Analysis on the Effectiveness and Mechanisms of Public Policies to Promote Innovation of High-Tech Startups in Makerspaces

Abstract

In the context of digital transformation and the rapid development of artificial intelligence, corporate innovation has become increasingly important in leading industrial development and improving national competitiveness. In 2015, China launched a “mass entrepreneurship and innovation program”; in response to the policy, many makerspaces have been established, resulting in clusters of high-tech startups. High-tech startups are the pioneers in innovation development. However, there is still a lack of empirical evidence for whether these firms’ innovative activities, capabilities, and performance can be effectively stimulated by public policy. Drawing from the institutional theory and resource-based view, this paper develops a model of policy perception on innovation response. Using a sample of 500 startups located in the three representative makerspaces in China, this work verifies the effectiveness of innovation and entrepreneurship incentives on startups’ innovative activities, capabilities, and performance, and successfully identifies the mediating role of policy adaptation in the policy perception-innovative responses’ link and the moderating role of makerspace support.

1. Introduction

In today’s business world, with the rapid development of digital technology and artificial intelligence, innovation-driven improvement of national competitiveness has increasingly become the focus of competition among the world’s major economies. Whether innovation can be stimulated through a top-down approach is controversial across the academic spectrum [1,2,3,4,5], policy incentives for innovation have been widely adopted in practice. The Chinese government has launched the “Mass Entrepreneurship and Innovation” program (EIP), which has aimed to stimulate entrepreneurship throughout society since 2015. Under the incentives of EIP, a growing number of makerspaces have been established, resulting in clusters of high-tech startups. However, whether the EIP incentives have effectively stimulated the innovative activities, capabilities, and performance of high-tech startups still lacks empirical evidence. The incentive mechanism of EIP also remains unclear. In this paper, we build a policy perception model to test the effectiveness and mechanism of EIP in high-tech startups in makerspaces.

The high-tech startups are particularly critical for building the innovation-driven growth model [6]. A startup is usually a newly formed company that starts with a small amount of capital and implements an innovative concept [7]. High-tech startups are typically characterized by being knowledge-intensive and innovative-driven, and producing and selling high-value-added products and services. They can promote the breakthrough or substantial improvement in relevant technology, services, or industry [8], but these types of entrepreneurships have the most difficulty succeeding [9]. They are the main target organizations of current government policy incentives. On the one hand, as startups, they usually have less path dependency and organizational inertia than established firms [10]; on the other hand, they engage in more innovative activities and can convert innovativeness into competitiveness compared to traditional companies, such that high-tech startups are a pioneer in technology, industry, and management innovation [11].

China’s innovation is characterized by the dual drive of the market and the government. In areas where the market is less developed, the policy plays a leading role [5]. According to the specific policies related to innovation and the survey of its implementation, the government can effectively support innovation by adjusting the allocation of resources, improving service efficiency, and reducing taxes and other operating costs [12]. However, in China’s bureaucratic system, the information is relatively opaque, and there is a lag in the flow of information [13]. On the one hand, the making, implementation, and interpretation of policies are based on the will of the leaders of the policy-making departments rather than the understanding of the policy text by the public or policy recipients. On the other hand, the central government’s policies are generally concise, leaving some room for personal interpretation. This leads to a wide variation in the perception and understanding of the policy by different startups. Therefore, the innovation incentive effects of EIP policy vary across startups due to the different corporate perception of the external innovation policy environment.

In this paper, we combine institutional theory and RBV to develop a model of policy perception and innovative responses of high-tech startups. Based on institutional theory, the external environment constitutes a special context for the survival and development of organizations and influences their strategic orientation [14,15]. Policy incentives form a regulative and normative institutional environment that encourages startups and innovation. Under the pressure of maintaining legitimacy within the particular institutional environment, the innovation decisions of high-tech startups are influenced not only by their own will but also by the context shaped by government policies. However, according to institutional theory alone, firms actively conform their actions to adapt to a homogeneous policy environment, while, in reality, uniformly issued policies do not have the same innovation incentive effect on different firms. The resource-based view (RBV) is a good complementary explanation for this phenomenon, as Barney [16] points out that heterogeneous firm characteristics and resources are the sources of their competitiveness. In our research context, startup policy adaptation linked the policy perceptions and variations in innovation responses and performance across startups in a similar external environment.

This study collects unique primary data by visiting 500 high-tech startups located in policy-supported makerspaces and aims to answer the following research questions through empirical analysis:

• Do policy incentives have the desired effect of driving innovative activities and enhancing the innovative capabilities and performance of high-tech startups?
• Through what mechanisms do policy incentives affect the innovativeness of high-tech startups?
• What is the role of makerspaces as platforms for policy implementation, service delivery, and entrepreneurial ecosystem construction, in the process of policy influencing the innovative behavior of startups?

This study advances the understanding of the research field of innovation and startups. First, it reveals the incentive effect of innovation policy on high-tech startups, which remains largely unexplored in the existing studies [1,4,17,18]. Unlike traditional manufactories, high-tech startups are more sensitive to government supportiveness. We emphasize that policy adaptation plays a key role in bridging the policy perception of EIP incentives and innovative responses of startups. Second, although existing studies uncovered how makerspaces directly drive corporate innovation, such as through knowledge spillover [19] or by providing multiple resources [20], little attention has been paid to enhancing the role of makerspaces in leveraging firms’ inherent capabilities. In this paper, we analyze and provide empirical evidence on the moderating role of makerspaces on policy incentive mediation mechanisms. Finally, we further explore the relationship between the outcome variables, identifying the mediating role of dynamic capabilities on the link between innovation activity and innovation performance. The conclusion emphasizes the importance of dynamic capabilities in innovation and further expands the application boundaries of RBV.

2. Background and Theoretical Foundation

Over the past decades, how to lead industrial and technological upgrading by promoting innovation to form national competitiveness has become a major concern for the research field and policymakers. In the field of innovation research, one stream focuses on technology-driven innovation and its impact on business society [21,22] while the other streams focus on the impact of the institutional context in driving innovation [2,12,17]. For instance, Li et al. [21] provided evidence that the application of digital financial services can promote the green innovation within the listed Chinese companies. Shao et al. [22] indicated that green technology innovation also drives the sustainability of society.

