Stochastic trends and multivariate cointegration tests in exchange rate models

Although it appears that exchange rates behave as random walk processes, the possibility remains that exchange rates may be part of a larger equilibrium system. Structural models of exchange rates (asset models) are primarily theories of long-run equilibrium, and substantial deviations in the short-run are possible. This dissertation gives a systematic application of maximum likelihood inference concerning cointegration vectors using the Johansen and Juselius (1990) procedure. The hypothesis of cointegration is given a simple parametric form in terms of cointegration vectors and their loadings. Univariate unit root tests are conducted for exchange rates and the fundamental determinants of exchange rates, that is, nominal money stocks, real incomes, short-term interest rates, long-term interest rates, and price levels of Canada, France, Germany, Italy, Japan, and the United Kingdom. The Purchasing Power Parity condition was analyzed and a cointegration relationship was found for France, Germany and Italy. Tests for the presence of cointegration relationships in the flexible price monetary model and sticky price monetary model were conducted and in each case two cointegration relationships were found for Germany. No cointegration relationship was found in the sticky price model for Italy. For Germany, the estimates and tests under linear restrictions on the cointegration vectors and the loadings are given for both models. In the case of Italy, the analysis was conducted only for the monetary model. A vector error correction model (VECM) was created for each country. For each model, an impulse response analysis using Monte Carlo integration was conducted. The effect on the spot exchange rate from a shock to the error term from each of the fundamental determinants was not different from zero and the effect on the exchange rate from a shock to itself took about two months to become insignificant. Spot exchange rates and forward exchange rates were cointegrated in all six countries. Thus, the foreign exchange market was found to be efficient