Three Essays on Non-linear Effects in Dynamic Macroeconomic Models

This thesis has aimed to analyse non-linearity in dynamic models. Attention has focused on the class of dynamic models that accommodate the possibility of distributional modification in the models. In chapter 1, I have studied the non-linear effects of policy shocks in the classical DSGE model. The analysis of such model is subject to two types of shocks, technology and monetary policy. I have extended the analysis of classical model by allowing for the distributional modification of monetary policy shock using WSN distribution. This study reveals the extent to which the distribution of macroeconomic variables may response to policy actions and outcomes involved. Moreover, in classical monetary model the long run behaviour of the level of inflation with respect to the inflation uncertainty has investigated. I have also analysed the dynamic model of AR-GARCH time series. I have investigated the possible non-linear and asymmetric effects of distributional assumptions on the behaviour of the QMLE of the parameters in AR(1)-GARCH(1,1) model. A Monte Carlo experiment is set up to evaluate the distributional misspecification in aforementioned model by applying both symmetric and asymmetric WSN distribution across a range of mean and volatility persistence. The other contribution in chapter 2 is computing the quantiles under distributional misspecification in AR-GARCH model. In terms of the accuracy of the estimated quantiles, I have implemented the bootstrap technique. In addition, in chapter 3 the attention has concentrated on the procedures with suitable technique for the analysis of unit root tests. The usefulness of bootstrap technique is investigated in the context of unit root test applying in stock indices and exchange rate series. I evaluate the popular unit root tests including Augmented Dickey Fuller(ADF) and Phillips Perron(PP) as well as DF-GLS. Furthermore, this chapter attempts to answer the question of how the difference in frequency of empirical data say, monthly, weekly, and daily might affect the unit root results.