A Study on the Impact of Boundary-Spanning Search on the Sustainable Development Performance of Technology Start-Ups

Abstract

Boundary-spanning search and ambidextrous learning are the key means for organizations to absorb and internalize external heterogeneous knowledge and play an important role in the sustainable development of enterprises. From the perspective of organizational learning and value co-creation, this paper constructs a theoretical model of the impact of boundary-spanning search on sustainable development performance and conducts multiple regression analysis and a bootstrap test based on the sample data of Chinese start-up technology enterprises. The results show that the following: (1) boundary-spanning search contributes to the sustainable development performance of technology start-ups; (2) the role of ambidextrous learning as a partial mediator between boundary-spanning search and the sustainable development performance of technology start-ups; (3) value co-creation positively regulates the relationship between boundary-spanning search and exploratory learning, while the regulation effect between boundary-spanning search and exploitative learning is not significant; (4) value co-creation strengthens the intermediary role of exploratory learning between boundary-spanning search and the sustainable development performance of start-up technology enterprises. The findings help reveal the mechanisms by which boundary-spanning search affects the sustainable development performance of technology start-ups and their boundary conditions.

1. Introduction

Driven by mass entrepreneurship and innovation strategy, Chinese technology enterprises have proliferated and gradually developed into the backbone of national and regional innovation development. At this time, sustainable development has put forward higher requirements for enterprises’ innovation and entrepreneurial activities, and ecological and environmental protection has been included as an important factor to be considered in enterprises’ innovation and entrepreneurial activities. In other words, companies need to pay attention to social and environmental benefits while focusing on economic benefits [1]. However, compared with mature companies, technology start-ups suffer from the dual endowment defects of “new entrant defect” and “weakness defect”, which expose the lack of internal innovation knowledge and the lack of success in innovation activities in the actual operation process [2]. Breaking the constraint of own resources has become the key to breaking the dilemma of the development of new technology start-ups. Boundary-spanning search is a knowledge acquisition strategy for enterprises to search for external novel knowledge to leverage existing knowledge innovation, catalyze the fusion of old and new knowledge and generate knowledge synergy effects [3]; this provides an important way for technology start-ups to break through the “growth dilemma” [4]. Currently, scholars generally agree that boundary-spanning search has a positive impact on improving business performance [5]; however, the path of this influence is mainly studied from the perspective of capacity building, trying to follow the logical line of “behavior-competence-performance” to explain the mechanism of action: an empirical study by Lavie verified that boundary-spanning search drives firm innovation through the bridging role of capability reconfiguration [6]; Wu et al. showed that firms with stronger dynamic capabilities can quickly integrate internal and external resources to enhance their competitive advantage [7]. However, the essence of boundary-spanning search is a process of searching for knowledge and information [8], the learning internalization effect of this knowledge is the direct contributor to the difference in performance results [9]. Little research has been conducted to uncover the mechanisms driving the sustainable development performance of technology start-ups from an organizational learning perspective.

In the process of organizational learning, due to the constraints of resource endowment, March proposed that organizational learning contains the dual attributes of exploration and exploitation [10]. Among them, exploratory learning aims to uncover the acquisition of potential new knowledge that differs significantly from the firm’s existing knowledge, while exploitative learning emphasizes the refinement, integration, enhancement, and improvement of existing knowledge [11], both of which will play an important role in the relationship between external knowledge absorption, integration and firm innovation [3]. However, it remains to be revealed whether both are beneficial for external knowledge integration and sustainable development performance improvement for technology start-ups facing “double defect” constraints. In addition, in the context of open innovation and digitalization, value co-creation has become an inevitable choice for newly created technology companies to expand their information, resource, and technology chains and tap into user value [12], which accelerates the process of connecting internal and external knowledge resources of the organization through closer interaction and cooperation with suppliers, users, research institutions, government agencies, and other relevant stakeholders, and then influences the effect of ambidextrous learning [13]. Therefore, uncovering the moderating mechanisms of value co-creation is crucial to revealing the boundary conditions of the impact of boundary-spanning search on the sustainable development performance of technology start-ups. Based on the perspective of organizational learning and value co-creation, this paper follows the logic of “behavior-learning-performance” and introduces ambidextrous learning and value co-creation as mediating and moderating variables, respectively, to investigate the mechanism of the boundary-spanning search for sustainable development performance of technology start-ups; empirical research methods such as multiple regression analysis and bootstrap test are used to test the proposed hypothesis on the survey data from Chinese start-up technology enterprises, to provide theoretical support for technology start-ups to break through the “growth dilemma”. The contribution of this paper will be in the following two aspects: First, based on the perspective of organizational learning, the relationship between boundary-spanning search and the sustainable development performance of start-ups is studied in depth according to the logical framework of “behavior-learning-performance”, with ambidextrous learning as the mediating variable and value co-creation as the moderating variable, The study is a breakthrough from the current perspective of capability constructs, and the research framework is closer to the real situation of knowledge acquisition, internalization and utilization than the “behavior-competence-performance” framework, providing a new way of thinking to reveal the mechanism of boundary-spanning search affecting the sustainable development performance of technology start-ups, and enriching the research on boundary-spanning search and organizational learning; Second, in the context of open innovation and digitalization, value co-creation has become an inevitable choice for start-up technology companies to overcome resource constraints and improve the efficiency of product marketability, the introduction of value co-creation into the theoretical model and the empirical testing of its moderating mechanism provide a more realistic analysis to clearly delineate the boundary conditions of boundary-spanning search on the sustainable development performance of technology start-ups, and provide useful reference for managers to enhance organizational learning and stimulate organizational creativity.

