B Post Level Term Experience Results

SOA Post-Level Term
Experience Results
Derek Kueker, FSA
Actuaries’ Club of the Southwest
June 19, 2014

Table of Contents
 Introduction to Post-Level Term

 Background
 Assumption Survey Results
 Experience Study Results
 Predictive Model
 Acknowledgements
2
Introduction to Post-Level Term
3
Introduction to Post-Level Term
10-Year Term: Premium Structure with Jump to ART
10-Year Term
45-Year Old, $500,000 Death Benefit
18,000
Super-Preferred Non-Smoker
16,000
14,000
12,000
$7,395
Annual Premium
Duration 11
10,000 Premium
8,000
20 times
6,000 Premium Jump
4,000
$375 Level
Premium
2,000
-
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
Level Period Post-Level Period
Duration
4
Background
5
Background
 SOA Sponsored Post-Level Term Reports

 2007 – Report on the Survey of Post-Level Premium Period Lapse and Mortality
Assumptions and Experience For Level Premium Term Plans
Assumptions for Level Premium Term Plans
 2010 – Report on the Lapse and Mortality Experience of Post-Level Premium
Period Term Plans
Assumptions for Level Premium Term Plans
 2014 – Report on the Lapse and Mortality Experience of Post-Level Premium
Period Term Plans
6
Assumption Survey Results
7
Overview
 Sent to top 100 term providers based on 2012 term insurance sales
 Responses from 41 companies
 Responses represented 62% of 2012 term sales
8
Product Distribution
Level Premium Term Product Mix by Level Period

Number of companies where product
Aggregate Distribution represents at least x% of individual
Product Level Period company's term sales
for Respondents
x=5% x=15% x=30%
5 Year Term 1.20% 3 1 0
10 Year Term 22.70% 42 36 8
15 Year Term 8.60% 32 6 1
20 Year Term 52.00% 44 44 42
25-30 Year Term 15.10% 33 23 9
Other 0.40% 5 2 1
9
Post-Level Premium Structure
 Dominant Structure is Jump to an ART Scale
Post-Level Product Design

Product Structure Responses
Premium jump to ART 40
Premium grade to ART 4
Jump to new level period 3
Face amount decrease 1
Product terminates 2
Flexible Premiums (Term UL) 1
10
Expected Changes to Post-Level Premium Structure
 No change expected for most term new business
 Minimal changes reported to inforce term business
Changes to Post-Level Premium Structure for Term New Business

No change 23
Grade premiums into an ART scale over 'x' years 3
Other 3
Changes to Post-Level Premium Structure for Term In Force

Implemented in
Description Considering
last 5 years
Lower post-level premiums 1 5
Grade into an ART scale 3 3
Other 0 2
11
Structure of Current & Guaranteed Ultimate Rates
 Trend towards increasing level of guaranteed ultimate rates
 Relationship between current and guaranteed continues to vary
Structure of Guaranteed Ultimate Rates
Description 2013 2009
% of 1980 CSO
Between 150-300% 2 8
% of 2001 CSO
Less than 200% 5 1
Exactly 200% 7 12
Between 200-300% 11 7
Exactly 300% 11 12
More than 300% 7 1
Other 3 5
Relationship between Current Ultimate and Guaranteed Ultimate Premiums

Description 2013 2009
Product has Guaranteed Rates only 14 15
Current Rates = Guaranteed Rates 15 12
Current Rates < Guaranteed Rates 14 16
Current Rates grade to Guaranteed Rates 1 2
12
Post-Level Lapse Rate Structure
 Most companies assume only one shock lapse, generally grading
down thereafter
Lapse Rate Trend By Duration

Responses
Description
10-Year Term 20-Year Term
Duration L Shock, then level 8 7
Duration L Shock, then grade down 12 10
2 Shocks, Duration L>=L+1 6 5
2 Shocks, Duration L+1>L 3 2
13
Post-Level Lapse Rate Assumption by Premium Jump
 Broad range of assumptions by premium jump
 Trend is somewhat inconsistent with experience
Premium Jump vs. Shock Lapse

