What to Do When the Actuary s Answer is Too High Timothy C. Mosler, FCAS, MAAA February 25, 2016
Agenda Tim Mosler s Background Disclaimer Basics of Reserve/Funding Analysis 3 Reactions When the Client Disagrees Loss Development Illustration Summary exhibit examples Common Issues Case Reserve Strengthening Declining Loss Cost in Recent Years Aggressive Settlements Questions 1
Tim Mosler Background Education B.S. in Mathematics from Florida Atlantic ACAS in May 2003, FCAS in May 2004 Experience National Council on Compensation Insurance (1996 2001) Towers Watson (2001 2014) Pinnacle Actuarial Resources (2014 ) Hundreds of discussions with clients about reserve analysis results! 2
Disclaimer This presentation is not intended to imply that any actuarial study is producing a result that is too high. The best estimate of reserves or funding in any given situation requires a formal actuarial analysis and an actuary s best estimate should be their own independent opinion. This presentation is focused on allowing the user of an actuary s report to understand the actuary s opinion and potentially help to refine the opinion. 3
Reserve/Funding Analysis at 12/31/2015 Hypothetical self insurance program Self Insures a portion of its liability lines Examples Auto liability, General liability, and Workers compensation Retention = $500,000 per claim Fiscal year runs 1/1 12/31 Typical results provided by the actuary Indicated Reserves at 12/31/2015 Projected Funding for 2016 Possibly projections for future years 4
Reserve/Funding Analysis at 12/31/2015 Actuarial Analysis Segregates data by accident period Ultimate losses are estimated for each accident period Based on a series of projection methods Indicated Reserves = Ultimate losses minus paid through 12/31/2015 2016 Funding = Ultimate loss estimate for 2015 (and possibly older years) with an adjustment for trend 5
Client Reactions to Actuarial Results Agree with the actuary Client may or may not have their own opinion Disagree with the actuary leads to three basic options Accept the result anyway Ask the actuary to change or to provide additional scenarios Challenge the actuary The remainder of the presentation assumes that, at least initially, there is disagreement and the client believes (or would like) a lower answer 6
Option 1 Accept the Result Could be the best option if The financial effect is small Time is limited There is a high level of trust in the actuary May still be best to get an explanation from the actuary 7
Option 2 Ask the Actuary to Change Re review of the selections There is a reasonable range of indications There is a range for each actuarial assumption that goes into creating the indication Asking creates a dilemma for the actuary Explicitly provide a range Does not affect income after the first year unless the point in the range changes Year 1 Reserve indication increases by $5M. Client only increase $2M Year 2 Reserve indication increases by $5M again. Now, client also has to increase $5M or go lower in the range. 8
Option 3 Challenge the Actuary Some combination of Stating that the answer is higher than expected Gathering data or requesting data from the actuary Identifying how the data conflicts with the actuary s recommended reserve Asking the actuary to revisit the selections in the analysis. 9
Loss Development Illustration 10
Incurred Loss Development Through 12/31/2007 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 11
Incurred Loss Development Through 12/31/2008 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 12
Incurred Loss Development Through 12/31/2009 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 13
Incurred Loss Development Through 12/31/2010 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 14
Incurred Loss Development Through 12/31/2011 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 15
Incurred Loss Development Through 12/31/2012 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 2012 16
Incurred Loss Development Through 12/31/2013 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 2012 2013 17
Incurred Loss Development Through 12/31/2014 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 2012 2013 2014 18
Incurred Loss Development Through 12/31/2015 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 2012 2013 2014 2015 19
Loss Development Projection of Recent Years $16.0 $14.0 $12.0 $10.0 Millions $8.0 $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 2012 2013 2014 2015 2014 Proj 2015 Proj 20
Projection of Recent Years Based on BF Method $16.0 $14.0 $12.0 $10.0 Millions $8.0 $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 2012 2013 2014 2015 2014 Proj 2015 Proj 21
Sample Ultimate Loss and Reserve Summary Hypothetical Client Development of Undiscounted Unpaid Loss & ALAE Selected Case Indicated Accident Ultimate Paid Loss & ALAE IBNR Unpaid Year Loss & ALAE Loss & ALAE Reserves Reserves Reserves 2007 $10,530,802 $10,530,802 $0 $0 $0 2008 9,650,319 9,566,616 14,170 69,534 83,704 2009 11,416,531 10,627,500 17,658 771,373 789,031 2010 12,855,939 11,192,425 277,880 1,385,633 1,663,514 2011 10,176,011 7,185,818 1,144,545 1,845,649 2,990,194 2012 13,396,963 9,261,710 740,088 3,395,165 4,135,253 2013 6,901,419 3,803,451 685,482 2,412,486 3,097,968 2014 7,229,633 2,535,051 461,103 4,233,480 4,694,582 2015 12,874,821 4,042,315 2,820,659 6,011,848 8,832,507 Total $95,032,440 $68,745,687 $6,161,586 $20,125,167 $26,286,753 22
Sample Claim Count Summary Hypothetical Client Summary of Claim Counts Accident Closed Open Reported IBNR Ultimate Year Claims Claims Claims Claims Claims 2007 1,082 0 1,082 0 1,082 2008 624 11 635 0 635 2009 590 8 598 0 598 2010 1,148 16 1,164 0 1,164 2011 972 54 1,026 0 1,026 2012 839 49 888 2 890 2013 876 117 993 2 995 2014 483 154 637 7 644 2015 279 701 980 55 1,035 Total 6,893 1,110 8,003 66 8,069 23
