Purchasing power parity under the gold standard

Long-run purchasing power parity (LRPPP), the basis of most open economy macroeconomic models, has proved difficult to back up empirically. However, there is one standout exception to the otherwise mixed results. Diebold, Husted, and Rush (1991) are consistently cited as having found strong evidence of LRPPP by using a fractionally integrated moving-average model whose restrictions are looser than those of traditional unit root tests. We propose structural change rather than fractional integration as a plausible behavior pattern for the data. Using the Bai-Perron (1998) test for multiple structural change, we find mean shifts in each of the real exchange rates. When those shifts are included in the model, the speed of mean reversion is greatly improved. We assert that quick mean reversion around an occasionally changing mean provides a more reasonable representation of the data than does fractional integration.