The relationship between intrapreneurial capabilities and development in high-tech SMEs in China

ABSTRACT

The deep-seated intersection between intrapreneurial activities and innovation presents a new paradigm with innovation at its core. This study explores the relationship between intrapreneurial capabilities and the development of high-tech small and medium-sized enterprises (HTSMEs) and examines the role and impact of ambidextrous innovation and knowledge heterogeneity. We employed a questionnaire survey of 649 HTSMEs in China, and the data are analysed using structural equation modelling. The results indicate that HTSMEs could achieve sustainable development by improving their intrapreneurial capabilities, while ambidextrous innovation plays a mediating role. When knowledge heterogeneity is introduced as a moderating variable, the results indicate that knowledge heterogeneity has a negative partial moderating effect on the relationship between intrapreneurial capabilities and ambidextrous innovation. This study contributes to our understanding of intrapreneurial capabilities by introducing balanced ambidextrous innovation and knowledge heterogeneity. Ultimately, the results suggest that HTSMEs should focus on building intrapreneurial capabilities and balance the needs of ambidextrous innovation as a path to achieving sustainable development. However, firms need to control the effect of internal knowledge heterogeneity in the innovation process.

KEYWORDS:

• Intrapreneurial capabilities
• ambidextrous innovation
• knowledge heterogeneity
• sustainable development
• SMEs

1. Introduction

Small and medium-sized enterprises (SMEs) are generally defined in China as those with less than 300 employees or operating revenues of less than 10 million yuan (State Council of the PRC, 2011). From the perspective of economic efficiency, sustainable development suggests that organisations rely on technological innovation to improve production efficiency and achieve sustainable and high-quality development (Halkos & Tzeremes, 2009). Thus, high-tech SMEs (HTSMEs) are those that achieve sustainable development by acquiring and transforming independent intellectual property rights into high-tech products or services through R&D activities engaged in by scientific and technological personnel (Innofund, 2016). SMEs, especially HTSMEs, are extremely important for economic growth; they not only create numerous jobs, but also provide the innovation needed to power society (Eggers, 2020). According to the National Bureau of Statistics of China (2020), by the end of 2019, HTSMEs in China accounted for 5.56% of all companies, 2.53% of total employment, 3.5% of operating revenue, and 3% of total profits. The contributions of these companies to GDP rose from 3.17% in 2015–3.55% in 2019. However, like other SMEs, lack of financial and technological support makes it challenging for HTSMEs to play to their strengths in the ever-changing external environment (Rizos et al., 2016). The external environment’s increased unpredictability because of COVID-19 makes the situation even worse. Technology-intensive enterprises, like HTSMEs, need to develop intrapreneurial capabilities to better adapt to the highly dynamic external environment (Honig & Samuelsson, 2021).

An organisation’s intrapreneurial capabilities are its abilities to quickly and creatively respond to changes in its internal and external environments, which can help it shape and adapt to a new environment and improve its core development ability (Klofsten et al., 2021). Klofsten et al. (2021) claim that intrapreneurial capabilities should constitute the interactions between dynamic capabilities and entrepreneurial orientation, and that studying it in both dimensions is necessary. Entrepreneurial orientation focuses on the processes, practices, and decision-making behaviour of new entrants. This can be realised through subjective actions, including intentions and behaviours of key individuals who create a business in a dynamic environment (Covin & Slevin, 1991). Unlike entrepreneurial orientation, intrapreneurial capabilities focus on internal innovation and development, an operation and development strategy in which established enterprises take advantage of new opportunities and create economic value (Adachi & Hisada, 2017; Klofsten et al., 2021; Pinchot, 1985).

The study of intrapreneurship focuses more on emerging intentions and the behaviours encouraged by leaders, which disrupt existing organisational business methods to ultimately achieve innovative performance (Antoncic & Hisrich, 2003; Pandey et al., 2021; Tseng & Tseng, 2019). For example, Yildiz et al. (2021) demonstrate, at both the employee and group levels, that absorptive capacity is vital for promoting intrapreneurship and innovation performance in firms. Moreover, Guerrero et al. (2021) claim that intrapreneurial capabilities not only directly promote a university’s outcomes but also positively mediate the relationship between its ordinary capabilities and outcomes. However, since the concept of intrapreneurial capabilities was introduced, research to date has not been extended to the level of corporate capabilities. Furthermore, research focused on HTSMEs in the field of intrapreneurship is limited. For example, in-depth exploration of which innovation path HTSMEs driven by intrapreneurial capabilities should choose to achieve sustainable development is required. To address the lack of research on intrapreneurial capabilities and sustainable development in HTSMEs, we analyse a sample of 649 HTSMEs using structural equation modelling (SEM). We find that (1) intrapreneurial capabilities can improve innovation performance, (2) ambidextrous innovation could play a mediating role in the relationship between intrapreneurial capabilities and innovation performance, and (3) knowledge heterogeneity may have a negative partial moderating effect on the relationship between intrapreneurial capabilities and ambidextrous innovation.

