Working Paper: CEPR ID: DP1730
Authors: Mark P. Taylor; Lucio Sarno
Abstract: Tests for long-run purchasing power parity (PPP) may lack power with sample periods corresponding to the span of the recent float, leading researchers to use more powerful multivariate unit root tests. We point out a potential problem with such tests: joint non-stationarity of real exchange rates may be rejected when only one of them is stationary. We suggest another test where the null hypothesis is violated only when all of the processes are stationary. This test is easily constructed and has a known limiting distribution. We investigate the finite-sample empirical performance of both tests using Monte Carlo techniques. Applying the tests to four major real exchange rates over the recent floating rate period, we find strong evidence of long-run PPP.
Keywords: real exchange rate; purchasing power parity; multivariate unit root test; test power; monte carlo simulation
JEL Codes: F31
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| real exchange rates (F31) | mean reversion (C22) |
| mean reversion (C22) | long-run PPP (F31) |
| nonstationarity (C22) | rejection of joint hypothesis (C12) |
| nominal exchange rates (F31) | price levels (E30) |