10 Year term
100%
90%
80%
Shock Lapse
70%
60%
50%
40%
30%
20%
10%
0%
0.0 10.0 20.0 30.0 40.0
Premium Jump
14
Post-Level Lapse rate Assumptions
 Today’s skewness assumptions do not always follow past experience
Monthly Lapse Skewness During Level Premium Period

Response Respondents
Lapses are uniformly distributed 18
Lapses occur on premium payment modes 10
Lapses occur at the end of the year 7
Other 4
No response 5
Monthly Lapse Skewness During Year of Shock Lapse

Lapses graded toward end of the year with shock in month 12 12
No response 7
Monthly Lapse Skewness During Post-Level Period

Lapses skewed to the beginning of L+1, Uniform thereafter 8
No response 14
15
Post-Level Mortality Methodology
 Dukes-MacDonald (or derivatives) and Flat Multiple are the primary
assumptions used in developing mortality after the shock lapse
Method of Developing Mortality Assumption

Method Responses
N/A - 100% shock lapse 9
Dukes-MacDonald or derivatives of Dukes-MacDonald 14
Canadian Institute of Actuaries Valuation Technique Paper #2 4
Flat Multiple 13
Other: Set by reinsurer/external consultant 3
Other: Internally developed method 3
16
Post-Level Mortality Assumptions vs. Cumulative Lapse Rate
 Mortality deterioration increases gradually with increasing lapse rates
Shock Lapse vs. Mortality Deterioration
10 Year term
800%
700% DM
Mortality Deteioration Duration 12
CIA
600%
Flat Multiple
500% Other
400%
300%
200%
100%
0%
30% 40% 50% 60% 70% 80% 90% 100%
Cumulative Lapse Rate Duration 10 through 11
17
Post-Level Mortality Assumptions vs. Premium Jump
 Mortality deterioration assumptions do not differ dramatically by
premium jump, inconsistent with experience
Premium Jump vs. Mortality Deterioration

1000% 10 Year term
Mortality Deterioration Duration 11
900%
800%
700%
600%
500%
400%
300%
200%
100%
0%
0.0 10.0 20.0 30.0 40.0
Premium Jump
18
Post-Level Mortality Assumptions vs. Prem Jump and Lapse Rate
 When split by lapse rate, it does appear assumptions vary by
premium jump
Shock Lapse vs. Mortality Deterioration by Premium Jump
600% 10 Year term
500%
1-10X Jump
11-20X Jump
Mortality Deteioration Duration 12
20X Jump or Greater

400%
300%
200%
100%
0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Cumulative Lapse Rate Duration 10 through 11
19
Post-Level Mortality Deterioration Structure
 Assumptions vary broadly by structure of mortality deterioration
Mortality Deterioration Assumption Trend By Duration

Responses
Description
10-Year Term 20-Year Term
Level all durations 7 6
Decreasing or grade to ultimate level 13 11
Increasing then decreasing 6 1
Generally increasing 1 3
20
Experience Study Results
21
Lapse Study Experience Results
Overview
2000-2012 Policy-Year Study
2014 Study 2010 Study Change
Number of Companies w/ Post Level Experience 36 25 144%
10-Year Term Post-Level Lapses with Premiums 436,307 87,544 498%
Post-Level Lapses without Premiums 258,030 170,171 152%
15-Year Term Post-Level Lapses with Premiums 37,673 12,191 309%
Post-Level Lapses without Premiums 1,364 1,359 100%
Source: http://www.soa.org/Research/Experience-Study/Ind-Life/Persistency/research-2014-post-level-shock.aspx
22
Lapse Study Experience Results – Jump to ART
Lapse Rates by Duration
T10 Jump to ART Lapse Rates by Duration
100% 1,000,000
90%
100,000
80%
Number of Lapses
70%
10,000
Lapse Rate
60%
50% 1,000
40%
100
30%
20%
10
10%
0% 1
6 7 8 9 10 11 12 13 14 15 16+
Lapse Rate Median Lapse Rate Number of Lapses (right axis)
23
Lapse Rates by Premium Jump
T10 Jump to ART
Duration 10 Lapse Rate by Duration 11/10 Premium Jump Ratio
100% 100,000
90%
80% 10,000
70%
Number of Lapses
Lapse Rate
60% 1,000
50%
40% 100
30%
20% 10
10%
0% 1
Duration 11/10 Premium Jump Ratio