2. Theoretical Background and Hypothesis Development

2.1. Theoretical Background

This paper provides a search and overview of the literature related to boundary-spanning search and technology start-ups in the Web of Science database. The search was performed with the keywords of boundary-spanning search, ambidextrous learning, and sustainable development. First, with regard to research on the definition of boundary-spanning search, it was found that Rosenkopf and Nerkar [14] were the first to introduce boundary-spanning search into the field of strategic management, and defined boundary-spanning search as the process of acquiring external heterogeneous knowledge across organizational and technological boundaries in a complex and dynamic environment for firms. Since then, scholars have conducted more and more studies on boundary-spanning search with different understandings of boundary-spanning search, and there is not yet a unified understanding of the definition and connotation of boundary-spanning search. Existing studies mainly define its concept and connotation from three perspectives: knowledge distance perspective, resource base perspective, and organizational learning perspective; the main views are shown in

Table 1. Definition of the concept and connotation of boundary-spanning search.

Research Perspectives	Representative Scholars	Define Meaning
Knowledge Distance Perspective	Rosenkopf & Nerkar (2001) [14]	Boundary-spanning search is derived from remote search, as opposed to local search, which is a search for new knowledge that is far away from the enterprise and in a different field.
Resource base perspective	Katila & Ahuja (2002) [5]	Boundary-spanning search is the process of searching for heterogeneous resources to acquire new knowledge, skills, and processes.
Organizational Learning Perspective	Roper (2017) [15]	Boundary-spanning search is not only a way to solve problems but also a way for organizations to learn, to learn in the environment, discover new ways to create value, or come up with new solutions to old problems.

.

The relevant research on the relationship between boundary-spanning and start-up technology enterprises mainly focuses on the following two aspects. Firstly, a portion of scholars’ results support the positive impact of boundary-spanning search on the innovation performance of start-ups. Kim and Park [16] investigated the relationship between exploratory search and innovation impact and showed that boundary-spanning search has a significant positive impact on innovation impact; the empirical findings of Choi [17] suggest that the search breadth of scientific knowledge has a positive impact on innovation disruption by enhancing the technological output of the sector. Secondly, the results from other scholars suggest that when search width and search depth exceed a certain threshold, they instead hinder organizational innovation [18]; the results of an empirical study by Ardito et al. [19] on Italian firms show an inverted U-shaped relationship with new product development performance as the boundary-spanning search width increases. In addition to this, Terjesen’s [20] study shows that the breadth search model is not conducive to the development of enterprise specialization, so it will reduce the innovation output. Although there are some differences in the research conclusions of scholars on the relationship between boundary-spanning search and enterprise-related performance, many scholars believe that boundary-spanning search is conducive to enterprise innovation, and innovation will stimulate the momentum of sustainable development of enterprises [21]. However, the existing research on the sustainable development of start-up technology enterprises is relatively small, and it is necessary to further explore the impact mechanism of boundary-spanning search on the sustainable development performance of start-up technology enterprises.

Elkington [22] divides the sustainability of enterprises into three bottom lines of economy, society, and environment, and believes that enterprises should abide by social and environmental benefits while ensuring economic growth. Sustainable development requires enterprises to continuously update and iterate new products, to ensure the triple benefits to economy, society, and environment. This process depends on the internalization and integration of external heterogeneous knowledge. Organizational learning theory points out that organizational learning is an important way for enterprises to quickly carry out business and achieve sustainable development. Especially for start-up technology enterprises, organizations can only continuously learn new external knowledge and integrate their knowledge to deal with the threats and challenges of the fierce external environment. Among them, ambidextrous learning is one of the focuses of research in the field of organizational learning. Ambidextrous learning includes both exploratory learning and exploitative learning [10]. Exploratory learning is the process of exploring the potential knowledge in the external market environment that has a large gap with the existing knowledge of the company with the orientation of complete differentiation, and the knowledge obtained through exploratory learning, although with high uncertainty [23], can add to the strengths of the company and help the company to capture new dynamics and construct new competitive advantages in the fast-changing technological environment [24]. Unlike exploratory learning, exploitative learning is a process in which companies acquire knowledge through repetitive in-depth learning based on previous experience with limited differentiation [25], and the knowledge gained through exploitative learning can provide strong support for companies to continuously optimize their products and services, consolidate their existing advantages, and gradually expand their market share [11]. Combining these views, this study defines exploratory learning as the process and capability of an organization to continuously expand its original knowledge base by entering a new field, and exploitative learning as the process and capability of fully exploiting existing knowledge and trying to create synergistic effects with new knowledge. According to the research of Wei and Guo [26], exploratory learning and exploitative learning should have a relationship with the sustainable development of enterprises, because enterprises can not only absorb a large amount of external novel knowledge through exploratory learning, but also continuously deepen the understanding of existing knowledge, to continuously increase their knowledge reserves; at the same time, rich knowledge reserves are conducive to improving their performance, upgrading and iterating green products, and optimizing existing products, to lay a solid foundation. This provides a theoretical basis for this paper to study the impact of boundary-spanning search on the sustainable development performance of start-up technology enterprises from the perspective of organizational learning.

