Andrew Kugler - Pricing Actuary, Group Benefit and Retirement Services, Manulife Financial. Barry Senensky President, Claim Analytics

Transcription

1

2 LTD Analytics New Ways of Looking at an Old Product Andrew Kugler - Pricing Actuary, Group Benefit and Retirement Services, Manulife Financial Barry Senensky President, Claim Analytics Erin Crump Director, Group Research and Development, Munich Re

3 Erin Crump Director, Group Research & Development, Munich Re

4 Agenda Munich Re LTD Benchmark Index Background Methodology Indicators Applications

5 Background 2008 economic crisis LTD and economy highly correlated Emerging disability trends Lack of timely market indicators Industry benchmarking Munich Re uniquely positioned to fill gap Overwhelming industry response LTD Benchmark Index launched July 2009

6 Methodology Simple and readily available claims metrics Three data points each month for entire Group LTD block of business (including ASO): Open Claims at end of month; New Claims during the month; and Terminated Claims during the month Full-time employment figures from Statistics Canada as proxy to the population insured for LTD

7 Methodology These data points allow us to calculate three indices: The LTD Benchmark Index measuring relative changes in open claims The Incidence Indicator measuring relative changes in new claims, and The Termination Indicator measuring relative changes in terminated claims Trending relative to participating insurers

8 MUNICH RE LTD BENCHMARK INDEX The Munich Re LTD Benchmark Index (the Index ), including the sub-indices of the LTD Incidence Indicator and the LTD Termination Indicator, is based on the voluntary submission of information from the Canadian Group ket and provides a basic metric that is based on the movement of aggregate LTD claims MUNICH data. The Index relies RE on basic LTD information BENCHMARK (total number of open LTD claims, number INDEX of new LTD claims and number of terminated LTD claims) from internal management reports of companies in the Canadian Group ket and is not independently verified by Munich Re. The Index does not address the actual costs of such claims and is subject to various qualifications. The underlying information upon which the Index is based may therefore not be complete or accurate for your purposes and no representation or warranty is given as to the accuracy of any of the information provided.

9 LTD Benchmark Index Munich Re Group LTD Benchmark Index Recession The Munich Re LTD Benchmark Index has increased over the last two quarters, ending at a new high of in December MR LTD Index 3-Month Moving Average 12-Month Moving Average 2013 Munich Re Group LTD Benchmark Index changes as of December 2013 Compared to Compared to Compared to Compared to Compared to Compared to Compared to Dec 2012 Open claims continue to rise as we see periods of increasing and decreasing employment The 3-month and 12- month moving averages also rose, ending at and in December, respectively MR LTD Index 12.1% 10.7% 6.0% 5.6% 3.8% 2.2% 2.2% 3-Month Moving Average 11.4% 10.0% 5.3% 4.9% 3.2% 1.6% 1.4% 12-Month Moving Average 10.5% 9.1% 4.4% 4.0% 3.1% 1.3% 0.7%

10 LTD Benchmark Index Incidence Indicator Incidence Rate Indicator Incidence Rate Indicator 3-Month Moving Average 12-Month Moving Average 2013 Munich Re Group LTD Incidence Indicator changes as of December 2013 Compared to Compared to Compared to Compared to Compared to Compared to Compared to Dec 2012 Incidence continues to be very volatile on a month to month basis October 2013 saw the highest level of new claims across the industry since we began tracking data Previous downward trends of the 3-month and 12-month moving averages have now reversed Incidence Rate Indicator 12.5% 10.9% 7.6% 4.8% 1.8% -1.3% 4.4% 3-Month Moving Average 15.8% 14.1% 10.7% 7.8% 5.0% 1.3% 0.3% 12-Month Moving Average 10.6% 9.0% 5.8% 3.0% 1.2% -2.3% -3.0%

11 LTD Benchmark Index Termination Indicator Termination Rate Indicator Termination Rate Indicator 3-Month Moving Average 12-Month Moving Average 2013 Munich Re Group LTD Termination Indicator changes as of December 2013 Compared to Compared to Compared to Compared to Compared to Compared to Compared to Dec 2012 Compared to tember 2013, terminated claims fell by 4.4%, while open claims increased by 1.0% The 3-month moving average fell over the quarter, from in tember to in December The 12-month moving average also fell, from in tember to in December Termination Rate Indicator -3.1% -4.4% -6.4% -5.9% -6.1% -7.7% -1.7% 3-Month Moving Average 0.2% -1.1% -3.2% -2.7% -2.9% -4.6% -2.9% 12-Month Moving Average 2.6% 1.2% -0.9% -0.4% -0.4% -2.0% -2.3%

