A relook at Portfolio Management practices in Non-Life Insurance

Over the last couple of weeks, again this word, “detariffication” crops up…. and the proposed nationalisation of the Motor Insurance compulsory ACT (third party bodily injury) scheme set to be implemented before end of 2010. Yes! This blog is inclined towards the Non-Life segment again…. There were lots of discussion on issues concerning best practices leading towards the detariff process – from making efforts in trimming down existing rigid clauses, warranties, circulars and practices, to working out actuarial analysis of the industry statistics (available with ISM) with an objective towards finding what’s in store ahead of time. Perhaps, we just put some focus here on actuarial analytics. Whether the industry is capable of finding its footing when two of the largest class of business eventually detariffed and can such analytics take us towards “Promised Land”?


Some looks like cracking in the face of RBC and will cracks further widen with detariffication?

The answer is probably a big NO.  😳 It is not difficult to predict  that both price and coverage terms would be severely pressured. While the actuarial analytics may be of help but these are unlikely to bring us anywhere – not when most actuarial works were being done on historical data (ie. founded on tariff) and are very costly to get the industry working on any predictive models to take understanding a step further. At the end of the day, what we thought should get us by in time of detariff is unlikely to be so…. If rates and terms are pressured by competition resulting in a downwards spiral, how would insurers cope within the Risk-based Capital (RBC) framework?

What then would be the next best thing?

Perhaps adopting the basics of Portfolio Management may do the job. Just what is Portfolio Management – Well, taking off those complicated stuff and just focus on a portfolio of risks that an insurer has underwritten into its book of business – how best these risks can be segregated into homogeneous classes for comparison and then subject to detailed time series analysis.

Going back to the basics of portfolio management – that is, splitting our underwriting books into numerous strategic portfolios and work all available statistics in the following two formats:

1.   Accidental Year Experience

Accidental Year Analysis

This portfolio management form is about matching of all losses occurring (regardless of when the losses are reported) during a given twelve-month period of time (usually a calendar year) with all premium earned (regardless of when the premium was written) during the same period of time. More specifically, the total value of all losses occurring (losses paid, plus loss reserves) during the defined twelve-month time period (i.e., the date of loss falls within the time period) is divided by the Earned Premium for this same exposure period to determine the (as at) Loss Ratio. As the experience is developing, loss reserves are used in the calculation, but the ultimate result cannot be finalized until all losses are settled. The most accurate method uses Exposure Earned as the denominator but for practical reasons, the Acounting Earned Premium is normally used.

If there are enough available statistics  collated over a reasonable long period of time, a loss development table can be built to determine how losses were paid and how loss reserves had build-up over the years. IBNR calculations by the actuary can also be strategically included as part of the table to provide an idea as to how the Ultimate Loss should look like.

From the loss development table, the management can build a viable Predictive Model (specific to portfolio) that has relevant RBC related calculations factored in. On top of these RBC driven factors, strategic market developments and internal efficiencies must be incorporated into the Model as well. Unlike the Life business segment which has matured “actuarial-driven” models, the Non-Life business certainly does not possess any. Thus it is ripe for the industry to work things out on our own – at least to build some internal underwriting models that can help put a range of MINIMUM PREMIUM in our underwriting guides.

The terms of reference are in page 2 of this post


Illustration of Accidental Year analysis table
Accident Year Claims Count Total Claims Incurred Claims Paid Claims Outs’tg Earned Premium Premium Deficiency factor Ultimate Loss (Actuarial)
2005
2006
2007
2008
2009
2010

 

Illustrating Accidental Year Loss Development Analysis table
Loss Development Years (Recording losses as they developed)
Accident Year 1st 2nd 3rd 4th 5th 6th
2005 1000 2000 3000 5000 5000 5000
2006 2000 2800 2800 2800 3000
2007 1200 2000 2000 2000
2008 2000 3000 4000
2009 2000 2300
2010 3000

 

2.   Policy Year Experience or Underwriting Year Portfolio

Underwriting Year Analysis
To the underwriting community, this is about the segregation of all premiums and losses attributable to policies having an inception or renewal date within a given twelve-month period. More specifically, the total value (losses paid plus loss reserves) of all losses arising from (regardless of when reported) policies incepting or renewing during the year is divided by the fully developed earned premium for those same policies to determine the (as at) Loss Ratio. The finally developed earned premium will always equal the written premium for those policies. Policy (or Underwriting)-Year Experience resembles Accident Year Experience in that, while the experience is developing, loss reserves are used in the calculation, but the ultimate result cannot be finalized until all losses are settled. Policy-Year Experience is different in that premiums earned from policies incepting during a one-year period of time will earn over the course of both the year of inception and a later year(s). Similarly, losses to be included will be occurring over the same extended time period. Also known as Underwriting Year.


Illustrating an UW Year Portfolio analysis table                            (page 1/2)

UW Year Gross Prem No of Policy Earned Prem Total Claims Incurred Claims Count Claims Paid Claims Outs’tg Severity Frequency Loss Ratio
2005
2006
2007
2008
2009
2010


(page 2/2)
UW Year Prem Deficiency factor Insurance Cyclical factor Incurred But Not Reported (IBNR) Min Prem range
2005
2006
2007
2008
2009
2010


Illustrating an UW Year (Loss) Development analysis format
UW Development years (Recording losses for an UW year as they develop)
UW Year 1st 2nd 3rd 4th 5th 6th
2005
2006
2007
2008
2009
2010


The above tables would be more appropriate and strategic for analysis if Pivotal formatted.

“Ultimately….would get us a range of minimum rates that are of RBC-compliant….”

