Testing for Long Memory Against ESTAR Nonlinearities
Autor: Heri Kuswanto, Philipp Sibbertsen
Nummer: 427, Aug 2009, pp. 41
JEL-Class: C12, C22
Abstract:
We develop a Wald type test to distinguish between long memory and ESTAR nonlinearity by using a directed-Wald statistic to overcome the problem of restricted parameters under the alternative. The test is derived from two basic model specifications where the first is the standard model based on an auxiliary regression and the second allows the parameter to appear as a nuisance parameter in the transition function. A simulation study indicates that both approaches lead to tests with good size and power properties to distinguish between stationary long memory and ESTAR. Moreover, the second approach is shown to have more power.
Zusammenfassung:
/N
Diskussionspapier als PDF-Datei herunterladen
BibTeX-Datensatz herunterladen