Sources of deviations from purchasing power parity after Bretton Woods

This dissertation discusses the sources of deviations from purchasing power parity under the recent floating exchange rate system. In addition, the study examines the validity of three variables, government spending, real world oil prices, and government debt, to explain the movements of nominal exchnge rates. Five bilateral exchange rates, the U.S. dollar-Canadian dollar, the U.S. dollar-Japanese yen, the U.S. dollar-German mark, the Canadian dollar-Japanese yen, and the Canadian dollar-German mark, are used in the empirical study. The data sets for all variables are from the period of post Bretton-Woods (1973:1-1996:4).

Cointegration tests of the model with Johansen's maximum likelihood method show that the variables in the model, real exchange rates, government consumption, real world oil prices, and government debt, are cointegrated. In addition, impulse response analyses and variance decomposition analyses show that the cointegrated relationships are reasonable. Because the cointegration relationship has the interpretation of the long-run equilibrium relationship, the model built with cointegrated variables shows a stable relationship among the variables.

The impact of the three variables on real exchange rates is analyzed with the results obtained from cointegration tests and impulse response functions. The cointegration tests and impulse response functions show conflicting signs on estimated coefficients of the variables of government spending and real world oil price. In regard to government debt, the two methods give the same signs. Cointegration tests and impulse response fiinctions of three variables with nominal exchange rates show conflicting and inconsistent resuUs across five cases. This implies that the three explanatory variables are not reliable in explaining the movements of nominal exchange rates.

Based on the variance decomposition analyses, government debt and government spending appear to be important variables in explaining the variation of the real exchange rates in cointegrating equations; the real world oil price appear to be a less important variable in explaining the variation of the real exchange rates. One of the most important findings of the research is that the U.S. federal government debt has a negative impact on the purchasing power of the U.S. dollar relative to foreign currencies. This can be explained by Barro's (1974) hypothesis that asserted that government debt was not net-wealth.