The two contrasting portfolio analyses would have been readjusted with parameters derived from actuarial calculation as adopted in the computation of General Insurance Liabilities per requirements in the RBC framework. Thus, the premium deficiency factors and the IBNR numbers are incorporated as part of the portfolio calculation. In the UW Year portfolio analysis, the portfolio cyclical factors are also included to improve on the outcome as well as assumptions made in the process of deriving the performance level of a portfolio.  Ultimately, these analyses should get us to a range of minimum rates that are of RBC complaint (within a prescribed confidence level range), both from “Supervisory” and “Internal” perspectives.

Nevertheless, it is important that individual insurer obtains similar portfolio analysis outputs from industry statistics; currently database are kept within ISM. Why is this industry level statistical important for members within the realm of such portfolio management? Simply puts, this industry-level statistics would be provide a much larger base for analysis and comparison with the individual insurer’s portfolio, thus enable underwriters to better make assumptions in deriving the range of minimum rating that correspond to the prescribed level of coverage. PIAM should take the lead in working out those available statistics within the ISM system into these two main formats of analyses.

Summing up, the methodology adopted should then be converted into some Portfolio Underwriting Tool (PUT) which the insurer can tie its usage with their own RBC parameters and internal Predictive models, if any. If the tools are tie with the RBC, then potential pressure on pricing is likely brought into control during the times of detariffication and over competition can be contained. I do not think the Central Bank is disagreeing with this proposal either! Ultimately, we still need to reproduce those portfolio analytical workings into “Tools” for operational efficiency…..”

Perhaps, the industry can also work towards putting the existing ISM-KMS system working at a more efficient and effective level. “KMS” simply means Knowledge Management Services – where currently, ISM operates both Fire and Motor statistical base. This, I supposed should be the better of all alternatives….. However, even if the industry statistics are efficiently analysed and packed for individual insurer usage, still there will be a vacuum in as far as internal statistical and portfolio analysis are concerned. Individual insurer would have to work on some forms of internal PUT – the usual ones may come from SAS and Business Objects but these can be both labour and costly to implement. The last time we had one (almost though….), the costs ran into almost a million! And, not mentioning the extensive works that are needed to get the parameters working within the Business Intelligence system. We have since turned to a lower and cheaper form – using the MS Excel sheet lah! And still have a long way to go, trying to discover and rediscover what the next best thing…. “Ultimately again…. we should spend to improve the ISM-KMS system for industry statistical analytics… and putting an internal business intelligence (BI) system in place for company-level portfolio analytics”

Do you think this works or something nice to have but impossible to get there? But then, it is not actually very difficult to get this PUT in place, an Excel sheet complete with various VBAs and macros would have done the job; but of course, we need those brains behind this great proposal!


Related Posts with Thumbnails

11 comments for “A relook at Portfolio Management practices in Non-Life Insurance

  1. koh cc
    September 15, 2011 at 11:11

    Is PIAM doing anything about pricing for both motor and fire insurance since these two major classes will be detariff soon under Competition Act early next year?

    Can you elaborate what PIAM is currently planning to do? It seems things are not being discussed openly, closed door most of the time??

  2. December 17, 2010 at 20:40

    My cousin recommended this blog and she was totally right keep up the fantastic work!

  3. January 28, 2010 at 00:52

    How to do this? How to calculate prem deficiency and earned premium?

    • January 30, 2010 at 02:17

      Well, premium deficiency is about the shortfall in premium charged when calculated as a portfolio.- hopefully I can learn something from the actuary as to how this is computed. Earned premium is premium already earned after the risk has gone through a specific time frame, usually link to policy period. Whatever that is not earned would eventually fall under this category of unexpired risk reserves…… where unearned premium is a subset of it.

  4. Jap-pan
    January 25, 2010 at 10:58

    Portfolio checks can be strategic if the industry can agree on the specific set of ratios for monitoring and adoption. From these established ratios, the identification of whether an insurer is able to continue writing a particular class of business over the medium to longer terms above the break-even position would be made easier. Tying min rates to RBC-calculated parameters should be deemed tactically correct – at least keeping unnecessary competitions in check. It is always a question of how are we going to do it.

    On the contrary, can the local players compete against the foreign-based insurers in the face of capital scarcity?

    • January 25, 2010 at 23:37

      Thanks for the comment…. getting to those minimum rates that are RBC-compliant or somewhere near there is indeed difficult. We can only visualise as of now! Let’s say, this posting is a beginning of a long and winding road(map) getting there. Perhaps we start by getting those portfolio analytical parameters in place. Sort to breakdown those RBC components (whether from the SCAR or ICAR computation) and have those components reworked backward aligning them with those derived from the portfolio analytics…….. LEt’s see what happens then!

      • January 27, 2010 at 09:02

        In the US and Europe markets, underwriting is already very much driven by actuarial computation. Most insurers employ a huge pool of actuary or trainees to work out both the industry and internal stats, then introduce a spreadsheet tool for the underwriters to use in their daily quotation. What we normally get is a rate too far out! To follow means to quote your way out of the market. Therefore, putting in discounting factors (sometimes to extent of 50%) is not uncommon – anything different from rate slashing? Of course, there will always be a limit as to what percentage you can slash.

        • January 27, 2010 at 22:07

          It is still a better practice to have quantitative analysis in place and then make final decision on available qualitative factors. Of course, the costs of getting to the quantitative data may be unjustifiably high, especially where qualified actuaries are engaged.

          • C
            January 30, 2010 at 10:38

            Rather than building with excel sheet this portfolio management stuff, perhaps it is better and more efficient if we consider SPSS Stats version 18.

            • January 30, 2010 at 22:41

              Thanks…. our malaysian insurance industry has yet to utilise such software for portfolio analysis and predictive modeling. Take your suggestion and relook at their latest version on statistics v 18…

Leave a Reply