再探臺灣菲利浦曲線的平坦化：變數衡量的角度

REVISITING THE FLATTENING OF TAIWAN’S PHILLIPS CURVE: A MEASUREMENT PERSPECTIVE

菲利浦曲線（Phillips curve）的平坦化顯示經濟活動與通膨之間的關係脫鉤，主要有貨幣政策或全球化等解釋；惟另有看法認為，用於估計菲利浦曲線的變數可能具有某些衡量上的因素，以致無法觀察到經濟活動與通膨之間的關係。Stock and Watson（2020）即認為，不同商品或服務之間的異質性甚大，彼此價格與景氣循環的關係皆可能不同。由於臺灣經濟長期以來歷經變遷，不僅各商品或服務價格與國內經濟活動之間的關係可能有異，亦可能隨時間而改變。爰此，本文藉由建構Stock and Watson（2020）之「循環敏感通膨」（cyclically sensitive inflation，CSI）指標的過程，從變數衡量的觀點探討臺灣的情形。CSI指標係對國內經濟活動順循環關聯較大的CPI組成項目賦予較大的權重，藉此衡量景氣循環產生的通膨壓力。實證結果發現，在不同的樣本期間下，經濟活動對不同的商品或服務價格亦產生不同的通膨壓力。在1987年至2000年，以「美容及衛生用品」或部分食物類的通膨壓力對國內經濟活動較為敏感；在2001年以後，經濟活動則與居住類項目（如房租）的價格變動較具順循環關係。接著，在考慮結構轉變且極大化具有順循環特性的CPI組成項目後，菲利浦曲線雖未全面失效，但抵換關係在2001年後仍有減弱的情形。由於2001年後的CSI指標已排除許多貿易財的影響，因此，除了全球化之外，可能尚有其他影響市場結構或廠商訂價行為的結構性因素。

The flattening of the Phillips curve suggests a decoupling of the relationship between economic activity and inflation. This has been primarily explained by factors such as monetary policy or globalization. However, some argue that the variables used to estimate the Phillips curve may have certain measurement issues, making it difficult to observe the relationship between economic activity and inflation. Stock and Watson (2020), for example, argue that there is significant heterogeneity among goods and services, and their price dynamics do not follow the same pattern over the business cycle. Given that Taiwan’s economy has changed over time, not only could the relationships between the prices of various goods and services and domestic economic activity differ, but they may also evolve over time. Therefore, this study constructs the“cyclically sensitive inflation”(CSI) index proposed by Stock and Watson (2020) to explore the flattening of Taiwan’s Phillips curve. The CSI assigns greater weight to CPI components that are highly correlated with pro-cyclical economic activity, thereby providing a means to measure inflationary pressures generated by the business cycle. Empirical results show that during different sample periods, economic activity exerts varying pressures on the prices of goods and services. From 1987 to 2000, domestic economic activity had a pronounced impact on the price of“beauty and hygiene products”and certain food items. From 2001 onward, economic activity became more closely linked to price changes in housing-related items such as rent. Furthermore, considering structural changes and maximizing the effect of the pro-cyclical CPI components in this CSI index, the Phillips curve, while not entirely invalidated, showed a weakening of the trade-off relationship after 2001. Since the CSI after 2001 largely excluded the effects of traded goods, this suggests other structural factors besides globalization affecting market structure or firms’pricing behavior.