Il giorno giovedì 16 febbraio alle ore 16 presso il Dipartimento di
Economia dell'Università di Perugia, la Dott.ssa Zehra Eksi, Assistant
Professor presso la WU-Vienna University of Economics and Business,
e ospite del Dipartimento di Economia in qualità di professore visitatore
nell'ambito del progetto YITP (Young Investigation Training Program), terrà
un seminario dal titolo
"EM Algorithm for Markov Chains observed via Gaussian Noise and Point
Processes Information"
*Abstract*:
In this paper we obtain an Expectation-Maximization (EM) algorithm for a
setting in which the state variable follows a finite-state Markov chain
observed via diffusive and point process information. Such a setting may
arise, for example, in the modelling of sovereign credit risk where
information stems from default history (jump component) and CDS spreads
(diffusive component). Obtaining the EM algorithm amounts to the derivation
of finite-dimensional filters for quantities such as number of jumps and
sojourn times associated to the Markov chain. In this context, we obtain
both exact and unnormalized filters for the related quantities. Moreover,
we compute discretized robust versions of the unnormalized filters. Next,
we introduce a novel goodness of fit test to check how well the estimated
model explains a given data set. Finally, we run a simulation study to test
speed and accuracy of the algorithm. In particular, we provide a comparison
for the estimates resulting from the robust and naive discretization and
the value of point process information (This is a Joint work with Camilla
Damian and Ruediger Frey)
Tutti gli interessati sono invitati a partecipare.
Dott.ssa Katia Colaneri
Department of Economics, University of Perugia
Via A. Pascoli 20, Perugia 06123- Italy
e-mail: katia.colaneri@unipg <katia.colaneri(a)gmail.com>.it