應用馬可夫轉換模型於貨幣危機研究的潛在缺點
Potential Pitfalls of Markov Switching Models in the Studies of Currency Crises
國立中正大學經濟系暨國際經濟研究所(Department of Economics and Graduate Institute of International Economics, National Chung Cheng University)
- 卷期
- 35卷2期
- 出版
- 2007年06月
- 頁碼
- 213–247
- 編號
- NTUEJ00006869
摘要
馬可夫轉換模型越來越常被應用於貨幣危機的研究。這個新方法宣稱可以避免過去方法的危機錯誤認定問題,同時更加可靠地找出貨幣危機的決定性因素。本文評估這個新方法的優缺點。我們以四個亞洲國家作為例子,說明這個新方法跟過去的方法相比,具有同樣的許多缺點。以預測的準確度與估計時間花費而言,馬可夫轉換模型很難說是更佳的選擇。
Abstract
A growing literature proposes the use of Markov switching models to study currency crises. The new approach promises to avoid the potential misclassifications in the construction of a crisis dummy that the previous approaches suffer, and to identify a set of reliable indicators of speculative attacks. This paper assesses the costs and benefits of the new approach. Using a sample of four Asian countries, we show that the new approach suffers many of the same drawbacks as the previous approaches. Taking the forecast accuracy and calculation efforts into account, Markov switching models can hardly be a better alternative.