Understanding the nature of short-run inflation dynamics has been pointed out in the literature as a very important issue for the adequate policy choice. The idea that sluggish price adjustment plays a central role in explaining monetary non-neutrality in the short-run dates back to Keynes. Since then, both theoretical modeling and empirical work have been developed showing the real effects of monetary policy when price rigidity is present. In the three essays of this dissertation, I study different issues related to inflation persistence. Empirical work developed here is applied to Brazil.

Most of the recent research in monetary policy has focused on the use of a single exogenously specified loss function to evaluate monetary policy performance. This literature has come to the conclusion that backward looking models are more difficult to control. Chapter 1 tests the validity of this conclusion, using both an exogenous loss function and a welfare theoretic endogenous lose function. Looking into the case where the policymaker is uncertain about the pricing behavior of the firms, the chapter also investigates the presence of robust policy rules. We find that conclusions vary markedly with different types of loss functions.

Chapter 2 investigates the patterns of price adjustments in Brazil. I derive directly from a large data set of the CPI price quotes, spanning approximately ten years until 2006, the main stylized facts that describe the behavior of price setters providing micro empirical evidences on the degree and features of price rigidity. I find that on average prices remain unchanged for 2.7 to 3.8 months, exhibiting, however, a large degree of product and sector heterogeneity. Data on the frequency and sign of price changes show that there is a strong symmetry between price increase and decrease. Conversely, as expected under a positive inflation environment, the magnitude of positive price changes compensates this effect. I also provide some insights on the determinants of the patterns of price adjustment. Results suggest that substantial disturbances to average inflation imposed a high enough cost of not adjusting prices and triggered more frequent price reviews.

Chapter 3 evaluates the dynamics of inflation in Brazil using a standard hybrid model for inflation and four alternative measures of marginal cost derived from different technology assumptions. Overall, results show that deviations from the standard Cobb Douglas production function, as Cobb-Douglas with overhead labor, constant elasticity of substitution (CES) and CES with overhead labor does not improve estimates for the reduced form parameters, nor for the structural parameters. In general, even for the standard Cobb-Douglas specification, estimates values are quite counterfactual, specially the estimates for the price stickiness parameter.