Mazzucato [2] pointed out that public policies can create and shape the market by implementing long-term strategies. In line with this view, the world’s leading economies are promoting innovation and high-technology development through policy incentives. In recent years, the United States passed the Innovation and Competition Act of 2021 to promote research and development in “key technology focus areas” including robotics, artificial intelligence, and cash energy. The U.K. established the Advanced Research and Invention Agency (ARIA) to provide funding for potential high-tech projects and signed a cooperation agreement with the U.S. to promote future technology cooperation in artificial intelligence technologies, digital standards development, and supply chain security. Since the EIP was launched by the Chinese government in 2015, a series of policy incentives and special subsidies have been provided to foster high-tech startups. Based on the statistics, the Chinese government’s budgetary expenditures and R&D appropriations have continued to rise since 2015, as Figure 1 shows. In addition, the number of makerspaces in Shanghai reached 8507 by the end of 2020 since the first makerspace was established in 2015.

Current studies have contributed to the research field of policy, incentives, and corporate innovation, which form the foundation of this study. Long and Liao [17] found that fiscal policy incentives (subsidies and tax credits) encouraged the innovation of manufacturers based on their survey of 265 Chinese firms, but they only focused on eco-product innovation. Zheng et al. [20] used a qualitative approach (fsQCA) to comparatively analyze how to effectively encourage the innovation capacity of makerspaces and summarized four synergistic elements: resource aggregation, cultural preservation, network connectivity, and platform services. Mohedano-Suanes et al. [18] identified the mediating role of market perception in the relationship between enterprises’ innovation and long-term performance. Sulistyo and Siyamtinah [23] indicated the positive influence of entrepreneurship, marketing capabilities, relational capital, and empowerment on SMEs’ innovation. Chen et al. [9] explored external and internal key success factors for high-tech startups entering the international market. According to Luo et al. [12], high-tech startups’ innovation can be effectively facilitated by government subsidies by mitigating the margin cost of the technology spillover effect and the risk of innovative uncertainty. In addition, the makerspace as a new phenomenon and an incubator of startups has drawn wide attention in recent years. Cuvero et al. [19] found, through a case study, that makerspaces promote innovation in high-tech startups through knowledge spillover. Wu and Ma [24] explored the characteristics of Shenzhen, China’s makerspace, from a cultural perspective, highlighting its importance in innovation and social dimensions. Based on the semantic analysis, Fu et al. [25] delved deeper into the unique connotation and extensions of Chinese makerspaces and their inspiration for the development of China’s creative industries. Most extant studies focus on the new concept or phenomena, macro-level analysis of governance, or operations of the government or platforms, but they lack an integrated model to interpret the effectiveness and mechanism of government policies on innovation in startups.

Based on the new institutional theory, organizations survive in and are shaped by the particular external institutional environment. External institutional pressures (both formal and informal) can have a significant impact on firms, and firms that do not meet institutional requirements will struggle to survive [14,15,26]. Organizations need to adjust and regulate their behavior to fit the specific external environment. Multiple policy and social factors have created a complex and dynamic environment for innovation and entrepreneurship. Startups in the makerspace are constrained by the particular context formed by EIP, where policy perceptions are crucial to stimulate their innovative responses. The promulgation of multiple policies aims to form a formal institutional environment to promote an innovation-friendly social system. For example, financial subsidies, tax refunds, and other material supports are designed to reduce R&D costs, and well-designed laws and regulations on intellectual property protection are designed to encourage substantive innovation.

The resource-based view indicates that sustainable competitiveness is built on valuable, scarce, inimitable, and irreplaceable resources [16]. On this basis, firms can develop dynamic capabilities that enable them to identify opportunities and adapt to the rapidly changing environment [27]. By leveraging and integrating the resources offered by the external environment and transforming them into firm-specific resources, high-tech startups can achieve sustained competitiveness. Due to the heterogeneity of firms, their ability to perceive the same external policies varies, which can lead to differences in the specific ways firms identify, utilize, and integrate the resources provided by the policies.

3. Hypotheses Development

3.1. Policy Perception and Innovative Activities

Under the layout of EIP, governments at all levels have issued various policies to support the whole process of innovation and entrepreneurship. Each region has also issued guiding policies based on local industrial characteristics and industrial planning. However, resources are limited, and policies are also limited as institutional environment resources for enterprises; in addition, not all entrepreneurs can accurately and effectively identify, access, and utilize policy resources. It has been shown that managerial cognitive biases can affect strategic business decisions.

The better a high-tech startup perceives external policies, the more likely it is to adopt innovative behavior. Based on the policy acceptance model proposed by Pierce et al. [28], policy perception refers to the cognitive process of the usefulness and ease of use of policies by startups in makerspaces. Specifically, the policy usefulness of the EIP is the practical perception of the startups that the particular policy solves the difficulties faced by the enterprises in the process of innovation and development. The ease of use of the EIP refers to the perception of startups regarding the threshold of access to innovation and entrepreneurship supportiveness, and the number of corporate resources and energies that will be consumed to respond to the policies. For startups, understanding the usefulness and ease of use of policies is quite subjective. Li et al. [29] suggested that both perceptions influence entrepreneurs’ attitudes and behavior toward innovation and lead to increased or decreased innovation. The perception of policy usefulness can improve the confidence and motivation of startups to innovate. The perception of policy ease of use enables startups to perceive the low threshold of government policy support for innovation and entrepreneurship and the ease of obtaining corresponding project support, which can also enhance the motivation and enthusiasm of enterprises to innovate and increase their innovative activities. In addition, startups engaging in innovative activities are risky. The higher ability of policy perception can reduce the startups’ psychological and actual costs of taking risks. EIP policies offer a variety of tax breaks and subsidy programs for startups. The existing study proved that tax deduction and financial subsidies pull/push corporate innovation and significantly improve the innovation performance of local firms [12,30,31]. Therefore, we propose the following hypothesis:

Hypothesis 1a (H1a).

Innovation and entrepreneurship policy perception has a positive impact on the innovative activities of startups in makerspaces.

3.2. Policy Perception and Dynamic Capability

Dynamic capabilities are critical for the innovation and development of startups. Teece et al. (p. 516, [27]) define dynamic capabilities as “the firms’ ability to integrate, build and reconfigure internal and external competencies to address rapidly changing environments”. In the process of transformation of resources into competitiveness, dynamic capability plays a critical role in identifying, capturing, and transforming technological opportunities. Compared to mature corporations, startups usually lack sufficient innovative resources. By leveraging dynamic capabilities, these firms can choose prospective directions to innovate, absorb various entrepreneurial and innovative resources, and allocate their limited resources to key sectors to achieve the optimal effect or capture the market share.

