 |
English
|
正體中文
|
简体中文
|
全文筆數/總筆數 : 18278/19583 (93%)
造訪人次 : 914755
線上人數 : 777
|
|
|
資料載入中.....
|
請使用永久網址來引用或連結此文件:
http://nhuir.nhu.edu.tw/handle/987654321/22159
|
| 題名: | 以向量自我迴歸模式探討台灣股價及國際油價之關聯性 |
| 其他題名: | Graduate Institute of Management Sciences, NanHua University |
| 作者: | 楊長霏
Yang, Chang-Fei |
| 貢獻者: | 管理科學研究所
丁誌魰
Chih-wen Ting |
| 關鍵詞: | Chow檢定;VAR向量自我迴歸模式;ECM誤差修正模型;Granger因果關係檢定
Granger causality methodologies;Chow test;error-correction model;Vector Autoregression |
| 日期: | 2005 |
| 上傳時間: | 2015-07-31 15:39:19 (UTC+8) |
| 摘要: | 本研究主要是以向量自我迴歸模式來探討台灣加權股價指數、八大類股指數、消費者物價指數與布蘭特原油價格之互動關係。本研究首先使用Chow檢定，決定研究期間為1990年5 月至2004 年12 月，共176筆月資料觀測值，透過單根檢定、共整合分析、衝擊反應函數、ECM誤差修正模型、變異數分解及Granger因果關係檢定，來檢驗研究變數間之關聯性。 茲將研究結果摘要於下: 一、單變量ARIMA方面：在ADF 單根檢定下，各變數之原始數列皆不為一穩定數列但經一階差分 [I(1)]後皆為一穩定數列，後為符合白噪音假設，故利用AIC 法則，決定最適落後期數。 二、多變量VAR方面： 1.兩變量之共整合分析在MAIC分析下發現九個模式中的三變量之共同遞延期數為1期。再經殘差檢定其遞延期數是否合適，若不合適再增加遞延期數，直到三個變量為被解釋變數的迴歸殘差皆符合白噪音。 2.經共整合檢定發現有三個模式具有長期共整合關係，後經ECM誤差修正模型檢定發現三模式並無短期動態均衡關係。 3.由變異數分解之數值的大小所分析出各模式變數間的相互關係，結果與本研究所做之Granger因果關係之結果相符，即相互影響程度小。
The purpose of this research is to apply the VAR model to analyze the relationship between Taiwan stock indices, industrial stocks indices, Consumer Price Index, and Brent crude oil prices. First ,the research uses Chow test to decide the research period and which includes 176 monthly data, from May 1990 to December 2004. The study then use ADF unit root test, Johansen Co-integration test, Impulse function, error-correction model (ECM), Variancede composition and causality test and Granger causality methodologies to examine the relationship among variables. Above tests indicate following findings: ARIMA: Based on the ADF unit root test, all series become stable after first difference, which means that all series are I(1). As for the optimal lag period, the study uses AIC measurement to solve this problem. VAR: 1. First, his study uses AIC to analyze the nine models. Second, based on error autocorrelation test, the result indicates that the lag period conforms to white-noise after conducting error autocorrelation tests. 2. The result from Johansen co-integration test indicates that three models show long-them equilibrium relationship. The study then adopts the ECM to test the co-integration relationship between variables. The result shows that there is no short-term equilibrium relationship exists in the three models. 3. By using the forecast error variance decomposition to analyze the interaction relationship of all the models, the results generally confirm to the results of the Granger causality methodologies. |
| 顯示於類別: | [企業管理學系(管理科學碩/博士班,非營利事業管理碩士班)] 博碩士論文-管理科學碩博士班
|
文件中的檔案:
| 檔案 |
描述 |
大小 | 格式 | 瀏覽次數 |
|---|
| 093NHU05457028-001.pdf | | 6043Kb | Adobe PDF | 1118 | 檢視/開啟 | | index.html | | 0Kb | HTML | 232 | 檢視/開啟 |
|
在NHUIR中所有的資料項目都受到原著作權保護.
|
