Estimation of a Time-Varying Parameter from Long Range Dependent Data

Abstract

We consider a time series regression with a time-varying parameter and long-range dependent data. No restriction is placed on the behavior of the time-varying parameter, allowing for both smooth changes and abrupt breaks.The time-varying parameter is estimated by a nonlinear orthogonal series estimator which is shown to have mini max estimation error and no spurious jumps in the large sample limit.