However, the construction of dynamic enterprise capabilities requires a keen awareness of the external environment. Policy perception is the basis for enterprises to quickly respond to policy changes and the starting point for opportunity identification. Only by identifying the development orientation and resource allocation tendency of EIP can the startups grasp and respond to the opportunity in time. On the contrary, if startups have deviations in the recognition of the policy environment when perceiving the key information related to innovation and entrepreneurship, they cannot effectively respond to external changes and build dynamic capabilities. Therefore, policy perception is the foundation of dynamic capabilities. In the rapidly changing and complex environment, startups need to constantly focus on and be aware of the relevant policies to improve their dynamic capabilities.

Hypothesis 1b (H1b).

Innovation and entrepreneurship policy perception has a positive impact on the dynamic capabilities of startups in makerspaces.

3.3. Policy Perception and Innovative Performance

Innovative performance is the outcome of new ideas or concepts that are implemented or commercialized by firms. Existing studies on innovative performance have yielded rich findings. Startups build broad alliances with the community, e.g., research institutes, universities, competitors, suppliers, etc., which is conducive to improving their innovative performance [32]. Network centrality is also critical to their innovative performance [33,34]. Combined with the view of Gunawan et al. [35], identifying potential opportunities in a specific social context and adapting to them is crucial for the survival and development of startups.

Entrepreneurs’ perceptions of EIP policies are the key factors for macro policies to act on startups’ actions and outcomes. Innovative performance is based on effective innovative behaviors. According to the policy acceptance model, firm behavior is influenced by external institutional and environmental factors, and when the entrepreneurial team or entrepreneur perceives that the policy is more supportive of the firm’s innovation, it is more favorable to carry out the innovation and achieve innovative performance. On the one hand, a series of policy incentives can increase the innovation enthusiasm of startups and enhance their innovation outcomes; on the other hand, relevant government policies, financial subsidies, and tax incentives can reduce the innovation input cost of startups, thus improving the potential innovation performance. Therefore, the policy perception of EIP policies reflects the managers’ perception of the external policy stimulus, which determines their attitudes and willingness towards innovation and is eventually reflected in the firm’s innovative performance.

Hypothesis 1c (H1c).

Innovation and entrepreneurship policy perception has a positive impact on the innovative performance of startups in makerspaces.

3.4. Mediating Effect of Policy Adaptation

Policy adaptation is an enterprise’s ability to adapt to the current, relatively stable, social environment and norms, as well as its ability to manage itself in future dynamic interpersonal relationships and complex situations. The EIP adaptation studied in this paper refers to the startups’ ability to interpret the innovation and entrepreneurship policies in their regions and adjust their operations and development based on them, to adapt to the macro policy environment, to reduce the risk of innovation and entrepreneurship, and to improve the success rate of entrepreneurship.

We argue that enterprises with a higher degree of perception of EIP policies have stronger policy adaptation ability. According to social cognition theory, the perception of the environment will influence the behavior of organizations and individuals. The perception of the policy makes startups clearer about the goals conveyed by the policy and enables them to make more accurate interpretations and judgments. On this basis, the policy adaptation ability of enterprises is improved, and the entrepreneurial team is better able to transition from “adapting to current policies” to “adapting to future policies”; i.e., the entrepreneurial team not only understands, adapts, and responds to current innovation and entrepreneurship policies in a more appropriate way, but also correctly anticipates the future policy trends and adjusts their responses in a timely manner.

Meanwhile, the more accurately the startups can perceive the policy, the easier it is to generate the subjective will to accept policy support and thus find ways to adapt to the innovation and entrepreneurship policy. D’Aveni and Smith [36] argue that in a highly competitive environment, competing for resources is a top priority for individual firms, and this is especially important for startups. In a rapidly changing and dynamic environment where opportunities are fleeting, EIP implies the government’s supportive attitude and industry direction for innovation and entrepreneurship. Startups with better policy adaptability are able to identify incentives and thus utilize the relevant opportunities and resources provided by the government. These startups are more likely to engage in innovative behaviors, foster their dynamic capabilities, and transform innovative behaviors into innovative performance.

Hypothesis 2 (H2).

Startups’ policy perception improves their policy adaptation.

Hypothesis 3a/b/c (H3a/b/c).

Startups’ policy adaptation mediates the effect of policy perception on innovative activities/dynamic capabilities/innovative performances.

3.5. Moderating Effect of Makerspace Support

Based on the characteristics and needs of innovation and entrepreneurship in China, makerspaces provide hardware equipment, value-added services, and capital financing to startups [37]. Extant studies indicate that makerspaces have provided various support to entrepreneurs and startups, such as business affairs, equity planning, financial accounting services, lower-than-market-rate space leasing, expert guidance, skilled workers at the product design and technology development stage, market development services, and financing services when investment funds are needed [19,35]. More than this, makerspaces help startups understand the policy orientation accurately and improve the policy adaptation by facilitating them to adjust strategies according to related policies, enhancing the mutual trust of government and startups. By complying with the relevant policies, makerspaces build a bridge between the startups and government and are critical for regional innovative economic development by constructing an innovative ecosystem [38]. Thus, a strong sense of support from the makerspace can be seen as a useful resource to amplify the positive effect of policy perception on startups’ innovation. Figure 2 Shows the theoretical research framework of this paper.

Hypothesis 4 (H4).

The startups’ perception of makerspace support has a positive moderating effect in the process of policy perceptions’ influence on policy adaptation. That is, the positive effect of policy perception on policy adaptation is enhanced in the case of high makerspace support, and the positive effect of policy perception on policy adaptation is weakened in the case of low makerspace support.

Hypothesis 5a/b/c (H5a/b/c).

The startups’ perception of makerspace support has a positive moderating effect on the indirect effect of policy perceptions on their innovative activities/dynamic capabilities/innovative performance. That is, when the makerspace support is high, the positive mediating effect of policy adaptation is enhanced and vice versa.

