Inferences about threshold effects in macroeconomic relationships

I investigate nonlinearities in macroeconomic relationships that can be described by threshold processes. In the first essay, I study asymmetries in the response of output to monetary policy shocks of different magnitude. Several studies have investigated this issue (Weise, 1999 Ravn and Sola, 2004 Lo and Piger, 2005) and have found mixed results with respect to the significance of this type of asymmetry. I argue that the mixed evidence in the empirical literature is explained by the exogenous imposition of the threshold that defines the size of a monetary policy shock. To address this issue, I utilize an econometric framework where the threshold is estimated from the data. In particular, I estimate an unobserved components model of output where the transitory component is driven by a threshold autoregressive (TAR) process and augmented by a monetary policy variable. When the threshold is estimated, there is strong statistical evidence that the effects of monetary policy on output vary with the size of the monetary shock. Moreover, the results from the model provide some support to the implications of menu-cost models: smaller monetary shocks trigger a larger response in output.The second essay studies the role of aggregation and firm heterogeneity in the context of asymmetries in the response of output to the size of monetary policy shocks within a Bayesian framework. In recent years, the implications of theoretical models with menu costs have been related to the empirical literature studying these asymmetries. However, the empirical results only partially support the implications of menu-costs models. I argue that this difference between theory and data could be caused by an aggregation problem. To the extent that the aggregation problem implies a distinction between an abrupt or a smooth change in regimes, I formally compare a threshold autoregressive (TAR) process and a smooth transition autoregressive (STAR) process within an unobserved components model. The Bayesian model comparison shows that the nonlinear dynamics are better described by a smooth transition between regimes, which suggests that firm heterogeneity and aggregation may play a role in explaining how the behavior of output varies with the size of the monetary shock.