 2021). TMT collective efficacy will be significantly improved. On the other hand, the higher the degree of similar interaction between the CEO and other members of TMT, the more they agree with each other (Buyl et al., 2017), and the risk conflict, innovation conflict, and cognitive conflict between the two will be reduced, thereby promoting the integration of team behavior (Li and Liao, 2014) and improving the efficiency of decision-making (Lin and Lin, 2019), which has a positive impact on entrepreneurial orientation (Figure 1). Thus, we propose that:

To evaluate the hypotheses of this study, we adopted a quantitative analysis method. Considering that there are no secondary public data for our selected variables, such as enterprise team and operation mechanism, this study adopted the questionnaire survey technique to collect data which is a valid prior research practice (Li et al., 2021; Chew et al., 2022; Xiabao et al., 2022). Cross-sectional surveys in China have mainly necessitated the translation of the original (English) questionnaire to simplified Chinese or Mandarin to help participants fully grasp questions and provide answers that represent their experiences or phenomenon of their enterprises (Xiabao et al., 2022). Borrowing insights from past approaches (Brislin, 1970), a team of bilingual translators performed a back-translation which consisted of an iterative procedure of repeated independent translation and back-translation and engaged different independent bilingual translators (Triandis and Brislin, 1984). Bilingual translators translated each instrument/item from the original English language to the Chinese language. Errors discovered were repeatedly checked until the team was convinced of a good similarity (concept equivalence) and the absence of grammatical problems between the two versions of the questionnaire. The questionnaire had three parts. The first part comprised control variables to help gather information on the samples’ demographic statistics. The second and third parts comprised widely used and well-validated measurement scale items retrieved from past studies, for instance, the works of Luo and Lin (2022) and Georgakakis et al. (2016). The items assessed our sample on the main constructs (see the measures section). The preliminary sampling phase focused on soliciting respondents’ feedback to confirm the suitability and clarity of the Chinese version of the questionnaire as well as ascertain whether the questions met the expected outcomes per their enterprises’ activities. For instance, participants were provided with definitions and brief descriptions of entrepreneurial orientation. Afterward, final concept equivalence checks were made to the Chinese questionnaire and administered to the participants through email and WeChat, a Chinese social media application. To encourage response, after sending out the questionnaires, we arranged for special personnel to make follow-up calls and serve reminders to facilitate respondents’ cooperation. In addition, we set “trap questions” in the questionnaire, such as “please select the second option from the left,” to ascertain whether the respondents were attentive and had a keen interest in the survey.

We chose China to conduct the research survey because China is a country with active entrepreneurship, fast economic growth, and is recognized as having a rapidly rising entrepreneurial ecosystem of domestic enterprises. In addition, considering the differences in the level of economic development and marketization among different regions in China, the levels of innovation and entrepreneurship are different to a certain extent. Therefore, some Chinese representative regions are eligible samples for research.

We engaged 10 CEOs and 210 TMTs from 10 enterprises, totaling 220 samples. The sample was drawn from a database of 1,000 enterprises in the high-tech industrial development zones of nine provinces in the Yangtze River Delta (Shanghai, Jiangsu, Anhui, Zhejiang), Pearl River Delta (Guangdong, Shenzhen), Bohai Rim (Shandong), Central China (Henan), and northwest China (Shanxi). This research collected the data through two main channels. First, we got in touch with the market supervision bureaus and tax authorities of each province and sent questionnaires to the corresponding entrepreneurial enterprises (e.g., intellectual property demonstration enterprises) through the market supervision bureaus and tax authorities of 16 provinces. Second, through an existing relationship network, we liaised with acquaintances to contact the senior management of the enterprises and sent questionnaires after their consent. The respondents of the questionnaire included CEOs and TMTs of the enterprises. Although we had less participation, all 220 (out of 1,000 copies of the questionnaire distributed) responses gathered were logical. This represents a 22% recovery rate.

To ensure the validity and reliability of the measurement tool, this study used the scales of high-quality academic journal publications (Luo and Lin, 2022) and made it conform to the Chinese context according to the suggestions of relevant professionals. In the survey, to avoid the likelihood of common method bias caused by self-reported data, the items of entrepreneurial self-efficacy were answered by TMT members while items of TMT collective efficacy were addressed by the CEO. The items measuring CEO–TMT interface and entrepreneurial orientation were addressed jointly by each enterprise’s CEO and TMT members. All items were scored on a 5-point Likert scale, with 1—strongly disagree and 5—strongly agree.