4. Methodology

4.1. Research Design

This study is dedicated to testing the implementation effects of innovation and entrepreneurship policies and exploring the mechanisms and boundary conditions of policy perceptions of startups on firm behavior. Since the variables of policy perception, policy adaptation, and sense of support for makerspaces mainly measure the subjective feelings and states of startups, data on these variables cannot be obtained through secondary data. Therefore, a questionnaire survey was conducted to obtain primary data for the empirical study.

The sources of the variable measures in this study were mainly based on direct quotations or modified quotations from established scales available in Chinese and English literature. The main work of constructing questionnaires is as follows: ① Literature review. We compiled the literature on policy perception, policy adaptation, innovation behavior, dynamic capabilities, innovation performance, and the sense of support in makerspaces [11,28,39,40,41]. By referring to the scales used by other scholars and comparing and analyzing them with the Chinese innovation and entrepreneurship context [29,42], we selected the most representative scales that are most relevant to this study. These scales are widely used and have been proven to have high reliability and validity in existing studies, especially in a Chinese context. ② Scale translation. Most of the scales referenced in the study were derived from literature published in authoritative journals. Although some of the scales have been translated and used by Chinese scholars, this study continues to translate the scales based on the original English version and modify and improve the scales in the context of the study’s research topic and innovative entrepreneurship in the Chinese context by referring to Chinese scholars [29,42]. Since the respondents are not experts in the research field, this study tried to use simple and easy-to-understand language in the translation of the scale so that the respondents could accurately understand the questions, thus ensuring the accuracy of the questionnaire information collection. ③ Measures. The five-point Likert scales, which ranged from totally disagree to totally agree, were used for all items in the measurement, and the respondents chose the most appropriate level from 1 to 5 options according to their actual feelings and experiences.

4.2. Sample and Data

Our survey targets the main team members and core managers of high-tech startups in makerspaces. Firstly, the main team members and core managers of the startups are more familiar with and understand the technological innovation, product development, and corporate strategy of their firm; secondly, the startup team members and core managers are able to make decisions on the technological innovation, product development, and corporate strategy of their firm, and have the influence to transfer their personal perceptions to the firm level to form firm behavior. To ensure the high return rate of the questionnaire and the reliability of the data, we visited 500 startups clustered in three large makerspaces in the city of Wuhan over 6 months. The three makerspaces were founded by government institutions, state-owned enterprises, and private-owned enterprises. We required that the questionnaire be answered by the founder or core managers in the focal startup. After the pre-study, the recovered data were analyzed for reliability and validity, and the question items of the variables were adjusted and improved to form the formal research questionnaire for this study, followed by a large-scale questionnaire study.

Due to the epidemic control, the research team adopted three ways to distribute the questionnaires: ① distributing paper questionnaires to the person in charge during the on-site research and collecting them, which accounted for 30% of all questionnaires, i.e., 150 copies; ② delegating the person in charge of the management or guidance department of the makerspace (e.g., district government or science and technology department) to distribute paper questionnaires and collect them on their behalf, which accounted for 30% of all questionnaires, i.e., 150 copies; ③ delegating the person in charge of the management or guidance department of the makerspace to distribute and collect the electronic questionnaires, which accounted for 40% of all questionnaires, i.e., 200 copies. A total of 500 questionnaires were distributed, and 447 questionnaires were finally collected, including 294 paper questionnaires and 153 electronic questionnaires, with an overall recovery rate of 89.4%. After excluding invalid or missing data questionnaires, 391 questionnaires were finally available for empirical analysis. In this study, SPSS 20 was used for descriptive statistical analysis, reliability testing of the data, and PROCESS plug-in in SPSS for mediating and moderating effects. AMOS 20 was used to complete the discriminant validity test of the model.

4.3. Variable Measures

The six key variables of our study are policy perceptions, policy adaptation, innovative activities, dynamic capability, innovative performance, and makerspace support. The measurement items were all developed based on a review of the English and Chinese literature. We measure firms’ policy perceptions in two dimensions by referring to the policy acceptance model developed by Pierce and Willy [28], namely, perceived usefulness and perceived ease of use. The former includes four items, and the latter includes three items. To adapt to the Chinese survey context, small changes have been made according to the suggestions by Li et al. [29]. We develop and measure policy adaptation using three items, referring to the survey used by Greenspan and Granfield [39] and Mahoney and Bergman [41] for individual social adaptive behaviors such as skills, adaptation, and coping styles in social situations. We focus on firms’ ability to identify, understand, and use current policies and adjust strategic plans to adapt to the future trend. The innovative activities of startups refer to a set of actions they adopted to enable technological innovation. We adopt the scales developed by Sok [43] and Scott and Bruce [15] to measure the innovative activities because they focus more on technological innovations, which are suitable for high-tech startups. Our measure of innovative activity includes five items, and it has good reliability and validity in testing. The dynamic capability of startups is measured according to Chien et al. [44], including six items that evaluate firms’ capabilities of learning and absorbing new knowledge and skills from the market, and sharing, utilizing, and exploiting new knowledge within the firm and among employees. In terms of innovative performance, we selected three items based on Zhu and Shi [42], which are high efficiency of product update, good market response to new products, and high market share of new products among similar products by startups. The perceived makerspace support is a notion of the perceived organizational support in the research field of organizational behavior; thus, we measured it by referring to Lynch et al. [40]. We selected the control variables based on existing literature, such as the critical features of focal entrepreneurs, startups, and makerspaces. The question items measuring the key variables are reported in Appendix A.

Referring to the selection and analysis of control variables in existing studies, this study selected ten variables that have an impact on policy perception and enterprise behavior as control variables to construct the model. In this study, individual-level variables—such as age, gender, education level, and job content—as well as organization-level variables—such as the firm size, age, and the ownership of the startups; the number of firms in the makerspace; the rating of the makerspace; and the institutions on which the makerspace is based—were used as control variables in the research model. Specifically, due to gender differences, managers have different ways of thinking and may have different acuities in perceiving the external environment and producing differences in policy perception. Due to different ages and experiences, managers’ understanding of the information conveyed by external policies may differ, which may have an impact on policy perception and policy adaptation. Due to different levels of education and different knowledge bases, there are differences in the acceptance and understanding of policy changes. Due to the different job contents of managers, the knowledge they receive differs greatly, especially the knowledge related to their professional fields; therefore, the policies they pay attention to are all different, and the dimension of understanding the information conveyed by the policies will be different. The larger the scale of the startup, the more likely it is to have more resources for innovative behaviors and stronger dynamic capabilities. The shorter the establishment time of the enterprise, the less experience and resources, and its innovative behavior and dynamic ability may be affected. The degree of constraint varies with the form of firm ownership.

