Abstract
The long-run purchasing power parity (PPP) hypothesis is re-examined using data for Turkey and its major trading partners during the period 1980–97, extending recent work by Telatar and Kazdagli. The empirical analysis suggests that, while conventional unit root tests do not enable the detection of mean reversion in real exchange rates and hence imply rejection of long-run PPP over the sample, using recently developed nonlinear modelling techniques, strong support is provided for the validity of long-run PPP as well as for theoretical models which predict nonlinear adjustment in real exchange rates. Results are in accord with previous evidence that PPP holds more closely on data for countries which have experienced unusually high inflation.