探討臺灣的菲利浦曲線及其通膨預測能力

TAIWAN’S PHILLIPS CURVE AND ITS INFLATION FORECASTING ABILITY

菲利浦曲線（Phillips curve）描繪失業率與通膨率之間的關係，若失業率與未來通膨之間存在穩定關係，則菲利浦曲線有助於預測未來的通膨情形。為瞭解臺灣的情況，本文透過縮減式的角度及滾動迴歸的方式，探尋長期以來臺灣的失業率（或失業率缺口）與未來通膨之間的關係是否穩定，並檢驗菲利浦曲線的通膨預測能力。實證結果發現，在2000年以前，失業率（或失業率缺口）與未來通膨之間呈現顯著的負向關係，且含有通膨的預測資訊；約至2001年以後，兩者關係減弱且轉呈不顯著，且通膨預測能力轉趨消失。儘管如此，若進一步檢驗2001年後菲利浦曲線在不同條件下的預測能力，則可發現當經濟活動下滑時，菲利浦曲線對未來半年及一年通膨率的預測績效將優於AR模型，且以經濟體遭受到重大衝擊（如2001年網路科技泡沫破裂、2008年全球金融危機等）之期間尤為明顯。因此，菲利浦曲線仍可能對未來一年內的通膨率具有預測資訊。此結果顯示，菲利浦曲線在特定條件下依然含有通膨的預測資訊，具有其參考價值。

The Phillips curve depicts the relationship between unemployment and inflation. If the relationship is stable, the Phillips curve can help predict inflation. This paper utilizes reduced-form rolling regressions to investigate whether the relationship between unemployment (or unemployment gaps) and future inflation is stable in Taiwan. The paper then further examines the forecasting ability of the Phillips curve. The empirical results show that before 2000, the inverse relationship between unemployment (or unemployment gaps) and future inflation was significant, and the Phillips curve contained useful information for inflation forecasting. However, the relationship became statistically insignificant after 2001, and the Phillips curve lost its forecasting ability. Nevertheless, the Phillips curve still can help predict inflation under certain conditions after 2001. When there is an economic slowdown, the Phillips curve outperforms AR models for the six-month and one-year ahead inflation forecasts; the forecasting ability is more potent during periods of severe downturns, such as the dotcom bubble or the 2008 global financial crisis. The findings suggest that the Phillips curve still serves as a useful reference for inflation forecasting in the case of Taiwan.