Loss triangle questions

[Packer16]

In reviewing P&C loss triangles, I noticed a difference between calander year and accident year results.  Does anyone know what the differences are?  and which one do you use to determine if the underwriting is sound?  A related question is for Life and Health firms is is there a similar metric for these companies?  TIA

 

Packer

[JEast]

It is my understanding that Calendar Year is January 1st whereas Accident Year may start April 1st since not all policies are signed on the company's calendar year.  In the link is a very brief explaination:  http://www.guycarp.com/portal/extranet/utility/glossary_a.html?vid=1, plus all the estimates for expenses, earned and unearned premiums, etcetera.

 

Irrespective, over the years I have found the 'triangles' to be nearly useless in making an investment decision.  Even so far as going to the Schedule Ps filed with the NAIC are not really helpful.  The reason is all the numbers are what people think (estimates) and we know what happens with estimates.  What can be helpful (slightly) is the paid claims because that is real monies going out the door and can be used for loss trends.  However, when you get into multi-lines the paid trend line is useless which takes you back to the Schedule P to break it out.  My 2¢.

 

Cheers

JEast

[racemize]

I think the modified triangles (that you have to do by hand) that shows the AY developments on a per year basis are somewhat useful.  The ones that are in the 10-k's do not seem very informative on their face though.

[onyx1]

In reviewing P&C loss triangles, I noticed a difference between calander year and accident year results.  Does anyone know what the differences are?  and which one do you use to determine if the underwriting is sound?  A related question is for Life and Health firms is is there a similar metric for these companies?  TIA

 

Packer

 

 

Calendar year results include losses for the current year plus loss development (reserve additions and releases) from previous years.  Accident year results are for the current year only.

 

Combined ratios (CR) are generally reported on a calendar year basis.  P&C insurers are still releasing reserves from the 2002-2006 hard market years which has the effect of lowering reported CRs.  Many investors understand this and want to know the accident year CR to judge incremental underwriting performance.  Recently, accident year ratios have been well above 100% but after reserve releases from prior years the P&C company reports a sub 100% CR and all appears well.  Once the well of excess reserves are exhausted, P&C companies won't be able to operate this way and will put additional pressure on underwriters to raise prices.

[Packer16]

Don't the triangles give you a sense how the future loss ratios may develop on average?  If there are alot of redundancies then the loss ratios may come in lower in the future than originally estimated than if there are deficiencies.  So if you focused on firms that only have recurring redundancies then you have some assurance that the underwriting is conservative over a cycle.  This metric in conjunction with low combined ratios appear to be a common characteristic of some the favorite firms discussed here including RLI, BRK and Lancashire.  Prem has stated that Fairfax has had on average 6 to 8% redundancies for FFH.  Is there a way this number can be pulled out of the loss triangles?  TIA

 

Packer

[twacowfca]

Don't the triangles give you a sense how the future loss ratios may develop on average?  If there are alot of redundancies then the loss ratios may come in lower in the future than originally estimated than if there are deficiencies.  So if you focused on firms that only have recurring redundancies then you have some assurance that the underwriting is conservative over a cycle.  This metric in conjunction with low combined ratios appear to be a common characteristic of some the favorite firms discussed here including RLI, BRK and Lancashire.  Prem has stated that Fairfax has had on average 6 to 8% redundancies for FFH.  Is there a way this number can be pulled out of the loss triangles?  TIA

 

Packer

 

Lancashire's reserve development has been very consistent and positive since their 2005 IPO.  Their actuaries reserved conservatively based on industrywide loss development even though their loss development was much less with their better underwriting.  As a result, their positive reserve development has given them an extra cushion for cat losses.

 

Montpelier Re has been the most admirable of all the Bermuda and London P&C insurers.  They more than fully reserved for all their huge IBNR 2011 cat exposure even though that put them in the red for last year and on watch for a possible downgrade.  Since then, they have experienced zero loss creep, the only insurer that has that honor.  :)

 

 

 

Yes.  FFH's annual reports, for example are a good case study for analyzing the quality of an insurance company's reported earnings.  Their accident year loss triangles a decade ago for recent acquisitions mostly showed alarming adverse reserve development.  Then, reserve development started to become generally positive half a dozen years ago as they exercised better control over their acquired businesses. 

 

That change in trend was a strong indicator of greatly improved underwriting discipline and a harbinger of increasing profits instead of losses.

[racemize]

Don't the triangles give you a sense how the future loss ratios may develop on average?  If there are alot of redundancies then the loss ratios may come in lower in the future than originally estimated than if there are deficiencies.  So if you focused on firms that only have recurring redundancies then you have some assurance that the underwriting is conservative over a cycle.  This metric in conjunction with low combined ratios appear to be a common characteristic of some the favorite firms discussed here including RLI, BRK and Lancashire.  Prem has stated that Fairfax has had on average 6 to 8% redundancies for FFH.  Is there a way this number can be pulled out of the loss triangles?  TIA

 

Packer

 

Note that FFH is nice enough to do the reserve triangle modification I mentioned above in their actual report (I haven't seen others do this), so you can just read it from the 10-k.