本文主要探討總體經濟變數對臺灣加權股價指數與8 大類股股 市熊市的預測力。利用Pagan and Sossounov (2003) 的方法認定熊市 並以probit 模型分析描述金融市場、整體物價水準、總體經濟活動、 貨幣政策、與財政政策等狀態之總體經濟變數的樣本內與樣本外預 測表現。主要實證結果顯示期間利差有助於預測大盤與各類股的熊 市，但僅在較短期與較長期較具預測力。在物價水準部分，消費者 物價指數 (consumer price index, CPI) 年增率僅對類股具有不同程 度的預測力。至於工業生產指數成長率與失業率變動等衡量總體經 濟活動的變數，相較於Chen (2009) 對美國股市的研究，我們的實證 結果則顯示前者不具預測力，而後者亦僅對大盤與部分類股的熊市 具微弱之預測力。最後，貨幣政策與財政政策變數雖有助於預測大 盤熊市，但僅有水泥與金融類股熊市同時對貨幣與財政政策具顯著 而強烈的反應；食品、塑化與營建類股熊市僅受貨幣政策影響，而 紡織、造紙、與機電類股熊市在短期僅受貨幣政策影響，但在長期 則僅受財政政策影響。

This paper examines the predictive power of macroeconomic variables for bear markets of the TAIEX and 8 industry indices in Taiwan. Based on Pagan and Sossounov’s (2003) modified Bry-Boschan method and probit models, the in-sample and out-of-sample results all reveal that the term spread is a better predictor, but only for shorter and longer forecast horizons. The annual percentage change in the consumer price index only contains information about future bear markets. The macroeconomic variables reflecting economic activity provide little information regarding future bear markets, marking a sharp contrast with the U.S. findings in Chen (2009). Finally, both fiscal and monetary polices have a great impact on the TAIEX as well as the cement and finance industries. However, while the food, plastic and chemical, and building materials and construction industries react only to monetary policy, the textile, paper and pulp, and electric machinery industries respond to monetary policy only in the short run, while responding to fiscal policy only in the long run.