本文以實質國民生產毛額成長率的變動來探討臺灣的景氣循環。我們首先了解傳 統的時間趨勢設定對景氣循環轉折點認定的影響，並利用雙狀態的馬可夫轉換模型來分析實 質國民生產毛額的年成長率。我們發現馬可夫轉換模型可適當地區分臺灣景氣高、低成長狀 態，亦可用於認定景氣循環的谷底時點。在預測經濟成長率方面，我們也發現此模型較其他 時間序列模型有更好的樣本外預測能力。

　　　　　In thsi paper we analyze business cycles in Taiwan usign real GNP data. We first examine the effects of different time trend specifications on turning point identification. A two-state Markov switching model is then employed to characterize the annual growth rate of real GNP. The empirical results suggest that this model is suitable for identifying the troughs of Taiwan's business cycles. The Markov switching model also compares favorably with linear time series models in outof-sample predictions.