Local Power of Andrews and Ploberger Tests Against Nearly Integrated, Nearly White Noise Process

We find that the Andrews and Ploberger’s (1996) tests have unit local power against the nearly integrated, nearly white noise process (ref. Nabeya and Perron (1994)). Therefore, compared with the stationary local alternatives, higher power is expected when testing against such process. Monte Carlo simulation confirms our results. We apply the tests to monthly SP500 stock returns and strongly reject the martingale difference hypothesis.

MoreLess

Year of publication:	2006-05
Authors:	Ai Deng Author-X-Name-First: Ai
Institutions:	Department of Economics, Boston University
Subject:	ARMA(1 \| 1) \| local power \| Nearly integrated \| nearly white noise process \| stock returns

More details

Series:	Boston University - Department of Economics - Working Papers Series.
Type of publication:	Book / Working Paper
Notes:	Number WP2006-027 08 pages
Classification:	C12 - Hypothesis Testing
Source:	RePEc - Research Papers in Economics

Persistent link: https://ebvufind01.dmz1.zbw.eu/10005795220