Duration 10 Lapse Rate Count Duration 10 Lapses
24
Lapse Rates by Duration & Premium Jump
T10 Jump to ART
Lapse Rate Duration 11/10
Dur 11 / Dur 12 /
100%
by Duration 11/10 Premium Jump Ratio Premium Jump
Dur 10 Dur 11
Ratio Band
90% 1.01x - 2x 79.5% 59.5%
2.01x - 3x 77.0% 46.7%
80%
3.01x - 4x 73.2% 44.9%
70% 4.01x - 5x 73.1% 42.7%
60% 5.01x - 6x 70.5% 45.4%
6.01x - 7x 73.5% 45.1%
50%
7.01x - 8x 73.2% 46.3%
40% 8.01x - 9x 74.5% 45.9%
30% 9.01x - 10x 78.7% 54.7%
10.01x-12x 79.1% 49.7%
20%
12.01x-14x 79.7% 49.4%
10% 14.01x-16x 82.9% 56.8%
0% 16.01x + 86.0% 59.8%
Subtotal Prem Data Available 59.4% 42.9%
No Prem Data Available 55.0% 43.1%
Grand Total 57.4% 42.9%
Duration
Duration 10 Lapse Rate 11/10 Premium
Duration Jump Ratio
11 Lapse Rate Duration 12 Lapse Rate
Dur 11 / Dur 10 Dur 12 / Dur 11
25
Lapse Rates by Duration and Issue Age
T10 Lapse Rates by Issue Age

100%
90%
80%
70%
Lapse Rate
60%
50%
40%
30%
20%
10%
0%
0-19 20-29 30-39 40-49 50-59 60-69 70+
Issue Age
Duration 6-9 Duration 10 Duration 11 Duration 12+

26
Lapse Rates by Duration & Gender
T10 Lapse Rates by Gender
100%
90%
80%
70%
Lapse Rate
60%
50%
40%
30%
20%
10%
0%
Male Female
Gender
27
Lapse Rates by Duration & Risk Class
T10 Lapse Rates by Risk Class
100%
90%
80%
70%
Lapse Rate
60%
50%
40%
30%
20%
10%
0%
Super- Preferred Non-Preferred Undiff Preferred Non-Preferred Undiff
Preferred
Non-Smoker Smoker Unk/Agg
Risk Class
28
Lapse Rates by Duration & Face Amount Band
T10 Lapse Rates by Face Amount Band
100%
90%
80%
70%
Lapse Rate
60%
50%
40%
30%
20%
10%
0%
< $50,000 $50,000 - $100,000 - $250,000 - $1,000,000 - $5,000,000 +
$99,999 $249,999 $999,999 $4,999,999
Face Amount
29
Lapse Rates by Duration & Premium Payment Mode
T10 Lapse Rates by Premium Payment Mode

