Vorlesen

Research Seminar Series in Statistics and Mathematics

Wirtschaftsuniversität Wien, Departments 4 D4.4.00812:30 - 14:00

Art Vortrag/Diskussion
SpracheEnglish
Vortragende/rEric Eisenstat (School of Economics, The University of Queensland, Brisbane, Australia)
Veranstalter Institut für Statistik und Mathematik
Kontakt katrin.artner@wu.ac.at

Eric Eisen­stat (School of Econo­mics, The Univer­sity of Queens­land, Bris­bane, Australia) about “Effi­cient Esti­ma­tion of Struc­tural VARMAs with Stochastic Vola­ti­lity”

The Insti­tute for Statis­tics and Mathe­ma­tics (Depart­ment of Finance, Accoun­ting and Statis­tics) cordi­ally invites ever­yone inte­rested to attend the talks in our Rese­arch Seminar Series, where inter­na­tio­nally renowned scho­lars from leading univer­si­ties present and discuss their (working) papers.

The list of talks for the summer term 2018 is avail­able via the follo­wing link:
Summer Term 2018

Abstract:

This paper deve­lops Markov chain Monte Carlo algo­rithms for struc­tural vector auto­re­gres­sive moving average (VARMA) models with fix coef­fi­ci­ents and time-va­rying error cova­ri­ances, modeled as a multi­va­riate stochastic vola­ti­lity process. A parti­cular benefit of allo­wing for time varia­tion in the cova­ri­ances in this setting is that it induces uniqueness in terms of funda­mental and various non-­f­un­da­mental VARMA repre­sen­ta­tions. Hence, it resolves an important issue in applying multi­va­riate time series models to struc­tural macro­eco­nomic problems. Although compu­ta­tion in this setting is more chal­len­ging, the condi­tio­nally Gaus­sian nature of the model renders effi­cient sampling algo­rithms feasible. The algo­rithm presented in this paper uses two inno­va­tive approa­ches to achieve sampling effi­ci­ency: (i) the time-va­rying cova­ri­ances are sampled jointly using particle Gibbs with ance­stry sampling, and (ii) the moving average coef­fi­ci­ents are sampled jointly using an exten­sion of the Whittle likelihood appro­xi­ma­tion. We provide Monte Carlo evidence that the algo­rithm performs well in prac­tice. We further employ the algo­rithm to assess the extent to which commonly used SVAR models satisfy their under­lying funda­men­tal­ness assump­tion and the effect that this assump­tion has on struc­tural infe­rence.



zurück zur Übersicht