Bayesian Value-At-Risk with Product Partition Models

ΟΙΚΟΝΟΜΙΚΟ ΠΑΝΕΠΙΣΤΗΜΙΟ ΑΘΗΝΩΝ

ΤΜΗΜΑ ΣΤΑΤΙΣΤΙΚΗΣ

ΠΕΜΠΤΗ 10/4/2014

12:00 – 13:00

ΑΙΘΟΥΣΑ 607, 6ος ΟΡΟΦΟΣ,

ΚΤΙΡΙΟ ΜΕΤΑΠΤΥΧΙΑΚΩΝ ΣΠΟΥΔΩΝ

(ΕΥΕΛΠΙΔΩΝΛΕΥΚΑΔΟΣ)

Erengul(Ozkok)Dodd

Faculty ofMathematical Sciences
University of Southampton

Modelling the time delay in Critical Illness Insurance using GB2 family of

distributions

ΠΕΡΙΛΗΨΗ (ΣΤΑΑΓΓΛΙΚΑ)

Critical Illness Insurance (CII) is subject to long delays between the diagnosis of a claim and the corresponding settlement. Consideration of the uncertainty associated with different models used in the estimation of the delay between diagnosis and settlement is crucial for financial purposes, especially when the date of diagnosis is missing from the data records. Using UK data supplied by the Continuous Mortality Investigation, we account for uncertainty in CII modelling by exploring various claim delay distributions such as generalised beta of the second kind (GB2) family of distributions in a generalised linear-type model setting. The non-recorded dates of diagnosis are included in the analysis as missing values using their posterior predictive distribution. The models are fitted under a hierarchical Bayesian framework using Markov chain Monte Carlo methodology. Variable selection using Bayesian methodology to obtain the best model with different prior distribution setups for the parameters is also discussed for a wide selection of error distributions.

ATHENSUNIVERSITY OF ECONOMICS & BUSINESS

DEPARTMENT OF STATISTICS

Thursday 10/4/2014

12:00 – 13:00

ROOM607, 6th FLOOR,

POSTGRADUATESTUDIESBUILDING

(EVELPIDON & LEFKADOS)

Erengul(Ozkok)Dodd

Faculty ofMathematical Sciences
University of Southampton

Modelling the time delay in Critical Illness Insurance using GB2 family of

distributions

ABSTRACT

Critical Illness Insurance (CII) is subject to long delays between the diagnosis of a claim and the corresponding settlement. Consideration of the uncertainty associated with different models used in the estimation of the delay between diagnosis and settlement is crucial for financial purposes, especially when the date of diagnosis is missing from the data records. Using UK data supplied by the Continuous Mortality Investigation, we account for uncertainty in CII modelling by exploring various claim delay distributions such as generalised beta of the second kind (GB2) family of distributions in a generalised linear-type model setting. The non-recorded dates of diagnosis are included in the analysis as missing values using their posterior predictive distribution. The models are fitted under a hierarchical Bayesian framework using Markov chain Monte Carlo methodology. Variable selection using Bayesian methodology to obtain the best model with different prior distribution setups for the parameters is also discussed for a wide selection of error distributions.