Optimal prediction model of cyclical downturns proposed by Neftci (1982) was widely used in application to predict turning points of different economies such as US, UK, Japan, and West Germany. I apply Neftci model to predict the occurrence of bear markets of Taiwan stock market by using exchange rate of NTD, monitoring indicator of economy, and turnover rate of TAIEX as predictors. The results show that turnover rate outperforms the other two predictors, which signaled bear markets almost concurrently on average and the variation of the signal time is the smallest, ranging from 4 months lead to 4 months lag. / Optimal prediction model of cyclical downturns proposed by Neftci (1982) was widely used in application to predict turning points of different economies such as US, UK, Japan, and West Germany. I apply Neftci model to predict the occurrence of bear markets of Taiwan stock market by using exchange rate of NTD, monitoring indicator of economy, and turnover rate of TAIEX as predictors. The results show that turnover rate outperforms the other two predictors, which signaled bear markets almost concurrently on average and the variation of the signal time is the smallest, ranging from 4 months lead to 4 months lag.

本文引用Pagan and Sossounovb（2003）針對Bry and Boschan（1971）景氣循環周期判定法修改後的法則，判定大盤與類股指數的牛市、熊市周期。將判定的周期結果畫成圖表，藉由簡單的圖表分析將可明確得知大盤周期與類股周期領先與落後的關係，並應用計量模型估計，找尋能夠顯著預測大盤周期變動方向的類股，或是檢驗大盤周期是否能夠預測類股周期方向；反之亦然。並且比較圖表分析與計量模型估計結果是否一致。 圖表分析與向量自我迴歸模型的實證結果一致，八大類股中，營建、金融、機電、塑化等四類股周期能夠顯著預測大盤周期走勢，其中以塑化類股最具預測能力；而大盤周期皆無法精準預測類股周期走勢。而羅吉斯迴歸模型結果也發現，營建、金融、機電、塑化等四類股周期能夠增加大盤周期走勢的預測機率；同樣的，大盤周期無法影響類股周期走勢的預測機率。