Based on dynamic capabilities theory (Teece et al., 1997), entrepreneurial orientation (Covin & Slevin, 1991; Miller, 1983), and knowledge-based theory (Conner, 1991; Foss, 1996), our study contributes to the literature in three aspects. First, it expands the research on intrapreneurial capabilities, which tightly integrates intrapreneurship and innovation. Second, our study proposes balanced ambidextrous innovation as a model for supporting intrapreneurial capabilities. Finally, it enriches the research related to knowledge-based theory by proposing that firms should consider knowledge heterogeneity during innovation activities because of its detrimental influence on ambidextrous innovation, especially exploratory innovation.

The remainder of this paper is organised as follows. Section 2 presents the literature review and hypotheses development. Section 3 describes the study design and research methodology. Section 4 reports the empirical results. Section 5 discusses the results and offers some implications.

2. Literature review and hypotheses development

2.1. Intrapreneurial capabilities and innovation performance

The current research on intrapreneurship involves two principal areas. The first is applying entrepreneurial thinking or creativity to the development of corporate strategy, most typically the study of entrepreneurial orientation (Kuratko & Audretsch, 2009). The second is developing a strategy to guide entrepreneurial activities (Anderson & Eshima, 2013). Intrapreneurship and innovation are enduring topics in business and organisational research. As a long-standing area of academic study, intrapreneurship has spawned several themes to describe the phenomenon of organisational entrepreneurship, such as intrapreneurship (Antoncic & Hisrich, 2003), corporate entrepreneurship (Ireland et al., 2009), entrepreneurial orientation (Covin & Wales, 2019; Miller, 1983), and strategic entrepreneurship (Kuratko & Audretsch, 2009). In critical periods of economic transformation and upgrading, many companies seek new growth momentum through intrapreneurship to maintain their competitive advantage. From the resource-based view, corporate intrapreneurial capabilities focus on a company’s resource endowment, which distinguishes it from its competitors and is often difficult to replicate and imitate in the short term (Antoncic & Hisrich, 2003). Dynamic capabilities theory emphasises a firm’s ability to leverage technical, organisational, and managerial resources to gain competitive advantage and adapt to dynamic environmental changes (Teece et al., 1997). However, unlike dynamic capabilities, intrapreneurial capabilities concentrate on improving a business’s agility and flexibility, identifying innovative ways to create value, and facilitating better access to external ideas (Teece, 2012).

The positive relationship between intrapreneurship and firm performance has gained wide academic acceptance in entrepreneurship research (Achtenhagen et al., 2010). Intrapreneurial capabilities have enhanced business performance through efficient and dynamic internal communication and marketing strategies (Butkouskaya et al., 2020). Moreover, intrapreneurial capabilities are reflected in the craftsmanship spirit of meticulousness and innovation, which is the key to success and the driving force for sustainable development (Anderson & Eshima, 2013). In particular, this is primarily reflected in enterprises’ continuous use of new or improved technologies to upgrade their products and services and to dynamically innovate their internal management processes and business operation models. Based on this argument, HTSMEs could improve their innovation performance by applying intrapreneurial capabilities. Hence, we propose:

H1. There is a positive relationship between intrapreneurial capabilities and innovation performance for HTSMEs.

2.2. Mediating role of ambidextrous innovation

Innovation is the most important driving force for enterprises to become highly competitive in domestic and international markets (Berraies & Abidine, 2019). March (1991) proposes the ambidextrous nature of organisational innovation, breaking it into the two core components of exploration and exploitation. Exploration and exploitation are critical aspects of innovation in a complex and rapidly changing business environment (Enkel et al., 2017; Lei et al., 2019; March, 1991). Exploitative innovation is a relatively small and micro-innovation method because it emphasises upgrading existing products and services to meet current market needs. In contrast, exploratory innovation emphasises implementing thorough innovation based on new knowledge and new skills. It requires enterprises to break away from their existing market and open new business areas to match the emerging needs of users and customers. Exploratory innovation focuses more on technological change (Berraies & Abidine, 2019; Gilson & Madjar, 2011). Notably, innovation must balance exploitation and exploration (Levinthal & March, 1993).

Intrapreneurial capabilities require enterprises to gain the abilities to be innovative, proactive, and take risks (Klofsten et al., 2021; Morris et al., 2008). Improving intrapreneurial capabilities will influence innovation in HTSMEs through these three abilities. First, if enterprises regard innovation as a core strategy, they will integrate innovation into their overall operations and emphasise aggressive product upgrading, pioneering market innovation, and advanced innovation (Chang et al., 2019). This will prompt both exploitative and exploratory innovation. Second, exploratory innovation increases a company’s operating costs and R&D risks, thus requiring it to have a higher level of risk-taking ability. Increased intrapreneurial capabilities increase the level of risk-taking, which allows firms to engage in more exploratory innovation activities. Last, exploitative innovation emphasises continuous accumulation of improved innovations that focus on existing customers’ changing needs (Lennerts et al., 2020). Consequently, the ability to be proactive can not only help enterprises identify opportunities and grasp the changing needs of their markets. but help them adjust their business direction in a dynamic and competitive environment to obtain a competitive market advantage (Erdil & Özdemir, 2016). In sum, improving the intrapreneurial capabilities of HTSMEs could promote ambidextrous innovation in terms of increased abilities of innovativeness, risk-taking, and proactiveness. Therefore, we propose the following:

H2a. There is a positive relationship between intrapreneurial capabilities and exploitative innovation for HTSMEs.

