Please enable JavaScript.
Coggle requires JavaScript to display documents.
CREDIBILITY (Comparaison of GACT vs LFCT (Z credibility estimator…
CREDIBILITY
-
Types of cred approach
GACT
1) Minimizes estimator for : EX(X), x= (estimator - actual )^2
2) Optimizes credibility weigh by examing
a) accuracy of relevent experience data
b) variability of subject experience data
LFCT
1) Actuary sets p and r
2) Partial credibility: same variance as full credibility, weight of relevel experience reduces variance because assumed accurate
3) Adjustment approaches
a) at each x - requires a lot of experience
b) aggregate - most common
-
-
Considération pour l'étude
1) At least 3 to 5 years of years of experience
2) Focus on most relevant ages - in pay status
3) Use appropriate relevant experience data
Critère dans selection de relevent experience data: :warning:
1) Should be recent
2) consistent with subject data '
a) gender, industry,collar
b) reflects improvement to time of study
c) shape
4) both should be either benefit or count dtaa
Ways to compare shapes
1) Ratio: actual to expected for each age
2) rates : standard vs actual at each age
If not same shape :
1) Sub groups : gender collar etc
2) age groupings
When update the study: :warning:
1) mortality studies trend/updates
2) Significant time since last stuy
3) Significant changes to
a) plan design (ex lump sums)
b) population
c) company specific shocks
LFCT
1) Ça prend 1082 décès pour que cred soit 100% selon probabilité de 90% et interval d'erreur de 5%
2) on veut que PR( (Estimé - valeur réelle) < estimé * erreur) ) > 90%
basé sur count weight