Macroeconomic Modelling and Forecasting Using Non-Stationary Data

Important aspects of macroeconomic modelling and§forecasting in the presence of non-stationarity are§examined in this book. Three forms of§non-stationarity are assessed: explosive,§structural-break, and unit root non-stationarity.§First, testing for unit-root non-stationarity in the§presence of explosive non-stationarity is considered.§Numerical difficulties are circumvented using§approximations before the finite-sample properties of§the unit-root test are assessed. Secondly the use of§model averaging given non-stationarity is§investigated. While model averaging can provide§competitive forecasts and parameter estimates,§selection is required, and often a single selected§model will perform best. Because averaging does not§avoid the need to select, methods of selection are§discussed. Third, regression models in the presence§of unit-root non-stationarity are estimated. Previous§empirical studies of monetary and fiscal policies§have made little reference to non-stationarity. A§cointegrated§vector-autoregressive model is used to combat this§and evidence for policy interactions is found.