Publication: Contrastes de momentos y de la matriz de información
Loading...
Files
Official URL
Full text at PDC
Publication Date
1994
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Facultad de Ciencias Económicas y Empresariales. Instituto Complutense de Análisis Económico (ICAE)
Citations
Exportar
Abstract
Se presentan los principales resultados de la literatura reciente sobre contrastes de momentos (m) y de la matriz de información, Los contrastes de momentos son un marco general para obtener diagnósticos de la especificación de modelos estimados bien por máxima verosimilitud o por el método de momentos. Los contrastes m pueden considerarse, bajo condiciones generales, contrastes de los multiplicadores de lagrange. Una fuente de condiciones de momentos en los que basar la construcción de los diagnósticos ro es la igualdad de la matriz de información. Se ilustra cómo, en el caso de regresión lineal, los contrastes basados en la igualdad de la matriz de información generan diagnósticos tanto conocidos como más novedosos. El comportamiento en muestras finitas de los contrastes es una consideración importante a la hora de su utilización, debiendo elegirse en cada circunstancia la versión más apropiada. Finalmente se señala el gran potencial de la igualdad de la matriz de información para generar baterías de diagnósticos para modelos para los cuales se dispone actualmente de una menor variedad de diagnósticos que para el caso de regresión lineal.
This paper presents Ihe main results of the recent literature on moment (m) tests and infonnation matrix tests. The m tests provide a general Cramework for deriving specification diagnostics fur models estimated by maximum IIkelihood or the method of moments. Under general conditions, m tests can be considered LM tests, A souree of moment conditions in which one could base the construction of m tests is the Information matrix equality. It is shown that in the case of linear regression, tests based on the information matrix equality generate a variety of diagnostics, some of them new while others are already familiar. Small sample considerations are important for the practical application of the tests; it is suggested that appropriate choice of the particular test will be an important practical decision. One of the most important implications of this framework is that il can generate batteries of diagnostics for models for which few diagnostics are presently available.
This paper presents Ihe main results of the recent literature on moment (m) tests and infonnation matrix tests. The m tests provide a general Cramework for deriving specification diagnostics fur models estimated by maximum IIkelihood or the method of moments. Under general conditions, m tests can be considered LM tests, A souree of moment conditions in which one could base the construction of m tests is the Information matrix equality. It is shown that in the case of linear regression, tests based on the information matrix equality generate a variety of diagnostics, some of them new while others are already familiar. Small sample considerations are important for the practical application of the tests; it is suggested that appropriate choice of the particular test will be an important practical decision. One of the most important implications of this framework is that il can generate batteries of diagnostics for models for which few diagnostics are presently available.
Description
UCM subjects
Unesco subjects
Keywords
Citation
Aitchison, J. Y S. D. Silvey (1958) «Estimation subject to restraints,» Annals of Mathernatical Statistics 29, 813-828.
Cavanagh, C. L. (1985) «Second order admissibility of likelihood based tests,» Harvard Institute of Economic Research Discussion papee 1148.
Chesher, A. y Spady, R. (1991) «Asymptotic expansions of the information matrix test statistic» Econometrica 59, 787-816.
COX, D. R. (1961) «Tests of separate families of hypotheses,» en Proceedings of the Fourth Berkeley Simnosium on Mathematical Statistics and Probability. Vol. 1 Berkeley: University of California Press, 105-123.
Cox, D. R. (1962) «Further results on tests of separate families of hypotheses». Journal of the Royal Statistical Society B, 406-424.
Davidson, R. y J. G. MacKinnon (1992) «A new form of the infonnation matrix test», Econometrica 60, 145-158.
Godfrey, L. G. (1988) Missnecification tests in econometrics, Econometric Society monograhs 16, Cambridge University Press.
Bausman, J. (1978) «Specification tests in econometrics», Econometrica 46, 1251-1272.
Kennan, J. y Neumann, G. R. (1988) «Why does tbe information matrix test reject too ofien? a diagnosis of some Monte Carla symptoms», Working Papers in Economics E-88-10, Hoover Institution, Stanford University.
Newey, W. (1985) «Maximum likelihood specification testing and conditional mament tests» Econometrica 53, 1047-1070.
Orme, C. (l990a) «The small sample peñormance of the informatian matrix test», Journal of Econometrics 46, 309-331.
Orme, C. (l990b) «Double and triple length regressions for the information matrix test and other conditional moment tests» University of New York, mimeo.
Pérez, T. (1989) «Dynamic information matrix tests and m tests: new results, simulations and an application», University of California San Diego Department of Economics Doctoral Dissertation.
Pérez, T. (1993a) «contrastes m y de la matriz de información dinámica con una aplicación a regresión lineal» Estadística Española. Vol 35, 133,291-317.
Pérez, T. (1993b) «El comportamiento en muestras finitas de los contrastes m y de la matriz de información dinámica: simulaciones», Estadística Española, Vol. 35, 132, 123-139.
Pérez, T. (1993c) «Comentario a "Guía para la estimación de contrastes ARCH"» Estadística Española. Vol 35, 132,64·67.
Rao, C. R. (1947) «Large sample tests of statistical hypotheses concerning several parameters with applications to problems of estimatian» Proceedings oí the Cambridge Philosophical Society 44, 50-57.
Taylor, L. W. (1987) «The size bias of the information matrix test», Economics Letters 24,63-67.
Tauchen, G. (1985) «Diagnostic testing and evaluation of maximum likelihood models», Journal of Econometrics 30, 515-444.
Theil, H. (1971) Principles of Econometrics. New York. Wiley.
White, H. (1982) «Maximum likelihood estimation of misspecified nodels», Econometrica 50, 1-25.
White, H. (1987) «Specification testing in dynamic models», en Advances in Econometrics, Fifth World Congress, Vol. 1, Bewley, T. (editor), New York, Cambridge University Press, 1-58.
White, H. (1994) Estimation Inference and Specification Analysis, Cambridge University Press.
Wooldridge, J. (1986) «Asymptotic Properties of Econometrics Estimators» University of California San Diego Department of Economics Doctoral Dissertation.
Wooldridge, J. (1990) «A unified approach to robust, regression-based specification tests», Econometric Theory 6, 17-43.