We measured entrepreneurial orientation based on Miller (1983),s conceptualization with the modifications introduced by Covin and Slevin (1989) and Covin and Miller (2014). Consistent with these studies, we also employed entrepreneurial orientation as a unidimensional construct involving three dimensions with nine items: innovation (measured with items EO1–EO3); proactivity (items EO4–EO6); and risk-taking (items EO7–EO9). This scale captures an enterprise’s entrepreneurial orientation-related attitude and has been frequently used as well as received high reliability and validity ratings (Covin and Wales, 2012). The Cronbach’s alpha (α) coefficient of the scale was 0.965.

For the measurement of entrepreneurial self-efficacy, this study utilized the 10 measurement items of Schwarzer et al. (1997) and McGee et al. (2009). We asked the participants to evaluate their perceptions of various behaviors, which are robust predictors of enterprise performance (Miller, 1983). Examples of items include: I was able to provide new ideas for existing products or services. The Cronbach’s alpha (α) coefficient of the scale was 0.949.

We measured TMT collective efficacy with four items adapted from Luo and Lin (2022). Sample items include: Our TMT is capable of handling heated competition with direct rivals. The Cronbach’s alpha (α) coefficient of the scale was 0.817.

The research project of Georgakakis et al. (2016) mainly discussed the moderating effect of CEO–TMT interface from five dimensions: career experience, political background, educational background, functional background, and social experience. There were five items in total. Sample items include: CEO and TMT share similar common tenure. The Cronbach’s alpha (α) coefficient of the scale was 0.915. Table 2 shows the reliability and validity of these variables.

We controlled for CEO gender, age, education, past performance, tenure in post, and asset-liability ratio as well as enterprise age, size, and type based on their possible influence on the occurrence of entrepreneurial orientation. CEO gender was controlled following the prior notion that it relates to entrepreneurial enthusiasm which can decrease with the age of CEOs (Levesque and Minniti, 2006). Similarly, controlling for the CEO’s educational level is consistent with the belief that it affects their entrepreneurship activities (Hamilton, 2012). Scholars also found that the past performance of an enterprise can increase the scale of redundant resources to shape innovative and entrepreneurial activities (Zahra et al., 2000). Enterprise age may affect the number of resources and opportunities that an enterprise obtains (Salunke et al., 2013) and consequently affects entrepreneurship. In addition, the asset-liability ratio of enterprises largely determines the ability of enterprises to obtain external resources, which can also affect entrepreneurial activities. Based on the resource-based theory, entrepreneurial orientation is driven by innovation, which generates resource demands in the process of innovation and drives enterprises to acquire different resources. Therefore, when the types of enterprises (high-tech and non-high-tech companies) are inconsistent, the impact on entrepreneurial direction is also different (Huang and Wang, 2013). We operationalized the control variables following extant studies (David et al., 1998; Zahra et al., 2000; Sinatoko Djibo et al., 2022; Xiabao et al., 2022). Accordingly, CEO gender was coded with 1 = representing male and 2 = representing female. Age was coded with values 1 to 4 (e.g., 1 = 25 and below; 2 = 26 to 35; 3 = 36 to 45; 4 = 46 and above; Xiabao et al., 2022). Past performance was operationalized as the 3-year average return on the enterprise assets or the net earnings divided by its assets (David et al., 1998; Zahra et al., 2000). Tenure in post was operationalized using the number of years the CEO has been served in his or her current rank or position (Simsek, 2007). Enterprise size was measured using the natural logarithm of the total number of employees of the enterprises (Casillas et al., 2010). Asset-liability ratio was operationalized as the aggregate of the enterprise’s existing financed loans (accrued and unpaid). We operationalized enterprise age using 1 = equal to or less than 4 years and 2 = otherwise and enterprise size with 1 = implying small; 2 = medium; and 3 = large (Sinatoko Djibo et al., 2022). Enterprise type was assessed using 1 = implying high-tech and 2 = non-high-tech companies (Huang and Wang, 2013).

