J
jonathans
Member
1. why should 200 claims and not 100 claims be used?
2. Where was is stated that they meant assumptions in pricing and not the standard credility theory assumptions?
Thanks!
-
We are pleased to announce that the winner of our Feedback Prize Draw for the Winter 2024-25 session and winning £150 of gift vouchers is Zhao Liang Tay. Congratulations to Zhao Liang. If you fancy winning £150 worth of gift vouchers (from a major UK store) for the Summer 2025 exam sitting for just a few minutes of your time throughout the session, please see our website at https://www.acted.co.uk/further-info.html?pat=feedback#feedback-prize for more information on how you can make sure your name is included in the draw at the end of the session.
1. why should 200 claims and not 100 claims be used?
2. Where was is stated that they meant assumptions in pricing and not the standard credility theory assumptions?
Thanks!
Hemant Rupani
Senior Member
1. Because partial credibility from classical credibility model is based on expected numbers... (you can check derivation given in the book). Derivation could not be possible with Actual Number of claims.
2. as it is given to using classical credibility theory there is no need to make the assumptions on credibility. I can recall from some past paper examiners' report where they stated that students should not make the assumptions for which full information are already given in the question.
2. as it is given to using classical credibility theory there is no need to make the assumptions on credibility. I can recall from some past paper examiners' report where they stated that students should not make the assumptions for which full information are already given in the question.
Q
Qayanaat
Member
Hi,
On the first question asked above by Jonathans:
Agree that the number of claims n should be the expected number of claims but since classical asks how much weight do you give to past data, should Z not be a function of expected #claims from past data?
Tailoring this to this qu: Since classical asks how much weight do you give to the £3200 (in this example this was based on 100 claims), should Z not be dependent on the fact that the £3200 was based on 100 claims?
Agree that that may not be the expected value, but in this case, could we not make the assumption that the observed number of claims is approx. expected value?
Secondly, the expected number of claims here is = 20% * 1000; where the 1000 is the number of vehicles to insure (i.e. not yet insured). Why are we basing the value of Z based on what the insurer is yet to insure?
Doing it this way would imply that we would give more weight to our past data (i.e. to the £3,200 estimate) if we were to insure a higher number of vehicles in the future, like let's say, we were to insure 2,500 vehicles, Z would be 96%, despite quality/quantity of past data not changing.
Appreciate all the help on this, thank you.
On the first question asked above by Jonathans:
Agree that the number of claims n should be the expected number of claims but since classical asks how much weight do you give to past data, should Z not be a function of expected #claims from past data?
Tailoring this to this qu: Since classical asks how much weight do you give to the £3200 (in this example this was based on 100 claims), should Z not be dependent on the fact that the £3200 was based on 100 claims?
Agree that that may not be the expected value, but in this case, could we not make the assumption that the observed number of claims is approx. expected value?
Secondly, the expected number of claims here is = 20% * 1000; where the 1000 is the number of vehicles to insure (i.e. not yet insured). Why are we basing the value of Z based on what the insurer is yet to insure?
Doing it this way would imply that we would give more weight to our past data (i.e. to the £3,200 estimate) if we were to insure a higher number of vehicles in the future, like let's say, we were to insure 2,500 vehicles, Z would be 96%, despite quality/quantity of past data not changing.
Appreciate all the help on this, thank you.
The square root rule is based on the expected number of claims and we often use the observed number of claims as an estimate for that.
Remember we are trying to come up with a premium for the 1000 vehicles we are going to insure, so the expected number of claims should be based on what we are going to insure not what we have insured in the past.
Remember we are trying to come up with a premium for the 1000 vehicles we are going to insure, so the expected number of claims should be based on what we are going to insure not what we have insured in the past.
Q
Qayanaat
Member
Hi Darren,
Okay thank you, I did not realise that the expected number of claims is based on what we are going to insure, and not based on past data.
Can you help me understand why that makes sense please?
As mentioned above, this method implies giving more weight to our past data if we were to insure a higher number of vehicles in the future, despite quality/quantity of past data not changing. Why is that?
Thank you.
Okay thank you, I did not realise that the expected number of claims is based on what we are going to insure, and not based on past data.
Can you help me understand why that makes sense please?
As mentioned above, this method implies giving more weight to our past data if we were to insure a higher number of vehicles in the future, despite quality/quantity of past data not changing. Why is that?
Thank you.
Hi Qayaant
Remember with classical credibility theory we are trying to work out how much weight we put on the past claims data for the risk we are pricing versus how much weight we should put on the ancillary data from other (hopefully similar!) risks.
Once we have sufficient past data to achieve the standard for full credibility we can determine our premium solely based on the data for that risk and we do not need to use the ancillary data.
If we haven't got enough past data for the risk of interest, we use the square root rule to help decide how much weight we should apply to the past data for the risk of interest versus the ancillary data.
Remember with classical credibility theory we are trying to work out how much weight we put on the past claims data for the risk we are pricing versus how much weight we should put on the ancillary data from other (hopefully similar!) risks.
Once we have sufficient past data to achieve the standard for full credibility we can determine our premium solely based on the data for that risk and we do not need to use the ancillary data.
If we haven't got enough past data for the risk of interest, we use the square root rule to help decide how much weight we should apply to the past data for the risk of interest versus the ancillary data.