原名「台灣學術線上」
包含TAO期刊庫 + TAO書籍庫 + 論文 + 史料文獻
首頁 | 關於TAO | 瀏覽 | 進階查詢 | 參考工具 | 會員服務 | 已購專書 | RSS服務 | 電子報 | FAQ  
查詢範圍:
   
查詢模式:
熱門查詢詞:
dvd賴氏肥料誘變
   
   
   
     
   
 
項次 書目
1
題名:以「HJ距離」衡量各種線性因子模型的模型誤設    
Measuring the Degree of Misspecification of Linear Factor Models with HJ Distance
著者:呂仁廣(Ren-Guang Lue)
出版地區:台灣
出版城市:台北市
學科:人文綜合 ; 社科綜合
關鍵字:HJ距離 ; 線性因子模型 ; 模型誤設 ; HJ distance ; Linear Factor Model ; Model Misspecification
刊名:醒吾學報
卷期:42期(2010.6)
頁碼:113-132
語言:繁體中文
摘要: 中文摘要PDF ; 英文摘要PDF

對於因子的組成成份,何種模型才能檢視金融市場的效率問題,一直是財務理論中的關鍵。根據半強式效率市場的定義,金融市場的投資大眾會運用所有已知的資料,來分析目前市場股價狀況,是故在此定義下,由於投資大眾無法獲得超額利潤,也就是我們如果想找出真正可以代表金融市場的線性因子模型,就必須利用到Cochrane(2001)文中所述的方法-一般動差法的正交條件(orthogonal condition)-來進行統計推論。但是一般動差法因為在數值求參數解的估計過程中,權數矩陣會不斷更新,以致於各模型間的配適程度將無法做比較,造成「何者才能代表樣本期間內效率市場下的真實定價模型」莫衷一是。本文遂使用由Hansen,Heaton and Luttmer(1995)和Hansen and Jagannathan(1997)所發展出來的「HJ距離」(HJ distance)檢定統計量,以進行這個統計推論的議題。我們將可以代表資產定價模型真正的隨機折現因子視為未知,並衡量模型的隨機折現因子和真實的隨機折現因子的「HJ距離」有多遠。文中將針對四種可以代表投資人最適投資行為的隨機折現因子來衡量模型誤設,並分別比照Jagannathan and Wang(1996)和Hansen and Jagannathan(1997)的做法,設計出兩種虛無假設,一是HJ距離為零的虛無假設,檢定各模型是否具有模型誤設;另一是HJ距離為對照組的HJ距離估計值,利用巢狀式檢定(nested test),檢定該模型是否較虛無模型(null model)為優。我們使用30種產業投資組合的美國股價季資料,樣本期間為1952年第2季至2002年第3季,實證結果顯示四種線性因子模型均具有模型誤設,模型誤設「程度最小」者為三因子模型。然而巢狀式檢定結果顯示,三因子模型並不顯著優於二因子模型,二因子模型也不顯著優於一因子模型。綜合言之,根據本文的分析結果,若我們從投資人投資最適投資行為的角度來分析金融市場,文獻上的線性因子資產定價模型都仍然有大小不一的模型誤設程度,是故它們都仍有被改進的空間。
The composition of the factor, what model to examine the efficiency of financial markets has been the key to financial theory for a long time. According to semi-strong form efficient market definition, every investor in financial markets will use all available information he has to predict stock prices. Therefore in this definition, because investors can not get excess profits, that is, if we want to find out the real linear factor model that can represent the financial market, we must use the general method of moments described in Cochrane (2001) for statistical inferences. However, because in GMM numerical optimizing process, the weighting matrix will be continuously updated, models can not be compared with each other. This fact results in no consensus about what model can represent the true one in efficient market during the sample period. Thus we use the HJ distance statistic developed by Hansen, Heaton and Luttmer (1995) and Hansen and Jagannathan (1997) to do this research. We take the true stochastic discount factor of asset pricing model as unknown, and measure how far the distance is between the model's stochastic discount factor and the true one. This paper use four popular stochastic discount factors and compare their misspecification. Using the approaches of Heaton and Luttmer (1995) and Hansen and Jagannathan (1997) respectively, we design two kinds of null hypotheses. The first null hypothesis is that HJ distance is zero, and we use it to test if there is model misspecification. The second is that HJ distance is that of compared model, and we use it to test if the nested model is better than the null model. We use 30 industry-portfolios in the sample period from 1952Q2 to 2002Q3, and find that there is misspecification in four models all. The minimum misspecification among four models is three-factor one. However, the results of nested tests show that three-factor model is not significantly better than the two-factor model, so is two-factor model than one-factor model. To sum up, there do exist misspecification in linear factor models, so we can still improve them.


    

本卷期目次
醒吾學報 42期 (2010.6)
數位媒體設計業就業能力分析研究/ 鄭聰興林峰羽
以「HJ距離」衡量各種線性因子模型的模型誤設/ 呂仁廣
歐洲聯盟睦鄰政策:高等教育中的田普斯計畫/ 王湘月
《嵇中散集》版本析論/ 黃怡婷
出售長期性資產損益與外在經濟環境因素和報導誘因間之關聯性研究/ 喬繼英
「台灣接單、委外生產」與勞動需求結構/ 喬中?
智慧資本與經營績效之關聯性-台灣高市價淨值比電子業之實証/ 張雯
梭羅借自然花木鳥獸的「重生」譬喻/ 涂成吉
Thoreau's Political Double Trios: Civil Disobedience and His Utopia/ Chen-Chi Tu
 
   
 
   

與TAO合作 | 隱私與版權聲明 | 聯絡方式 | 下載Adobe Reader
地址:台北市中正區(100)北平東路30-12號3樓
電話:(02)2393-6968 傳真:(02)2393-6877
Email: service@wordpedia.com
Wordpedia Family: 學校、企業版入口 | 遠流影音館
Copyright©2011 Wordpedia Co., Ltd. All Rights Reserved.