財經訊息對匯率的影響-以歐元為例

現今資訊網路科技普及，各種資料取得容易，藉由大量數據的應用，透過量化模型，運用有限時間掌握精準且有效的資訊，已成為未來各個領域不可避免的發展趨勢。匯率對一國經濟具重大影響力，對於匯率預測之研究較少以財經訊息探討其關聯性，因此本文以文字探勘研究財經訊息對歐元匯率的影響。以2015年11月3日至2016年11月4日國內某大型金融機構每日的財經專業報告作為研究資料進行實證研究。採用中文詞知識庫小組(Chinese Knowledge and Information Processing, CKIP)的中文斷詞系統(China Word Segment System,CWSS)將資料予以斷詞，並利用集群分析及貝氏網路篩選出具代表性的關鍵詞，且另外加入利率、股價指數、黃金等數據資料作為自變數。因為樣本期間發生了英國脫歐的非預期政治事件，以鄒檢定(Chow Test)檢驗確實產生結構性變異，因此加入虛擬變數(Dummy Variable)，再以普通最小平方法(OLS)進行迴歸分析。實證結果顯示，短期內匯率受各種相關訊息影響甚鉅尤其是政治面(留歐、脫歐)的消息，而加入虛擬變數後，發現影響匯率變動的因素，仍回歸經濟面(升息、降息、通膨、匯率收貶、匯率收升、美股)的消息對匯率較具顯著性之影響。本研究結合經濟指標的數值資訊與財經訊息的文字資訊量化後，利用計量方法對匯率進行分析預測，期能作為貨幣當局匯率的決策者、企業及投資人的參考，使其更能掌握時機，適度評估審慎運用避險工具，即時採取適當的避險措施，制定風險控制策略，達成降低投資風險之目標。

關鍵字

財經訊息；匯率；文字探勘；集群分析；貝氏網路；最小平方法(OLS)

並列摘要

Nowadays, as the popularization of Internet Information Technology, it is easy to obtain all kinds of information. It is an inevitable trend to analyze the large amount of data to master accurate and effective information. Less researches focus on the relationship between financial text information and exchange rate prediction. Therefore, this study investigates the impact of financial information on Euro exchange rate with text mining. This study applies a large domestic financial institution's daily financial professional report, from November 3, 2015 to November 4, 2016. The Chinese Word Segment System (CWSS) is used to decipher the data in Chinese Knowledge and Information Processing (CKIP). Moreover, we filter out the representative of the key words by using the cluster analysis and Bayesian network. In addition, we also consider the interest rates, stock index, gold and other data as financial variables. Because of the events of the Brexit, we also employee the Chow Test to capture the structural change, thus apply the dummy regression analysis to examine the relationship between the text information and Euro exchange rate. The empirical results show that short-term exchange rate is influenced by financial text and financial numerial related information during the political news (Bremain, Brexit). It is found that the economic key words (raising interest rates, interest rate cuts, inflation, exchange rate depreciation, exchange rate rise, US stocks) would influences the exchange rate significantly. Based on the numerical information of financial indicators and the quantification of the text information of financial information, this study uses the measurement method to analyze and forecast the exchange rate, which can be used as the reference of the decision makers, enterprises and investors of the currency exchange rate. Consequently , they can grasp the timing, evaluate the use of hedging instruments appropriately and prudently, take appropriate hedging measures immediately, and formulate risk control strategies to achieve the goal of reducing investment risk.

並列關鍵字

Financial Information ； Exchange Rate ； Text Mining ； Cluster Analysis ； Bayesian Network ； Ordinary Least Squares (OLS)

參考文獻

李沃牆(2015)。Fed升息搖擺對全球金融市場及臺灣的衝擊與因應。會計研究月刊，360，27-31。doi：10.6650ARM.2015.360.27.

陳裕菘、謝邦昌、李勝輝、陳郁婷(2014)。運用文字探勘與資料採礦技術建立匯率預測模型-以人民幣兌新台幣為例。Journal of Data Analysis，9(1)，133-146。

葉長城(2016)。美國升息對其國內及當前全球經濟之可能影響研析。戰略安全研析期刊，129，23-32。

周濟群、連子杰(2011)。運用文字探勘與XBRL技術提升企業資訊擷取與整合效益之研究。當代會計，12(1)，85-114。doi：10.6675/JCA.2011.12.1.04.

鄭美幸、詹志明(2002)。灰色理論與時間序列模型在匯率預測績效上之比較。台灣金融財務季刊，3(2)，95-104。

國際替代計量

財經訊息對匯率的影響-以歐元為例

全文下載

主題瀏覽