資訊品質如何影台灣股市之群聚交易

機構典藏 > College of Business and Management > Graduate Institute & Department of Banking and Finance > Research Paper > Item 987654321/76226

Please use this identifier to cite or link to this item: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/76226

Title:	資訊品質如何影台灣股市之群聚交易
Other Titles:	The Impact of Information Quality on Herding Behavior of Taiwan Stock Market
Authors:	林蒼祥;蔡蒔銓
Contributors:	淡江大學財務金融學系
Keywords:	雜訊;交易成本;群聚行為;限價委託簿;Noise;transaction cost;herding;search model;order book
Date:	2010
Issue Date:	2012-05-03 20:45:40 (UTC+8)
Abstract:	「資訊品質如何影台灣股市之群聚交易」－中文摘要台灣股票市場中，個人投資人占大多數，文獻上認為此類投資人缺乏資訊，投資報酬偏低。在股票市場波動劇烈時，蕭朝興、王子楣(2008)發現，個人投資人之委託有群聚現象，可能包含資訊內涵。然而根據Easley, Hvidkjaer, and O’Hara(2002)，資訊品質市場引發之逆選擇，會提高股票之風險溢酬。本計畫針對各類股票之資訊品質，探討其是否影響各類投資人委託單之群聚，並藉其結果進一步研究，逆選擇之現象是否因群聚加重或減輕。如果逆選擇因群聚而減輕，則市場資訊揭露越有效率，越對占市場多數之個人投資人有利。反之如果逆選擇因群聚而加重，則應在市場波動大時降低交易成本，以提昇資訊之效率。研究之結果，有助於擴大資本市場規模，因此對於主管機關或證券交易所未來對交易制度之規劃，將具重大之參考意義。 Madhavan(1992)指出，價格發現將投資人潛在需求轉換成實質交易，這就是交易機制的功能。O’Hara(2001)認為，日內交易之複雜性，使價格發與交易機制均需要改變。許多探討資訊品質之文獻，著重於日或跨週以上之影響，因此為探討資訊交易在極短期間之變化。本計畫不但檢視日內交易中資訊品質之改變，並企圖了解其對各類型交易者之日內群聚交易之影響。本計畫進一步研究，因資訊品質不佳所衍生之逆選擇，是否顯現於各類型交易者之交易委託上。而群聚現象之強弱，是否影響逆選擇之結果，更是對交易機制未來發展之重要參考。本文將建構資訊交易比率實證模型，計算出日內各區間資訊及非資訊交易比率，來瞭解交易頻率變化對市場交易者日內交易的影響，並在控制相關日內交易變數下，進行交易頻率對市場績效影響之分析。先前研究僅觀察交易頻率改變造成市場日內績效影響，鮮有探討交易頻率如何影響日內資訊交易者行為的。由於市場績效是資訊交易造成之結果，因此欲使市場流動性變大、波動變小及價格發現速度更快，首先應瞭解交易行為，而非市場績效。瞭解資訊及非資訊交易者之日內群聚行為變化，對設計出使資訊及非資訊交易進行之日內交易，對達到更佳之交易機制將有莫大之幫助。因此，在本文中我們將以資訊交易實證模型計算獲得之資訊交易機率，觀察交易頻率改變後之日內群聚行為之差異，瞭解市場績效的真正成因。 Easley, et al. (2002) 提出之訊息交易比例(PIN, the probability of information-based trading)，採用的是交易與報價資料，根據Lee and Ready (1991)之模型，推算買單與賣單，並以之估計是否有投資者有掌握內部訊息。Chakravarty, Kalev and Pham(2005)發現，在市場多頭時期，私有資訊交易者以較大規模委託單進行交易；在市場空頭時期，則以小型委託單來進行交易。因此本計畫希望以委託單為基礎，建立訊息交易比例模型，以免除推算買賣單之缺點。過去文獻對於資訊不對稱的探討，常以報價驅動市場為主，本計畫採用與成交資料配對後之委託資料，而非僅就成交資料建立模型，也反映我國股市採用委託單驅動之交易現實。研究中並對資訊品質之代理變數如市值、週轉律及市價淨值比等，進行交叉分析。對於日內之各極短區間，如開盤後或收盤前，亦探討資訊品質與群聚之交互作用。各股報酬率之高低，以及股市數於空頭抑或多頭，也一併考慮於分析之中。本研究將採用台灣證券交易所之每日統計資料，以及日內「委託檔」、「成交檔」資料庫，衡量資訊品質及群聚現象。本研究之樣本期間為民國 94 年1 月1 日至民國95 年12 月31 日止，共2 年。本計畫參考Lakonishok, Shleifer and Vishny（1992）之群聚指標，以Patterson and Sharma(2006)之日內群聚指標為基礎，與Easley, et al. (2002)採用之PIN，探討其間之關係及其意涵。本計畫之貢獻有二，首先是檢討日內交易之資訊品質與群聚之關係，可釐清資訊品質之測度效率；第二是了解群聚如何影響逆選擇，可促進交易機制之有效改進。 The Impact of Information Quality on Herding Behavior inTaiwan Stock Market－ Project Abstract Individual investors account for a majority of stock market trading over years. Literatures often consider them as lacking information and therefore unable to realize satisfactory returns. Hsiao and Wang (2008) found that orders of individual investors in Taiwan tend to herd during periods of high volatility. According to Easley, Hvidkjaer and O’Hara (2002), quality of information induces adverse selection problem and hence raises risk premium of stock returns. This project aims therefore at investigating information quality of stock trading in Taiwan, with a specific purpose of identifying how information quality affects herding of investor orders. Furthermore, this project attempts to find out if the adverse selection phenomenon