集團企業股權網絡結構與研發決策的跨層次分析

How the Structure of Equity Ties within Business Groups Affects R&D Expenditure?Applying Hierarchical Linear Model

本研究引用創新決策及社會網絡理論，以股權連帶為研究主軸，探討各集團企業之子公司透過交叉持股互相持有股權，如何影響其研發決策，並探討最有利創新的網絡結構究竟以何種形式存在？本研究以2002到2005年台灣209個集團，963個分子公司為樣本，研究結論發現，在子公司層次，若處於較中心的位置，會加速創新的機會；在集團整體層次，網絡密度愈高，使後續的創新決策將受到限制；在高度集中化網絡，對創新決策也有負向效應。此外，就個體與整體層次的組合而言，研究發現，最有利創新的網絡結構有二：「高中心性的子公司」搭配「低密度的集團企業整體連結」，以及「高中心性的子公司」與「低集中度的集團企業整體連結」的組合。在研究方法上，本研究採用階層線性模式（Hierarchical Linear Models, HLM），針對各個階層分別建立迴歸模式，再將組內與組間的變異因素分別考慮，因此可以增進參數估計的精確度。

This study investigates how equity ties between firms within business groups facilitate the innovation opportunities and/or create constraints. To develop our hypothesis, this research combines innovation studies and network theory. Using samples from 963 firms within 209 business groups in Taiwan between 2002 and 2005, we find that the greater the individual centrality of an affiliate within its group, the greater the innovative activity of the affiliate. In aggregate, a greater overall density and centralization of investment ties of business groups decreases individual affiliates' innovativeness. Finally, the most innovative affiliates tend to be central businesses in the loosely-connected or less-centralized groups. The results have implications for the more general issue of multi-business firm innovation. Since nested data structures exist in a business group, this paper employs Hierarchical Linear Models (HLM) to manage nested data and build separate regression models for each hierarchy. Investigating variations from both within and between hierarchies, this approach improves the accuracy of parameter estimation.