Table 1. Definition of control variables.

No.	Variables	Definition
1	Age	The age of interviewee
2	Gender	Gender of interviewee, woman = 1, man = 2
3	Education	Highest degree of the interviewee, ranging from 1 to 3
4	Job content	Interviewee’s responsibility in the startup
5	Firm size (ln)	Number of employees in the startup
6	Firm age	Number of months since the startup was founded
7	Ownership	Nature of ownership of the startup: private, public, or mixed ownership
8	NFM	Total number of firms in the makerspace
9	Makerspace rating	Level of makerspace recognized by the government: district, city, provincial, or national makerspace
10	Supporting institutions	Type of organization to rely on: state-owned enterprise, private-owned enterprise, government, or research institution

shows the definition of control variables.

5. Empirical Analysis

5.1. Reliability Analysis

Reliability analysis is used to analyze the internal consistency, stability, and reliability of the scale, and we use Cronbach’s alpha coefficient method to test the scale reliability. When the alpha is between 0.7 and 0.8, the reliability of the scale is good. When the alpha coefficient is greater than 0.9, the scale reliability is very high. When the Cronbach’s alpha after the item deleted (CAID) is higher than before, deleting the item can improve the reliability of the questionnaire; otherwise, the item should be retained. In this paper, all the key variables involved in the model were subjected to reliability analysis, and the results are reported in

Table 2. Results of scale reliability analysis.

Variables	Item	CAID	Cronbach’s α
Policy perception	PP1	0.867	0.893
	PP2	0.845
	PP3	0.861
	PP4	0.877
Policy adaptation	PA1	0.851	0.899
	PA2	0.856
	PA3	0.861
Makerspaces support	PMS1	0.933	0.939
	PMS2	0.921
	PMS3	0.925
	PMS4	0.924
	PMS5	0.923
Innovative activities	IB1	0.495	0.599
	IB2	0.480
	IB3	0.494
	IB4	0.833
	IB5	0.548
Dynamic capabilities	DC1	0.865	0.881
	DC2	0.856
	DC3	0.861
	DC4	0.858
	DC5	0.857
	DC6	0.870
Innovative performance	IP1	0.808	0.865
	IP2	0.811
	IP3	0.804

.

Accordingly, Cronbach’s alpha was greater than 0.8 for all the key variables except for innovative activities; the data generally had good reliability. Interestingly, the reliability coefficient of the firm’s innovative activities is 0.599, which is between effective and ineffective. It is evident that startups do not recognize their own behavior as innovative, but such behavior is in line with the intention and will of Chinese innovation and entrepreneurship policy implementers and receives policy support. In social science studies, the general requirement of Cronbach’s alpha is no less than 0.55 [45]. Therefore, the reliability coefficient of the innovative behavior of startups is not high, but it does not affect the correlation between the variables of the whole model.

5.2. Validity Analysis

The validity analysis is to verify the data validity for empirical analysis. As introduced in the previous text, the content validity is ensured by building each index on the existing theory and measurements from related literature, and with the advice of management experts, appropriate modifications were made according to the purpose of the study.

The confirmatory factor analysis (CFA) is conducted to test the scale construct validity. The KMO coefficients were all greater than 0.6 and the Bartlett’s spherical test was significant at the p < 0.001 significance level, indicating that the data are suitable for validated factor analysis. As reported in

Table 3. KOM and Bartlett sphericity of the key variables.

No.	Variables	KMO Coefficient	Chi Square	p Value of Sphericity Test
1	Policy perception	0.824	367.582	0.000
2	Policy adaptation	0.753	303.249	0.000
3	Makerspaces support	0.902	384.753	0.000
4	Innovative activities	0.809	327.847	0.000
5	Dynamic capabilities	0.884	313.687	0.000
6	Innovative performance	0.739	526.593	0.000

, the KOM values of all key variables are above 0.7, and the P value of the sphericity test is significant, indicating that CFA is appropriate for the validity test. The factor analysis was performed by AMOS to verify the model fit of the six-factor, five-factor, four-factor, three-factor, two-factor, and one-factor models, respectively; among these, the six-factor model had the best fit, so the six-factor model was used in this paper. Based on the test, the goodness-of-fit of our model is proven (χ2 = 573.064, df = 284, χ2/df = 2.018, NFI = 0.922, IFI = 0.959, TLI = 0.953, CFI = 0.959, RMSEA = 0.051) according to the criteria given by Hair et al. [46]. Therefore, this study has discriminant validity between variables, and a six-factor model is appropriate.

Then, we further analyzed the convergent validity. Convergent validity should fit three criteria: ① the standardized factor loadings should be greater than 0.5 and significant, ② the composite reliability (CR) should be greater than 0.7, and ③ the average variance extracted (AVE) should be greater than 0.5.

Table 4. Results of confirmatory factor analysis.

Variables	Item	Loadings	CR	AVE
Policy perception	PP1	0.847	0.902	0.732
	PP2	0.865
	PP3	0.904
	PP4	0.921
Policy adaptation	PA1	0.913	0.895	0.621
	PA2	0.827
	PA3	0.863
Makerspaces support	PMS1	0.922	0.937	0.809
	PMS2	0.942
	PMS3	0.919
	PMS4	0.908
	PMS5	0.939
Innovative activities	IB1	0.525	0.698	0.532
	IB2	0.553
	IB3	0.667
	IB4	0.823
	IB5	0.728
Dynamic capabilities	DC1	0.865	0.881	0.728
	DC2	0.856
	DC3	0.861
	DC4	0.858
	DC5	0.857
	DC6	0.870
Innovative performance	IP1	0.808	0.865	0.734
	IP2	0.811
	IP3	0.804

Note: CR = Composite Reliability, AVE = Average Variance Extracted.

reports the results of confirmatory factor analysis, showing that all key variables reached the criteria. The AVE values of all variables ranged from 0.532 to 0.809, indicating that the indicators can explain more than 50% of the variance in the construct [47]. Therefore, the measurement of our scales has good convergent validity.

