New bootstrap methods for financial and economic time series

The bootstrap is a promising simulation tool that can help to solve complicated statistical problems with no tractable solution. Specifically, the fundamental idea of the bootstrap is to use re-sampling methods to approximate otherwise unknown properties of an estimator. This thesis investigates bootstrap methods for financial and economic time series to do forecasting. The results are presented in three self-contained parts which include theory, simulations, and empirics for the implemented bootstrap method.