A simple one-sided test when the covariance matrix has non-negative eigenvectors
Files
(Published version)
Date
2011
Authors
Lu, Z.H.
Editors
Advisors
Journal Title
Journal ISSN
Volume Title
Type:
Working paper
Citation
Statement of Responsibility
Conference Name
Abstract
This paper proposes a simple, but somewhere most powerful test involving a one-sided hypothesis. Our test is constructed under an additional condition imposed on the covariance matrix of the restricted function. The imposed condition is shown to be guaranteed in the case of no more than two restrictions, as well as in other cases such as when every element of the covariance matrix is positive. Most importantly, we show that our test is unbiased and consistent whereas some other somewhere most powerful tests proposed in the literature are biased and inconsistent. We also compare the performance of our test with that of the Chi-bar squared (X2) test. Our results suggest that our test is more powerful than the X2 test under certain alternatives. We discuss an application of linear restrictions in linear models, including simulation studies.
School/Discipline
Dissertation Note
Provenance
Description
Access Status
Rights
Copyright [2011] Zeng-Hua Lu. Not for distribution without permission of author.