- 學位論文
以泰勒法則為基礎之TVP匯率預測模型
Forecasting Exchange Rates with Taylor Rule Fundamentals and Time-Varying Parameters Model.
摘要
本文旨在利用泰勒法則基礎模型，搭配貝氏TVP模型進行匯率預測，本研究使用之資料期間為1997年1月至2016年12月，共120筆月資料，針對10個國家的匯率，分別預測其一個月、一季、半年與一年之後的走勢，實證結果在10個國家中，有6個國家的匯率可以利用泰勒法則基礎模型預測。 若以台灣的實證結果來看，泰勒法則基礎模型可以預測國內1個月後之匯率，而3、6、12個月後的預測結果並不理想。 本研究有兩項主要發現：(1)亞洲的開發中國家(如：馬來西亞、菲律賓)，其匯率走勢亦可使用泰勒法則基礎模型進行預測；(2)本文在設定模型時將預期通貨膨脹率納入考量，在模型中以延後1期之通貨膨脹率進行匯率預測，但實證結果發現其效果並不顯著。
並列摘要
This paper forecasts exchange rate using Taylor rule fundamentals model with time-varying parameters model estimated by Bayesian method. The data of this paper consists of monthly exchange rate from 1997M1 to 2016M12, on up to 10 countries relative to the U.S. dollar, respectively, forecasts in four forecasting horizons (1,3,6,12). This paper finds strong evidence of exchange rate predictability with Taylor rule fundamentals model for 6 out of 10 countries. There are two findings in this study: (1) Asian developing countries (e.g., Malaysia and Philippines) whose exchange rate can also be predicted by using the Taylor rule fundamentals model; (2) In this paper, consider expected inflation rate is an important factor when the government makes monetary policy. Therefore, this paper replaces inflation rate with expected inflation rate. However, the empirical results found that the effect is not significant.