H2b. There is a positive relationship between intrapreneurial capabilities and exploratory innovation for HTSMEs.

Ambidextrous innovation is an important value-creation activity, and the core competitiveness of enterprises has gradually changed from occupying traditional resources to acquiring knowledge resources and cultivating continuous learning and innovation ability (Liao et al., 2018). Most listed companies, international enterprises, and SMEs implement an ambidextrous innovation strategy to survive in fierce competition or obtain excellent innovation performance (Alayo et al., 2022; Buccieri et al., 2020; Lin & Chang, 2015). Although exploratory and exploitative innovation represent two fundamentally different enterprise innovation learning paths, enterprises need to strike a balance between the two (He & Wong, 2004; March, 1991), which Tushman and O’Reilly III (1996) refer to as innovation ambidexterity. Thus, it can be claimed that HTSMEs should apply both exploratory and exploitative innovation as an innovation strategy to promote innovation performance. Therefore, we propose the following:

H2c. There is a positive relationship between exploitative innovation and innovation performance for HTSMEs.

H2d. There is a positive relationship between exploratory innovation and innovation performance for HTSMEs.

From the above discussion and hypotheses (H2a), (H2b), (H2c), and (H2d), we posit:

H2e. Ambidextrous innovation mediates the relationship between intrapreneurial capabilities and innovation performance for HTSMEs.

2.3. Moderating role of knowledge heterogeneity

Based on knowledge-based theory (Conner, 1991; Foss, 1996), enterprise innovation depends on creating new knowledge, which necessitates innovative enterprises to have high knowledge creation ability and requires that the external environment continuously provide heterogeneous knowledge (Corsaro et al., 2012). Rodan and Galunic (2004) define knowledge heterogeneity as the degree of diversification in the knowledge, know-how, and skills available to individuals in a network. Furthermore, Nonaka (1994) and Nonaka and von Krogh (2009) divide knowledge into explicit and tacit knowledge. Organisational knowledge is created through continuous dialogue between the two. Therefore, we can divide knowledge heterogeneity further into explicit knowledge heterogeneity and tacit knowledge heterogeneity.

Heterogeneous knowledge enables enterprises to combine diverse types of knowledge and stimulates enterprises to produce new ideas, which promotes innovation (Granstrand, 1998; Roper & Love, 2018). However, owing to its cost, knowledge heterogeneity has a negative impact on firms’ innovation activities (Lin & Chen, 2005). Specifically, for some complex knowledge resources, knowledge transfer and sharing are quite difficult, which is not conducive to innovation. Furthermore, excessive knowledge heterogeneity leads to an increase in coordination, R&D, and integration costs. Given the increase in knowledge heterogeneity, HTSMEs must absorb more heterogeneous resources, which limits the redeployment and combination of relevant knowledge. From this discussion, we propose the following:

H3. Knowledge heterogeneity negatively moderates the relationship between intrapreneurial capabilities and ambidextrous innovation for HTSMEs.

We summarise the proposed relationships among intrapreneurial capabilities, ambidextrous innovation, knowledge heterogeneity, and innovation performance for HTSMEs in Figure 1.

Figure 1. Conceptual model.

3. Method

3.1. Sample and procedures

We first asked HTSMEs in Beijing to participate and complete an anonymous online questionnaire. Then, to improve the research and data quality, we collected samples of HTSMEs from the Jiangsu and Zhejiang provinces. All participants were informed that the data obtained were for research purposes only. In total, the questionnaire was distributed by e-mail and WeChat to about 1,000 people. We collected 713 responses (response rate = 71%) from HTSMEs through online distribution. Based on a logic question (e.g. ‘Please choose “2” for this question’), we deleted 37 invalid questionnaires, leaving 649 for analysis. The respondents included managers of different genders (S.D. = 0.5) and ages (S.D. = 0.85), as well as enterprises of diverse nature (S.D. = 0.96), age (S.D. = 0.79), and size (S.D. = 1.15). Additionally, 54.5% (45.5%) of respondents were men (women), with 52.5% between 26 and 35 years old. Regarding the enterprises, 37.1% were more than 10 years old, approximately 63% were non-state-owned enterprises, and about 71% had fewer than 200 employees. Based on these statistics, we believe that the respondents have strong representativeness.

3.2. Measures

For the 27 core items in this study, we adopted a seven-point Likert scale (1 = ‘Strongly disagree’ and 7 = ‘Strongly agree’). All items were developed in Chinese using the translation–back translation procedure.

