我还是把它贴在这吧,非常着急
As discussed in class, the high persistence in the linear time series can be explained from
two perspectives. One is the long memory model, i.e.,(1-B)d xt = u+ et; (-1 < d <=1) where et are i.i.d. standard normal random variables.
Another perspective is to assume there is some structural changes in the model, for instance, Xt=Ut+AtXt-1+et;where et are i.i.d. standard normal random variables and (Ut; At) are piecewise constantwith unknown break points. Which viewpoint do you agree (or you have different opinions)?
Write an short paper to explain your conclusion. (The paper should contains introduction,
theoretical or simulation analysis, real data analysis and conclusion, and it should NOT
exceed 10 A4 pages.)