Affordable Access

Publisher Website

Score test of fit for composite hypothesis in the [formula omitted]model

Authors
Journal
Journal of Statistical Planning and Inference
0378-3758
Publisher
Elsevier
Publication Date
Volume
139
Issue
2
Identifiers
DOI: 10.1016/j.jspi.2008.04.033
Keywords
  • [Formula Omitted]Model
  • Conditional Distribution
  • Efficient Score Vector
  • Data-Driven Test Of Fit
  • Square-Root Consistent Estimator
  • Martingale Difference Array
  • Central Limit Theorem
  • Monte Carlo Simulations
  • Ged Family

Abstract

Abstract A score test of fit for testing the conditional distribution of the stationary GARCH ( 1 , 1 ) model conceived by Bollerslev [1986. Generalized autoregressive conditional heteroskedasticity. J. Econometrics 31, 307–327] is proposed. The null hypothesis asserting that the noise distribution belongs to the specified parametric class of distributions is considered. Exploiting the pioneer idea of Neyman [1937. Smooth test for goodness of fit. Skand. Aktuarietidskr. 20, 149–199] and the device proposed by Ledwina [1994. Data driven version of Neyman's smooth test of fit. J. Amer. Stat. Assoc. 89, 1000–1005], the efficient score statistic and its data-driven version are derived for this testing problem. The asymptotic null distribution of the score statistic is established. Replacing the nuisance parameters with their square-root consistent estimators results in the data-driven test statistic. It is proved that in that case the asymptotic behaviour of the test statistic remains unchanged under appropriate regularity conditions and under discretization of the estimators. Computer simulations of the critical value and the power performance of the test for several alternatives in the case of generalized error distribution family serving as a null distribution are also presented.

There are no comments yet on this publication. Be the first to share your thoughts.