Relational Characteristics Of Alliance Portfolio And Its Impact On Firms' Dualistic Innovation Performance

With the accelerated implementation of the ‘Belt and Road Initiative' and the integration between big data technologies and the manufacturing industry,the Chinese government has proposed the strategic goal of promoting industrial upgrading.In the ever-changing market environment,not only should Chinese enterprises make full use of the existing technologies to ensure market position,but they should also constantly explore new technologies to seize the opportunities for breakthroughs.The enterprises establish cooperation vertically or horizontally,with different types of partners for technology innovation.Therefore,they can obtain relational rents through alliance portfolio and improve the dualistic innovation performance through building technological innovation alliance led by core enterprises.Through literature review,this study built the theoretical framework on how the relational characteristics of alliance portfolio affect firms' dualistic innovation performance.Social network analysis and case study were conducted to explore the characteristics and accumulation of social capital of innovation alliances.Besides,we elaborate the specific connotations of characteristics of an alliance portfolio at the binary level and individual level.Furthermore,based on patent data,enterprises attributes data,external environmental data of Chinses automobile industry,a series of studies on how the relational characteristics of alliance portfolio affect dualistic innovation performance were conducted as follows:First,this study divided the relational characteristics of alliance portfolio into four dimensions,namely the tie breadth and tie strength in the binary level and the horizontal ties and vertical ties in the individual level.It then followed by building a theoretical model on how the relational characteristics of alliance portfolio can affect dualistic innovation performance in the form of direct effects,mediating effects,and moderating effects.Second,a theoretical analysis and an empirical test were conducted based on forecasting of the dualistic innovation by Machine Learning(ML)methods.We compared the prediction performance of the IML(i.e.,Decision Tree,Support Vector Machine,Multi-layer perception)method,the EML(i.e.,Bagging,Random Forest,Random Subspace and Multi Boosting)methods and the HEML(i.e.,RS-Multi Boosting)method.Besides,seven evaluation criteria were selected to evaluate the prediction accuracy of the eight ML methods,including the average accuracy,Type I error,Type II error,recall ratio,precision ratio,‘F-measure' rate and ROC curve.The empirical results reveal that the hybrid ensemble machine learning(HEML)method has better prediction results than those of the ensemble machine learning(EML)and the individual machine learning(IML)methods.More specifically,RS-Multi Boosting is the best method of the eight approaches to forecast firms' dualistic innovation performance.Third,the study introduced the methods of empirical research and technical alliance construction.We built 14 innovation alliance networks based on patent data of the Chinese automobile industry during 2001-2014,and described the accumulation and characteristics of social capital.Based on the patent data,standardization data and firms' ownership data,the econometric model was used to empirically test how the relational characteristics of alliance portfolio affect dualistic innovation performance.Forth,the study introduced the standardization capability of the core enterprise into the framework on how the relational characteristics of alliance portfolio affect firms' dualistic innovation performance,which reveals the “black box” of this relationship.We empirically test the mediating effects of the standardization capability of core enterprises,and the results show that the standardization capability of core enterprises indeed has partial mediating effects between the tie breadth of the alliance portfolio and firms' dualistic innovation performance.Fifth,the study introduced the technological diversification of core enterprises as the moderating variable,analyzed the effects of contingency factors in depth.The empirical results show that the technological diversification of core enterprises weakens the influence of vertical tie strength on promoting exploitative innovation.Besides,the technological diversification of core enterprise strengthens the influence of horizontal tie strength on promoting exploratory innovation.However,technological diversification has no significant moderating effects on the relationship between vertical tie strength and the exploratory innovation as well as the horizontal tie strength and the exploitative innovation.The innovation of this paper is mainly reflected in the following three aspects: 1)From a different perspective,this paper discusses how the relational characteristics of alliance portfolio affect the dual innovation performance of enterprises,further divides the relational characteristics of alliance portfolio from binary level and individual level,and constructs the overall analysis framework of the relational characteristics of alliance portfolio,which is a further development of collaborative innovation theory and social capital theory.2)This paper has made many innovations in data collection and empirical research methods,including using patent data to measure the social capital of the alliance subject and the dualistic innovation performance of enterprises.At the same time,AI theory and machine learning method are introduced into the evaluation process of technological innovation performance for the first time,which can be used to evaluate binary innovation performance of enterprises based on historical data in practice.3)This paper has discussed the direct influence of the relational characteristics of alliance portfolio on the dualistic innovation performance of enterprises,the mediating effect of core enterprise's standardization capability and the moderating effect of core enterprise's technology diversification degree.It not only opens the black box of the role of the relational characteristics of alliance portfolio on dualistic innovation performance,but also further considers the possible influence of contingency factors on the basis of existing research.The research conclusions of this paper provide practical guidance and management suggestions for enterprises to choose technology innovation alliance on the basis of differentiated technology,how to effectively manage alliance partners and alliance relations,and how to improve enterprise technology innovation performance.