本文主要檢視總體變數對台灣股票市場的預測能力。我們首先以馬可夫轉換模 型定義台股市場中熊、牛市兩種狀態。之後的迴歸分析結果顯示期間利差(TS)、 通貨膨脹率(IR)、製造業生產指數(IPIMIG)、M1B 成長率(M1BG)、重貼現 率(DRR)、匯率(DEX) 以及政府負債(DPD) 分別對1個月至兩年後的台股 熊市機率具有不同的預測力。大致上以政府負債變動及貨幣政策對台灣股票市 場的預測能力較佳, 而在期間利差在係數上, 與理論以及美國實證結果不一致。 另外, 與其他預測股票報酬率的文獻比較, 本文以總體變數預測熊市機率的 結果較佳。在頑強性檢視方面, 我們發現以較近期的樣本期間估計, 無論國內外 的總體變數, 預測力皆有隨樣本期間不同而改變的可能。最後, 本文實證結果亦 顯示以總體變數預測台灣股票市場較一般買入持有策略之報酬率佳。 綜觀台灣股票市場, 在結構、波動或是管理方面皆有其小型經濟體的特徵, 尤 可見得政府政策左右股市的痕跡。
The aim of this paper is to investigate the predictive power of macroeco- nomic variables on Taiwan stock market. The study covers the period from 1971.1 to 2011.12, using Markov-Switching model to identify the bear and bull stock market, finding that term spread(TS), inflation rate(IR), indexes of industrial production-manufacturing index annual growth rate(IPIMIG), M1B annual growth rate(M1BG), changes in exchange rate(DEX), and changes in public debt(DPD) have different in-sample and out-of-sample predictive powers at one to 24-month-ahead horizon. It’s noteworthy that changes in public debt and monetary policy shock have more impact on Taiwan stock market. Moreover, this paper shows that macroeconomic variables have better performance in predicting bear stock market than predicting stock return. In the robustness check, employing multivariate models show that pre- dictive power may change in different sample period. Finally, this paper also conducts a market-timing strategy, and finding that it outperforms buy-and-hold strategy. According to the aforementioned, we can find the characteristics of small open economy in Taiwan stock market, and the authorities still play a role in it.