Actuarial Outpost
 
Go Back   Actuarial Outpost > Exams - Please Limit Discussion to Exam-Related Topics > CAS > CAS Exams > Exam 8 - Advanced Ratemaking
FlashChat Actuarial Discussion Preliminary Exams CAS/SOA Exams Cyberchat Around the World Suggestions

ACTUARIAL SALARY SURVEYS
Contact DW Simpson for a Personalized Salary Survey

Reply
 
Thread Tools Search this Thread Display Modes
  #1  
Old 10-26-2013, 03:04 PM
mrdomer07 mrdomer07 is offline
Member
CAS
 
Join Date: Oct 2007
Studying for 8
College: Notre Dame
Favorite beer: Goose Island BCBS Cherry Rye
Posts: 284
Default 2012 Q5 vs 2011 Q2 - Couret & Venter

In 2011 Q2, we are asked to use SSE to evaluate whether the credibility technique is better than the raw data. In the solution, we compute the SSE of the credibility procedure vs. the holdout and the SSE of the raw data vs. the holdout. The former has a smaller SSE so we choose the credibility procedure.

In 2012 Q5, we are asked to determine whether the credibility procedure should be used. However, in this solution we compute the SSE of the credibility procedure vs. the holdout and the SSE of the holdout vs. 1.00 (which I presume is the relativity of Hazard Group F fatal claims in total by definition). The latter wins so we do not use the credibility procedure.

Why do we compare raw data vs. holdout in one solution and holdout vs. 1.00 in the other? This has bugged me for awhile and I've gone back through Couret & Venter a few times but can't seem to find why these problems are different.
Reply With Quote
  #2  
Old 10-26-2013, 03:24 PM
chicken_po_boy's Avatar
chicken_po_boy chicken_po_boy is offline
Member
CAS
 
Join Date: Nov 2008
Location: 401 Poydras New Orleans, LA 70130
Studying for Nothing Ever Again
College: been there, done that
Favorite beer: You gonna finish that?
Posts: 2,102
Default

I see your point. I would have compared to the HG average in both cases and ignored the raw. But perhaps the most complete way to answer is to make both comparisons: cred vs. HG avg (1.00), and cred vs. raw.

Raw should almost always have higher SSE, because the frequencies are so volatile the sample means make horrible estimates.
Reply With Quote
  #3  
Old 10-27-2013, 10:32 AM
dmbaldwi dmbaldwi is offline
Member
CAS
 
Join Date: Feb 2012
Posts: 94
Default

If C&V had used the even as hold out and odd as basis for cred procedure, wouldn't the raw data produce a lower sse? At one point I had reasoned this didn't matter or was an incorrect observation, but I can't remember why now?
Reply With Quote
  #4  
Old 10-27-2013, 11:25 PM
mrdomer07 mrdomer07 is offline
Member
CAS
 
Join Date: Oct 2007
Studying for 8
College: Notre Dame
Favorite beer: Goose Island BCBS Cherry Rye
Posts: 284
Default

The study manual I was using did not show the Sample 2 solution for 2011 Q2 but I just looked it up myself. In that solution, they also do the holdout vs. 1.00 calculation which makes me feel a little better. At least there is a common thread between the 2011 and 2012 Couret & Venter solutions now.

However, it still worries me for cases like 2012 Q5 where doing raw vs. holdout would lead to one conclusion (use credibility procedure) and 1.00 vs. holdout would lead to another conclusion (reject credibility procedure). Anyone have any thoughts?
Reply With Quote
  #5  
Old 10-27-2013, 11:38 PM
stoneking stoneking is offline
Member
CAS
 
Join Date: May 2008
Location: Canada
Posts: 128
Default

I thought to answer this type of question, as chicken_po_boy has mentioned, is to always compare SSE between the holdout with whatever is provided, and the HG relativities (which is unity) as well.
Reply With Quote
  #6  
Old 10-27-2013, 11:53 PM
ebradford ebradford is offline
Member
CAS AAA
 
Join Date: May 2003
Posts: 156
Default

The only think I can tell is there is some very slightly difference in wording between the two. In 2011 they word it this way:

"The technique produced a raw predicted relativity...

demonstrate whether the credibility technique produced an improved estimate"

Although not clear, since they didn't mentioned anything about hazard group I'm guessing they wanted you to show whether the credibility technique was better than the raw technique. However even if it was better that doesn't mean you should be "use" it to predict something like 2012 said.

Honestly I think they've just been really sloppy with things up until 2012. I've found multiple errors on the 2010 exam problems where they give you an ELF but expect you to use it like a LER and where the answer shows some expenses used as a percent of net premium but the question explicitly says they are a percent of Standard Premium.

I really appreciate the commentary in 2012, it's helped me understand why I keep losing points on questions I know really well.
Reply With Quote
  #7  
Old 10-28-2013, 12:18 PM
Joshua0317's Avatar
Joshua0317 Joshua0317 is offline
Member
CAS
 
Join Date: Apr 2012
College: UW - Madison Alum
Favorite beer: Hoegarden
Posts: 1,038
Default

It is my assumption that they weren't clear enough they just accepted the comparison to raw in 2011. They didn't necessarily provide any clue that they wanted you to compare to the HG and raw so they were bound to just accept a comparison to either or both.
Reply With Quote
  #8  
Old 08-15-2019, 03:36 AM
adamjohari adamjohari is offline
CAS
 
Join Date: Jun 2019
Posts: 5
Default

Was looking at the solutions from TIA. Is there a reason that Fatal Freq is using 10m/100 * 0.0002 * 0.006 * 1 ??

Thought it would be
10m/100 * 0.0002 * 0.006 * 1.05

since the HG was selected in part (a)

Thanks
Adam
__________________
8 9
Reply With Quote
  #9  
Old 08-15-2019, 04:42 AM
adamjohari adamjohari is offline
CAS
 
Join Date: Jun 2019
Posts: 5
Default

My bad. The HG would have the value of 1 I believe. Shouldn't be taking the rels from the Holdout Sample.

Thanks
Adam
__________________
8 9
Reply With Quote
Reply

Tags
beers, fun, study

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off


All times are GMT -4. The time now is 03:09 AM.


Powered by vBulletin®
Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
*PLEASE NOTE: Posts are not checked for accuracy, and do not
represent the views of the Actuarial Outpost or its sponsors.
Page generated in 0.18592 seconds with 11 queries