2.2. Boundary-Spanning Search and Sustainable Development Performance of Technology Start-Ups

Boundary-spanning search originates from the theory of organizational search, which refers to the process of repositioning and reorganizing various types of knowledge and creating new knowledge in an organization [13]. Rosenkopf and Nerkar [14], who focused on the process of acquiring heterogeneous knowledge across organizational and technological boundaries by enterprises in a complex and dynamic environment, proposed the concept of boundary-spanning search, and defined it as an external search activity conducted by enterprises across organizational boundaries and knowledge bases based on the knowledge distance perspective. Subsequently, many scholars have defined the boundaries of boundary-spanning search from organizational and knowledge perspectives, where organizational boundaries mainly include corporate boundaries [14] and industrial boundaries [5], while knowledge boundaries include cognitive boundaries [27], geographical boundaries [28], and temporal boundaries [29]; in addition to the above unidimensional boundary definition, Choi [17] and Flor [30] focus on the combination of organizational and knowledge boundaries and the combination between different knowledge boundaries to define the boundary of boundary-spanning search in a multidimensional combination. This paper defines boundary-spanning search as a heterogeneous knowledge resource search activity conducted by an organization across the boundary of organization and knowledge base to discover and create value.

Sustainable development sets higher standards for corporate process and product innovation, thus meeting social and environmental standards on top of achieving economic gains. The knowledge base view suggests that external knowledge sources are the key factors that stimulate innovative thinking in organizations and influence the innovation process in enterprises [31], and the search for this knowledge is beneficial in breaking the shackles of innovative thinking, improving production processes, and enhancing technological output. Firstly, it enables organizations to establish cooperative relationships with multiple subjects, promotes the sharing of knowledge among multiple parties, and provides a strong guarantee for companies to maintain sustained competitiveness [32]; secondly, the “creative collision” with external knowledge can break through the inherent thinking, timely revise the trajectory of technological development, and even form new practice [33]; third, enterprises search for scarce and complementary resources across organizational boundaries and have more opportunities to access emerging technologies and capture new market opportunities, thus helping them to develop technology portfolio advantages and synergy effects [34]. Therefore, in the scenario of scarce resource endowment, boundary-spanning search has a positive effect on both the acquisition of complementary knowledge and the improvement of existing knowledge systems by enterprises. As such, we formulate the following hypothesis:

Hypothesis 1 (H1).

Boundary-spanning search has a positive effect on the sustainable development performance of technology start-ups.

2.3. Ambidextrous Learning and Sustainable Development Performance of Technology Start-Ups

Organizational learning theory suggests that organizations can quickly acquire knowledge in R&D, marketing, and other areas using learning, which is an effective means for companies to expand their existing knowledge base and promote growth [35]. On the one hand, through exploratory learning, enterprises can acquire new knowledge that is distinct from the existing knowledge base, which will have a positive impact on opening up sources of creativity, improving the efficiency and quality of new product development, and broadening new product development channels [36]; in addition, exploratory learning processes break through existing learning models in iterative experimentation and change, which will effectively enhance the ability of companies to adapt to their environment [37]. Additionally, it helps companies to develop new products in line with the economic and social development process, thus benefiting sustainable development performance. On the other hand, exploitative learning emphasizes continuous and repeated refinement, replication, and reinforcement of the original knowledge bundle to achieve innovative use of the knowledge resources at hand by deeply exploring the multiple values of the existing knowledge resources, which not only facilitates enterprises to promptly capture the current market needs but will also effectively enhance the efficiency of internal resource use [38]. In addition, the more frequently an enterprise adopts exploitative learning, the more efficient it is in applying and integrating existing knowledge, which can promote rapid iterative innovation of its products and services [25]. That is, doing so allows enterprises to meet the dual requirements of enterprise economic benefits and energy cost consumption at the same time, thus to a certain extent meeting the requirements of sustainable development of enterprises. In summary, the characteristics of ambidextrous learning such as “learning from the past” and “knowing the new” promote sustainable development and performance improvement in terms of expanding the knowledge base, improving organizational adaptability, enhancing resource efficiency, and accelerating iterative innovation. As such, we formulate the following hypothesis:

Hypothesis 2a (H2a).

Exploratory learning contributes to the sustainable development performance of technology start-ups.

Hypothesis 2b (H2b).

Exploitative learning contributes to the sustainable development performance of technology start-ups.

