本文應用Teräsvirta(1994)所提出的線性檢定方法,配適出最適當的非線性雙門檻平滑狀態轉換模型和顯著的雙門檻,再檢定轉換變數門檻值內、外的區間後發現,門檻內的購買力平價說不會成立,而在門檻外的購買力平價說會成立,我們認為是因為名目匯率存在中立區間而導致上述現象的存在,而阻礙PPP成立之障礙又為中立區間存在之原因,阻礙PPP成立之障礙包括市場中存在交易成本、關稅、投資人異質性預期和央行干預等,以經濟學角度可視其為一機會成本;當市場失衡且套利之利得大於套利之機會成本時,隱含其偏離PPP的程度夠大,此時若套利行為能正常運作,則套利行為將使失衡之經濟體重新回復至原均衡水準,使得PPP成立;而當市場失衡且套利之利得小於套利之機會成本時,隱含其偏離PPP的程度尚不夠大,套利者不會套利,使得PPP不會成立。本研究利用非線性平滑狀態轉換模型來檢驗日本購買力平價說,藉由非線性平滑狀態轉換模型與PPP的結合,對PPP的檢定提供了另一種解釋。 In this paper, we examine the Purchasing Power Parity(PPP) in Japan, applying linearity tests by Teräsvirta (1994). Our results clearly reject linear model in favor of the nonlinear smooth transition switching model with two thresholds of transition variables. We respectively examine whether the PPP will be hold or not in the band of two thresholds or out of the band of two thresholds. We find if data in the band, the PPP will not be hold. Instead, if data out of the band, the PPP will be hold. We think this phenomenon result in nominal exchange rates exist the neutral band. The obstacles which impede PPP hold is also the main reason for existing the neutral band of nominal exchange rates. These obstacles include in the presence of transaction costs, tariff, heterogeneous expectation and interference by central bank. These obstacles also are called opportunity costs if on the economic views. When market is out of the equilibrium, and the gains from arbitrage are greater than the costs from arbitrage. The activities of arbitrage can bring the market to the equilibrium if activities of arbitrage can normally work. In this case, PPP will hold. By the contraries, if the gains from arbitrage are less than the costs from arbitrage, arbitrager won’t make arbitrage. So the market is still out of the equilibrium, the PPP will not be hold. We use the non-linear smooth transition switching model to examine the PPP, offering another explanation for PPP.