Ensaios econométricos sobre a teoria quantitativa da moeda e a curva de Phillips

Abstract

This thesis discusses the application of established relationships in macroeconomics under more empirical frameworks. The first chapter deals with an empirical analysis of the quantity theory of money in Brazil, considering both its weak version, which investigates the relationship between money and nominal income, as its strong version, which deals with the purely nominal relationship between money and prices. Under the methodology of Vahid and Engle (1993), one investigates the existence of short-term equilibrium relationships between these variables and, if possible, common cycles are extracted from the series to determine their correlation with the individual cycles and identify the existence of causal relations between them. In cases where the money is represented by its restricted version, as defined by the Central Bank of Brazil, it is detected the presence of a common trend and a common cycle in both versions of the basic theory. Therefore, their results provide considerable insight into the nature of the interaction between the series and therefore that the observed pattern of comovement is consistent with the assumptions of the quantity theory of money. The second chapter investigates the relationship between the inflation rate and the output gap in Brazil under the methodology of unobserved component model, in the spirit of Harvey (2011). First a bivariate structure is constructed from the series of inflation and output in order to identify a high correlation between the estimated disturbances of cyclical components and thus ensure that inflation is directly affected by the output gap, represented here by his cycle. Moreover, we investigate the presence of common trends and cycles in this structure. An alternative modelling is also considered where a univariate model of inflation is directly affected by the output gap. The advantage of this structure is that it allows the parameter that measures the effects of the output gap on inflation may vary over time. Three different specifications to the dynamic process of this coefficient are considered: an AR(1), a random walk and a smoothing spline model. The first one features better fit and greater predictive power. Three different specifications to the dynamic process of this coefficient are considered: an AR(1), a random walk and a smoothing spline model. The first one presents better fit and greater predictive power. The time-varying regression coefficient indicates that although the relationship of the Phillips curve is valid, it presents a flattened shape during the study period, indicating a low sensitivity of the inflation rate under the impact arising from changes in the gap of product.