2.4. Boundary-Spanning Search and Ambidextrous Learning

In response to rapid changes in the external environment, the two characteristics of boundary-spanning search, breadth, and focus, drive the process of organizational learning [15]. On the one hand, boundary-spanning search promotes the embedding of enterprises into multi-node innovation networks and multi-subject innovation ecosystems expands the diversified knowledge sources of enterprises, beneficially breaks down inter-organizational knowledge boundary barriers, and provides a wider range of differentiated knowledge for organizations to break through existing knowledge and technology limitations and stimulate breakthrough innovation results [39], thus enhancing the efficiency of exploratory learning. On the other hand, enterprises rely on themselves and focus on the existing knowledge resource base, and they search for the required knowledge in the multidimensional technology field by crossing organizational and knowledge boundaries, which is beneficial to the docking and reorganization of different types of knowledge, forming the cross-border intermingling of multidisciplinary and multi-dimensional knowledge, refining and improving the existing enterprise technology growth paths and operation method [40], promoting the effective marketization of current knowledge and, thus, enhancing the relevance and effectiveness of exploitative learning activities. In summary, boundary-spanning search activities enhance the efficiency and effectiveness of ambidextrous learning by expanding a wide range of stable social network relationships and conducting targeted focus mining. As such, we formulate the following hypothesis:

Hypothesis 3a (H3a).

Boundary-spanning search positively influences the effect of exploratory learning.

Hypothesis 3b (H3b).

Boundary-spanning search positively influences the effect of exploitative learning.

2.5. The Mediating Role of Ambidextrous Learning between Boundary-Spanning Search and Sustainable Development Performance of Technology Start-Ups

In the context of open innovation, enterprises can acquire richer external heterogeneous knowledge through boundary-spanning search, but they usually need to digest, assimilate, integrate or innovate this knowledge before applying it to value-added aspects such as improving product and service quality, expanding new markets or optimizing management processes [41], and this process relies on organizational ambidextrous learning to internalize knowledge resources [11]. Specifically, exploratory learning can integrate the new and completely differentiated knowledge acquired into the original knowledge system to stimulate “creative collisions” [42], giving rise to creative ideas and thoughts for breakthrough innovations such as product and process innovation and business model innovation, which in turn can provide powerful support for enterprises to explore blue ocean markets and pre-empt new opportunities [38]. Exploitative learning is more of a progressive innovation activity [43], such as product improvement and channel expansion, through “gap-filling” and repetitive in-depth learning of existing knowledge to expand the scope of application of existing technologies and make them more suitable for the external environment [44], thus converting the knowledge gained into benefits. In summary, ambidextrous learning is the process of digesting and assimilating the knowledge resources gained and putting them into action, acting as a bridge between boundary-spanning search and the sustainable development performance of technology start-ups. As such, we formulate the following hypothesis:

Hypothesis 4a (H4a).

Exploratory learning mediates the process of the impact of boundary-spanning search on the sustainable development performance of technology start-ups.

Hypothesis 4b (H4b).

Exploitative learning mediates the process of the impact of boundary-spanning search on the sustainable development performance of technology start-ups.

2.6. The Moderating Role of Value Co-Creation between Boundary-Spanning Search and Ambidextrous Learning

Value co-creation is a new paradigm for enterprises to dig deep into the potential value of users and pull the production ecosystem operation by interacting with them, which is characterized by multiple elements, diverse subjects, and variable processes, and helps enterprises cross inter-organizational barriers and widely absorb heterogeneous knowledge, thus affecting the process of cross integration of internal and external knowledge [45]. Putz found that frequent information exchange and adequate communication helped to include consumers in the design of product development and were beneficial in scaling up the business to capture future market trends [46]; Gronroos found that the value co-creation process in which suppliers provide the organization with the technology and product resources needed for production, as well as the latest market feedback, helps to bridge the knowledge gap, improve the existing knowledge system, and promote organizational learning activities [47]. In addition, through value co-creation activities with universities and research institutions, companies can better carry out organizational learning activities and quickly understand and even master cutting-edge technological information, thus promoting the integration of old and new knowledge and structurally compensating for the weak innovation capacity [48]. Therefore, value co-creation has become an important factor influencing the acquisition and transformation of external knowledge.

However, in reality, the willingness of value co-creation is influenced by many factors such as value fit, benefit distribution mechanisms, and task difficulty [49], and the value co-creation scenarios faced by different firms vary, thus affecting the efficiency of the uptake of knowledge gained from boundary-spanning search. When in a high-value co-creation scenario, the degree of cooperation between enterprises and the external environment is stronger, and it is easier to understand and transform the shared knowledge of each other, while for frontier technology knowledge and extrapolated hotspot knowledge, the form of cooperative innovation can also be used to promote the internalization of knowledge absorption of both parties [50], thus benefiting the organization to efficiently dock with external knowledge resources and enhance the knowledge transformation effect. On the other hand, in the low-value co-creation scenario, there are large differences in values between enterprises and multiple subjects, and there are biases in the understanding of heterogeneous knowledge transformation, which to some extent weaken the extent of the role of boundary-spanning search on ambidextrous learning, which in turn affects the efficiency of external knowledge transformation [13]. It can be seen that the interaction process with external multiple subjects can limit or expand the absorption and transformation of the searched knowledge by the enterprise, which is manifested in the difference in the efficiency of the transformation and integration of knowledge by the organization under different value co-creation scenarios. As such, we formulate the following hypothesis:

Hypothesis 5a (H5a).

