Portmanteau statistics for partially nonstationary multivariate AR and ARMA models
by Tai Man Tang
viii, 64 leaves : ill. ; 30 cm
This thesis studies the residual autocorrelation functions of partially nonstationary multivariate autoregressive models and partially nonstationary multivariate autoregressive moving-average models. Their limiting distributions are derived under full rank and reduced rank estimations for both models. These limiting distributions are used to construct the portmanteau statistics. It is shown that these statistics asymptotically follow same x2 distributions. Some simulation results are reported and two numerical examples are given to illustrate the methods.
Permanent URL for this record: https://lbezone.hkust.edu.hk/bib/b803235