Abstract:    The prior researches show that the density of knowledge network has positive effects on product innovation. Does that mean the higher density of intra-group knowledge network within the group, the higher capability of product innovation the group has? After reviewing the extant literatures, we find that the answer to the above question is not clear. As a matter of fact, the related literatures in this field are still comparatively rare although business groups offer the perfect scenarios for whole network research. Consequently, we know little about the effect of intra-group knowledge network structure on the capability of product innovation from the perspective of whole network up to date. In order to make up the aforementioned gap in the extant literatures, this research has been designed and done.
   With the aim to explore the contingent effects of the density of intra-group knowledge network on the group′s capability of product innovation, this paper carefully decomposes the focal constructs, identifies the key contextual variable, develops corresponding hypotheses and tests them so as to reveal the underlying mechanism. Firstly, we decompose the independent variable, the density of intra-group knowledge network, into two dimensions, the density of intra-group technology knowledge network and the density of intra-group market knowledge network. We also decompose the dependent variable, the capability of product innovation, into two dimensions, the capability of incremental product innovation and the capability of radical product innovation. We demonstrate the possible effects of technology network density and market knowledge density on the capability of incremental product innovation separately, and the effects of technology network density and market knowledge network density on the capability of radical product innovation as well. Four hypotheses of main effects are developed subsequently (H1a, H1b, H2a and H2b). Secondly, based on the extant literatures and the practices of business groups, we identify a key contextual variable of geographical dispersal that may impact on the aforementioned effects, and view it as a moderator in this research. The possible moderating roles of geographical dispersal are discussed in detail, and four corresponding hypotheses of moderating effects are developed as the results (H3a, H3b, H4a and H4b).
    In the empirical analysis, we have done the following three aspects. First, for the measurement, the density of technology knowledge network and the density of market knowledge network are calculated by the formula of density of whole network, and all the other variables are measured by the items developed and used in the extant literatures. The testing shows that the measurements of variables are reliable, they have high reliability and validity. Second, in the implementation of data collection, many precautions are taken to prevent and reduce the undesirable negative effects of common method biases as possible as we can. Third, after that, we analyze the correlations of the focal variables and use hierarchical regression model to test those hypotheses with the unique data of 173 samples.
    After doing all these, two major findings emerge as follows. (1) Both the density of intra-group technology knowledge network and the density of intra-group market knowledge network formed by group affiliated firms can contribute significantly to the group′s capability of incremental product innovation, and geographical dispersal positively moderates the aforementioned relations. The results show that H1a, H1b, H3a and H3b are empirically supported. That means a group with high density of intra-group technology knowledge network and high density of intra-group market knowledge network usually has stronger capability of incremental product innovation. Especially, for those groups whose affiliated firms are geographically dispersed, they can benefit more from the high density of intra-group technology knowledge network and the high density of intra-group market knowledge network in terms of groups′ incremental product innovation. (2) There is an inverted U shape of relationship between the density of intra-group technology knowledge network and the group′s capability of radical product innovation, and so is the relationship between the density of intra-group market knowledge network and the group′s capability of radical product innovation. However, geographical dispersal only plays positive moderating role in the latter. The results show that H1b, H2b and H4b are empirically supported, while H4a has not passed the test. So, this finding reveals that it′s not always the higher of density the better of radical product innovation. Only when the density of whole knowledge network is below the threshold, can the increase of density and geographical dispersal make the groups more capable of radical product innovation.
    This research has three theoretical implications. The first, it expands the researches on knowledge network and innovation from the perspective of whole network in business groups by exploring relationship between the density of intra-group knowledge network and the group′s capability of product innovation. The second, it helps us to understand the controversial results of the prior researches on the relationship between the density of knowledge network and innovation. It reveals one of the reasons that probably results in the controversial opinion is due to the distinct attributes of different dimensions of knowledge and innovation. If we discuss them at the level of dimensions separately, the divergence maybe will not be so significant any more. The third, it offers an alternative explanation of the geographical dispersal of a group′s R&D network. The effect of geographical dispersal on product innovation can′t be discussed in an isolation way. It should be discussed according to situation of knowledge network structures.
    Lastly, there are two limitations and the related future research. First, the moderating role of geographical dispersal in between the density of intra-group technology network and the capability of radical product innovation is not supported. Although this paper makes the explanation about the reason why it′s not supported, it′s better to make further testing if there is sound condition in the future. Second, this paper focuses relationship of the density of intra-group knowledge network and the capability of product innovation. However, network density is only one the characteristics of network structure. There are still many effects of other network structure characteristics (e.g. structural hole, centrality, and clan) on groups′ capability of product innovation left for future researches.

Key words:  business group, knowledge network density, geographical dispersal, product innovation capability, whole network