Testing Misspecification in Vector Time Series Models with Exogenous Variables
Journal of the Royal Statistical Society: Series B (Statistical Methodology) · 1984 · Volume 46 · Issue 2
DOI: 10.1111/j.2517-6161.1984.tb01303.x
Abstract
In this paper we consider the application of hypothesis testing procedures toARMAXsystems. The score, or Lagrangian multiplier, principle is used to obtain mis-specification tests appropriate to these models. The structure of the information or sensitivity matrix is investigated in order to discuss the identifiability of alternative models and is used to show the existence of a set of admissible alternatives that are locally equivalent to one another. This notion is then developed to suggest why the score test is to be regarded as a pure significance test.
Keywords
autoregressive, moving average, exogenous variables, misspecification test, identifiability, diagnostic check
Enable JavaScript for citation tools and additional paper interactions.