is alleviated or acerbated by order herding. If herding causes investors’ selection to be less adverse, then it implies information dissemination in this market is efficient and benefits the individual investors as the majority in it. However, if adverse selection situation worsens as order herds, then exchange authority should consider lower trading or related costs to raise efficiency of information dissemination. The results of this research project will help broadening local capital market, and could therefore serve as a valuable reference for the regulating bodies or exchange authority in conceiving future plans. This project plans to construct empirical model for proportion of intra-day information-based trading for sub-intervals within a day with the objective of extracting the influence of trading frequency on investor orders and selection adversity. Having controlled relevant factors governing trading within intra-day intervals we would expect to identify how investment performance is linked to information quality in this framework. Previous studies have covered the effects of trading on performance, but not so much on intra-day information-based herding behavior. As market performance is also a consequence of information-based trading, to achieve a trading mechanism with better market liquidity, lower volatility and faster price discover, we must lean more on information quality rather than investment performance. The understanding of intra-day herding behavior of information- and non-information-based investors would contribute to a more efficient trading mechanism. So our empirical information-based trading model would discern herding intensity of various investors with the change of trading frequency, and find out the real cause of market performance under it. The project will employ daily as well as intra-day order book and transaction file database of the Taiwan Stock Exchange to measure information quality and order herding. Data period starts from January 1, 2005 to December 31, 2006. Our herding intensity comes from the intra-day runs-based measure from Patterson and Sharma(2006), which compares with Lakonishok, Shleifer and Vishny（1992）which have been used in longer periods primarily. The herding measure will be analyzed against the PIN measure of Easley, et al. (2002) to produce subsequent findings。We believe there are two contributions that could be brought about by this project. First, the examination of intra-day relation between information and order herding helps clarifying greatly the measurement efficiency of information quality with very short but crucial periods. Second, the understanding of how order herding influences selection adversity furthered by this study lends itself to the effective improvement of trading mechanism in practice.
Relation:	資訊品質如何影台灣股市之群聚交易
Appears in Collections:	[Graduate Institute & Department of Banking and Finance] Research Paper [Graduate Institute & Department of International Business] Research Paper

Files in This Item:

File	Size	Format
index.html	0Kb	HTML	85	View/Open

Loading...