Value co-creation positively moderates the impact of boundary-spanning search on exploratory learning.

Hypothesis 5b (H5b).

Value co-creation positively moderates the impact of boundary-spanning search on exploitative learning.

From the above analysis, it is clear that on the one hand, the boundary-spanning search can enhance the sustainable development performance of enterprises through ambidextrous learning activities, while on the other hand, value co-creation can enhance the effect of boundary-spanning search on ambidextrous learning. Logically, value co-creation may play a moderating role in the mediating mechanism of boundary-spanning search to enhance sustainable corporate performance through ambidextrous learning, i.e., there is a mediating effect of being moderated. As such, we formulate the following hypothesis:

Hypothesis 6a (H6a).

Value co-creation moderates the mediating role of exploratory learning between boundary-spanning search and sustainable development performance of technology start-ups, i.e., the higher the intensity of value co-creation, the stronger the mediating effect of exploratory learning.

Hypothesis 6b (H6b).

Value co-creation moderates the mediating role of exploitative learning between boundary-spanning search and sustainable development performance of technology start-ups, i.e., the higher the intensity of value co-creation, the stronger the mediating effect of exploitative learning.

In summary, this study constructs a theoretical model of boundary-spanning search affecting the sustainable development performance of technology start-ups with boundary-spanning search as the dependent variable, based on the relevant research results of organizational learning theory and value co-creation theory, with ambidextrous learning as the mediating variable and value co-creation as the moderating variable (see Figure 1).

3. Research Design

3.1. Sample Selection and Data Collection

The data were collected by questionnaires, mainly from Chinese technology start-ups, and the data was collected from managers or technical leaders in charge of R&D departments of the companies. The survey enterprises mainly selected new technology enterprises in Beijing, Shanghai, Qingdao, and other cities. The sample enterprises are mainly subordinate to new medicine and biotechnology, new energy and materials, electronics information, precision instruments, and other industries. A total of 450 questionnaires were distributed through on-site interviews and Internet return visits from July to November 2021; 352 were returned, and 289 valid questionnaires were finally obtained by excluding those with obvious errors and those with serious missing data, with a valid return rate of 64.2%. The characteristics of the sample are described in

Table 2. Sample description.

	N=	Percentage
Firm age (years)
≤3	135	46.7%
3–8	154	53.7%
Number of employees
≤50	51	17.6%
51–100	65	22.5%
101–150	70	24.2%
151–200	103	35.6%
Industries
New Medicine and Biotechnology	79	27.3%
New Energy and Materials	56	19.4%
Electronics Information	43	14.9%
Precision Instruments	111	38.4%

Note. N = 289.

and

Table 3. Respondents’ statistics.

	N=	Percentage		N=	Percentage
Gender			Educational background
male	178	61.6%	undergraduate	19	6.6%
female	111	38.4%	master	207	71.6%
Age			doctor	63	21.8%
<30	20	6.9%	Position
30–40	215	74.4%	Chairman	45	15.6%
41–50	44	15.2%	General manager	79	27.3%
>50	10	3.5%	Senior management	165	57.1%

Note. N = 289.

.

3.2. Questionnaire Design

First, to ensure the reliability and validity of the questionnaire, all the items used in this study were derived from the mature scales of previous studies, and then, the questionnaire items were pretested and evaluated, and the formulation of the items was revised and improved according to the feedback results to ensure that the designed scale met the requirements of academic research. The variable measures were scored on a five-point Likert scale (from 1 = strongly disagree, to 5 = strongly agree). The sources of the measurement scales used in this paper are specified as follows.

(1)

Boundary-spanning search mainly draws on the measurement scale of Laursen [9] and others and abridges the original scale to determine the four questions that are most relevant to the implementation of boundary-spanning search strategies of enterprises, such as companies will constantly try new knowledge.

(2)

Ambidextrous learning is based on Chung’s [51] scale. The exploratory learning consists of three measures, including questions such as “Companies dare to accept new demands beyond existing products/services”. Exploitative learning is based on four items, such as “Investing resources in the application of existing technologies to improve efficiency”.

(3)

Value co-creation is based on the scale developed by Ballantyne [52]. It contains four questions, such as “Customers will discuss our products with friends and family in their lives”.

(4)

Sustainable development performance draws on Bansal’s [53] scale, which consists of six questions. For example, “We make many efforts to protect the environment”.

(5)

Control variables. Drawing on the findings of previous studies, the age of the firm, firm size, and R&D intensity was effectively controlled [54].

3.3. Reliability Validity Test

The statistical software SPSS 26.0 and AMOS 24.0 were used to test the reliability and validity of the final scales. The results showed that the Cronbach’s α coefficient of each variable was higher than 0.7. This fully shows that the scales have good internal consistency, which can be utilized to prepare for subsequent analysis; the validation factor analysis revealed that the five-factor model fitted the data well (χ²/df = 1.77, NFI = 0.926, TLI = 0.960, CFI = 0.966, RMSEA = 0.052), as shown in

Table 4. Measures and validation.

