Actuarial Outpost
 
Go Back   Actuarial Outpost > Actuarial Discussion Forum > Property - Casualty / General Insurance
FlashChat Actuarial Discussion Preliminary Exams CAS/SOA Exams Cyberchat Around the World Suggestions


Reply
 
Thread Tools Search this Thread Display Modes
  #11  
Old 03-15-2019, 05:47 PM
jerrytuttle jerrytuttle is offline
Member
CAS
 
Join Date: Oct 2001
Posts: 280
Default

Hopefully there is a data field on why each policy was canceled: non-payment of premium, a significant false statement on the app, loss experience, etc. And did the policyholder cancel, and why?

I don't know if it is still true, but in Texas non-standard auto, some insurers would issue a policy for a term of one month. Policyholders might buy insurance in January, not buy it in February, buy it again in March, and so on, and insurers were content with that.
__________________
Thanks,
Jerry
Reply With Quote
  #12  
Old 03-26-2019, 09:22 AM
Starscream's Avatar
Starscream Starscream is offline
Member
CAS
 
Join Date: May 2008
Location: Cybertron
Studying for 7 for the millionth time.
College: Eastern Michigan
Favorite beer: Samuel Smiths Taddy Porter
Posts: 77
Default 2c

I would stick with earned, but you'll want to look more closely at what is going on with your data.

Your negative exposures may be due to cancellations that are back dated. Look at a record with a negative exposure, and see if it overlaps with another.
For instance, if your annual term ends, you'd have a record with exposure 1. Lets say you call your agent and say "i want my policy cancelled, i got a new policy a month ago". They would backdate that cancellation and another record would be produced with -1/12 exposure.
__________________
Real songs are about deals with the devil, far off lands, and where you find smoke in relation to water
Reply With Quote
  #13  
Old 03-26-2019, 10:39 AM
ALivelySedative's Avatar
ALivelySedative ALivelySedative is offline
Member
CAS
 
Join Date: Dec 2013
Location: Land of the Pine
College: UNC-Chapel Hill Alum
Favorite beer: Red Oak
Posts: 3,198
Default

Quote:
Originally Posted by ShundayBloodyShunday View Post
I suggest you first understand how your data is encoded to the database and work on understanding how to best aggregate your data.

If you're aggregating on a calendar year basis, then a policy written on 12/1/20X8 and subsequently cancel on 3/1/20X9, and you're only pulling records for 20X9, the data is going to show a negative premium only because the record with the "positive" premium is in the prior year.
This is clearly true, but do you actually do anything about it? Either loss ratio or pure premium will de facto result in a 0, since you can't have a loss on negative earned premium/exposure. You could eliminate records with earnings less than zero, but I've never seen this mentioned in practice. The only 'true' way to avoid this scenario is to aggregate by policy year, which has its own drawbacks, and I've never heard this advocated in a modeling discussion either.
__________________
Stuff | 6 | ACAS | FCAS stuff
Reply With Quote
  #14  
Old 03-26-2019, 02:24 PM
itGetsBetter itGetsBetter is offline
Member
CAS AAA
 
Join Date: Feb 2016
Location: Midwest
Favorite beer: Spruce Springsteen
Posts: 270
Default

Quote:
Originally Posted by ALivelySedative View Post
This is clearly true, but do you actually do anything about it? Either loss ratio or pure premium will de facto result in a 0, since you can't have a loss on negative earned premium/exposure. You could eliminate records with earnings less than zero, but I've never seen this mentioned in practice. The only 'true' way to avoid this scenario is to aggregate by policy year, which has its own drawbacks, and I've never heard this advocated in a modeling discussion either.
Are you saying that you have never heard of policy level modeling being advocated for? From my experience, policy level aggregation is the best for modeling for both pricing and underwriting because there is a perfect match between exposures and the losses that result from them. We want to segment our best policies from our worst policies and assign relativities accordingly. In order to do this, we need to know which policy characteristics lead to policy level or peril level losses.
Reply With Quote
  #15  
Old 03-26-2019, 02:29 PM
ALivelySedative's Avatar
ALivelySedative ALivelySedative is offline
Member
CAS
 
Join Date: Dec 2013
Location: Land of the Pine
College: UNC-Chapel Hill Alum
Favorite beer: Red Oak
Posts: 3,198
Default

Quote:
Originally Posted by itGetsBetter View Post
Are you saying that you have never heard of policy level modeling being advocated for? From my experience, policy level aggregation is the best for modeling for both pricing and underwriting because there is a perfect match between exposures and the losses that result from them. We want to segment our best policies from our worst policies and assign relativities accordingly. In order to do this, we need to know which policy characteristics lead to policy level or peril level losses.
No. 'Policy year' meaning policy-year effective date aggregation of premiums and losses, as opposed to calendar/accident year.
__________________
Stuff | 6 | ACAS | FCAS stuff
Reply With Quote
  #16  
Old 03-26-2019, 06:32 PM
itGetsBetter itGetsBetter is offline
Member
CAS AAA
 
Join Date: Feb 2016
Location: Midwest
Favorite beer: Spruce Springsteen
Posts: 270
Default

Quote:
Originally Posted by ALivelySedative View Post
No. 'Policy year' meaning policy-year effective date aggregation of premiums and losses, as opposed to calendar/accident year.
When I clean data to prepare to design an underwriting or pricing GLM, I pull policy year premiums and policy year losses and hook them up to policy characteristics for individual policies.
Reply With Quote
Reply

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 10:26 PM.


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 0.35121 seconds with 11 queries