Abstract: Scholars are increasingly paying more attention to the problem that how enterprises' management innovation is affected by the relationship between enterprises and geographical agglomeration degree. By discussing the impact of network reciprocity levelon management innovation and analysis of intermediate role of learning capability and the controllable role of agglomeration, R&D investment and enterprise size, we propose a theoretical model among networkreciprocity level, learning capability and management innovation, and use structural equation model to test the research hypothesis, which are based on 458 samples of high-tech enterprises and private technological enterprises in Guangdong, China. Theempirical findings show that learning capability plays a complete intermediate role on the relationship between network reciprocity level and management innovation, and exhibits a more positive intermediate role in the enterprise network of low agglomeration;meanwhile, the relationship among networkreciprocity level, learning capability and enterprises' management innovation is also affected by the size of the enterprise scale and the level of R&D investment. The research extends and deepens the understanding and recognition that whether and how external organizations affect the enterprises' management innovation in industrial clusters and network.

Key words: Network reciprocity, learning capability, management innovation, agglomeration degree