100%
90%
80%
70%
Lapse Rate
60%
50%
40%
30%
20%
10%
0%
Annual Semi-Annual Quarterly Monthly
Premium Payment Mode
30
Lapse Rates (duration 10) Premium Jump and Issue age
T10 Jump to ART
Duration 10 Lapse Rate
by Issue Age and
100% 100,000
90%
80% 10,000
70%
Lapse Rate
Number of Lapses
60% 1,000
50%
40% 100
30%
20% 10
10%
0% 1
<40 40-49 50-59 60+ <40 40-49 50-59 60+ <40 40-49 50-59 60+ <40 40-49 50-59 60+ <40 40-49 50-59 60+
1.01x - 3x 3.01x - 5x 5.01x - 7x 7.01x - 9x 9.01x +
Duration 10 Lapse Rate Duration 10 Lapses
31
Lapse Rates (duration 10) Premium Jump and Face Amount Band
T10 Jump to ART
Duration 10 Lapse Rate
by Face Amount and
100% 100,000
90%
80% 10,000
70%
Lapse Rate
Number of Lapses
60% 1,000
50%
40% 100
30%
20% 10
10%
0% 1
<100k 100k- 250k- 1M+ <100k 100k- 250k- 1M+ <100k 100k- 250k- 1M+ <100k 100k- 250k- 1M+ <100k 100k- 250k- 1M+
249k 999k 249k 999k 249k 999k 249k 999k 249k 999k
1.01x - 3x 3.01x - 5x 5.01x - 7x 7.01x - 9x 9.01x +
Duration 10 Lapse Rate Duration 10 Lapses
32
Skewness of Lapses
T10: Lapse Skewness by Month
Companies without Grace Period Adjustments
100%
12
90%
11
80%
10
70% 9
Proportion of Lapses
60% 8
7
50%
6
40%
5
30%
4
20% 3
10%
2
1
0%
6-9 10 11 12+
Duration
33
Lapse Rate
0%
100%
10%
30%
40%
50%
60%
70%
80%
90%
20%
1.01 - 2.00
2.01 - 3.00
3.01 - 4.00
4.01 - 5.00
5.01 - 6.00
6.01 - 7.00
7.01 - 8.00
8.01 - 9.00
9.01 - 10.00
Comparison of RGA to SOA
10.01 - 11.00
11.01 - 12.00
12.01 - 13.00
13.01 - 14.00
14.01 - 15.00
Premium Jump
15.01 - 16.00
16.01 - 17.00
17.01 - 18.00
18.01 - 19.00
19.01 - 20.00
20.01 - 21.00
21.01 - 22.00
22.01 - 23.00
23.01 - 24.00
24.01 AND UP
10-YR Term Duration 10 - Jump to ART
0
5,000
10,000
15,000
20,000
25,000
35,000
40,000
45,000
30,000
Lape Count
Rate
Rate
Count
Count
SOA Lapse
SOA Lapse
RGA Lapse
RGA Lapse
34
Mortality Study Experience Results
Overview
2000-2012 Calendar Year Study
2014 Study 2010 Study Change
10-Year Term Post-Level Claims with Premiums 2,651 382 694%
Post-Level Claims without Premiums 729 381 191%
15-Year Term Post-Level Claims with Premiums 432 145 298%
Post-Level Claims without Premiums 12 13 92%
35
Mortality Study Experience Results – Jump to ART
Mortality by Duration
T10 Jump to ART
450% Mortality by Duration 10,000
400%
350%
1,000
Mortality Relative to
Durations 6-10
300%
Number of Deaths
250%
100
200%
150%
10
100%
50%
0% 1
6 7 8 9 10 11 12 13 14 15 16+
A/E Mortality Median A/E Mortality Number of Deaths (right axis)
36
Mortality by Duration & Premium Jump
T10 Jump to ART
Duration 11 Mortality Relative to Level Period
2000%
by Duration 11/10 Premium Jump Ratio 10,000
1800%
1600%
1,000
1400%
Number of Deaths
Mortality Relative to
1200%
Durations 6-10
1000% 100
800%
600%
10
400%
200%
0% 1
1.01x - 2.01x - 3.01x - 4.01x - 5.01x - 6.01x - 7.01x - 8.01x - 9.01x - 12.01x- 15.01x- 18.01x +
2x 3x 4x 5x 6x 7x 8x 9x 12x 15x 18x
Relative Mortality Ratio Number of Deaths (right axis)
37
Mortality by Duration & Issue Age
T10 Jump to ART
2008 VBT Mortality Ratios
by Issue Age
400%
350%
300%
250%
200%
150%
100%
50%
0%
0-29 30-39 40-49 50-59 60+
Issue Age
Duration 6-10 Duration 11+ Post-Level Relative to Dur 6-10
38
Mortality by Duration & Gender
T10 Jump to ART
by Gender
350%
300%
250%
200%
150%
100%
50%
0%
Male Female
Gender
39
Mortality by Duration & Risk Class
T10 Jump to ART
550% by Risk Class
500%
450%
400%
350%
300%
250%
200%
150%
100%
50%
0%
Super- Preferred Non-Preferred Undiff Preferred Non-Preferred Undiff
Preferred
Non-Smoker Smoker Unk/Agg
Risk Class
40
Mortality by Duration & Face Amount Band
T10 Jump to ART
by Face Amount Band
500%
450%
400%
350%
300%
250%
200%
150%
100%
50%
0%
< $50,000 $50,000 - $100,000 - $250,000 - $1,000,000 +
$99,999 $249,999 $999,999
Face Amount Band
41
Mortality Study Experience Results
Mortality by Monthly Duration
T10 Duration 11
2008 VBT Mortality Ratios by Month
Companies without Grace Period Adjustments
350% 175
300% 150
250% 125
Number of Deaths
200% 100
150% 75
100% 50
50% 25
0% 0
1 2 3 4 5 6 7 8 9 10 11 12
Months since start of Duration 11
2008 VBT Mortality Ratios Number of Deaths (right axis)
42
Mortality Deterioration by Shock Lapse
T10 Jump to ART
Dur 10 Shock Lapse vs. Dur 11 Mortality Deterioration
by Company
1500%
1350%
Duration 11 Mortality Relative to Duraitons 6-10
1200%
1050%
900%
750% >=35 claims
<35 claims
600%
Expon. ()
450%
300%
150%
0%
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Duration 10 Shock Lapse
43
Mortality Deterioration by Shock Lapse
10-YR Term Duration 11 - Jump to ART

