Purchasing power parity (PPP) is considered as an important theory of explaining how exchange rate varies in the long run. Most of empirical studies in the past adapted linear cointegration method to test the purchasing power parity. However, there are papers point out that exchange rate exists non-linear cointegration and unexplainable bias might exist in testing the purchase power parity theory while using linear cointegration test. The methodology of this study is based on an application of ESTR ECM proposed by Kapetaniosetet al. to enhance the inadequate of linear cointegration test. We analyze the dynamic adjusting behavior of short-term nominal exchange rate with ESTR ECM model while the non-linear cointegratoin exists. The empirical result indicates that the purchase power parity between Taiwan and its major trading countries is confirmed. Among the trading countries, American, Japan and Hong Kong are suitable for using linear error correction model and non-linear error correction model for Singapore and Korea.

The methodology is based on an application of nonlinear ESTR ECM by Kapetanios et al. (2006) to analyze the short-run dynamic adjustment to long-run equilibrium in Taiwan money demand function. We take consideration of Taiwan as a small open economy system, the exchange rate could be included in money demand function. The result indicate that using ESTR ECM to analyze the adjustment behavior of money demand function in Taiwan is better than linear ECM. Our findings point out that the public adjusts at any time for holding money and the speed of adjustment for real balances depends on the size of deviation.