Factors and Items	Loading
Boundary-spanning search (Cronbach’s α = 0.892, CR = 0.923, AVE = 0.892)
1. Companies will constantly try new knowledge	0.791
2. Companies are more willing to enter new technology areas on their initiative	0.953
3. Companies are committed to seeking new knowledge to break through the limitations of their existing knowledge	0.9
4. Enterprises pursue the improvement and perfection of existing technologies	0.814
Exploratory learning (Cronbach’s α = 0.722, CR = 0.817, AVE = 0.774)
1. Companies dare to accept new demands beyond existing products/services	0.95
2. Companies are often trying to develop and develop completely new products/services	0.746
3. From time to time, companies will take advantage of new opportunities in new markets	0.6
Exploitative learning (Cronbach’s α = 0.93, CR = 0.941, AVE = 0.898)
1. Investing resources in the application of existing technologies to improve efficiency	0.821
2. Ability to incrementally improve and resolve existing customer issues	0.93
3. Consolidate development process skills for existing products	0.895
4. Frequently adjust procedures, rules, and policies to improve company operations	0.933
Value co-creation (Cronbach’s α = 0.75, CR = 0.81, AVE = 0.732)
1. Enterprise employees can solve problems for customers	0.9
2. Positive interaction and communication between customers and employees	0.646
3. Customers will discuss our products with friends and family in their lives	0.623
4. Customers will talk about our products with others on other platforms	0.69
Sustainable development performance (Cronbach’s α = 0.845, CR = 0.875, AVE = 0.736)
1. We are very concerned about the work related to environmental protection	0.86
2. We make many efforts to protect the environment.	0.739
3. Enterprises actively undertake environmental projects	0.673
4. Companies frequently review relevant environmental performance	0.714
5. In the same industry, our company can obtain good economic performance	0.721
6. By protecting the environment, our company has gained social recognition	0.693

, the combined reliability (CR) of each variable was higher than 0.8, with good structural and combined validity; AVE values were higher than 0.5. And

Table 5. Table of descriptive statistics and correlation coefficients.

Variables	1	2	3	4	5	6	7	8
Firm age	-
Firm size	0.406 **	-
R&D intensity	0.08	0.227 **	-
Boundary-spanning search	−0.158 **	−0.103	0.06	0.892
Exploratory learning	−0.015	0.002	0.152 **	0.260 **	0.774
Exploitative learning	−0.038	−0.018	0.176 **	0.177 **	0.133 *	0.898
Value co-creation	−0.036	0.064	−0.026	−0.041	0.014	0.011	0.732
Sustainable development performance	0.002	0.056	0.274 **	0.295 **	0.531 **	0.220 **	0.043	0.736
Mean	3.57	2.63	3.28	3.972	4.045	3.825	3.977	3.596
SD	0.719	1.057	0.772	0.53	0.517	0.341	0.653	0.672

Note: ** p < 0.01, * p < 0.05. The square root of AVE on diagonal.

shows that the square root of AVE is higher than the correlation coefficient between the main variables, indicating good discriminant validity.

4. Empirical Analysis

4.1. Common Method Deviation Test

To avoid the possible influence of homology errors, this study used Harman’s one-way method to test the possible bias of the common method on the results. Five factors were finally obtained by the test, with a total explanation of 72.54%, and the first factor explained 27.42% of the variance (less than 40%), indicating the absence of serious homoscedasticity bias.

4.2. Correlation Analysis

The correlation test between variables provides a preliminary test of the relationship between the main variables. The Pearson correlation coefficient of each variable was obtained using the mean value of the question term as the value of the latent variable. Table 5 shows that boundary-spanning search has a significant positive relationship with exploratory learning (r = 0.260, p < 0.01), exploitative learning (r = 0.177, p < 0.01), and sustainable development performance (r = 0.295, p < 0.01), while exploratory learning has a significant positive relationship with sustainable development performance (r = 0.531, p < 0.01), exploitative learning and sustainable development performance (r = 0.220, p < 0.01) also had a significant positive relationship, and the correlation analysis provided a basis for further testing the proposed hypotheses.

4.3. Main Effects Test and Analysis

This study used hierarchical regression to test the hypothesis.as shown in

Table 6. Results of main effects and mediating effects regression analysis.

Variables	Exploratory Learning		Exploitative Learning		Sustainable Development Performance
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8	Model 9	Model 10
Firm age	−0.017	0.018	−0.034	−0.012	−0.02	0.019	−0.012	−0.014	0.011	0.021
Firm size	−0.028	−0.011	−0.047	−0.037	0.002	0.021	0.015	0.01	0.026	0.025
R&D intensity	0.16 **	0.138 *	0.19 **	0.176 **	0.275 ***	0.251 ***	0.195 ***	0.242 ***	0.188 ***	0.227 ***
Boundary-spanning search		0.254 ***		0.161 **		0.285 ***			0.169 **	0.263 ***
Exploratory learning							0.501 ***		0.458 ***
Exploitative learning								0.177 **		0.134 **
R2	0.025	0.087	0.036	0.061	0.075	0.154	0.32	0.106	0.346	0.171
Adjusted R2	0.014	0.074	0.026	0.048	0.066	0.142	0.31	0.093	0.334	0.156
F	2.39 *	6.747 ***	3.513 *	4.594 **	7.758 ***	12.92 ***	33.419 ***	8.38 ***	29.914 ***	11.661 ***

Note: *** p < 0.001, ** p < 0.01, * p < 0.05.