1200% 160
140
1000%
RGA Claim
120
Count
800%
100 SOA Claim
Claim Count
A/E 08VBT
Count
600% 80
RGA A/E
60
400% SOA A/E
40
200%
20
0% 0
1.01 - 3.01 - 5.01 - 7.01 - 9.01 - 11.01 - 13.01+
3.00 5.00 7.00 9.00 11.00 13.00
Premium Jump
44
Predictive Model
45
Predictive Model
 Multivariate Lapse Rate Model – T10 Duration 10 Shock Lapse

 Model:
 Generalized Linear Model (GLM)
 Target variable follows distribution in the exponential familly
 Response variable = observed lapse count
 Follows a Poisson distribution
 Benefits:
 Elimination of possible bias from a uni-variate approach
 Systematic way of controlling lapse assumption complexity
 Transparent insight into true drivers of lapse rates
 Distribution of target variables
46
Predictive Model – Model 1
Model Parameter Validation of Results
Actual Predicted
Data Actual /
Variable Type Coefficient Factor P-Value Lapse Lapse
Proportion Predicted
Rate Rate
Intercept - -2.676 <2.0E-16

Issue Age Numerical 0.0552 <2.0E-16
(Issue Age)^2 Numerical -0.000316 <2.0E-16
Risk Class Categorical
Super-Pref NS 0 1.00 0 11.3% 82.6% 82.6% 100.1%
NS -0.06736 0.93 <2e-16 76.9% 69.1% 68.8% 100.4%
SM -0.01296 0.99 0.0339 11.8% 63.5% 63.5% 100.0%
Face Amount Categorical
<50K 0 1.00 0 1.2% 14.9% 17.9% 83.2%
50-100K 0.600 1.82 <2.0E-16 3.9% 61.8% 62.7% 98.5%
100K-250K 1.262 3.53 <2.0E-16 51.9% 67.5% 67.4% 100.1%
250K-1M 1.559 4.75 <2.0E-16 36.9% 74.4% 73.8% 100.8%
>1M 1.585 4.88 <2.0E-16 6.1% 80.7% 80.4% 100.4%
Premium Mode Categorical
Annual 0 1.00 0 22.9% 85.8% 85.3% 100.5%
Semi/Quarter -0.1365 0.87 <2.0E-16 24.5% 74.9% 75.1% 99.7%
Monthly/BiWeekly -0.3506 0.70 <2.0E-16 45.4% 56.9% 56.6% 100.5%
Other/Unknown -0.08481 0.92 <2.0E-16 7.2% 85.3% 84.8% 100.7%
Cross Term Mixed
Issue Age:Face Amt <50 0 0
Issue Age:Face Amt 50-100K -0.001288 0.3162
Issue Age:Face Amt 100-250K -0.01074 <2e-16
Issue Age:Face Amt 250K-1M -0.01653 <2e-16
Issue Age:Face Amt >1M -0.0172 <2e-16
47
Model Predicted vs. Actual Lapse Rate

