Kalman-Filter方法估計匯率泡沫現象

A Kalman-Filtering Analysis of Exchange Rate Bubbles

10.6989/JN.201112.0365

王昭偉(Zhao-Wei Wang)；王春源(Chun-Yuan Wang)

匯率泡沫 ； 狀態空間模型 ； 卡爾曼濾波 ； Exchange Rate Bubbles ； State Space Model ； Kalman-Filter

南亞學報

31期（2011 / 12 / 01）

365 - 388

繁體中文

過去文獻中，將泡沫（bubbles）定義為資產價格偏離市場基本面的部分。本文以Flood and Hodrick（1990）的泡沫理論模型及Meese（1986）的貨幣模型為基礎，設定了包含泡沫項的匯率決定方程式，並採用狀態空間模型，運用卡爾曼濾波以最大似然法進行遞推估計。研究對象為中國自2005年7月匯改政策以來至2010年11月的人民幣／美元，及1990年1月至2010年11月的日圓／美元的匯率泡沫現象。本文估計結果顯示，中國及日本所估計的匯率泡沫項皆存在且泡沫參數皆顯著。分析泡沫項的走勢及泡沫參數的梯度，都能符合樣本期間匯率的變動型為，合理解釋了匯率偏離市場基本面的情形。

In the past literatures, ”bubbles” is defined as the deviation of an asset's price from the value of its market fundamentals. Based on the bubble's model of Flood and Hodrick(1990) and the monetary model of Meese(1986), this paper sets up a exchange rate determination equation including a bubble size term, then we adopt a state space model and make a MLE recursive estimation in use of Kalman-Filter. We use the samples of the exchange rates of RMB/U.S. dollars from 2005/07 when China adopted the revolution policy of exchange rate system to 2010/11, and the exchange rates of Japanese Yen/U.S. Dollars from 1990/01 to 2010/11 to examine if there are exchange rate bubbles. The estimation outcome show that there are exchange rate bubbles and the bubbles' parameters are significant. By analyzing the trend of the bubbles and the gradients of the bubbles' parameters, the estimation result corresponds to the behavior of exchange rate during the sample, and the deviation of exchange rate from its market fundamentals can be explained reasonably.