‹#›
Modeling Challenges: Structural limitations in frequency calculation
>Frequency of operational loss events is generally country specific and
particularly institution
specific –No institution
would have large history of operational losses or it would not be there!
>Therefore, internal data needs to be
combined with external data in order to establish reliable probabilities–External operational data may distort
complete calculations
and calculated
probabilities may reflect a picture which has nothing to do with institution!
>Even in presence of external data,
frequency of high impact events is too low to model some credible statistical pattern – Tail prediction dilemma!