12 Applications How to use: To identify emerging industry LTD trends on a timely basis To monitor and understand the trends in your book of business relative to broader industry trends How not to use: To compare the trends in your book of business to the industry without normalizing for changes in exposure To draw conclusions around LTD profitability you may be priced appropriately for the trends you see in the Index

13 Applications Participant Benchmarking MR Group LTD Benchmark Index Comparison with Ins. Co. experience Jun Dec Jun Dec Jun Dec Jun Dec Jun Dec Jun Dec Jun Dec 2013 MR LTD Index Ins. Co. LTD Index

14 Applications New to Terminated Ratio 120% New to Terminated Ratio 115% 110% 105% 100% 95% 90% 2007 Jun Dec Jun Dec Jun Dec Jun Dec Jun Dec Jun Dec Jun Dec 2013 Industry New / Terminated Industry New / Terminated - 3 Mth MA Industry New / Terminated - 12 Mth MA

15 Applications Benchmark vs Other Indicators 116 Incidence versus Expected Aging Jun 2007 Dec 2007 Jun 2008 Dec 2008 Jun 2009 Dec 2009 Jun 2010 Dec 2010 Jun 2011 Dec 2011 Jun 2012 Dec 2012 Jun 2013 Dec 2013 Incidence Indicator - 12 month MA Average Age Expected Incidence

16 Applications LTD Benchmark Special Edition Annual edition of the Benchmark Index which provides further detail on: Average claim duration Termination rates in relation to change in definition Claims by region/province Claims by primary diagnosis

17 Andrew Kugler Pricing Actuary, Group Benefit and Retirement Services, Manulife Financial

18 Agenda: LTD Pricing Analytics Agenda: Traditional data and analytics Potential considerations for data and analytics to enhance Pricing Presentation Goal: Consider the different ways carriers might price LTD by looking at the data already available them.

19 Traditional Sources of Data 3 general data categories Client Specific Block ( Manual ) External (reinsurers, CIA studies, etc.) Traditionally data above reflective of explicit LTD experience (incidence and terminations by age/gender/province/etc )

20 Traditional Pricing Analytics Pricing blends Manual and Client Experience together using a credibility methodology Manual LTD Factor view developed by age, gender, industry & a number of other pricing factors (depending on credibility). Experience Incorporate client specific risk characteristics Year of Incurral Analysis Review group s termination and incidence experience

21 Potential LTD Analytics Consider other sources of data that correlate with LTD risk Consider pricing to incorporate cause of disability Consider a variety of pricing techniques

22 Pricing with Other Data Sources Consider LTD correlation with the following: Short Term Disability experience Health experience (Drug, Paramed, etc) Client/plan member participation in mental health and wellness programs. Need to consider data privacy rules

23 Consider Pricing using Cause of Disability LTD Termination assumptions traditionally incorporate age / gender / duration / province. Consider: Diagnosis and Industry Actual to Expected Experience by Cause of Disability, Total Terminations Cause of Disability Canada Québec Non-Québec Mental and Nervous, 92.1% 104.4% 80.8% Musculo-skeletal 92.4% 91.5% 92.9% Neoplasms (mostly cancers) 128.3% 92.7% 149.6% Accidents 118.8% 107.7% 124.8% All Other Identified Causes 100.8% 96.2% 102.9% Not Stated or Unknown 99.0% 113.3% 91.3% Total 100.0% 100.0% 100.0% Source: CIA Group Long Term Disability Termination Study October 2011 (page 35)

24 Consider Other Pricing Techniques Consider behaviour by funding arrangement and taxability of benefit Non-refund arrangement, non-taxable, employee paid Refund arrangement, taxable, employer paid Year of Incurral Amount of years to use Allocation of weight to each year Need to think about consistency to customers Can be limited in capturing trend

25 Other Pricing Techniques Cont d Linear or Multi-variable regression More robust Can capture trend Can be complicated Consider Credibility Formula Z x (Experience Rate) + (1 Z) x (Manual Rate) Consider a demographic / plan design factor to account for changes not reflect in experience Consider more exposure for full credibility for early claim durations and less exposure for full credibility in later claim durations

26 Where can all of this lead? Improved risk management Reduced risk class subsidies and pricing that is more competitive for healthy risks Improved profitability Cautions: Be wary of privacy compliance, complexity and consistency.

27 LTD Benchmarking Using Predictive Modeling Barry Senensky President, Claim Analytics

28 Overview of Claim Analytics Claim Analytics Inc. specializes in building predictive models for Insurance Companies Main focus of expertize has been building models to assist in the management of disability claims Has been benchmarking management of disability claims since 2009 In Canada and US Just completing sixth study

29 Challenges of Benchmarking LTD Claim outcomes are impacted by many factors such as elimination period, claimant age, diagnoses, region.. Directly comparing companies with different books of claims can very easily be misleading.

