Academic journal article The Journal of Developing Areas

Nonlinear Threshold Unit Root Test and Ppp in Transition Countries

Academic journal article The Journal of Developing Areas

Nonlinear Threshold Unit Root Test and Ppp in Transition Countries

Article excerpt

(ProQuest: ... denotes formulae omitted.)

INTRODUCTION

Unlike most previous studies that relied upon regression analysis to test the well-known theory of Purchasing Power Parity theory (PPP here after), most recent studies use unit root tests to establish mean reverting properties of the real exchange rate. While standard and old unit root tests assumed that adjustments of the real exchange rates over time are linear, more recent studies have demonstrated that the adjustment could be non-linear in nature. As such, non-linear unit root tests have produced relatively more support for PPP. Foreign exchange market intervention, cost of transactions, different speed of adjustment of individual commodity prices to changes in the exchange rate, structural breaks, etc. are among factors that are said to contribute to non-linear movement in the real exchange rates. Bahmani-Oskooee et al. (2008) discuss these issues and empirically compare a few linear unit root tests to non-linear test and show that non-linear tests provide more support for the PPP. Bahmani-Oskooee and Hegerty (2009) provide a nice review of the literature, especially in developing and emerging markets.1

However, the finding of nonlinear adjustment does not necessarily imply nonlinear mean reversion (stationarity). As such, stationarity tests based on a nonlinear framework must be applied. The powerful nonlinear threshold unit root test of Caner and Hansen (2001) is designed to achieve these two goals in one step, i.e., simultaneously investigate non-stationarity and nonlinearity. Liu et al. (2012) have already applied this test to eight transition countries of Bulgaria, the Czech Republic, Hungary, Latvia, Lithuania, Poland, Romania, and Russia with not much success. The only country in which PPP was supported was Romania. The reason could be the fact that they applied the test to the real bilateral exchange rate between each country's exchange rate and the U.S. dollar. Clearly if the U.S. is not major trading partner of any of the countries in the sample, PPP will fail. A better exchange rate will be the real effective exchange rate that includes almost all major trading partners in its construction. These rates that are constructed and published by the International Monetary Fund for certain countries have a major deficiency in that the trade weights used in calculating the weighted average of indexes of real bilateral exchange rates are fixed and borrowed from a base year.2 Given that transition countries have become increasingly open and have increased their trade via liberalization programs, a much better measure of the real effective exchange rate would be the one that allows trade weights to change over time. Fortunately such rates are now constructed and published by the Bank for International Settlements (BIS).

Therefore, it is the main purpose of this paper to test the PPP for 14 transition countries for which the BIS constructs monthly real effective exchange rates that allows trade weights to vary over time. Our sample includes Austria, Bulgaria, Croatia, Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Russia, Slovenia, and Slovak Republic. While the data used in this study are explained in Section II, the method that is based on threshold unit root test of Caner and Hansen (2001) is explained in Section III. In Section IV we report our findings. In summary we find that threshold unit test strongly rejects the unit root process for 5 of these 14 transition countries examined, while the traditional unit root tests such as the ADF, PP, and KPSS did not lead to rejection. Section V concludes the paper.

DATA

As mentioned before our sample included 14 transition countries: Austria, Bulgaria, Croatia, Cyprus, Czech Republic, Estonia, Hungary, Latvia, Lithuania, Poland, Romania, Russia, Slovenia, and Slovak Republic. The data are monthly real effective exchange rate over the period January 1994 to June 2012. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.