本文探討台灣家族及非家族企業上市後股權分散之路徑及動機是否相同。從初步統計可發現兩類企業內部人持股在初步公開上市後逐漸且穩定地分散。 使用Pobit模型發現，家族之永續傳承考量下，家族成員認定家族企業為家族系統之衍生，因此內部人持股降低因素並非由代理因子所造成。而歸結多種方式分析，可發現市場因子，即企業股票之流動性為主要且一致造成股權分散之因素。文章後段使用不同的家族企業認定方法，我們仍然得到相同地結論。 文章最後階段反向思考家族控制進入及離開企業之決定因素，可發現代理因子在非家族企業轉變為家族企業中扮演主要角色，而另一方面，市場因子則在家族企業轉變為非家族企業扮演主要角色。

本論文希望透過股權分散的變化，來預測下週或是下月的股價報酬率。因為，假設流通在外股數不變下，當股份從眾多持有少量股票的散戶，集中至持有大量股票的大股東或法人時，集保總人數會下降。而我們認為可以透過觀察集保總人數的變動率，來推測法人及大戶整題而言對該股票的預期方向，因為交易往往伴隨著成本，假設投資人為理性下，法人及大戶會買進或賣出股票往往是因資訊的流入，使得他們認為需要調整持股部位。 透過迴歸模型中的普通最小平方法，當週集保人數減少1%，下週股價報酬率平均而言顯著上升0.055%，年化後的報酬率為2.86%。而當月集保人數減少1%，下個月股價報酬率平均而言顯著上升0.074%，年化後的報酬率為0.89%。由上述結果，可以推斷集保總人數的變化，在統計上會顯著影響下週或下月的股價報酬率。另外，欲瞭解此指標是否適用於所有產業，發現使用上需避開造紙工業及資訊服務業，因該產業之迴歸分析結果為顯著正相關，無法與原先的假設推斷一致。 / Several researches discuss the predictability of the movement of institutional investors and major shareholders because they probably have better ability to analyze the stock or to gather relevant information. However, it is difficult to observe what they are doing especially major shareholder that don’t have obligation to disclose their holding rate. As a result, we wonder if the change in shareholders’ number implies the view of major players in stock markets. The empirical result demonstrates that decrease in shareholders’ number accompanied by positive stock return next period in a significant level which is in accordance with the economic intuition that we suppose the total number of shareholders could reflect the transition of stock shares from many individual investors to some major shareholders. Furthermore, there are systematic cross-industrial differences in the stock return reactions to the prior change of shareholders’ number. In spite of that, all of the industries with negative relationship reach significant level but the industries with positive relationship don’t.

本研究透過不同股權結構來探討傳統代理問題與核心代理問題對於研發支出投入程度之影響。 本研究將具有金字塔結構或交叉持股結構之企業分類為股權集中之企業，而將不具有金字塔結構或交叉持股結構之企業分類為股權分散之企業。實證分析顯示，在股權分散情況下，傳統代理問題的降低有助於研發支出之提升。其中，董事長兼任總經理、經理人持股比率、董監事持股比率、大股東持股比率、法人持股比率及獨立董事席次比率皆對研發支出之投入有正相關影響。在股權集中情況下，核心代理問題的降低也有助於研發支出之投入。 此外，相較於股權分散之企業，股權集中之企業有著較高的研發支出投入，代表股權集中之企業較有辦法投入較多資源於研發支出上。而透過交乘項之實證分析後，發現相對於股權集中之企業，若股權分散之企業的傳統代理問題能降低，那麼會更有效地提高研發支出以提升企業經營績效。同理也得知，在相對於股權分散之企業，股權集中之企業之核心代理問題若能降低，那麼會更有效地提高研發支出以提升公司價值。 / Based on ownership structures, this study investigates the effects of type I and type II agency problem on R&D expenses for innovation investments. This study classifies both pyramid ownership structure and cross-holding ownership structure as ownership-concentrated structure while others ownership-dispersed structure. The empirical results indicate that under ownership-dispersed structure, the reduction in type I agency problems increases research and development (R&D) expenses in innovation investments. The ways that reduce type I agency problem and boost innovation investments include CEO duality, executive ownership, corporate ownership, block ownership, institutional ownership, and the percentage of board seats held by independent directors. Under the case of ownership-concentrated structure, reduction in type II agency problems lead to increase in R&D expenses for innovation investments. Moreover, comparing with ownership dispersed firms, the ownership concentrated firms spend more in R&D expenses. However, if the dispersed ownership firms can decrease type I agency problems then they spend more in R&D relative to the ownership concentrated firms. By the same token, comparing with ownership-dispersed firms, the reduction in type II agency problems of ownership-concentrated firms will more effectively increase R&D expenses to raise the company value.