. Firstly, based on the introduction of control variables, the effect of boundary-spanning search on ambidextrous learning and sustainable development performance was tested, and the results showed that all three paths were positively significant (model 2, β = 0.254, p < 0.001; model 4, β = 0.161, p < 0.01; model 6, β = 0.285, p < 0.001); i.e., H1, H3a, and H3b were supported by the data. Then, the effect of ambidextrous learning on corporate sustainable development performance was verified, and the results were also positive and significant (model 7, β = 0.501, p < 0.001; model 8, β = 0.177, p < 0.01); i.e., H2a and H2b were supported by the data. Finally, boundary-spanning search was simultaneously with exploratory learning and exploitative learning, respectively, into the regression model with sustainable development performance as the dependent variable; the results show that ambidextrous learning is still significantly positive for sustainable development performance (model 9, β = 0.458, p < 0.01; model 10, β = 0.134, p < 0.01), while the regression coefficients of boundary-spanning search are reduced to 0.169 and 0.263, respectively, but still remain significant, indicating that exploratory learning and exploitative learning partially mediates between boundary-spanning search and sustainable development performance, and H4a and H4b are supported by the data.

To further test the mediating effect played by exploratory learning and exploitative learning between boundary-spanning search and sustainable development performance of technology start-ups. Using the PROCESS program in SPSS to conduct the bootstrap test, by taking 5000 bootstrap samples, as shown in

Table 7. Bootstrap intermediation effect test.

Paths	EFFECT	BOOTSE	BOOTLLCI	BOOTULCI
Boundary-spanning search-exploratory learning-sustainable development performance	0.148	0.04	0.075	0.23
Boundary-spanning search-exploitative learning-sustainable development performance	0.028	0.012	0.006	0.054

, the confidence intervals of exploratory learning and exploitative learning at the 95% level are (0.075, 0.230) and (0.006, 0.054), respectively, both of which do not include 0. The results of the above study reconfirmed that ambidextrous learning plays a mediating role in the relationship between boundary-spanning search and corporate sustainable development performance, and hypotheses H4a and H4b were again supported.

4.4. Moderating Effect Test

First, to avoid the effect of multicollinearity, exploratory learning, exploitative learning, and value co-creation was centered and interaction terms were calculated, and second, the moderating effect was tested by adding control variables, value co-creation, and interaction terms to the independent variable box in turn, with exploratory learning and exploitative learning as dependent variables. As shown in model 12 in

Table 8. Results of regression analysis of the moderating effect of value co-creation.

Variables	Exploratory Learning		Exploitative Learning
	Model 11	Model 12	Model 13	Model 14
Firm age	−0.017	0.016	−0.034	−0.01
Firm size	−0.028	−0.011	−0.047	−0.039
R&D intensity	0.16 **	0.14	0.19 **	0.177 **
Boundary-spanning search		0.252 ***		0.162 **
Value co-creation		0.032		0.024
Boundary-spanning search × value co-creation		0.13 *		−0.004
R2	0.025	0.104	0.036	0.061
Adjusted R2	0.014	0.085	0.026	0.041
F	2.39 *	5.483 ***	3.513 *	3.073 **

Note: *** p < 0.001, ** p < 0.01, * p < 0.05.

, the correlation coefficient between the interaction term of boundary-spanning search and value co-creation on exploratory learning was 0.13 (p < 0.05). Value co-creation significantly and positively moderated the relationship between boundary-spanning search and exploratory learning, and hypothesis H5a was tested, while the interaction term between boundary-spanning search and value co-creation had an insignificant effect on exploitative learning, and hypothesis H5b was not tested.

Figure 2 shows the moderating effect of value co-creation on exploratory learning at one standard deviation above and one standard deviation below the mean level, respectively. The abscissa LOW indicates boundary-spanning search at a low level, and HIGH indicates boundary-spanning search at a high level. That is, value co-creation effectively increases the positive effect of boundary-spanning search on exploratory learning, indicating that value co-creation positively moderates the relationship between boundary-spanning search and exploratory learning.

4.5. Moderated Mediating Effects Test

The bootstrap program is used in the process plug-in to check the adjusted mediation, that is, assume H6a and H6b; the mediating effects at different levels of value co-creation are shown in

Table 9. Mediating effects of ambidextrous learning under the different intensity of value co-creation.