Common method bias was tested by Harman’s single-factor test because it is the most commonly adopted test in studies, and the bias of the CMV method can be large when the source of the data is single. The Harman test yielded a total of four factors with eigenroot values greater than one, and the four factors were able to explain a total of 65.444% of the total variance, with the equation contributions of the four factors being 19.093%, 18.279%, 16.453%, and 11.618%, respectively. There were no cases where one factor alone explained most of the variance, indicating a low degree of common method bias.

5.3. Descriptive Analysis

Descriptive analysis and correlations of variables are shown in

Table 5. Descriptive analysis and correlations.

No.	Variables	Mean	S.D.	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1	Age	2.03	0.73
2	Gender	1.52	0.50	−0.088
3	Education	2.14	0.43	0.225 **	−0.128
4	Job content	1.55	1.11	−0.085	0.172 **	−0.086
5	Firm size (ln)	2.80	1.26	−0.092	−0.087	0.022	0.014
6	Firm age	22.42	8.94	−0.098	−0.040	0.038	−0.019	−0.007
7	Ownership	2.87	0.49	0.018	−0.129 *	−0.035	−0.025	0.021	0.030
8	NFM	1.77	0.90	0.148 **	−0.001	0.027	0.001	−0.069	0.005	0.058
9	Makerspace rating	2.88	1.80	−0.085	−0.047	0.088	−0.023	−0.006	0.001	0.192 **	0.038
10	Supporting institutions	3.65	1.54	−0.098	−0.033	−0.095	−0.055	0.156 **	0.088	−0.087	0.004	0.000
11	Innovative activities	3.53	0.92	−0.065	0.035	0.013	0.005	0.031	0.029	−0.022	0.008	0.049	0.039
12	Innovative performance	3.45	0.79	−0.045	0.038	0.046	−0.031	0.012	0.023	0.028	0.050	0.086	0.026	0.547 **
13	Dynamic capabilities	3.71	0.68	−0.044	0.026	0.001	−0.027	−0.017	0.004	−0.026	0.002	0.023	0.015	0.556 **	0.489 **
14	Policy perception	3.75	0.72	−0.085	0.083	−0.013	−0.068	−0.006	0.020	0.017	0.018	0.097	0.006	0.425 **	0.483 **	0.543 **
15	Policy adaptation	3.69	0.77	−0.057	0.037	0.040	−0.032	−0.025	0.002	0.051	0.016	0.036	0.011	0.424 **	0.508 **	0.521 **	0.480 **
16	Makerspace support	3.85	0.79	−0.071	0.023	0.014	−0.047	−0.042	0.018	0.040	0.080	0.080	0.041	0.432 **	0.446 **	0.526 **	0.432 **	0.524 **

Note: * p < 0.10, ** p < 0.05.

. The mean of the firm age is 22.42 months, indicating that our sample firms are mostly startups. The policy perception has a significant positive correlation with innovative activities (0.43), dynamic capability (0.48), and innovative performance (0.54). The policy adaptation is positively correlated with policy perception and outcome variables. The correlations of the main variables are initially consistent with the hypotheses presented in the study, laying the foundation for regression.

By conducting empirical analysis, this study examines the main effect of policy perception on innovative activities/dynamic capabilities/innovative performance (H1a/b/c), the mediating effect of policy adaptation (H2, H3a/b/c), the moderating effect (H4), and the moderated mediating effect (H5a/b/c). The results are reported separately in the following sub-sections.

5.4. Results of Mediating Effect

In

Table 6. PROCESS analysis of mediating effect.

	Innovative Activities (Y1)		Dynamic Capabilities (Y2)		Innovative Performance (Y3)		Policy Adaptation (M)
Control Variables	M1	M2	M4	M5	M6	M7	M8
Age	0.546	−0.033	−0.008	−0.005	−0.019	−0.015	−0.013
Gender	−0.037	−0.006	−0.043	−0.037	0.016	0.024	−0.024
Education	−0.013	0.008	0.011	−0.015	0.089	0.054	0.103
Job content	0.039	0.020	0.007	0.003	0.002	−0.004	0.016
Firm size	0.025	0.000	0.000	0.000	0.000	0.000	0.000
Firm age	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Ownership	0.000	−0.080	−0.069	−0.084	0.041	0.021	0.059
NFM	−0.062	0.009	0.012	0.014	0.051	0.053	−0.005
Makerspace rating	0.008	0.000	−0.003	−0.002	0.003	0.003	−0.002
Supporting institutions	−0.000	−0.000	0.000	−0.000	−0.000	−0.000	0.000
Independent variables
Policy perception	0.546 ***	0.323 ***	0.524 ***	0.339 ***	0.529 ***	0.276 ***	0.736 ***
Mediating variables
Policy adaptation		0.302 ***		0.251 ***		0.343 ***
R	0.434	0.472	0.548	0.586	0.494	0.551	0.685
R2	0.189	0.223	0.300	0.343	0.244	0.303	0.469
MSE	0.704	0.676	0.336	0.316	0.490	0.453	0.327
F	8.013	9.048	14.758	16.431	11.090	13.695	30.401

Note: *** p < 0.01.

, model 1 tested H1a, the effect of policy perception on innovative activities; model 4 tested H1b, the effect of policy perception on dynamic capabilities; and model 6 tested H1c, the effect of policy perception on innovation performance. To measure the mediating effect of policy adaptation, model 8 reported the effect of policy perception on the policy adaptation, and model 2/5/7 resulted by adding mediators into the main effect.

As shown, the policy perception significantly facilitates startups’ innovative activities (β = 0.55, p < 0.01). Similarly, the effects of policy perception on startups’ dynamic capabilities (β = 0.52, p < 0.01) and innovative performance (β = 0.53, p < 0.01) are also confirmed by empirical evidence. The H1a/b/c is supported by the result. Model 8 shows that startups’ policy perception improves their policy adaptation (β = 0.74, p < 0.01) and supports H2. By adding the policy adaptation to the regression model, the mediating effect of policy adaptation is tested, and the H3a/b/c are all supported according to model 2/5/7. The total effect consists of a direct effect and an indirect effect through the mediating role of the policy adaptation. The indirect effect of policy adaptation on innovative activities/dynamic capabilities/innovative performance is 0.22/0.19/0.25, respectively.

5.5. Results of Moderating Effect

As shown in

Table 7. Regression results of moderating effect of perceived makerspace support.