Case Reserve Changes Scenario 1 Scenario 1: Consistent adjustment to a large set of claims across all years Example Case reserves didn t account for or consistently understated several components of the final costs. Now they are better reflections of the final cost It s expected that IBNR will now be lower relative to the case reserves. What if the actuary has it at the same level or higher? Data to review: Average case reserves = Total case reserves / Total O/S claim count in this year s report and last year s report If the case reserve strengthening was significant, this average will be significantly higher Incurred development factors or just the factor for the latest year if the actuary agrees that there was strengthening, then it should be lower 24
Case Reserve Strengthening $14.0 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 2012 2013 2014 2015 25
Case Reserve Changes Scenario 1 (cont d) Actuarial Concern is that the strengthening masks the real change in experience Possible Resolutions BEST Provide data before and after the case reserve adjustment, actuary creates indications based on both data sets Adjustment to historical incurred loss development factors More reliance on paid indications 26
Case Reserve Changes Scenario 2 Scenario 2: There was a review of the large open claims in older years that lead to several case reserves being adjusted higher. Claims staff (or the TPA) indicates that there s a higher degree of confidence in the case reserves now. In the actuarial report, the IBNR is now higher but we thought it would be lower. Harder to challenge than Scenario 1 27
Case Reserve Changes Scenario 2 (Cont d) Actuarial Concerns The strengthening sounds like normal loss development isn t the case reserve always a best estimate based on current information? Several claims in more recent years could have the same development as they get older. Should the development factors be higher? 2 ways to look at a significant change in the level of case reserves: Average Case Reserve Optimistic View Pessimistic View Significantly Higher than Prior Analysis Significantly Lower than Prior Analysis Case reserves are overstated we need less IBNR Claims are less serious than we thought we need less IBNR Claims are more serious than we thought we need more IBNR Case reserves must be understated we need more IBNR 28
Case Reserve Changes Scenario 2 (Cont d) Data to Review and/or Provide Same two items Average case reserves = Total case reserves / Total O/S claim count across years If this is significantly higher, it supports that the case reserve increase was atypical. Incurred development factors or just the factor for the latest year Are they significantly higher than in the prior analysis? Important additional item provide qualitative information on the claims and the uniqueness of the case reserve adjustment for those claims Possible Resolutions Adjustment to give less weight to the years affected by the strengthening More reliance on paid indications 29
Effective Loss Control Example Several loss control initiatives have been effective with significant reductions in the number of claims incurred. The actuary continues to give significant weight to older years experience in the projection of the recent years. 30
Effective Loss Control Illustration $14.0 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 2012 2013 2014 2015 31
Effective Loss Control Projection $14.0 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 2012 2013 2014 2015 2014 Proj 2015 Proj 32
Effective Loss Control Projection with the BF Method $14.0 $12.0 $10.0 $8.0 Millions $6.0 $4.0 $2.0 $0.0 0 12 24 36 48 60 72 84 96 108 Months of Development 2007 2008 2009 2010 2011 2012 2013 2014 2015 2014 Proj 2015 Proj 33
Claim Count Effective Loss Control Actuarial concern: The little claims have been eliminated but the more serious claims are left. There will be higher development. Accident Year Before Loss Control Average Initial Reserve Average Ultimate Total Initial Reserve Total Ultimate Claim Count Accident Year After Loss Control Average Initial Reserve Average Ultimate Total Initial Reserve Total Ultimate 1,000 1 1 1,000 1,000 100 1 1 100 100 1,000 5 5 5,000 5,000 100 5 5 500 500 1,000 10 11 10,000 11,000 100 10 11 1,000 1,100 100 50 100 5,000 10,000 100 50 100 5,000 10,000 3,100 21,000 27,000 400 6,600 11,700 LDF = 1.29 LDF = 1.77 Loss Amounts are in $000 s 34
Effective Loss Control Data to review or request: Average unpaid per O/S claim (Increase in paid) divided by (the decrease in case reserves) It s a one year approximation of the unpaid to case ratio Trend assumption in the BF method Trend implied by actual experience Actual experience has a downward trend, but BF method probably assumes a positive trend Potential resolutions Slightly more weight to loss development methods instead of BF methods Movement on the trend to reflect the favorable experience Enhanced analysis Primary layer and excess layer Lost time and medical only 35
Settlement Example Concerted effort to settle claims earlier in their development. It seems like the actuary s report is not reducing the reserve enough to reflect how few claims there are now. Actuarial Concern: Less serious claims are the ones that get settled. The remaining claims are serious and will have higher development. 36
Settlement Data to review: Closed Claim Count divided by Reported Claim Count by year compared to the year at the same age in last year s report AY 2015 has 75% closed at 12/31/2015 and AY 2014 had 60% closed at 12/31/2014 demonstrates settlement efforts were effective Average unpaid (case + IBNR) per O/S claim Possible resolutions: Less reliance on methods based exclusively on paid losses Use of methods that are based directly on case reserves or open claim counts 37
Thank You for Your Time and Attention Tim Mosler tmosler@pinnacleactuaries.com 678 894 7254 Commitment Beyond Numbers 38