This dissertation is divided in four papers about monetary policy. Initially, we evaluate how inflation expectations are formed in BrazilOur results suggest that data do not follow what rational or adaptive expectations would predict. Using two different methods to measure the informational readjustment frequency in Brazil, we find evidence that inflation expectations seem to behave more closely to what sticky information theory would predict. The second paper investigates the inflation expectations transmission inside the Brazilian financial market. We propose a model in which financial market agents, when they build their own forecasts, not only observe to recent data, but also use lagged expectations from the best forecasters. Our model is evaluated empirically and we can show that the weight given to the best forecasters grows as we get closer to the release of inflation data. In the third paper, we study monetary policy and exchange rate impacts on inflation in Brazil. Our results indicate that it takes from six to twelve months until price structure returns to the one that existed before the shock. The fourth paper investigates central bank´s credibility using individual financial market agents´ expectations. Our hypothesis is that expectations´ heterogeneity for long term horizon comes from different beliefs about central bank´s aversion to inflation. Using this argument, we build a credibility index using Markov Chains for Brazil.