Use of semi-parametric methods to estimate the fractional parameter model (O, d, O) ARFIMA   Theoretical and experimental study

Abstract

The property of the stationarity time series models plays an important role in building these models . And when the condition of stationary is not hold , then the positive integer differences of the observation are used . But in some time series , the effect of (d) parameter on distant observations decays hyperbolically as the lag increases which is called as long –term processes . In this study , the fractional differences will be dealt with this model which lie in the closed interval (0.5 , 0.5) . The goal of this thesis is to study ARFIMA (0 , d , 0) model by using the method of estimating of fractional parameter in semi parametric approach . In addition to , a comparison between these methods by using the simulation technique .