Intrapreneurial capabilities (INC): We measured intrapreneurial capabilities on the seven-item scale developed by Miller (1983), Covin and Slevin (1991), and Guerrero et al. (2021). INC 1–3 measured innovativeness (a sample item is ‘many new products or services are introduced’). INC 4 and 5 measured risk-taking (a sample item is ‘more likely to take bold and swift action to achieve business goals’), and INC 6 and 7 measured proactiveness (a sample item is ‘initiate action ahead of industry competitors’).

Ambidextrous innovation: We divided ambidextrous innovation into exploitative innovation (EXI) and exploratory innovation (EXR). Ambidextrous innovation involves an eight-item scale adopted from Volberda and Lewin (2003) and Cao et al. (2009). EXI 1–4 measured exploitative innovation (a sample item is ‘improve existing products and services’), and EXR 1–4 measured exploratory innovation (a sample item is ‘test new products and services in established markets’).

Knowledge heterogeneity: We divided knowledge heterogeneity into explicit knowledge heterogeneity (EXK) and tacit knowledge heterogeneity (TAK) following Nonaka (1994) and Jehn et al. (1999). EXK 1–3 cover explicit knowledge heterogeneity (a sample item is ‘the professional skills of employees vary greatly’), and TAK 1–3 comprise tacit knowledge heterogeneity (a sample item is ‘employees’ values at work vary greatly’).

Innovative performance (INP): We applied the six-item scale for innovative performance developed by Lovelace et al. (2001) and Jansen et al. (2006) (a sample item is ‘provide new products and services faster than competitors’).

4. Results

4.1. Confirmatory factor analyses

We used six maturity scales and conducted first-order confirmatory factor analysis (CFA) individually. Table 1 shows the fitness test criteria for the CFA (Bagozzi & Yi, 1988; Bearden, 1983; Bentler & Bonett, 1980; Hair et al., 1998; Hu & Bentler, 1999; Jackson et al., 2009). The CFA results obtained using AMOS 23.0 in Table 2 indicate that the absolute and relative fit indices for the six-factor model were within the ideal range (χ²/df = 2.932, CFI = 0.961, TLI = 0.956, RMSEA = 0.055, and SRMR = 0.027). Furthermore, the six-factor model fits the data model better than the competing models, indicating that the six proposed constructs have good discriminant validity. Thus, the results of the measurement model are acceptable.

Table 1. Evaluation indexes and criteria for fitting the structural equation models.

Index name		Evaluation criterion
Absolute fit index	χ^2	Smaller values are better
	χ^2/df	< 3
	GFI	>0.9
	RMR	<0.05 good, 0.05–0.08 acceptable
	SRMR	<0.05 good, 0.05–0.08 acceptable
	RMSEA	<0.05 good, 0.05–0.08 acceptable
Relative fit index	NFI	>0.9
	TLI	>0.9
	CFI	>0.9

Table 2. Model fit for the CFA.

Model	χ2/df	CFI	TLI	RMSEA	SRMR
Six-factor model (INC, EXR, EXI, EXK, TAK, INP)	2.932	0.961	0.956	0.055	0.027
Five-factor model (INC+EXR, EXI, EXK, TAK, INP)	3.513	0.949	0.942	0.062	0.032
Four-factor model (INC, EXR+EXI, EXK+TAK, INP)	3.969	0.938	0.932	0.068	0.032
Three-factor model (INC+EXR, EXI+EXK, TAK+INP)	9.665	0.819	0.801	0.116	0.115
Two-factor model (INC+EXR+EXI, EXK+TAK+INP)	10.353	0.803	0.785	0.120	0.116
One-factor model (INC+EXR+EXI+EXK+TAK+INP)	10.841	0.792	0.773	0.123	0.116

Table 3 shows that the standardised factor loading of each item for INC and the other factors is more than 0.6, the SMC and AVE are above 0.5, and the C.R. is more than 0.8 (Fornell & Larcker, 1981; Hair et al., 1998), indicating that each item achieved very significant convergent validity with the factor for the CFA; therefore, we retained all items.

Table 3. CFA results.

