Abstract:    In order to meet the requirements of open innovation, firms in the high-tech industry will actively seek external partners and establish strategic alliances. An alliance is an inter-organizational cooperative relationship governed by a formal contract, which involves the exchange of resources and knowledge, organizational commitment and joint development. Firms′ strategic alliance partners will also form alliance relationships with other kinds of organizations, including government, universities and public research organizations. These different organizations and the alliance relationships between each other form the strategic network. 
    A large number of studies have shown that the strategic alliance network will affect the technological innovation performance of firms. Organizations can obtain certain resources from the embedded network for internal technological innovation activities, and there are mainly three different types of resources: resources, knowledge and information obtained directly by partners through their own alliances. For the focal firm, the strategic network has structural advantages, as well as intangible benefits such as power, reputation and trust that the focal firm can obtain from the whole network. The impact of the network on innovation can be understood as how the composition of the network and the structure of the network affect the focal firm′s technological innovation. The impact of network structure on the technological innovation performance of enterprises is more complicated. There are two different mechanisms for the impact of interorganizational network structure on firm performance: network closure and network openness. However, current empirical research does not have a stable conclusion, which may be because the theoretical boundary is not clear. 
   This paper argues that which mechanism works depends on the innovation tasks that the firm faces and whether the source of knowledge is within the boundaries of ego network. Bases on three firm level databases: Recap, USPTO and Compustat, we construct a longitude data of biotechnology and pharmaceutical industry from 1990 to 2001. We use social network analysis method, fixed effect model and random effect model to test our hypotheses. The findings suggest that increase of density of strategic network and increase of structural holes index has a negative influence on firm′s technological innovation quantity and a positive influence on firm′s technological innovation quality. We also found network structure has a moderating effect on the relation between direct/indirect ties and innovation outcomes. In detail, in a dense and closed network, the increase in direct connections will have a positive impact on the quality of innovation and the quantity of innovation outcomes; the increase in indirect connections will have a negative impact on the quality of innovation and the quantity of innovation outcomes. In a sparse, open network, the positive impact of increased direct connections on innovation quantity and innovation quality will weaken; the negative impact of increased indirect connections on innovation quality will also weaken. 
   Based on the existing analysis, this research further reveals the complex relationship between network structure and innovation. The main points are as follows: First, the nature of the innovation task determines how structural holes affect the technological innovation of an enterprise. The role of structural holes in organizational technological innovation has been controversial, and current studies showed both negative empirical results and positive results. This research provides a way to measure the innovation output of firms using multiple indicators rather than only patent numbers, and found that the inconsistency of structural holes effects on technological innovation depends on the inconsistency of innovation tasks. It allows us to better understand the relationship between structural holes and innovation. Secondly, existing studies have mostly analyzed the impact of network structure on innovation from the perspective of ego networks and direct connections, ignoring the embeddedness of enterprises in the global network. We find that, within the boundaries of ego networks, the effects of the network structure can be analyzed from two aspects: intra-network effects and extra-network effects. The effects of the two are different: closed networks are more conducive to the use of exploitive search, which is mainly reflected in the problem-solving process of the focal firm, thereby improving the benefits that the firm obtains from direct ties; on the other hand, the open network is conducive to the firm′s exploratory search, helping the focal firm to acquire more new knowledge, thereby increasing the benefits that the focal firm can get from indirect ties. Finally, this research design adopts the perspective of dynamic networks, focusing on the impact of changes in network structure characteristics on technological innovation, breaking the pre-assumed assumption of static networks. Based on panel data, the paper is able to examine the impact of changes in network characteristics on technological innovation outcomes of enterprises. Existing studies have mostly analyzed the impact of static network indicators on the number of technological innovations in a particular year rather than the effects of network dynamic. 
    This research also has some implications for the practice of corporate strategic alliance networks. First, corporate managers should not only care about who they are allied with, but also the relationships between their alliance partners, and even the indirect relationship between alliance partners and third parties. These connections construct the network structure. Second, managers should also consider whether the network the firm embedded in is dense, closed, or sparse and open. Third, when analyzing the impact of an embedded network on their own companies, managers must not only focus on the positive effects of the network, but also the negative effects of the network.

Key words: strategic alliance, network structure, innovation quantity, innovation quality, ego network boundary