Trending claims question

[ciza5]

When using the burning cost (or frequency severity approach) the notes outline a two step approach to trending claims,
1.inflate to current values
2. Inflate to midpoint of exposure period which rates apply

Does this exposure period in part two refer to average claim date rather than average policy inception date? If so is it always avegrage claim date we inflate to?

Thanks,

 

[Darren Michaels]

You should always inflate from the average date of claim payment in your historical data to the average date of claim payment for claims that will arise from the period in which the premium rates you are deriving will apply.

In some questions you can ignore the delay between the claim event being notified/reported to the insurer and the claim being settled (the settlement delay) or at the very least you are assuming that the settlement delay is unchanged, but make sure that you write this down as one of your assumptions!

 

[Hello1991]

Darren, if there is a settlement or reporting delay given in the question as you mentioned below, we then project from the average payment date in the historical data to the new average payment for claims that will arise from the period in which the premium rates you are deriving will apply.
So therefore we do not project from old average claim date to the new average claim date in case there is a delay of some kind. Hence we always project from average payment dates. Is this correct?

 

[Darren Michaels]

Hi Hello 1991

Yes strictly speaking you project from average payment date to average payment date, especially if the question gives you information about reporting and settlement delays.

If the question doesn't give you anything on these then you need to make suitable assumptions. In some cases you may be able to assume no event, reporting or settlement delays, which means you can project from date of claim event to date of claim event.

 

[Hello1991]

Thanks for that Darren.

I just have a similar query in relation to adjusting data for 12 months of data. In Q6 ST8 April 2012, the data is only given for 10 months of the most recent underwriting year. Therefore an adjustment is made to premium data to gross it up. There is no adjustment to incurred claims data as we are given the development profile at the same date.

By contrast for the example question in the revision booklet (Subject 9B 1992, Q1,paper 2), the data for the most recent year is also not fully compiled. However there is an adjustment made to the claims amounts and number of claims here, which is not consistent with the above? Is there a general rule of what adjustments to make if data for most recent year is not fully compiled? Thanks for your help.

 

[code9063]

In CH12 Q12.8 claims are projected to the average occurrence date although there is a 10 month delay. What's the reason for this?

 

[Darren Michaels]

Hi there

I think the answer to this really depends on what information/data you are given in the question.

In ST8 April 2012 Q6, you did not need to "manually" adjust the claims data because the development pattern you were provided with allowed for the fact that the claims data was to 31 March 2012 in each case.

In 9B 1992, Q1,paper 2, you needed to manually adjust both the number of claims and the total claims cost because this was a shorter-tailed class where you weren't given any development pattern to apply.

Remember whatever you do you need to make sure that your claims and exposure data are consistent. For accident years you should look at incurred/ultimate claims on earned exposures and earned premiums and for underwriting years you need to look at ultimate claims (on earned and unearned exposures) as well as written premiums.