Long memory and fractional cointegration with deterministic trends

We discuss the estimation of the order of integration of a fractional process that may be contaminated by a time-varying deterministic component, or subject to a break in the dynamics of the zero-mean stochastic component, and the estimation of the cointegrating parameter in a bivariate system gener...

Full description

Bibliographic Details
Main Author: Iacone, Fabrizio
Published: London School of Economics and Political Science (University of London) 2006
Subjects:
Online Access:http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.428012
id ndltd-bl.uk-oai-ethos.bl.uk-428012
record_format oai_dc
spelling ndltd-bl.uk-oai-ethos.bl.uk-4280122015-06-03T03:22:51ZLong memory and fractional cointegration with deterministic trendsIacone, Fabrizio2006We discuss the estimation of the order of integration of a fractional process that may be contaminated by a time-varying deterministic component, or subject to a break in the dynamics of the zero-mean stochastic component, and the estimation of the cointegrating parameter in a bivariate system generated by fractionally integrated processes and by additive polynomial trends. In Chapter 1 we review the theoretical literature on fractional integration and cointegration, and we analyse a situation in which a fractional model reconciles two apparently conflicting economic theories. In Chapter 2 we consider local Whittle estimation of the order of integration when the process is contaminated by a deterministic trend or by a break in the mean. We propose a simple condition to assess whether the asymptotic properties of the estimate are unaffected by the time-varying mean, and a test, with asymptotically normal test statistic under the null, to detect if that condition is met. In Chapter 3 we discuss local Whittle estimation when the zero-mean stochastic component is subject to a break: we show that the estimate is robust to instability in the short term dynamics, while in presence of a break in the long term dynamics only the highest order of integration is consistently estimated. We propose a test to detect that break: the limit distribution of the test statistic under the null is not standard, but it is well known in the literature. We also propose a procedure to estimate the location of a break when it is present. In Chapter 4 we consider a cointegrating relation in which a nonstationary, bivariate process is augmented by a deterministic trend. We derive the limit properties of the Ordinary Least Squares and Generalised Least Squares estimates: these depend on the comparison between the deterministic and the stochastic components.330.015195London School of Economics and Political Science (University of London)http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.428012http://etheses.lse.ac.uk/1937/Electronic Thesis or Dissertation
collection NDLTD
sources NDLTD
topic 330.015195
spellingShingle 330.015195
Iacone, Fabrizio
Long memory and fractional cointegration with deterministic trends
description We discuss the estimation of the order of integration of a fractional process that may be contaminated by a time-varying deterministic component, or subject to a break in the dynamics of the zero-mean stochastic component, and the estimation of the cointegrating parameter in a bivariate system generated by fractionally integrated processes and by additive polynomial trends. In Chapter 1 we review the theoretical literature on fractional integration and cointegration, and we analyse a situation in which a fractional model reconciles two apparently conflicting economic theories. In Chapter 2 we consider local Whittle estimation of the order of integration when the process is contaminated by a deterministic trend or by a break in the mean. We propose a simple condition to assess whether the asymptotic properties of the estimate are unaffected by the time-varying mean, and a test, with asymptotically normal test statistic under the null, to detect if that condition is met. In Chapter 3 we discuss local Whittle estimation when the zero-mean stochastic component is subject to a break: we show that the estimate is robust to instability in the short term dynamics, while in presence of a break in the long term dynamics only the highest order of integration is consistently estimated. We propose a test to detect that break: the limit distribution of the test statistic under the null is not standard, but it is well known in the literature. We also propose a procedure to estimate the location of a break when it is present. In Chapter 4 we consider a cointegrating relation in which a nonstationary, bivariate process is augmented by a deterministic trend. We derive the limit properties of the Ordinary Least Squares and Generalised Least Squares estimates: these depend on the comparison between the deterministic and the stochastic components.
author Iacone, Fabrizio
author_facet Iacone, Fabrizio
author_sort Iacone, Fabrizio
title Long memory and fractional cointegration with deterministic trends
title_short Long memory and fractional cointegration with deterministic trends
title_full Long memory and fractional cointegration with deterministic trends
title_fullStr Long memory and fractional cointegration with deterministic trends
title_full_unstemmed Long memory and fractional cointegration with deterministic trends
title_sort long memory and fractional cointegration with deterministic trends
publisher London School of Economics and Political Science (University of London)
publishDate 2006
url http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.428012
work_keys_str_mv AT iaconefabrizio longmemoryandfractionalcointegrationwithdeterministictrends
_version_ 1716804812518981632