Variable	Item	Unstandardised loading	Standardised loading	S.E.	C.R. (t-value)	SMC	P	C.R.	AVE
INC	INC1	1.000	0.808			0.653		0.898	0.557
	INC2	1.009	0.760	0.048	20.930	0.577	***
	INC3	0.956	0.758	0.046	20.869	0.575	***
	INC4	0.971	0.740	0.048	20.256	0.548	***
	INC5	0.964	0.690	0.052	18.544	0.476	***
	INC6	0.969	0.689	0.052	18.513	0.474	***
	INC7	0.915	0.773	0.043	21.384	0.597	***
EXI	EXI1	1.000	0.868			0.753		0.926	0.757
	EXI2	0.919	0.840	0.033	27.654	0.705	***
	EXI3	1.019	0.888	0.033	30.561	0.789	***
	EXI4	1.035	0.884	0.034	30.331	0.782	***
EXR	EXR1	1.000	0.905			0.819		0.916	0.732
	EXR2	1.002	0.883	0.031	32.090	0.780	***
	EXR3	0.907	0.832	0.032	28.684	0.692	***
	EXR4	0.818	0.797	0.031	26.474	0.635	***
EXK	EXK1	1.000	0.747			0.558		0.843	0.641
	EXK2	1.086	0.821	0.058	18.736	0.674	***
	EXK3	1.107	0.832	0.059	18.779	0.692	***
TAK	TAK1	1.000	0.737			0.543		0.849	0.653
	TAK2	1.063	0.796	0.056	18.921	0.634	***
	TAK3	1.181	0.885	0.062	19.176	0.783	***
INP	INP1	1.000	0.857			0.734		0.956	0.774
	INP2	1.136	0.886	0.037	30.938	0.786	***
	INP3	1.166	0.902	0.036	31.992	0.813	***
	INP4	1.112	0.834	0.040	27.646	0.696	***
	INP5	1.242	0.913	0.038	32.830	0.834	***
	INP6	1.159	0.883	0.038	30.706	0.780	***

Note: N = 649; ***p < 0.001.

4.2. Nonresponse bias

We tested nonresponse bias by comparing the gender and age variables of the early respondents to those of the later respondents (Armstrong & Overton, 1977). An achi-square test of the 437 early respondents and 212 late respondents showed that they do not differ significantly (p > 0.05) regarding either gender or age. Thus, nonresponse bias is not a pervasive issue in this study.

4.3. Common method bias

To test for potential common method bias, we used Harman’s single-factor test (Podsakoff & Organ, 1986). The first factor accounts for 32.4% (<50%) of the total variance, and the top three factors explain 67.6% of the total variance. Therefore, common method bias is not a critical problem in this study.

4.4. Descriptive analyses

We report the means, standard deviations, zero-order Pearson correlations, and scale reliabilities for each variable in Table 4. INC is positively correlated with EXI (r  = 0.836, p < 0.001), EXR (r = 0.894, p < 0.001), and INP (r = 0.903, p < 0.001). Moreover, EXI and EXR are positively correlated with INP (r = 0.903, p < 0.001; r = 0.902 < 0.001). EXI is positively correlated with EXK (r = 0.099, p < 0.05) and TAK (r = 0.109, p < 0.05), while EXR is positively correlated with EXK (r = 0,093, p<0.05) and TAK (r = 0.132, p < 0.01). Further, the reliability test results in Table 4 indicate that the Cronbach’s α of each factor is above 0.8, which is significantly higher than 0.7. These results provide preliminary support for the above hypotheses.

Table 4. Descriptive analyses.

	1	2	3	4	5	6	7	8	9	10	11
1. Sex	1
2. Age	0.044	1
3. Nature	−0.041	0.117***	1
4. Enterprise age	0.05	0.165***	−0.032	1
5. Size	0.021	0.040	0.074	0.375***	1
6. INP	−0.034	0.157***	0.092*	0.036	0.036	0.953
7. INC	−0.069	0.114**	0.054	0.052	0.050	0.903***	0.896
8. EXI	−0.016	0.123**	0.074	0.04	0.055	0.902***	0.836***	0.926
9. EXR	−0.001	0.148**	0.074	−0.02	−0.004	0.931***	0.894***	0.920***	0.917
10. EXK	0.07	0.074	0.058	0.052	−0.003	0.099*	0.061	0.099*	0.093*	0.841
11. TAK	0.078*	0.085*	0.087*	0.059	0.003	0.104*	0.086	0.109*	0.132**	0.777***	0.846
Mean	0.55	2.12	1.80	2.11	2.58	4.73	5.48	5.40	5.25	4.14	3.77
S.D.	0.50	0.85	0.96	0.79	1.15	1.60	1.05	1.51	1.52	1.49	1.54

Note: Cronbach’s α in bold; N = 649; ***p<0.001; **p < 0.01; *p < 0.05.

4.5. Hypotheses testing

We used an SEM to test the hypothesised relationships among the independent variables, mediation variables, and dependent variables in the theoretical model. As multiple mediation effect hypotheses are in the model, we employed AMOS 23.0 to calculate the path model using the bootstrap method to test the mediation effects. Table 5 shows that the absolute and relative fit indices of the SEM for the main and mediation effects of the four variables (INC, EXR, EXI, and INP) reached the desired values, indicating that the model fit is satisfactory.

Table 5. Path analysis.

Hypothesis	Independent variable		Dependent variable	Standardised coefficient	S.E.	C.R.	P	Hypothesis supported
H1	INC	→	INP	0.396	0.077	7.994	***	Yes
H2a	INC	→	EXI	0.843	0.062	19.556	***	Yes
H2b	INC	→	EXR	0.876	0.064	21.156	***	Yes
H2c	EXI	→	INP	0.276	0.053	5.711	***	Yes
H2d	EXR	→	INP	0.330	0.058	5.728	***	Yes
χ^2/df=2.705 GFI=0.935 CFI=0.977 RMSEA=0.051 SRMR=0.021

Note: N  = 649; ***p < 0.001.

