Data for the U.S. and the Euro-area during the post-Bretton Woods period shows that nominal and real exchange rates are more volatile than consumption, very persistent, and highly correlated with each other. Standard models with nominal rigidities match reasonably well the volatility and persistence of the nominal exchange rate, but require an average contract duration above 4 quarters to approximate the real exchange rate counterparts. I propose a two-country model with financial intermediaries and argue that: First, sticky and asymmetric information introduces a lag in the consumption response to currently unobservable shocks, mostly foreign. Accordingly, the real exchange rate becomes more volatile to induce enough expenditure-switching across countries for all markets to clear. Second, differences in the degree of price stickiness across markets and firms weaken the correlation between the nominal exchange rate and the relative CPI price. This correlation is important to match the moments of the real exchange rate. The model suggests that asymmetric information and differences in price stickiness account better for the stylized facts without relying on an average contract duration for the U.S. larger than the current empirical estimates.