Abstract:  Cross-border M&A is an important way for enterprises to break through the path dependence and resource bottleneck for opening innovation. In the first half of 2018, the amount of Cross-border M&A transactions in China dropped by 25% compared with the second half of 2017, while cross-border M&A in high-tech industry is still in the leading position. Cross-border M&A by Chinese enterprise is experiencing a period of improving quality and increasing efficiency. However, it is difficult to explain and guide the key issues of synergistic development of integration and innovation quality improvement of Chinese cross-border overseas M&As by using the theory of ownership advantage based on multinational enterprises from developed countries and the existing research on the performance of Chinese cross-border M&A. Based on knowledge integration theory and network embedding theory, we have constructed the research framework of post-merger integration, network embeddedness balance and innovation quality of enterprises in the context of Chinese cross-border M&As. High level of knowledge integration ensures the frequency of knowledge exchange and promotes the effective transfer of tacit knowledge and complementary knowledge, while the differences among experts, technological capabilities and incentive systems and governance structures increase the coordination cost of knowledge integration. Thus, we put forward the first hypothesis H1: Post-merger knowledge integration has an inverted U-shaped relationship with the innovation quality of acquiring firms. Network embedding balance is the coordination of local and overseas network knowledge, while minimizing the cost of coordination and the difficulty of knowledge transfer. Focusing only on local or overseas knowledge may lead to incomplete knowledge loci and imperfect understanding of cultures and institutions of host countries. Therefore, we propose H2a to assume that the internal and external network embedding balance can optimize the inverted U-shaped curve relationship between post-merger knowledge integration and innovation quality. It’s confrontational relationship between enterprises pursuing foreign networks or local networks due to limited resources. Geographical proximity, low cost of local cooperation and industrial agglomeration make traditional enterprises to depend more on local network. The specific advantages of firms established in domestic networks based on human capital or raw materials cannot be transferred and replicated to overseas markets, which will inevitably reduce the ability of knowledge integration and continuous innovation after cross-border M&A. Therefore, we propose H2b: when domestic network dependence is higher than foreign network dependence, internal and external network embedding balance plays a stronger role in optimizing the inverted U-shaped relationship between post-merger knowledge integration and innovation quality. The concept of industrial clock speed is to describe the dynamic change rate of industry determined by endogenous factors such as products, processes and organizational structure. When the industrial clock speed is fast, it is difficult for enterprises to maintain the competitive advantage based on an existing innovation. The direction of industrial innovation and the rapid dynamic change of new product make it more difficult for enterprises to build new technological advantages only by independent R&D. The combination of local and international partners can effectively improve the speed of knowledge flow and the heterogeneity of knowledge scope. Therefore, we propose H2c: when cross-border M&As occur in industries with fast clock speed, the balance of internal and external network embedding plays a stronger role in optimizing the inverted U-shaped relationship between post-merger knowledge integration and innovation quality.This paper uses empirical study to test hypotheses. Our sample consists of technology-sourcing overseas M&As with Chinese listed companies as acquiring firm in the manufacturing industry from 2001 to 2016 obtained from the BVD Zephyr database. Event study and multiple regression models are used to test the nonlinear relationship between knowledge integration and innovation quality in the process of overseas M&A innovation, and compares different integration modes corresponding to improving domestic innovation quality and international innovation quality. The results show that the relationship between the knowledge integration and domestic innovation quality in a cross-border M&A presents an inverted U-shaped curve. Knowledge integration and the quality of international innovation are negatively correlated. We further examine the moderate effect of network embeddedness balance and industrial clock speed and found that network balance can optimize the inverse U-shaped relationship between knowledge integration and innovation quality; the optimization effect of network balance is stronger when the domestic network efficiency is higher than foreign network efficiency. Moreover, the optimization effect of network balance is also stronger when the clock speed of the M&A industry is quicker. In this paper, GLS regression method is used to correct the heteroscedasticity and sequence correlation problems of random items in OLS model. Except for the differences in the significance of individual variables, the regression results of the main explanatory variables are consistent. It is proved that the results in this paper are robust. Therefore, the conclusions of this paper are as follows: First, the choice of knowledge integration degree in overseas M&A needs to balance the two goals of improving domestic innovation quality and international innovation quality. Moderate integration can best improve the quality of domestic innovation, while low integration can best improve the quality of international innovation. Second, despite the inverted U-shaped relationship between knowledge integration and domestic innovation quality, network embedding balance can optimize this relationship, especially when domestic network dependence is higher than foreign network dependence; third, in industries with fast industrial clock, the sensitivity of overseas knowledge integration is lower, and the balance of internal and external network embedding can enhance post-merger innovation. Our study contributes to the related literature. Breaking through the overseas M&A theory which only considers the binary relationship between isolated M&A parties, this paper starts from the global network embedding theory, focusing on how balance between domestic and foreign network embedded by acquiring firm enhance the global innovation advantage after overseas M&A. This paper, for the first time, provides a theoretical basis for Chinese enterprises to concern the improvement of innovation quality instead of "patent bubble" and "innovation illusion" in the process of knowledge integration of cross-border M&As based on the perspective of embeddedness balance between domestic network and foreign network, and help Chinese enterprises to go into the world innovation center by enhancing the quality of foreign investment.

Key words: cross-border knowledge integration, innovation quality, local and foreign network balance, world innovation center, industry clockspeed