This thesis investigates the consequences of model uncertainty and persistence in consumption habits and price-setting behaviour in a New Keynesian model.As the decision of a central bank is made in an uncertain environment, the analysis of uncertainty is essential in monetary policy models. In opposition to Bayesian methods, which always require an a-priori density function this thesis uses a robust control approach. Uncertainty is represented by a vector-shock-process, distorting the transition equations. Given this process, optimal interest rates under uncertainty are derived by minimizing the loss function of the distorted model, and compared to optimal policy in the approximating model. Unlike a Bayesian planner, who would only be concerned about calculated risk, a robust planner faces unorganised uncertainty and cannot express his beliefs in probabilistic statements. A robust control theorist builds a set of possible models around the reference model and chooses the optimal policy in the worst case scenario.In the New Keynesian literature the Keynesian assumption of rigid prices is combined with modern macroeconomic methods. Models are based on detailed microfoundations, which are log-linearized around an equilibrium to achieve a simple, linear model, that only depends on inflation, the output gap, and the interest rate, in the case of a single closed economy. For deriving the optimal monetary policy, an investigation of the persistence in output and inflation dynamics is crucial. Because of the controversial debate on the importance of expectations, the consequences of persistence and the interaction with uncertainty is analysed. Therefore, the persistence in macro-variables is justified by the model’s microfoundations of private sector behaviour. This allows to study the influence of private sector behaviour.Thus, chapter 2 starts with explaining the relationship of monetary policy and different types of uncertainty, before chapter 3 discusses the New Keynesian perspective on monetary policy. Subsequently, the fourth chapter illustrates the robust control methodology. The fifth chapter summarizes and categorizes the existing literature regarding the investigation of model uncertainty in a closed economy, before chapter 6 studies a closed hybrid New Keynesian economy under different preferences of the central bank, and for different assumptions on private sector behaviour. Model dynamics are shown for the benchmark model, for the case of uncertainty, and for the case of an unfounded fear against model misspecification. All results are derived for a discretionary policymaker, and a policymaker, who acts under commitment to the optimal interest rate response. Furthermore, different types of simple rules are analysed. Since New Keynesian models have been extended to two-country economies during the last years, due to increased international interdependencies, chapter 7 gives a summary of this evolution, before chapter 8 studies a two-country New Keynesian framework. The conclusions are given in chapter 9.The analysis shows that central banks should react stronger to changes in inflation and output under uncertainty. Furthermore, it is shown that private sector behaviour is of much more importance for the conduct of optimal monetary policy than a general model uncertainty for a given persistence level. Additionally, the study illustrates that monetary policy under commitment is a policy on the razor’s edge with respect to uncertainty, since commitment can be destabilizing, when the fear against model misspecification is unfounded (commitment to the wrong model). This questions the advantage of a commitment solution and shows that protection against the worst case can only be achieved at costs of average performance.