	Innovative Activities	Dynamic Capabilities	Innovative Performance	Policy Adaptation
Control Variables	M9	M10	M11	M12
Age	−0.302	−0.005	−0.015	−0.005
Gender	−0.006	−0.369	0.024	−0.022
Education	0.008	−0.015	0.054	0.090
Job content	0.020	0.003	−0.001	0.015
Firm size	0.000	0.000	0.000	0.000
Firm age	0.000	0.000	0.000	0.000
Ownership	−0.080	−0.084	0.021	0.044
NFM	0.009	0.014	0.053	−0.019
Makerspace rating	0.001	−0.002	0.003	−0.028
Supporting institutions	−0.000	−0.000	−0.000	0.000
Independent variable
Policy perception	0.323 ***	0.339 ***	0.276 ***	0.236
Mediating variable
Policy adaptation	0.302 ***	0.251 ***	0.343 ***
Moderating variable
Makerspace support				−0.222
Interaction
Policy perception * Makerspace support				0.104 **
R	0.472	0.586	0.551	0.703
R2	0.223	0.343	0.303	0.494
MSE	0.676	0.316	0.453	0.314
F	9.048	16.431	13.695	28.264

Note: * p < 0.10, ** p < 0.05, *** p < 0.01.

, the interaction term between policy perception and makerspace support has a significant positive effect on policy adaptation (β = 0.104, p < 0.01), so the positive effect of policy perception on policy adaptation is moderated by the perceived makerspace support by startups. The positive effect of policy perception on policy adaptation is enhanced in the case of high perceived makerspace support and weakened in the case of low perceived makerspace support. The moderating effect of the perceived makerspace support is shown in Figure 3. In the case of high perceived makerspace support, the slope of the linear relationship is steeper, and the intensity of the effect is stronger; in the case of low perceived makerspace support, the slope of the linear relationship is flatter, and the intensity of the effect is weaker. H4 is supported. The startups’ perception of makerspace support has a positive moderating effect on the process of policy perceptions’ influence on policy adaptation. According to our analysis, the effect of policy perception on policy adaptation is significant in the case of high perception of makerspace support (β = 0.735, 95% CI = [0.611, 0.860]) and in the case of low perception of makerspace support (β = 0.548, 95% CI = [0.431, 0.665]).

5.6. Results of Moderated Mediating Effect

Moderated mediation exists when the mediating process that connects the relationship between the independent variable and the dependent variable by the mediating variable is influenced by the moderating variable [48]. One standard deviation above/below the centralized mean was used to represent contexts with high/low perceived makerspace support.

The indirect effect of policy perception through policy adaptation on the innovative activities of startups in the makerspace is significant in the case of high makerspace support perception (β = 0.222, 95% CI = [0.143, 0.316]), and the indirect effect of policy perception through policy adaptation on the innovative activities of startups in the makerspace is also significant in the case of low makerspace support perception (β = 0.166, 95% CI = [0.103, 0.239]). However, the difference between the two cases of high and low perception of makerspace support is also significant (β = 0.031, 95% CI = [0.008, 0.068]), indicating that the mediating process of policy adaptation linking policy perception and innovative activities of startups in the makerspace is influenced by their perceived makerspace support; thus, there is a moderated meditating effect. H5a is supported. The startups’ perception of makerspace support has a positive moderating effect on the indirect effect of policy perceptions on their innovative activities.

Similarly, the indirect effect of policy perception through policy adaptation on the dynamic capabilities of startups in the makerspace is significant in the case of high makerspace support perception (β = 0.184, 95% CI = [0.119, 0.264]), and the indirect effect of policy perception through policy adaptation on the dynamic capabilities of startups in makerspace is also significant in the case of low makerspace support perception (β = 0.138, 95% CI = [0.085, 0.201]). The difference between the two cases of high and low perception of makerspace support is significant (β = 0.026, 95% CI = [0.006, 0.055]), indicating that the mediating process of policy adaptation linking policy perception and dynamic capabilities of startups in makerspace is influenced by their perceived makerspace support; thus, there is a moderated meditating effect. H5b is supported. The startups’ perception of makerspace support has a positive moderating effect on the indirect effect of policy perceptions on their dynamic capabilities.

The indirect effect of policy perception through policy adaptation on the dynamic capabilities of startups in the makerspace is significant in the case of high makerspace support perception (β = 0.252, 95% CI = [0.161, 0.349]), and the indirect effect of policy perception through policy adaptation on the dynamic capabilities of startups in makerspaces is also significant in the case of low makerspace support perception (β = 0.188, 95% CI = [0.116, 0.263]). The difference between the two cases of high and low perception of makerspace support is also significant (β = 0.036, 95% CI = [0.008, 0.076]), indicating that the mediating process of policy adaptation linking policy perception and innovative performance of startups in makerspaces is influenced by their perceived makerspace support; thus, there is a moderated meditating effect. H5c is supported. The startups’ perception of makerspace support has a positive moderating effect on the indirect effect of policy perceptions on their innovative performance.

5.7. Further Test

The three outcome variables proposed in this paper, i.e., innovative activities, dynamic capabilities, and innovation performance, also have an “action-capability-performance” chain from a theoretical point of view. Therefore, this paper further tested the mediating effect of dynamic capabilities on innovative activities and innovation performance of these startups. The results are shown in

Table 8. Relationship of outcome variables.

	Innovative Performance		Dynamic Capabilities
	Model 13	Model 14	Model 15
Age	−0.021	−0.013	−0.016
Gender	0.066	0.058	0.006
Education	0.081	0.075	−0.003
Job content	−0.011	−0.013	−0.020
Firm size	0.000	0.000	0.000
Firm age	0.000	0.000	−0.000
Ownership	0.104	0.104	−0.022
NFM	0.040	0.041	0.006
Makerspace rating	0.005	0.005	−0.001
Supporting institutions	0.000	0.000	0.000
Innovative activities	0.469 ***	0.200 ***	0.414 **
Dynamic capabilities		0.649 ***
R	0.558	0.726	0.559
R2	0.312	0.526	0.312
MSE	0.446	0.308	0.330
F	15.586	34.999	15.630

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

. Model 13 tested the causal relationship between innovative activities and innovation performance, and the results are significant (r = 0.469, p < 0.001). After adding the variable of dynamic capabilities of startups, model 14 shows that firm innovative activities still have a significant positive effect on innovation performance (r = 0.200, p < 0.001), but the degree decreases, which, combined with the results of model 15, indicates that the effect of innovative activities of startups on innovation performance is partially mediated by dynamic capabilities. Specifically, by analyzing the mediating path and the effect of dynamic capabilities in startups, the total mediating effect of dynamic capabilities is 0.469. The direct and indirect mediating effects of dynamic capabilities are 0.200 and 0.269, respectively, indicating that dynamic capabilities explain about 27% of the contribution of innovation activities to innovation performance.

