Abstract
This empirical note applies the threshold unit root test proposed by Caner and Hansen (Citation2001) to test the validity of long-run Purchasing Power Parity (PPP) in Mainland China and Taiwan over the period of January 1986 to October 2009. The empirical results indicate that PPP holds true for the two areas under study, and the adjustment towards PPP is found to be nonlinear.
Acknowledgements
We are grateful to Bruce Hansen for making available his Matlab codes for the TAR model, which were modified for this exercise. We thank an anonymous referee and the editor (Professor David Peel) of this journal for their several helpful comments, suggestions and time spent in reading this article. These all make this article more valuable and readable.
Any remaining errors are our own.
Notes
1Reasons for the nonlinear adjustment are the presence of transaction costs that inhibit international goods arbitrage, and official intervention in the foreign exchange market may be such that nominal exchange rate movements are asymmetric (see Taylor and Peel, Citation2000; Taylor, Citation2004; Juvenal and Taylor, Citation2008). Kilian and Taylor (Citation2003) also suggested that nonlinearity may arise from the heterogeneity of opinion in the foreign exchange market concerning the equilibrium level of the nominal exchange rate: as the nominal rate takes on more extreme values, a great degree of consensus develops concerning the appropriate direction of exchange rate moves, and traders act accordingly.
2The real exchange rate series of a country at time t is defined as , where St is the nominal exchange rate of home country per dollar, and denote the CPIs of home country and the United States, respectively.
3As R 1T has more power than R 2T , we only report the results of R 1T in our study.
4Kapetanios et al. (Citation2003) also proposed a testing procedure to detect the presence of nonstationarity against nonlinear but globally stationary Exponential Smooth Transition AutoRegressive (ESTAR) process. They constructed t-statistic of test by regressing the following auxiliary equation based on Taylor series using ordinary least squares:
In this framework, the null hypothesis and the alternative hypothesis are expressed as δ = 0 (nonstationary) against δ < 0 (nonlinear ESTAR stationary). Kapetanios et al. (Citation2003) showed that t-statistic of the parameter of interest, that is, δ does not have an asymptotic normal distribution and thus one must resort to simulations for asymptotic critical values. Wu and Lee (Citation2008) applied this test on both the bilateral real exchange rates and the real effective exchange rate for G-11 and Asian countries and found weak evidence on the nonlinear mean-reverting adjustment of real exchange rates in both fixed and floating regimes. Following the suggestion of an anonymous referee, we also add Kapetanios et al.’s (Citation2003) test in our study. The results reported in also indicate that the real exchange rates are nonstationary for both Mainland China and Taiwan.