100% 80,000
90%
70,000
80%
60,000
70%
50,000
60%
Exposure
Lapse Rate
Exposure
50% 40,000
Predicted
40%
Actual 30,000
30%
20,000
20%
10,000
10%
0% 0
0-19 20-29 30-39 40-49 50-59 60-69 70+
Issue Age
48
100% 40,000
90%
35,000
80%
30,000
70%
25,000
Lapse Rate
Exposure
Exposure
60%
50% Predicted 20,000

Actual
40%
15,000
30%
10,000
20%
5,000
10%
0% 0
0-19 20-29 30-39 40-49 50-59 60-69 70+
Issue Age

100% 18,000
90% 16,000
Exposure
80%
14,000
Predicted
70%
Lapse Rate
Actual 12,000
Exposure
60%
10,000
50%
8,000
40%
6,000
30%
4,000
20%
10% 2,000
0% 0
49
Premium Jump
Acknowledgements
50
Acknowledgements
Special Thanks
 Contributing Companies
 SOA and SOA Staff
 Project Oversight Group

 Tatiana Berezin
 Jeff Beckley
 Tony Phipps
 Brian Carteaux
 David Wylde
 RGA Contributors  Richard Xu

 Derek Kueker  Scott Rushing
 Tim Rozar  Minyu Cao
 Michael Cusumano  Kathleen Broom
 Susan Willeat
51

B Post Level Term Experience Results

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

B Post Level Term Experience Results

Uploaded by

Copyright:

Available Formats

SOA Post-Level Term

Actuaries’ Club of the Southwest

June 19, 2014

 Introduction to Post-Level Term

 SOA Sponsored Post-Level Term Reports

Level Premium Term Product Mix by Level Period

Post-Level Product Design

Changes to Post-Level Premium Structure for Term New Business

Changes to Post-Level Premium Structure for Term In Force

Relationship between Current Ultimate and Guaranteed Ultimate Premiums

Lapse Rate Trend By Duration

Premium Jump vs. Shock Lapse

Monthly Lapse Skewness During Level Premium Period

Monthly Lapse Skewness During Year of Shock Lapse

Monthly Lapse Skewness During Post-Level Period

Method of Developing Mortality Assumption

Cumulative Lapse Rate Duration 10 through 11

Premium Jump vs. Mortality Deterioration

20X Jump or Greater

Cumulative Lapse Rate Duration 10 through 11

Mortality Deterioration Assumption Trend By Duration

Lapse Rate Median Lapse Rate Number of Lapses (right axis)

Duration 11/10 Premium Jump Ratio

T10 Lapse Rates by Issue Age

Duration 6-9 Duration 10 Duration 11 Duration 12+

Duration 6-9 Duration 10 Duration 11 Duration 12+

Duration 6-9 Duration 10 Duration 11 Duration 12+

T10 Lapse Rates by Premium Payment Mode

Premium Payment Mode

Duration 6-9 Duration 10 Duration 11 Duration 12+

Duration 10 Lapse Rate Duration 10 Lapses

Duration 10 Lapse Rate Duration 10 Lapses

A/E Mortality Median A/E Mortality Number of Deaths (right axis)

Relative Mortality Ratio Number of Deaths (right axis)

750% >=35 claims

10-YR Term Duration 11 - Jump to ART

 Multivariate Lapse Rate Model – T10 Duration 10 Shock Lapse

Intercept - -2.676 <2.0E-16

Model Predicted vs. Actual Lapse Rate

50% Predicted 20,000

Model Predicted vs. Actual Lapse Rate

 SOA and SOA Staff

 Project Oversight Group

 RGA Contributors  Richard Xu

You might also like