根據Krugman(1991)提出匯率目標區之想法,當匯率於中心匯率附近波動時,政府干預匯率的可能性很低;但當匯率逐漸偏離中心匯率,接近目標區之上下限時,政府對匯率進行干預的機率便會增加,以防止匯率超出匯率目標區所設定之範圍。 過去,已有許多文獻針對Krugman(1991)所提出之匯率目標區模型進行實證研究,但大部分之實證結果均拒絕他在匯率目標區模型中之設定或推論。本文依據Lundbergh and Terasvirta(2003)所提出之STARTZ模型,利用非線性模型來描述匯率在目標區內的波動情形,並進一步檢定STARTZ模型是否能支持Krugman匯率目標區模型設定之前提假設。 本研究以歐洲匯率機制會員國中參加經濟貨幣同盟,同時為G7成員中之德國、法國及義大利對ECU匯率當做研究對象,樣本期間為1987年1月14日至1989年12月29日之匯率日資料。 由實證結果中可以發現,義大利及法國之估計結果支持STARTZ模型可以用來描述該國匯率在匯率目標區內的波動情形,但德國之估計結果卻不支持。在檢定是否支持Krugman匯率目標區模型之前提假設時,發現實際估計出之匯率目標區間遠小於官方所公告目標區範圍,因此,檢定結果並不支持Krugman(1991)中官方公告之匯率目標區間之公信力。
According to the exchange rate target zone model of Krugman(1991), government won’t intervene the exchange rate as long as the exchange rate is around the central parity. When the exchange rate reaches the edge of the band, the probability of government interventions will increase to prevent the exchange rate from being out of the band. The Krugaman’s exchange rate target zone model has been tested in different ways in many empirical studies. But most of the empirical results reject Krugman’s implications. In this paper, the smooth transition autoregressive target zone(STARTZ)model proposed by Lundbergh and Terasvirta(2003)is used to fit the dynamic behavior of the exchange rate that implied by Krugman(1991). Furthermore, we want to test whether or not the empirical results of the STARTZ model can support the hypothesis of the Krugman model. The time series we studied are 691 daily observations from 14th January 1987 until 29th December 1989 of the exchange rates of Germany, France, and Italy against European Currency Unit(ECU). The sample countries in this paper are not only the members of European Monetary System(EMS)but also the members of G7. The empirical results for Italy and France estimated by the STARTZ model can be used to describe the dynamic behavior of an exchange rate in the target zone model, but Germany’s result is not. We also found that the zones we estimated are narrower than the zones declared by the government. Hence, our empirical results do not support the credibility hypothesis of Krugman.