4.5.1. Main effects

The results in Table 5 indicate that the main effects among the four variables INC, EXR, EXI, and INP were all significant (p < 0.001) and positive in the theoretical model. Regarding the effect of INC on EXI and EXR, the effect of INC on EXR was relatively large (β = 0.876, p < 0.001). Additionally, the effect of EXR on INP was relatively large (β = 0.330, p < 0.001). Hence, the results support H1, H2a, H2b, H2c, and H2d.

4.5.2. Mediation effects

To test for mediation effects in the model, following Baron and Kenny (1986) and Zhao et al. (2010), we find (1) a significant correlation between the independent variables (i.e. intrapreneurial capabilities) and mediating variables (i.e. ambidextrous innovation), (2) a significant correlation between the mediating variables and the dependent variables (i.e. innovation performance), (3) that the relationship between the independent and dependent variables becomes insignificant when the mediating variable is present, and (4) a significant correlation between the independent and dependent variables. Thus, per the results in Table 5, INC positively affected EXI and EXR (β = 0.843, p < 0.001; β = 0.876, p < 0.001). Furthermore, there were positive effects among EXI, EXR, and INP (β = 0.276, p < 0.001; β = 0.330, p < 0.001). Additionally, we see a positive correlation between INC and INP (β = 0.396, p < 0.001). These results indicate ambidextrous innovation mediated the relationship between intrapreneurial capabilities and innovation performance for HTSMEs, which supports H2e.

Second, following Mackinnon (2008) and Hayes (2009), we further tested the mediation effect using the bootstrap method. The results in Table 6 show that after bootstrapping 1,000 samples to compute bias-corrected and percentile confidence intervals, the intervals of both the total and indirect effects between the minimum and maximum did not contain zero; however, the difference between these two indirect effects contained zero.

Table 6. Total and mediation effects.

Path	Point estimation	Bootstrapping
		Bias-corrected 95% CI		Percentile 95% CI
		Lower	Upper	Lower	Upper
Total effect: INC→INP	0.918	0.892	0.941	0.890	0.938
Indirect effect 1: INC→EXI→INP	0.233	0.144	0.320	0.150	0.322
Indirect effect 2: INC→EXR→INP	0.289	0.166	0.411	0.166	0.411
Difference	−0.056	−0.222	0.146	−0.222	0.146

Last, one could argue that there was no significant difference between the two mediating effects. In addition, when the two mediating effects were significant, the independent variable INC still had a significant effect on INP, indicating that the two mediating variables were partial mediators. The total effect of the independent variable INC on INP was 0.918, and EXI and EXR accounted for 25% and 31% of the total effect, respectively. Hence, these results provide further support for H2e.

4.5.3. Moderating effects

According to Table 7, our results using SPSS 23 indicate that the moderating effects of EXK and TAK in the effect of INC on EXI and EXR were both negative. Among them, the moderating effect of EXK and TAK in the effect of INC on EXR reached significance at the 0.01 and 0.05 levels (β = −0.066; β = −0.045), respectively, while the effect of INC on EXI reached significance at the 0.1 level (β = −0.050; β = −0.036). We applied the control variables sex, age, nature of enterprise, age of enterprise, and size of enterprise as dummy variables, given their categorical nature.

Table 7. Moderating effect.

Dependent variable		EXI			EXR
		Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	Constant	−0.499*	−0.873**	−0.694**	−1.359***	−1.847	−1.702
Control variables	Sex (1)	0.111	0.089	0.097	0.162*	0.133	0.137
	Age (1)	−0.098	−0.084	−0.103	−0.191	−0.173	−0.193
	Age (2)	−0.214*	−0.202*	−0.219*	−0.162	−0.145	−0.161
	Nature of enterprise (1)	−0.055	−0.054	−0.039	−0.034	−0.032	−0.007
	Nature of enterprise (2)	−0.056	−0.056	−0.0048	−0.160	−0.160	−0.146
	Age of enterprise (1)	0.040	0.051	0.052	0.202*	0.217*	0.223*
	Age of enterprise (2)	0.003	0.009	0.012	0.047	0.054	0.066
	Size of enterprise (1)	−0.066	−0.062	−0.053	0.098	0.103	0.110
	Size of enterprise (2)	−0.091	−0.079	−0.087	0.129	0.145	0.137
	Size of enterprise (3)	0.014	0.026	0.019	0.255*	0.271**	0.265*
Independent variable	INC	1.102***	1.122***	1.110***	1.185***	1.212***	1.194***
Moderator variable	EXK		0.061^+			0.079**
	TAK			0.037			0.073**
Interaction	INC*EXK		−0.050^+			−0.066**
	INC*TAK			−0.036^+			−0.045*
Goodness of fit	Adjusted R^2	0.579	0.583	0.580	0.662	0.671	0.668
	△R^2		0.004	0.001		0.009	0.006
	F	81.904****	70.707***	69.869***	116.517****	102.564***	101.268***
	VIF	[1.015, 1.689]

Note: N = 649; ***p < 0.001; **p < 0.01; *p < 0.05; ⁺p < 0.1.

