1. Session 5.4: Results of the Preferred Class Structure Analysis JARON ARBOLEDA, ASA, MAAA CINDY MACDONALD, FSA, MAAA, CFA April 5, 2016

2. Joint SOA/MIB Initiative 10/2015- Preferred Class Structure report 3/2016 - Preferred Class Structure Part 2 report SOA website - http://www.soa.org/Research/Experience- Study/Ind-Life/Mortality/2016-preferred-class-structure- 2.aspxhttp://www.soa.org/Research/Experience- Study/Ind-Life/Mortality/2016-preferred-class-structure- 2.aspx Written Report Excel files with pivot tables This ppt will be added 2

3. Agenda Data Preferred class structure evolution Actual to Expected results Next steps Feedback/questions 3

4. Data

5. Source of Data Majority of the data is from the Statistical Agent submissions of NY State from 2009 to 2013 and KS State from 2011 to 2013; a total of 88 companies in 2013; exposure requirement 11 KS companies also submitted data for years 2009 and 2010; requested by the SOA MIB, the Statistical Agent for NY and KS, is currently working on the 2014 data call Periodic updates to this Preferred study every few years 5

6. Preferred Data 85 companies For NS, preferred class structures of 2, 3 or 4 For SM, only preferred class structure of 2 Companies submitted other PCS (e.g. NS PCS 5 or 6) but were deemed unreliable or, upon further analysis, were actually PCS 3 or 4 Issue ages 18 and over Excluded post level term experience in mortality displays 6

7. What changed since prior report / Data Considerations Issues of 1990 and later; from 1981 in previous report; criteria in the early years were not consistent with the criteria used in the later period 2009-2013; previous report included common companies in calendar years 2010-2012 only From issue age 18; previous report limited preferred cases from issue age 25 only Post level term and extended term / reduced paid up experience excluded in report; were in previous report Added issue age field in the pivot table Includes analysis on non-preferred nonsmoker/smoker business 7

8. Historical evolution of preferred class structures

9. The Pivot Tables Total number of companies in database = 85 Started displays when at least five companies counted Small number of companies in earlier years and categories not shown Presentation by number of companies, amount of insurance and number of death claims Overall results plus results by size band and major plan codes Values given for males, females and combined Depending on a company’s distribution of business, a company could be counted in different quartiles 9

10. By Policy Issue Year NT-2 already significant by 1990; started early in the 1980s; increased steadily but started dropping off by 2008/09 NT-3 showed up in the late 1980s NT-4 showed up in the early 1990s NT-2 drop off in 2008/09 could be the likely result of companies introducing NT-3 and NT-4 TB-2 had the same pattern as NT-2 – started in the mid- 1980s, increasing steadily without a drop off Partial years affect amount of insurance values beyond 2009 10

11. By Duration The pattern by duration is similar to that by issue year, in reverse but not quite In the 2013 submission year 2013 issues are all in duration 1; 2012 issues are generally in duration 2 In the 2012 submission year 2012 issues are also in duration 1, 2011 issues are generally in duration 2 11

12. By Issue Age By exposure, age bands 30-49 had the highest exposure Lesser exposure from 18-29 Dropping off sharply from age 50 12

13. By Gender Not that much difference between males and females with regard to number of companies offering preferred policies By exposure, males are slightly higher Higher A/E ratios for females though in some general categories 13

14. By Product More companies reported term issues on a preferred basis than any other plan Preferred policies were first issued on term insurance Not evident on the tables in the report because no companies showed up issuing term in the early 1990s? – term policies were short term policies so by 2009-2013 calendar years they have already expired Term, Perm, UL, ULSG, VL, VLSG 14

15. By Policy Size First issued above $100,000 Companies started issuing at higher amount bands To accommodate requests from sales people, preferred policies were issued below $100,000 15

16. Non-Preferred NS and SM NT-1 or TB-1 also called Preferred Class Structure of 1 For cases that do not qualify for preferred rates Before preferred policies were issued 16

17. Actual to Expected Results

18. A/E Results Gender – male, female, both Preferred class structures Non-tobacco (NT), Tobacco (Tob) NT2: class 1 & 2 NT3: class 1, 2, & 3 NT4: class 1, 2, 3, & 4 Tob2: class 1 & 2 Issue age, duration, or issue year Count or amount Plans – All, perm, term, UL, ULSG, VL, VLSG Size – <$100k, $100-249k, $250-499k, $500-999km, $1-2.49 mil, $2.5+ mil Expected basis = 2008 VBT Removed any data points with claims <=35 18