30 Challenges of Benchmarking Our Solution: Use predictive modeling to normalize for differences between books of claims But what does this really mean?

31 Normalize Using Predictive Modeling Create a universe of claims Build a predictive model for each company Model to predict claim outcomes based on each company s historical claims experience Each claim in universe is run through each company s predictive model Result is unbiased prediction on how well each company would perform on the universe of claims

32 Limitations Accuracy and Credibility of data Requires lots of work!

33 2014 Benchmarking Study

34 Goals of 2014 Benchmarking Compare LTD claimant recovery experience between participating companies Identify areas where there is potential to improve claim practices Quantify the potential economic impact value of these improvements Identify Trends

35 Approach 2011 vs Study 2014 Study focussed on a company s relative performance based on a universe of typical claims focus on relative performance based on each company s claims measured performance using percentages measures performance using dollars includes year over year trends in performance separate Quebec claims from the rest of Canada include a survey section

36 Methodology 1.Calculate the expected present value of net benefits for claims approved by the company in Done on a claim by claim basis Using predictive models Based on the company s actual LTD claims experience

37 Methodology (cont d) 2. Recalculate the expected present value of net benefits for claims approved in Done on a claim by claim basis Using each of the other carrier s predictive models developed based on that carrier s specific LTD claims experience

38 Methodology (cont d) 3.Performance is measured as the difference between The company s expected claim cost for its own 2012 portfolio of claims and; The expected claim cost for that same portfolio of claims if the claims were managed by the other carriers.

39 Predicted Claim Cost-2012 Claims Predicted Cost of Company D's 2012 LTD Claims Rest of Canada All Claims 80.0 $ Millions A B C D E F G H I J Cost Ave The predicted cost of net benefits for claims approved by Company D with benefit commencement date in 2012 is $2.6 million less than the industryaverage predicted cost for the same portfolio of claims

40 Predicted Claim Cost vs. Average Company D's 2012 LTD Claim Cost vs. Average Rest of Canada $ Millions TOTAL Term 0-12 m Term m Term 30+ m Q/CPP Perf Company D has higher than average termination rates during the first 30 months of benefits and this results in a $9.5 million reduction in claim cost. Company D has a lower than average rate of CPP offsets, which results in a $6.9 million increase in claim cost.

41 Predicted Claim Cost vs. Best Performer 10.0 Company D's 2012 LTD Claim Cost vs. Best Performer Rest of Canada $ Millions TOTAL Term 0-12 m Term m Term 30+ m Q/CPP Perf Company D has higher termination rates during the first 12 months of benefits, as well as in the month period. In aggregate, this results in a total $3.3 million decrease in claim cost. Company D has lower termination rates beyond 30 months, resulting in a $0.5million increase in claim cost. Company D has a lower rate of CPP offsets, which results in a $9.6 million increase in claim cost.

42 Breakdown by Age Company D's 2012 LTD Claim Cost vs. Average by Age (years) Rest of Canada $ Millions Perf % of Claims 19% 26% 40% 15%

43 Breakdown by Elimination Period Company D's 2012 LTD Claim Cost vs. Average by Elimination Period (days) Rest of Canada $ Millions Perf % of Claims 47% 48% 5%

44 Breakdown by Diagnosis $ Millions Company D's 2012 LTD Claim Cost vs. Average by Diagnosis Rest of Canada Cancer Circulo Back Nervous Injury Mntl Hlth Musculo Other Perf % of Claims 18% 8% 9% 6% 12% 16% 15% 16%

45 Breakdown For Cancer Only $ Millions Company D's 2012 LTD Claim Cost vs. Average Cancer Claims Rest of Canada TOTAL Term 0-12 m Term m Term 30+ m Q/CPP Perf

46 Breakdown by Region Company D's 2012 LTD Claim Cost vs. Average by Region Rest of Canada $ Millions BC Alta Prairies Ont itimes Other Perf % of Claims 10% 15% 40% 26% 8% 1%

47 Breakdown by Industry $ Millions Company D's 2012 LTD Claim Cost vs. Average by Industry Rest of Canada Services Other Edu/Hlth/Pu ba Manu Finance Trade Perf % of Claims 9% 5% 15% 8% 8% 55%

48 Conclusion Predictive Modeling can be a highly effective tool for benchmarking the effectiveness of LTD Claims Management in an organization Can be used to identify areas of opportunity Where else can these techniques be used to enhance benchmarking?

49 Questions?