The four negative moderating effects indicate that as the moderating variables EXK and TAK increase, the influence of the independent variable INC on the moderating variables EXI and EXR decrease. We followed Dawson (2014) and separated the sample into a low group (mean minus one standard deviation) and high group (mean plus one standard deviation). Based on the four moderating effects in Figures 2–5, the coefficients for the effect of the independent variables on the moderating variables for the low group were significantly higher than those for the high group. Hence, the results partially support H3.

Figure 2. Moderating effect of EXK in the relationship between INC and EXI.

Figure 3. Moderating effect of EXK in the relationship between INC and EXR.

Figure 4. Moderating effect of TAK in the relationship between INC and EXI.

Figure 5. Moderating effect of TAK in the relationship between INC and EXR.

5. Discussion and conclusion

5.1. Discussion

Based on the empirical analysis of the questionnaire data, we have several findings. First, intrapreneurial capabilities have a positive effect on innovation performance. This finding enhances the research on intrapreneurship and supports the argument that intrapreneurial capabilities facilitate innovative development (Guerrero et al., 2021; Klofsten et al., 2021; Marques et al., 2021; Urbano et al., 2013). Specifically, intrapreneurial activities primarily involve the three dimensions of innovativeness, risk-taking, and proactiveness, which can help companies rediscover opportunities to achieve innovation and ultimately improve their performance and development. HTSMEs can explore unknown knowledge areas, although their innovation investment inevitably incurs high risks, high costs, and uncertain returns (Peneder, 2008). Moreover, HTSMEs make complete use of all available information and resources to develop first-mover technologies to enhance their innovation advantage and gain growth momentum (Covin & Slevin, 1989). Several studies based on Chinese data and the practical experiences of SMEs in China verify our findings. Liu and Wang (2020) found that proactiveness, innovativeness, and risk-taking could positively affect the performance of HTSMEs in China. For example, intrapreneurial activities, in terms of innovation and venture capital, accelerated the development of Xi’an Airlines in China (Yang et al., 2017). Moreover, owing to its successful intrapreneurship, DJI, founded as a high-tech SME in China, is currently the most successful drone manufacturer globally (JaeDo, 2018). From DJI’s establishment, in 2006, to 2012, no consumable drone existed in the market (Shang et al., 2021). DJI grasped the development opportunity and invested and developed consumable drones that could carry professional photography equipment (JaeDo, 2018; Shang et al., 2021); sales increased from more than 3 million yuan in 2010–200 million yuan in 2012 (Shang et al., 2021).

Second, our study constructed a model that considers intrapreneurship, ambidextrous innovation, and innovation performance; moreover, ambidextrous innovation had a mediating effect in the relationship between intrapreneurial capabilities and innovation performance. Thus, our study points out that improved intrapreneurial capabilities stimulate innovative outcomes through the path of balanced ambidextrous innovation. This finding integrates and enriches the studies of intrapreneurship (Ambos & Tatarinov, 2022) and balanced ambidextrous innovation (Levinthal & March, 1993). However, our result differs from Morandi Stagni et al.’s (2021) view that a company should focus on one aspect of ambidextrous innovation. In particular, intrapreneurial capabilities, composed of innovativeness, risk-taking, and proactiveness, promoted both exploitative innovation and exploratory innovation. Furthermore, ambidextrous innovation can encourage innovative development of SMEs, which meet the needs of emerging and existing markets. According to a similar case about a bio-3D printing technology SME in China, an ambidextrous innovation strategy can help companies identify opportunities for intrapreneurial activity to gain a competitive advantage and iterate innovation (Fan & Liang, 2021).

Last, knowledge heterogeneity has a negative partial moderating effect on the relationship between intrapreneurial capabilities and ambidextrous innovation for HTSMEs. This result differs from the those of prior studies (Huang & Chen, 2010; Oerlemans et al., 2013), which claim that knowledge heterogeneity can facilitate innovation activities in organisations, especially concerning exploratory innovation. Our study reveals that heterogeneous knowledge has a negative effect on innovation, as it easily generates conflicts and makes transferring and assimilating knowledge difficult. Furthermore, excessive heterogeneity leads to an increase in coordination, R&D, and integration costs. Organisations intending to absorb external knowledge should ensure that it is similar to their internal knowledge because a high degree of knowledge similarity can facilitate innovation (Ahuja, 2000; Lin & Chen, 2005). Therefore, in the relationship between intrapreneurial capabilities and ambidextrous innovation, knowledge heterogeneity has a negative moderating effect on ambidextrous innovation, as well as higher adverse effects on exploratory innovation than exploitative innovation. As exploratory innovation consumes more corporate resources than other innovation activities (Morandi Stagni et al., 2021). Low knowledge heterogeneity could support an organisation’s ability to combine knowledge and thus, supply more knowledge resources (Lin & Chen, 2005). Therefore, to perform exploratory innovation activities more efficiently, organisations should control its knowledge heterogeneity. For example, Tsai (2021) found that the local companies in Taiwan formed by people with different knowledge backgrounds did not result in better innovation. Given the above, our study contributes meaningfully to knowledge-based theory research.

