There is a broad range of literature on long memory or long range dependent processes, especially on fractionally integrated processes. Throughout the thesis different models generating long memory processes, particularly the extensively discussed fractional ARIMA model, are studied.
One of the main parts analyzes structural breaks versus long memory. To shed new light on this problem I make heavy use of an error duration model, which gives a nice view on stochastic processes in general and on short versus long memory in specific.
After presenting various estimators and tests for long range dependence, I compare short and long memory models for financial data (Dow Jones Index, Alcoa Inc., and EUR / USD exchange rate).
My contribution is a model for time-varying (long) memory and herewith I try to unify the concurring views of long memory and structural breaks.