6. Discussion

6.1. Theoretical Implications

This study provides Chinese evidence on whether policy incentives are effective in stimulating innovation, bridging the gap in the existing literature that focuses mainly on developed regional markets [49]. Moreover, by combining institutional theory with RBV theory, this study innovatively views policy not only as a given contextual feature but also as a scarce resource that can be directly absorbed and utilized by firms. This perspective could be enlightening for related research on government policies. Based on the policy-making perspective, Mazzucato [2] identified four key issues (i.e., directionality, evaluation, organization, risk, and rewards) in framing market-creating policies. We further reveal the organizational responses in a given policy context and the underlying mechanisms that induce these responses. The purpose of China’s innovation and entrepreneurship policy is to manage innovation in technology for the state, rather than for individual production technology. From this perspective, our study finds that Chinese innovation and entrepreneurship policy formulation and implementation are successful and extremely effective, which indicates that the proper design of innovative policies is critical for encouraging startups and their innovations.

In addition to providing a positive answer to the controversial academic question of whether top-down policy incentives have a catalytic effect on the innovation activities of startups, firms’ policy perception capabilities are critical for transmitting the incentive effects of policies. This adds to the existing literature on the specific mechanisms of how policy stimulates corporate innovation. Startups can gather information and resources from the external context and engage in innovative activities, improving dynamic capabilities and thus improving innovative performance. This implies that variation in innovation behavior and performance of startups in similar environments is largely attributable to the heterogeneity of their policy adaptation capabilities. Given that Mohedano-Suanes et al. [18] identified the mediating role of market perception in the relationship between enterprises’ innovation and long-term performance, this study completes the “environment-behavior-effects” chain by adding the missing link from external pressures to internal innovation behavior. The revelation of policy adaptation mediation mechanisms also indicates that, to a large extent, the capabilities and innovation orientation of startups in a government-dominated environment are policy-driven.

This paper also revealed the linkage among the three outcome variables, which not only provided an in-depth understanding of the policy implementation effects but also enhanced the existing understanding of the antecedents of innovation performance based on previous findings [12,17]. The significant relationship among the three outcome variables in the further test shows that the innovative activities of startups can improve their innovative performance by nurturing and developing dynamic capabilities. Thus, there is a dynamic facilitative relationship between firms’ innovative actions, capabilities, and performance under the influence of policies. At the national level, innovative performance can be stimulated by effective policies but is unlikely to improve substantially in the short term.

6.2. Practical Implications

This study also provides inspiration for policymakers. First, consistent with the findings of Zubeltzu-Jaka et al. [31], our study suggests that the policy as a context is critical for startups. Entrepreneurs and managers should improve their policy acumen, including not only financial subsidies and tax incentives directly related to “mass innovation and entrepreneurship”, but also indirectly related to industrial deployment, environmental protection, scientific and technological research and development support, and other important policies. Government officials can facilitate startups to innovate by establishing a perceivable and adaptable policy context, improving the policy usefulness and ease of use from the perspective of target firms. Second, a well-established makerspace has a significant positive impact on the innovative development of startups, which provides empirical evidence for the conceptual findings of Cuvero et al. [19]. Moreover, this positive impact of the makerspace is accomplished not only through knowledge spillover revealed by existing studies [19] but also by promoting a deep understanding of firms’ policy perceptions and translating their adaptation to the policy environment into concrete innovative actions, capabilities, and performance.

6.3. Limitations and Future Research

This paper also has some limitations. A major concern is that we collected primary data from startups in three representative makerspaces in Wuhan, China, which may limit our findings to startups in central China, where entrepreneurial and innovative resources and opportunities are more scarce compared to developed regions. However, based on an institutional perspective, we believe that our findings are valid in the vast majority of regions where the government is dominant and has the power to make key resource allocation decisions.

We provide two directions for future studies. One direction is to improve the research in the narrow innovation scope by providing more evidence for different types of innovation. The difference in the survival of innovative startups with or without policy support is also worth examining. The other is to extend the research perspective to a broader range of innovative behavior and to integrate it with frontier technological developments. For example, Jia et al. [50] assessed the impact of the application of AI technologies on innovation performance from the perspective of employee creativity. Besides policy and markets, technology is probably the most important driver of innovation in the coming period, and it deserves focused attention.

7. Conclusions

This study uses a sample of 500 startups located in the three representative makerspaces in China to conduct an empirical analysis at the micro-firm level, exploring the mechanism of the policy influence on the behavior of startups in makerspaces. The findings indicate the following:

• Chinese policy incentives indeed drive innovative activities and enhance the innovative capabilities and performance of high-tech startups. When the startups in makerspaces perceive that the government’s policies are supportive of their own innovation, their confidence and motivation to innovate are enhanced, and their innovative activities, capabilities, and performance increase. When the startups perceive that the threshold of government EIP support is low and it is easy to obtain the corresponding project support, this will also improve their enthusiasm and capability to innovate.
• Based on the findings of the mediating effect, policy adaptation, as the startups’ interpretation of the innovation and entrepreneurship policies in their region, is the foundation for startups to adjust their own operations and development to adapt to the macro policy environment, reduce innovation and entrepreneurship risks, and improve the success rate of entrepreneurship. Therefore, innovation and entrepreneurship policy perception can positively influence the policy adaptation of startups in makerspaces, thus promoting the innovative activities, dynamic capability, and innovative performance of the startups in makerspaces.
• Makerspace support—as the platform for policy implementation, service delivery, and entrepreneurial ecosystem construction—is influential for startups’ innovative behaviors as it provides key resources based on policies such as finance, technology, and experts. The positive effect of policy perception on policy adaptation is enhanced when the perception of makerspace support is high, while the positive effect of policy perception on policy adaptation is weakened when the perception of makerspace support is low. In addition, the three dependent variables in the model are empirically tested under the “behavior-ability-outcome” paradigm. That is, based on the dynamic capability perspective, the innovative behavior of startups can enhance their dynamic capability, which in turn enables them to achieve higher innovative performance.