This study’s constructs/items reliability and validity estimation aligns with extant studies conducted in China (Lewis, 2017; Sinatoko Djibo et al., 2022) and abroad (Luo and Lin, 2022). Accordingly, SPSS-AMOS statistical procedures were employed. We conducted a sampling adequacy test using the Kaiser–Meyer–Olkin Measure (KMO) technique. The KMO score of 0.867 was higher than the minimum acceptable value of 0.7 (Kaiser, 1974). This indicated that the sample size of this study was sufficient. Next, we conducted an exploratory factor analysis (EFA) to decipher constructs/items’ reliability and validity. At the items level, all factor loadings were above the threshold of 0.6 for significance (Hair et al., 2017). The constructs/variables level validity analysis was estimated using Cronbach’s alpha (α), composite reliability (CR), and average variance extracted (AVE) tests. For each of the tests (see Table 2), all the Cronbach’s alpha (α) and composite reliability (CR) values were greater than the recommended benchmark of 0.7 (Hair et al., 2017). The AVE scores for all constructs were higher than 0.50 (ranging between 0.583 to 0.760; Fornell and Larcker, 1981). For discriminant validity, we calculated the square root of constructs AVEs reported in Table 2. As shown in Table 3, the square root of AVEs aligned with the given cut-off margin of not lesser than 0.7 for validity. Additionally, we employed structural equation modeling using SPSS-AMOS software to test our model’s goodness of fit to the data. The results: χ²/df, NFI, CFI, RMSE, and IFI were 2.136, 0.911, 0.950, 0.072, and 0.951, respectively, showed a good fitting degree with the data (Lewis, 2017; Sinatoko Djibo et al., 2022).

To reduce common method bias, the methods of pre-precaution and post-check were adopted, including the respondent information hiding method and reverse item design method. In addition, different variables and dimensions of the same variable were separated to reduce the connection between the two. We used one of the most common methods to examine the issue of common method variance (CMV). Precisely, we used Harman (1967)’s single-factor test to evaluate CMV. The exploratory factor analysis was performed on all constructs items as a whole. The value of 25.433% generated was less, thus, the homology deviation problem was not significant in this study. This result indicates that no serious CMV exists between variables in this study. We also conducted the multicollinearity test. The generally accepted rule of thumb to judge the existence of multicollinearity is that the variance inflation factor (VIF) must be greater than 10 (Khan et al., 2020). Based on the results, the highest VIF value was 1.651 and the lowest tolerance value was 1.179. Therefore, multicollinearity does not seem to be a serious problem in our dataset.

Table 3 shows the descriptive statistics, mean, standard deviation, and correlations between entrepreneurial self-efficacy, TMT collective efficacy, CEO–TMT interface, and entrepreneurial orientation.

Hierarchical regression analysis was conducted following a similar method employed by prior studies, for instance, Yamini et al. (2020). This methodology allows researchers to enter variables orderly depending on the priorities or proposed relationships of variables (Osborne, 2000). Although multicollinearity issues were not a major concern in this study, we mean-centered the independent and moderator variables (Aiken et al., 1991). The results of the hierarchical linear regression analysis are shown in Table 4. In Model 1, we estimated only the effect of the control variable on the dependent variable. Among the control variables estimated, age, education, asset-liability ratio, enterprise age, and enterprise size had statistically significant effects on entrepreneurial orientation. CEO gender, past performance, tenure in post, and enterprise type were invalid. The explained variance of Model 1 was significant, and R² was 0.152 (F change = 4.188, p < 0.01). To test the relationship between entrepreneurial self-efficacy and entrepreneurial orientation, the independent variable (entrepreneurial self-efficacy) was added to the regression Model 2. As demonstrated in Model 2, we had a significant positive relationship between entrepreneurial self-efficacy and entrepreneurial orientation (β = 0.313, p < 0.01). In Model 2, the explained variance was significant, and R² was 0.243 (F change = 25.029, p < 0.01). TMT collective efficacy was included in Model 3 to estimate Hypothesis 1. In Model 3, the coefficient of β = 0.271, p < 0.01, and explained variance was significant with an R² of 0.305 (F change = 18.583, p < 0.01). The CEO–TMT interface was further added in Model 4 to estimate Hypothesis 2. In Model 4, the coefficient of β = 0.364, p < 0.01 explained variance was significant, and R² was 0.401 (F change = 33.048, p < 0.01).

In Model 5, to test the moderating effect of TMT collective efficacy on the relationship between entrepreneurial self-efficacy and entrepreneurial orientation as stated in Hypothesis 1, three interaction terms were added to the estimation in Model 5. From the regression results, the interaction term of entrepreneurial self-efficacy and TMT collective efficacy had a significant positive relationship with entrepreneurial orientation, and the standardized regression coefficient was 0.101 (p < 0.01). This result indicated that the positive relationship between entrepreneurial self-efficacy and entrepreneurial orientation is positively moderated by TMT collective efficacy. Therefore, Hypothesis 1 is supported.

In Model 6, to test the effect of CEO–TMT interface as stated in Hypothesis 2, the three-way interaction was added to the regression analysis in Model 6. From the regression results, the three-way interaction between entrepreneurial self-efficacy, TMT collective efficacy, and CEO–TMT interface had a significant positive relationship with entrepreneurial orientation. Moreover, the standardized regression coefficient was 0.228 (p < 0.01), indicating that in higher CEO–TMT interface teams, higher TMT collective efficacy can enhance the positive relationship between entrepreneurial self-efficacy and entrepreneurial orientation. Therefore, Hypothesis 2 is supported. To better understand how CEO–TMT interface affects the influence of TMT collective efficacy on the relationship between entrepreneurial self-efficacy and entrepreneurial orientation.

