B3-B4. NCCI & ISO experience rating plans

Question 1

Why NCCI made the 3 adjustments in 1998

Answer 1

1. reduced medical only losses
   -reduces financial incentive for an employer to not report med only losses in an effort to manipulate its Mod.
   Since Med only are very small portion of total loss will not affect predictive accuract
2. increase excess loss weights
   - increases incentive for safety due to Mod more responsive to experience
   - Recognized xs loss are predictive of severity
3. Inflation sensitive split point
   - prevents the impact of primary losses on Mod from eroding over time due to higher portion being in excess layer

Question 2

Why a no-split plan does not apply well to work comp

NCCI

Answer 2

Problem:

• no-split plan forces the Mod to be a linear function of the losses
• but work comp losses have heavy tails
• therefore, no-split plan produce greater squared error in estimations

Solution:

• split losses into primary and excess
• then the linear constraint applies separately on primary and excess losses
• therefore, this produces smaller squared error in estimations

Question 3

Impacts on credibility when NCCI moved from multi-split to single-split plan

Answer 3

Small risks:

• increased primary credibility
• increased excess credibility => even the smallest risks have some excess credbility (before they had 0 excess credibility)

Large risks:

• decreased primary credibility
• increased excess credibility

Question 4

How NCCI stabilizes losses in experience rating

Answer 4

1. limit to any single accident, multiple claimant, and to the range of Mod
2. split losses between primary/excess
3. credibility weight actual/expected
   - give more weight to larger risks actual experience
   - do not do that for small risk bc : the larger the risk , the more stable their expected loss are, if actual experience for small risks was given the same weight as for large risks, then the mods for small risks would be erratic
   - stabilize using a ballast value

Question 5

What are the assumptions behind K and Ballast being constant ( NCCI)

Answer 5

1. Large risks are more stable than small ones
2. The increased stability follows from the law of large numbers
   * *3. Variance of LR is inversely proportional to premium

These assumptions are not true

Question 6

What are the problems with the assumptions of K and Ballast being constant ( NCCI)

Answer 6

1. the variance does not decrease that fast with larger risks
2. larger risks display more variation than the law of large numbers would predict

by exemple : WC loss experience does not follow
the law of large numbers due to the long tailed nature
of the business. Also, large insureds have more loss
ratio variance due to diverse operations and exposures

Thus, large risks would have too much credibility and small risks too little

Small risk with credit mod will become preferred
business because their premium is higher
because of low credibility.
Because of that, in a competitive market, rates will
go down for those preferred risks (demand
stays the same while offer increases will push
down prices).
The same will happen to large risk with debit mod
(they will pay too much, offer increases while demand stays equal, and rates decrease).

Question 7

Why apply basic limits to loss and also MSL to loss+ALAE

ISO

Answer 7

1. prevents random large loss not predictive of futur loss from having too much influence on the mod factor
2. makes the rating plan more responsive to frequency of losses rather than severity

Question 8

Why MSL increases when size of risk increases

ISO

Answer 8

larger risk => larger expected loss

If MSL increases with size of risk:

• allows more credibility to actual exp
• proportion uncapped / capped remains stable => impact on the mod factor stable

Question 9

What are the circumstances when entities are combined for Experience Rating?

What is the purpose of such combination?

Answer 9

When entities share common majority ownership, they are combined. This prevents the ownership from moving loss experience from one company to another for purposes of manipulating the experience mod.

Question 10

An actuarial student learning about the NCCI experience rating plan makes a suggestion to replace all Fatal claims used in the experience rating calculation with an average value of $50,000. Provide 2 responses to the student defending why this change should not be made

Answer 10

• A loss limit already exists in the plan, which limits the impact of large Fatal claims.
• NCCI studies have shown that excess losses are predictive, so making this change might hurt the predictive value of the plan.
• This change might lower the safety incentive for insureds that have Fatal claims above the average value.
• This change would mean that insureds with a Fatal loss in their experience that was lower than the proposed average amount would be charged for losses above their actual experience, which would not be fair.