B.06.Venter

Card Set Information

Author:
Exam8
ID:
166017
Filename:
B.06.Venter
Updated:
2012-08-14 14:42:22
Tags:
experience rating
Folders:

Description:
Experience Rating - Equity & Predictive Accuracy
Show Answers:

Home > Flashcards > Print Preview

The flashcards below were created by user Exam8 on FreezingBlue Flashcards. What would you like to do?


  1. Experience rating
    • using historical loss experience of insd to determine P
    • use latest 3yrs (skip most recent years, incomplete)
    • calculate E(L) based on insd size and manual classification rates
    • blend actual & expected losses
    • compare to expected losses to get credit/debit mod
    • apply mode to manual premium to get standard premium
  2. 3 goals of experience rating
    • safety incentive
    • predictive accuracy
    • enhances mkt competition
  3. Experience rating formulae
    • mod = (A + K) / (E + K), K = cred constant
    • (-) for skewed heavy tailed dist, substantial diff btwn un/constrained estimate
    • multi-split formula: ZpAp + (1-Zp)Ep + ZeAe + (1-Ze)Ee
    • set Ze = E / (E + K) and Zp = E / (E + B)
    • mod = (Ap + WAe + (1-W)Ee + B) / (E + B)
    • W = (E + B) / (E + K)
    • (+) primary & xs loss heavy tailed → more predictable
  4. B & K should increase for larger firms
    • inverse pattern was used to (1) prevent large swings for small insds and (2) provide self rating for largest insd
    • too much weight was allocated to exp of larger insds, too little for smaller
    • → small ind w credit mod preferable
    • → large firms w credit mod less profitable
  5. NCCI modifications to exp rating formula
    • B & K increase as prem incr
    • primary-excess split simplified (L<5K = primary)
  6. Testing performance of rating plan
    • identify risk diff: quintiles test on manual loss ratio, looking for incr LR by bucket w wide dispersion
    • correct risk diff: quintiles test on standard loss ratio, looking for equal LR by group, and no trend in data

What would you like to do?

Home > Flashcards > Print Preview