Intermediate Variables	EFFECT	BOOTSE	95% Confidence Interval
Exploratory learning	0.064	0.051	[−0.027, 0.178]
	0.150	0.040	[0.078, 0.236]
	0.249	0.061	[0.137, 0.380]
Exploitative learning	0.132	0.137	[−0.156, 0.040]
	0.262	0.120	[0.006, 0.053]
	0.412	0.248	[−0.007, 0.0917]

. For the mediating effect of exploratory learning, the indirect effect of boundary-spanning search acting on firm sustainable development performance through exploratory learning was significant when value co-creation was at a high level (r = 0.249, the BOOTSE = 0.061, confidence interval [0.137, 0.380]); when value co-creation is at a low level, the indirect effect does not reach a significant level (r = 0.064, BOOTSE = 0.051, confidence interval [−0.027, 0.178]), therefore, the mediating effect of exploratory learning on the relationship between boundary-spanning search and corporate sustainable development performance is influenced and positively moderated by value co-creation, and hypothesis H6a is tested. Similarly, for the mediating effect of exploitative learning, the confidence intervals at both high and low levels of value co-creation include 0, i.e., the indirect effect of boundary-spanning search on corporate sustainable development performance through exploitative learning is not significant, and thus, the mediating effect of exploitative learning on the relationship between boundary-spanning search and corporate sustainable development performance is not moderated by value co-creation, and hypothesis H6b is not passed.

5. Research Conclusions and Recommendations

5.1. Research Conclusions

In the context of open innovation, enterprises’ boundary-spanning search for heterogeneous knowledge outside the organizational boundary can not only enrich the existing knowledge base but also enhance the ability of enterprises to develop new products and meet the requirements of sustainable development of enterprises. However, integrating the searched external heterogeneous knowledge into its knowledge system still needs to be further internalized and integrated. With reference to the research of Nonaka and Takeuchi [55], exploratory learning is conducive to absorbing new external knowledge, and exploitative learning is conducive to enriching the existing knowledge system. Both learning methods can improve knowledge output, which is conducive to stimulating enterprise innovation behavior, and innovation provides a constant driving force for enterprises. Therefore, from the perspective of organizational learning, this paper brings ambidextrous learning into the analysis framework, and uses the questionnaire sample data from the managers of Chinese start-up technology enterprises, this paper explores the mechanism of boundary-spanning search on the sustainable development performance of technology start-ups in the logic of “behavior-learning-performance” and analyzes the mediating and moderating roles of ambidextrous learning and value co-creation. First, boundary-spanning search has a significant contribution to the sustainable development performance of technology start-ups; second, exploratory learning and exploitative learning play a partially mediating role between boundary-spanning search and the sustainable development performance of technology start-ups; third, value co-creation positively moderates the impact of boundary-spanning search on exploratory learning and strengthens the exploratory learning between boundary-spanning search and the sustainable development performance of technology start-ups. This may be because exploitative learning aims to strengthen the existing knowledge and skills, and this learning activity reflects the extension of the existing paradigm of the company, while the absorption of external novel knowledge is less efficient [10], which means the advantages of multi-level interaction and multi-coupling of value co-creation are not well reflected. Therefore, the moderating effect of value co-creation in the influence of boundary-spanning search and exploitative learning is not reflected, but it does not negatively affect the exploitative learning activities of enterprises as a result.

5.2. Management Insights

The findings of the study have implications and management implications for the implementation of boundary-spanning search strategies and ambidextrous learning activities in the development of companies:

Firstly, in this study, we build an external knowledge monitoring system covering the information on “fashionable, frontier, and outreach”. First, we strengthen the monitoring and search for new opportunities in the market, accurately control new market demands, continuously optimize existing technologies and products, and improve the sustainable development performance of enterprises; second, we actively locate industry-leading enterprises, internationally and domestically renowned experts, and research teams as well as leading users to timely search for and obtain cutting-edge knowledge of industry development, broaden the cognitive boundary of enterprises, and lay the foundation for enterprises to develop new products and open up new markets; third, we lay out the search for knowledge in extended fields, focus on the cross-border integration of multi-disciplinary and multi-level knowledge, and stimulate the breakthrough of innovative ideas and thoughts.

Secondly, we build an organizational learning mechanism that takes both internal and external knowledge into account. On the one hand, through the classification and arrangement of acquired knowledge, we carry out targeted ambidextrous learning activities of exploration and utilization, efficiently absorb and integrate the searched knowledge promptly, stimulate the synergistic effect of internal and external knowledge, to truly transform the searched knowledge resources into market revenue. On the other hand, we actively create a good organizational learning atmosphere and innovation environment, enhance the relevance and efficiency of organizational learning through centralized specialized knowledge training, creative thinking training, etc., carry out efficient creative reconstruction of internal and external knowledge of enterprises, and realize the integration and innovation of internal and external resources.

Finally, we build a value co-creation ecosystem with a solid core and extensive outreach. On the one hand, we establish a solid core layer of value co-creation with downstream customers, end-users, and leading users, and promote the active participation of core layer members in the whole product life cycle by establishing network communities and designing reasonable incentives mechanisms. On the other hand, we can build a value co-creation support layer including suppliers, enterprises in the same industry, research institutions, government departments, and other stakeholders through joining technology alliances and cooperative innovation, to promote effective value co-creation behavior, grasp market opportunities more flexibly and improve the efficiency of external heterogeneous knowledge transformation.