有效評估潛在產出與產出缺口一直是執行貨幣與財政政策的重要環節。過去文獻中用以估計潛在產出的方法大多假設在潛在產出與非加速通貨膨脹失業率（NAIRU）中存在一般單根（regular unit root）。然而，這樣的假並不適用於估計臺灣未經季節調整的潛在產出。有鑑於此，本文以Watson's（1986） decomposition與Apel and Jansson's（1999a）systemapproach為例，以季節單根取代一般單根，從而考量更多的季節波動於模型中。同時，為了驗證模型的可靠性（robustness）以及更為有效評估臺灣的潛在產出與NAIRU，我們亦嘗試對NAIRU與失業缺口（unemployment gap）作不同的模型設定。實證結果顯示在一般單根假設下，亦即模型未考量季節波動時，資料本身的季節波動嚴重干擾估計結果且無法提供有用的訊息；反之，當以季節單根取代一般單根後確實更能有效評估臺灣潛在產出與產出缺口，並提供更為攸關的訊息供臺灣政府執行政策之參考。

Measuring potential output and the output gap have long been an important task for conducting monetary and fiscal policies. There exist several methods for this purpose and a partial list includes the univariate detrending method, the multivariate filtering approach, and the structural VAR system approach. One common feature of all these methods is assuming the existence of a unit root for the unobserved potential GDP. While this assumption is appropriate for the cases of US and most European countries where macroeconomic data are seasonally adjusted, it does not fit the Taiwanese economy. Almost all of Taiwan's macroeconomic data are seasonally unadjusted, and the seasonal unit root as well as richer dynamics have to be embedded in the model. In this paper, we analyze the impact of seasonality on various potential output measures. To check robustness and investigate how sensitive the results are to further changes in the specification of the NAIRU and the unemployment gap, distinct classes of NAIRU and unemployment gap concept are implemented. Empirical analysis confirms the importance of seasonal behavior. Switching from a regular unit root to a seasonal unit root improves the efficiency of measuring potential GDP and output gap for Taiwan and provides more relevant information in conducting monetary and fiscal policies.