ch8 doubts

keshav1996keshav1996
in F-08-ExpectedClms
  1. One of the situations where expected claims method works is when there's an operational change (and historical data may no longer be relevant). But, in such case, then how do the ultimate claims and the ECRs thus calculated (based on historical data in paid/reported triangle) make sense for selecting a final apriori LR for expected claims method?
  2. When both EP and OLEP are given in data and we are asked to calculate ultimate claims for an intermediate AY (and NOT the most recent AY, so EP and OLEP can differ) using expected claims method, should we multiply final selected ECR with EP or OLEP of that AY (and why)?
  3. Pop Quiz B doubt (1) - Suppose we have data for AYs 2020-2025 and we wish to select ECR for 2024 using simple average of 3 most recent years' ECR. Why should we take AY 2025's ECR (a future year) in the average?
  4. Pop Quiz B doubt (2) -An as extension, suppose we have to find ultimate claims and thus select ECR for both - AY 2024 and AY 2025 using same criteria as above. Now, which 3 years' ECRs will we select in the average (assuming there's no trend) for each of the AYs- 2024 and 2025 (and why)?

Thanks

Keshav

Comments

  • GrahamGraham

    Hi Keshav,

    1. You're right that using the historical data to calculate the ECR when there has been an operational change doesn't fully make sense. What would be done in such a situation is to first calculate an ECR with historical data, but recognize that it may not be accurate and adjust it using judgment. For example, you might predict that the operational change would cause the loss ratio to decrease by 5% so that's the adjustment you would make to the ECR. It is probably still not very accurate but there aren't many good options when changes render your historical data invalid.
    2. Is there a particular exam problem you are referring to where this happened? It's easier for me to make sure I accurately address your question if we can refer to a specific situation.
    3. Reserving is not an exact science so there's no single correct answer. The reason you might use AY2025 when estimating AY2024 is just that AY2025 is more recent than going all the way back to AY2022. If you thought AY2025 was very different from AY2024 then maybe you wouldn't use it. In a real-life reserving situation there are many things to consider. This pop quiz is just a test to make sure you understand the basics of how to do the method.
    4. Again there's no single correct answer. You could do AY2024 first using 21-23 then use 22-24 to estimate 25. Or just base the ECR on 21-23 then use that to estimate both 24 and 25. The key is to do what makes sense for the given situation. On the exam, as long as you give a valid reason for your choices, you would get credit. There is often more than one acceptable way to solve a reserving problem since it's all based on estimation and judgment anyway.


  • keshav1996keshav1996

    Hi Graham,

    1. This relates to Spring 2014 (Q19) - https://www.battleacts5.ca/pdf/Exam_(2014_1-Spring)/(2014_1-Spring)_(19).pdf

    However, the only difference is that we've originally been asked in question to calculate ultimate claims for the latest year where OLEP and EP are bound to be same. My question is - what if we have to calculate ult claim for a year other than the most recent year (and thereby OLEP and EP can be different)? In that case should the final selected ECR be multiplied by OLEP of that year or EP?

    (According to me that depends on the 'as at date' given, irrespective of the year. So, in the above ques if we were to calculate ultimate claims for 2012 (instead of 2013), but as at 31 Dec 2013, we must still use the OLEP (and NOT EP) of 2012 to reflect the recent rates as on given date. But I'm not sure. Please correct me.)

  • GrahamGraham

    Hmm...I think I would probably do as follows: (I think you have the right idea. I just need to think through the steps myself.)

    • eliminate 2012 data from consideration
    • use the development method on the remaining data in the "normal way" to get non-trended ultimate losses for 2010, 2011, 2013
    • now trend those losses to Dec 31, 2013, use OLEP for 2010, 2011, 2013, then select an appropriate ECR at Dec 31, 2013 to use for 2012
    • multiply this 2012 ECR by the OLEP for 2012 to get ultimate losses for 2012 at Dec 31, 2013
    • detrend those losses back 1 year to 2012 and that's the answer

    Of course in this particular problem, you didn't have to do any of that because they wanted you to calculate 2013. I think they gave you both EP and OLEP to test whether you knew to use OLEP. (The regular EP was not used in the solution.)

Sign In or Register to comment.