19. Within Structure - NT 2 All Plans, Male & Female 19

20. Within Structure - NT 3 All Plans, Male & Female 20

21. Within Structure - NT 4 All Plans, Male & Female 21

22. Within Structure - Tob 2 All Plans, Male & Female 22

23. Across Structures - Class 1 All Plans, Male & Female 23

24. Across Structures - Class 2 All plans, Male & Female 24

25. Across Structures - Class 3 All plans, Male & Female 25

26. Within and Across Structure A/E’s in worst class in each structure drop across issue ages Other classes, A/E’s are relatively flat across issue ages Class 1 and 2 similar results in NT-3 and NT-4 Class 1 and 2 in NT-2 and Tob-2 may seem odd BUT different expected basis 26

27. Preferred wear-off

28. Across Structures, Class 1 & 2, All Plans Duration 28 Class 1 Class 2

29. Across Structures, Class 3 & 4, All Plans Duration 29 Class 3 Class 4

30. Duration vs Issue Year Across Structures, Class 1, All Plans 30

31. Preferred Wear-Off Some classes seem to show some evidence of possible preferred wear-off Class 1 in Tob-2, Class 2 in NT-2 and Tob-2, and Class 3 in NT-3 Other classes are less clear Class 1 in NT-2, Class 3 in NT-4, and Class 4 in NT-4 Other classes are not showing much evidence of any preferred wear-off Class 1 and 2 in NT-3 and NT-4 Duration vs Issue Year – similar patterns, but reversed 31

32. Plan & Size Comparison

33. A/E’s by Plan 33 No data points removed; by count

34. A/E’s by Size 34 No data points removed, by count

35. A/E’s by Size & Issue Age – NT2 35 Removed any data points with claims <=35; by count

36. A/E’s by Size & Issue Age – NT3 36 Removed any data points with claims <=35; by count

37. A/E’s by Size & Issue Age – NT4 37 Removed any data points with claims <=35; by count

38. A/E’s by Size & Issue Age – Tob2 38 Removed any data points with claims <=35; by count

39. By Plan & By Size By plan, no consistent pattern of A/E’s Term policies in NT-2, class 2 has the highest A/E’s A/E’s, across all issue ages, vary inversely with size A/E’s decrease with issue age in the highest class in a structure 39

40. Actual to Expected Quartile Analysis

41. A/E Quartile Analysis Within a structure – across classes A/E’s & claims Expected = 2008 VBT All plans Male vs female Issue age groups and duration groups by count Quartiles based on company ranking Within the group, ranked by A/E; if same A/E, ranked by exposure At least 5 companies in a quartile. Removed any data points with claims <=35 41

42. Quartile A/Es -NT2 Structure All Plans, Male vs Female, Issue Age 42 Male Female

43. Quartile A/Es –NT3 Structure All Plans, Male vs Female, Issue Age 43 Male Female

44. Quartile A/Es –NT4 Structure All Plans, Male vs Female, Issue Age 44 Male Female

45. Quartile A/Es –Tob2 Structure All Plans, Male vs Female, Issue Age 45 Male Female

46. Quartile A/Es -NT2 Structure All Plans, Male vs Female, Duration 46 Male Female

47. Quartile A/Es –NT3 Structure All Plans, Male vs Female, Duration 47 Male Female

48. Quartile A/Es –NT4 Structure All Plans, Male vs Female, Duration 48 Male Female

49. Quartile A/Es –Tob2 Structure All Plans, Male vs Female, Duration 49 Male Female

50. A/E Quartile Analysis Conclusions Variance across companies increases with class Early durations have the highest variance across companies Excluded data included small companies in Q1, Dur 1-5 and younger issue ages, with no claims and a 0% A/E ratio 50

51. Actual to Expected Quartile Analysis – by Size & by Plan

52. A/E Quartile Analysis - by size & plan Each structure/class By size: <$100k, $100-249k, $249-499k, $500-999k, $1-2.49M, $2.5M+ By plan: Level term, perm, UL/ULSG, VL/VLSG Both male & female, by count Quartiles based on company ranking Within the group, ranked by A/E; if same A/E, ranked by exposure At least 5 companies in a quartile. Removed any data points with claims <=35 52

53. Quartile A/Es by Size - NT2/Class 1, Male & Female, Issue Age 53

54. Quartile A/Es by Size - NT2/Class 2, Male & Female, Issue Age 54

55. Quartile A/Es by Plan - NT2/Class 1, Male & Female, Issue Age 55

56. Quartile A/Es by Plan - NT2/Class 2, Male & Female, Issue Age 56

57. A/E Quartile Analysis by Size & Plan 57 Smallest size group seem to have more variance across companies, at least in class 2 Class 1 has lower variance across companies than class 2 By plan, no clear conclusions across companies

58. Next Steps 58 In 2016, older and younger age mortality Future preferred structure analysis updates Additional data will be added each year. A/E’s using 2015 VBT Remove smallest companies in quartile analysis

59. Questions/Feedback