5.2. Theoretical implications

This study offers several important theoretical implications regarding intrapreneurial capabilities and innovation. First, it broadens the research field of intrapreneurship. Based on the concept of intrapreneurial capabilities (Klofsten et al., 2021), this study points to a positive relationship between intrapreneurial capabilities and innovative development. Furthermore, this study further expands the study of intrapreneurship from the individual level (i.e. employees or women; Huang et al., 2021; Zhang et al., 2020) to the organisational level. Second, the study empirically constructed a model of how intrapreneurial capabilities affect innovation, which emphasises the mediating role of balanced ambidextrous innovation (March, 1991; Tushman & O’Reilly III, 1996). In this study, we believe that balanced ambidextrous innovation is a reasonable path for intrapreneurial capabilities to have a positive impact on innovation performance. Finally, it further explores knowledge-based theory (Conner, 1991; Foss, 1996). It enriches the knowledge heterogeneity literature by noting the negative correlation between intrapreneurial capabilities and ambidextrous innovation (Williams & O’Reilly III, 1998), especially exploratory innovation.

5.3. Practical and policy implications

This study’s results have several practical implications. First, for SMEs to become ‘specialised and new’ and exhibit enhanced intrapreneurial capabilities, HTSMEs should focus on business innovation, acting as risk-takers and pioneers of new strategies, thus revolutionising their industry. Although failure is possible, HTSMEs can gain potential consumers by launching new products and services before their competitors do so. Second, HTSMEs should fulfil market demand to achieve efficient innovation activities, which requires them to maintain good relationships with consumers and keep abreast of market conditions. Ultimately, HTSMEs need to update their knowledge base according to the changing environment. However, the heterogeneity between the old and new knowledge bases should not be overly significant because increased heterogeneity can erode innovation activities.

Our results also have some policy implications. The government should play a guiding role in the development of enterprises (Ruttan, 2001, 2006), which could help HTSMEs implement their intrapreneurial and innovation strategies. As a short-term strategy, the government should subsidise HTSMEs to reduce their risks and costs of HTSMEs intrapreneurial and innovation activities, thereby encouraging HTSMEs them to boost economic recovery. The government’s long-term goals should create demand for HTSMEs. Demand-based policies can reduce costs and risks for enterprise innovation and business activities (Te Kulve et al., 2018). For example, standing orders from the government can continuously enhance the revenue of HTSMEs. Moreover, the government should bridge the gap between HTSMEs and scientific research institutions. HTSMEs have access to the latest technology; furthermore, they can help scientific research institutions test this technology, ultimately facilitating a win–win situation.

5.4. Limitations and future research

Although this study applied scientific procedures and criteria, it has some limitations. First, to make the findings more generalisable, future studies should collect a sample from more diverse geographic areas, especially to identify lagging areas of high-tech industries. In addition, panel data and more factors would be used to explore the high-quality development of HTSMEs over a long period of time. Second, in an era dominated by knowledge innovation, developing technological self-reliance, comprehensive innovation, and intrapreneurship are new requirements for building intrapreneurial capabilities. Thus, studies of intrapreneurial capabilities in high-tech industry development should include more dimensions to intrapreneurship capabilities, which may require developing a new scale to meet the requirements of the knowledge economy era (Klofsten et al., 2021). Finally, the mediators and moderators in the research model should be expanded to add factors that may better reflect the transformation and development of HTSMEs. In particular, it is necessary to further explore how HTSMEs could balance exploratory innovation and exploitative innovation, so as to help them carry out intrapreneurial activities and finally achieve development.

Acknowledgement

All authors thank the editor and anonymous reviewers for their constructive comments and suggestions, which improved the quality of this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data presented in this study are available on request from the authors. The data are not publicly available due to privacy concerns.

Additional information

Funding

This work was supported by the Fundamental Research Funds for the Central Universities [Grant Number 2021YJS065].

Notes on contributors

Jianjun Sun

Mr. Jianjun Sun is a PhD candidate at School of Economics and Management, Beijing Jiaotong University. His research focuses on business management (E-mail address: [email protected]).

Shuxiang Wang

Dr. Shuxiang Wang is a Professor at School of Economics and Management, Beijing Jiaotong University. His research focuses on business management (E-mail address: [email protected]).

Fang Yuan

Dr. Fang Yuan is a lecturer at Business School, University of Jinan. Her research focuses on business management and financial management (E-mail address: [email protected]).