臺灣股市羊群效應及領頭羊之績效分析

領頭羊（Bellwether）定義為羊群效應（Herd behavior）中首先做出行為的領導者。文獻中說明羊群效應的投資行為是非理性的跟隨，而跟隨者受限於投資資訊獲得的不對稱性，寧願放棄自己對買進股票或賣出股票的決策；而選擇加入領頭羊出現後所形成的羊群效應中。本研究使用個體結構（micro structure）以時間資料來定義領頭羊，同時分析臺灣股市（2007-2016）年間，所有上市股票在某特定時間點買進或賣出股票最多者為領頭羊。而後循時間序列進行價格結算統計分析所定義之領頭羊的投資報酬率。研究方法中將領頭羊的投資報酬率分為二階段統計：（一）領頭羊出現後造成股票價格上漲、（二）領頭羊出現後造成股票價格下跌。領頭羊的身分類別分為五類：（一）自營商（證券公司自行買賣）、（二）臺灣投資信託基金、（三）外資（QFII）、（四）散戶（一般投資人）、（五）臺灣一般法人等。最後以六項執行條件統計結算領頭羊的投資報酬率。結論也證明個股股票價格上漲階段羊群效應中領頭羊之投資報酬率大於個股股票價格下跌時之投資報酬率。因下跌階段部份股票持有者，持有之股票未做賣出（disposition effect）。本研究所採用的個體結構分析，在金融監管上可追蹤分析領頭羊的行為是否在入市前後有內線交易、操控股票價格…等，可用于健全金融市場，創造公平交易之市場機制。

關鍵字

羊群效應；資訊不對稱；實證檢驗

並列摘要

The bellwether is the leader of a sheep herd who tends to take certain actions first before others do. According to relevant literature,the investment behaviors in the sheep herd are based on blind following. The followers are subject to information asymmetry during investments. Thus,they tend to give up on their decisions to purchase or sell certain stocks,choose to follow the bellwether,and induce the so-called “Herd Behavior.” The most significant part of this study is that this study employs the temporal data to identify the bellwether from the micro-structural perspective and analyzes which stock underwent the highest amount of transactions in Taiwan’s stock market sometime during 2007 and 2016 to identify the bellwether at a certain time. And then,this study sequentially analyzes the Title of Thesis：Bellwether Herding Behavior of Total pages：61 Taiwan Stock Market and The Performance Key word：Herd Behavior,Asymmetric information,empirical test Name of Institute：Executive Master’s Program of Business Administration (EMBA) in Management Sciences, Tamkang University Graduate date：June, 2023 Degree conferred：Master Name of student：Tsang Tsz kin oji Advisor：Dr.Shui-Lien Chen 曾子鍵陳水蓮博士 Dr. Iee-Fung Wu 吳怡芳博士 Abstract The bellwether is the leader of a sheep herd who tends to take certain actions first before others do. According to relevant literature,the investment behaviors in the sheep herd are based on blind following. The followers are subject to information asymmetry during investments. Thus,they tend to give up on their decisions to purchase or sell certain stocks,choose to follow the bellwether,and induce the so-called “Herd Behavior.” The most significant part of this study is that this study employs the temporal data to identify the bellwether from the micro-structural perspective and analyzes which stock underwent the highest amount of transactions in Taiwan’s stock market sometime during 2007 and 2016 to identify the bellwether at a certain time. And then,this study sequentially analyzes the settlement prices and conducts statistical analysis to ascertain the rates of return from the previously identified bellwethers. This study conducts statistical analysis on the rates of return from the previously identified bellwethers in two phases when： (1) stock prices go up after the bellwethers emerge； and when (2) stock prices go down after the bellwethers emerge. The identities of the bellwethers can be classified into 5 categories： (1) dealers (securities companies that conduct all transactions on their own),(2) entities that offer investment trust funds,(3) foreign investment entities (QFII),(4) retail investors (general investors),and (5) general corporations in Taiwan. Finally,this study calculates the rates of return from the bellwethers based on 6 execution conditions. The literature that states the rates of return are generally higher when individual stock prices go up is confirmed,as when individual stock prices are going down,some shareholders may not sell the shares in their hands and that induces the so-called disposition effect. The micro-structural analysis adopted by this study can also be implemented in financial regulations to detect if the bellwethers are committing inside trading,rent-seeking or price manipulation before and after they enter the stock market； thus,the micro-structural analysis can be used to perfect the financial market and create a mechanism for fair trading in the market.