Unit:
Κατεύθυνση Στατιστική και Επιχειρησιακή ΈρευναLibrary of the School of Science
Author:
Georgopoulou Vasiliki
Supervisors info:
Μελιγκοτσίδου Λουκία, Αναπληρώτρια καθηγήτρια, Τμήμα Μαθηματικών, ΕΚΠΑ
Μπουρνέτας Απόστολος, Καθηγητής, Τμήμα Μαθηματικών, ΕΚΠΑ
Σιάννης Φώτιος, Επίκουρος καθηγητής, Τμήμα Μαθηματικών, ΕΚΠΑ
Original Title:
Bayesian unit root testing
Translated title:
Bayesian unit root testing
Summary:
The current dissertation analyzes the Bayesian unit root tests in autoregressive processes
as an alternate to the classical autoregressive unit root tests. The Bayesian approach
to unit root testing was mainly motivated by the power and size distortions
of the classical tests under the Dickey-Fuller distribution. Initially, basic principles
and theory concerning time series and Bayesian analysis are introduced which are
then followed by the structure of the classical autoregressive tests and the Dickey-
Fuller distribution. The subsequent chapter is devoted to the methods applied in
the Bayesian unit root testing. Simulation results and conclusions based on two
Bayesian methods are finally reported.
Main subject category:
Science
Keywords:
unit root testing, stationarity, autoregressive models, bayesian methods