Finally, this research used the graphical slope plot suggested by Aiken et al. (1991). Figure 2 demonstrates the slopes for the relationship between entrepreneurial self-efficacy and TMT collective efficacy. Also, Figure 3 illustrates this interaction, reflecting, and confirming our hypothesis regarding the difference in the effect of both types of efficacy on entrepreneurial orientation depending on CEO-TMT interface.

Consistent with Yamini et al. (2020), we conducted a robustness test of the model and our major results to check the validity of the data and hypotheses. Specifically, the robustness test examined the robustness of the evaluation method and index interpretation ability. To do this, we used a bootstrap test to artificially increase the data sample size to 1,000 observations. As reported in Table 5, the standard errors did not change in any meaningful way. As a result, we concluded that the results of our study are reliable.

This study also provides important practical implications for enterprises and their top management teams. First, our findings on the positive relationships between entrepreneurial self-efficacy, TMT collective efficacy, and entrepreneurial orientation serve as a guide for CEOs to exercise control while providing support for their TMTs to accomplish enterprise entrepreneurial orientation objectives. In this light, for example, the CEO can provide more intrinsic motivation such as inspirational incentives and personalized care to improve TMT collective efficacy. In addition, CEOs need to acknowledge or view TMT members as important strategic resources with task execution propensities that complement a CEO’s attributes for enterprise entrepreneurial orientation development rather than a sole focus on a CEO’s value over that of TMTs. Prior studies’ practical implications shed light on the duty of CEOs in establishing a conducive work environment. Precisely, in the field of entrepreneurship, the CEO’s confidence or conviction affects his/her environment, as a result, aids work incumbents to achieve goals such as the documented outcomes of entrepreneurial behavior (Chen et al., 1998; Hand et al., 2020). Consequently, there is a need for unrelenting cooperation, effective communication, and periodic intra-enterprise training on task execution to enhance individual-to-individual level exchanges and augment team-level efficacies for the cultivation and performance of enterprise entrepreneurial orientation.

Second, reflecting on the positive and negative influence of CEO–TMT interface, this study provides the following standards for enterprises to build and optimize the benefits of CEO-TMTs as well as mitigate the negative consequence of retrogression in collective efficacy and entrepreneurial orientation. Thus, CEOs should not compromise on the resource pool of TMT candidates in their recruitment and selection. For instance, TMT candidates’ past working experiences, skills, and abilities should be the topmost criteria for external and internal recruitment and selection. This implies that CEOs should partake in recruiting TMTs and ensure there is a cognitive similarity between incoming TMT member (s) and the incumbent CEO to reduce possible conflicts that arise in collaboration (TMTs and CEOs). Driven along with the emphasis, the CEO could foster a highly ethical and inclusive work environment for the sustainable development of the TMT members’ efficacy. Cumulatively, these practical implications would serve as a guiding mechanism for effective job crafting for the efficient operation of the CEO and TMT to attain a long-term impact on their entrepreneurial orientation.

This study also has limitations that need to be further addressed in subsequent studies. Firstly, this study takes entrepreneurial enterprises in the context of China as its research object to test its hypothesis. As our subject, China meets all the requirements of our study, and the results strongly support our hypothesis. However, future research can consider the difference in the impact of entrepreneurial self-efficacy on entrepreneurial orientation in trans-national and trans-cultural contexts. Also, researchers can further carry out cross-cultural comparisons in multiple countries to clarify the difference in entrepreneurial self-efficacy in different cultural contexts. Doing so could provide results that could be more generalized.

Secondly, although the empirical study uses cross-sectional data and verifies the validity and reliability of its method, researchers could use other data types to extend the empirical contribution of this study. Nevertheless, the cross-sectional data are suitable and also consistent with similar investigations in the literature (Ma et al., 2022). Similarly, considering the different stages of an enterprise’s life cycle, the management decisions have different emphases which influence the entrepreneurial orientation level of alienation (Kesidou and Carter, 2018). As a result, future studies can explore the impact of entrepreneurial self-efficacy on entrepreneurial orientation through longitudinal sequence data or panel data.

Lastly, further exploration can be made based on our conceptual model to provide new perspectives on other crucial external mechanisms for the development of entrepreneurial self-efficacy and entrepreneurial orientation.