Abstract:    At present, enterprises are in a dynamic environment where technologies, products and services are rapidly iterating, and the artificial intelligence technology is deeply integrated with traditional industries. It is difficult to maintain a competitive advantage by relying on existing knowledge and resources for closed innovation activities. As an important way for enterprises to obtain heterogeneous knowledge and break through the ability trap as well as the innovator′s dilemma beyond the existing knowledge boundary, boundary-spanning search has attracted extensive attention from academia and managers. Since the concept of "boundary-spanning search" was proposed by Rosenkopf et al in 2001, much more efforts have been devoted to relevant research for almost 20 years, and a large amount of literature has been accumulated. 
   In this study, the meta-analysis method was introduced into the quantitative integration of 53 papers, 26,812 samples, and 55 effect sizes in the field of boundary-spanning search and firm innovation performance, and the following three conclusions were drawn. 
    First, the boundary-spanning search behavior can promote the firm innovation performance. Second, country differences, industry characteristics, innovation ability can adjust the contribution of boundary-spanning search to the firm innovation performance. Specifically, compared with other countries, boundary-spanning search has a greater positive effect on the firm innovation performance in China; compared with non-manufacturing enterprises, boundary-spanning search is more effective in improving firm innovation performance for manufacturing enterprises; compared with high-innovation enterprises, boundary-spanning search brings a more significant improvement to the low-innovation enterprises. Third, the type of boundary-spanning search across knowledge boundaries determines the degree of improving the innovation performance. Among the various types, for different search scope, expanding the search width has a greater positive effect on innovation performance than strengthening the search depth; for different geographical boundaries, local search takes greater effect for innovation performance promotion than non-local search; in terms of knowledge types, the impact of market knowledge is more effective than technical knowledge on innovation performance in boundary-spanning search. 
    Theoretically, this study quantitatively integrated multiple independent studies in the field of boundary-spanning search, and concluded the positive impact of boundary-spanning search on firm innovation performance. Furthermore, differences in the impact of boundary-spanning search on innovation performance by search types and different enterprise characteristics such as industry and innovation ability are also discussed. In particular, the classification and integration of the boundary-spanning search research in the context of China, while providing theoretical reference and practical basis for boundary-spanning activities in China, has also developed a study on the mechanism of environmental factors in boundary-spanning search. Moreover, this study also found that the boundary-spanning search has a more significant positive effect on the firm innovation performance under higher economic activity and higher innovation demand. 
    In practice, this study provides empirical evidence and decision-making basis for enterprises to strategically arrange the focus and methods of boundary-spanning search in order to obtain sustained external heterogeneous resources, and to break through the constraints of capability potential and resource thresholds. In addition, companies should fully consider the heterogeneous impact of boundary-spanning search on firm innovation performance, and rationally allocate limited searches resources and attention based on their own industry characteristics, innovation capabilities, resource needs, etc. as well as the external conditions such as the development speed of the region. Finally, to realize the upgrade to "intelligent manufacturing" in China, enterprises need to cultivate an open culture, take the initiative to continuously learn external technology, products, and market knowledge through boundary-spanning search, and effectively integrate internal and external knowledge to promote innovation capabilities.

Key words: boundary-spanning search, firm innovation performance, meta-analysis, moderating effect