Spring 2013 Q7 vs Spring 2016 Q1
Hi Graham - Here are two similar concept questions with a slight variation that I'm struggling to understand the reason why they are different. I did Spring 2013 Q7 first, felt good about it. Then eventually got to Spring 2016 Q1, solved it using a similar approach and realized it was wrong after checking the solution.
I get in Spring 2013 Q7, that we are provided with ultimate losses evaluated at pre-July 1, 2011 benefit levels. And I drew the parallelogram method to help me understand it graphically. My takeaway was that we did not need to pro-rate the benefit changes because by the time we hit Policy Year 2013 (time period for the rate indication) all of those benefit changes would be fully accounted for so use the full benefit amounts in on-leveling each accident year.
I applied a similar logic to Spring 2016 Q1, which I thought was appropriate since it asks you to calculate an on-leveled loss ratio for the purposes of a rate indication so by the time we hit Policy Year 2017, all our benefit changes would be fully accounted for. Only real difference for this problem is that ultimate losses provided are not at the pre-benefit level for each year. Where did I go wrong in my thinking and why is that statement such a difference maker? Thanks.
Comments
The 2013 problem was easier precisely because the given loss information was "pre-benefit". So to get to post-benefit levels, you can simply apply the given benefit level changes to all the data.
The difficulty with the data in the 2016 problem is that each year contains loss data at different benefit levels. In other words, you have to figure out what portion of the loss data for each year is at which benefit level and then make the adjustments accordingly. Much more difficult!
I think you're more likely to see a problem like the 2016 question on the exam. (The 2013 problem was way too easy for 3 points.)
Hi Graham - If "pre-benefit" is just another way of saying unadjusted historical losses then it's not clear to me why we wouldn't on-level the losses via parallelogram just as we would when we are provided unadjusted historical premiums and have to on-level the premium.
It might help if you are able to walk me through an example for premium if they provided "pre-rate change" earned premiums by calendar year, and when it's appropriate to on-level with the parallelogram-like approach vs applying rate change to all the data like they did in the 2013 problem. My concern for the exam is that they will use a similar concept on the premium side that provides data at pre-rate change (or benefit) levels and I'd easily get confused.
I feel like you're starting to go down a rabbit hole here so let me try to pull you back.
In general, stay focused on firmly understanding specific problems because future exam problems will almost always be variations on these. If you're having trouble understanding the solution to this 2016 problem, leave it for a week or so then come back to it. Things often make more sense after you've let it settle in your brain.
I realize I haven't specifically answered your question, but your question feels a little too abstract for me to be able to provide a useful answer. Like I said, the best strategy is to stay connected to specific examples because if you understand those, you can tweak those methods to fit whatever you see on the actual exam.
I'll try not to get too bogged down on this topic. But just so you know, I do understand using the parallelogram method in the 2016 problem and in general how to apply it to premium and losses. I don't fully grasp why it's not needed in the 2013 problem. And I suspect it has to deal with this notion of "pre-benefit" levels. But we can move on, thanks!
Oh, ok. That's good. I looks like you understand this topic very well.
I think you actually can apply the parallelogram method to the 2013 problem but because none of the data has been subjected to the new levels, the ALL (Average Loss Level) is 1.0 everywhere so it's super-simple to bring the losses to CLL (Current Loss Level). You just multiply everything by the current loss level.
Thanks! That comment on the Average Loss Level being 1.0 is exactly what I was missing. It makes so much more sense now.
Awesome! 😁