Marginal Likelihood and Unit Roots

M.K. Francke, A.F. de Vos

    Research output: Contribution to JournalArticleAcademic


    We develop new tests for the hypothesis of unit roots that are based on the marginal likelihood of the general linear model. The marginal likelihood allows the incorporation of invariance arguments in the likelihood function. It turns out that marginal likelihood tests for unit roots appear to be more powerful than other unit root tests. For some basic models power functions almost coincide with the power envelopes, even in small samples. General correlation structures can be incorporated, either by standard likelihood procedures or by adjustments of the test statistics on the basis of asymptotic distributions. © 2006 Elsevier B.V. All rights reserved.
    Original languageEnglish
    Pages (from-to)708-728
    JournalJournal of Econometrics
    Publication statusPublished - 2007


    Dive into the research topics of 'Marginal Likelihood and Unit Roots'. Together they form a unique fingerprint.

    Cite this