A small open economy model which incorporated. the rational expectations - natural rate hypothesis was constructed. The resulting model indicated that to the extent that foreign monetary policies may affect the relative price of domestic to foreign goods, the domestic unemployment rate was not neutral with respect to foreign monetary policies.

Using Canada as the small open economy and the United States as the rest of the world, the weak open economy version of the natural rate hypothesis was empirically tested for both the flexible and the fixed exchange rate periods. The empirical methodology employed was the Granger causality test. The results based on the goodness of fit test indicated that for all the exchange rate regimes, the Canadian unemployment rate was rtot Granger caused by any causal variables considered. However, the results based on a comparison of the postsample forecasting performance were different. For the first flexible exchange rate. period, 1953-1962, we found that the Canadian unemployment rate was Granger caused by the foreign price level. However, this result was consistent with the weak open economy version of the natural rate hypothesis. We also found that the foreign unemployment rate Granger caused theCanadianunemployment rate in the second flexible rate period, 1970-1979. But, this did not damage the natural rate hypothesis since the natural rate hypothesis does not preclude real variables from having a systematic effect on the Canadian unemployment rate.

The tests for rationality with the Canadian price level equations for both the flexible and the fixed exchange rate periods were inconclusive. First, we were not able to detect a breakpoint in the foreign and domestic money rules. The alternate tests of rationality which examined whether the relevant variables were included in the Canadian price level equations were also inconclusive because of a high degree of multicollinearity between the variables.