critical illness insurance - institute and faculty of …...critical illness insurance: cmi...
TRANSCRIPT
© 2010 The Actuarial Profession www.actuaries.org.uk
Health and care conference 2011Dave Heeney and Dave Grimshaw, CMI Critical Illness Committee
Critical Illness insuranceAC04 diagnosis rates
20 May 2011
Critical Illness insurance:CMI 2003–2006 diagnosis rates
Agenda
• Summary of past work
• AC04 Diagnosis rates (Working Paper 50)
• 2003-2006 Cause-specific rates
• Analysis of subsets of data
• Future work
1
CMI Critical Illness – Outputs
• 2005 Results for 1999, 2000, 2001, 2002
WP14 – Initial methodology (Grossing-up factors)
Dec 05: WP18 – Feedback on WP14 & future work
• 2007 2003 (Revised) and 2004 (Unadjusted) Results
WP28 – Towards improved methodology
• 2008 WP33 – A new methodology (Adjusted Results)
1999-2004 Adjusted Results
2005 Unadjusted and Adjusted Results
• 2009 2006 Unadjusted and Adjusted Results
2003-2006 Unadjusted Results
• 2010 WP43 – Diagnosis Rates (Accelerated 1999-2004)
2003-2006 Adjusted Results
Draft 2003-2006 diagnosis rates
• 2011 WP50 – Diagnosis Rates (Accelerated 2003-2006) 2
• „Unadjusted Results‟ / WP14 methodology
– Actual Settled Claims v Expected Diagnosed Claims
– Mismatch ... „Grossing-up factors‟
• „Adjusted Results‟ / WP33 methodology
– Actual Settled Claims v Expected Settled Claims
– Match A & E, but presented using settlement timing
• Diagnosis Rates / WP43 methodology
– Derive from „Adjusted Results‟ / WP33 methodology
– Smoothed, fitted diagnosis rates for claims settled in 99-04
CMI Critical Illness – Methodology
3
CMI Critical Illness - WP33 Methodology
• CMI CI data / analysis problem:
– Claims collected by year of settlement; diagnosis date often
unknown; material lag from diagnosis to settlement
• Start with the known in-force and settled claims
0
50000
100000
150000
200000
250000
300000
1995 1996 1997 1998 1999 2000 2001 2002 2003
In Force at 1 Jan
0
100
200
300
400
500
600
1995 1996 1997 1998 1999 2000 2001 2002
Settled Claims
4
• From known in-force, estimate prior years in-force
– Roll back known data (over time, age and duration)
– Add back an estimate of business exiting before start date
In Force
0
50000
100000
150000
200000
250000
300000
350000
1995 1996 1997 1998 1999 2000 2001 2002 2003
CMI Critical Illness - WP33 Methodology
5
• From the in-force, estimate exposure in each year, then estimate
diagnosed claims by year (at each age & duration) using an initial
set of claim rates
In Force
0
50000
100000
150000
200000
250000
300000
350000
1995 1996 1997 1998 1999 2000 2001 2002 2003
0
100
200
300
400
500
600
700
1995 1996 1997 1998 1999 2000 2001 2002
Diagnosed Claims
CMI Critical Illness - WP33 Methodology
6
• From estimated diagnosed claims by year, estimate settled
claims by year (by age & duration) using an assumed claim
development distribution (CDD)
0
100
200
300
400
500
600
700
1995 1996 1997 1998 1999 2000 2001 2002
Diagnosed Claims
0
100
200
300
400
500
600
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Estimated Settled Claims
NB Max interval from diagnosis to settlement = 2 years in this illustration
CMI Critical Illness - WP33 Methodology
7
• Compare estimate of expected settled claims in investigation
period with known settled claims by year, age and duration
• Produces „adjusted‟ results (Actual Settled Claims/Expected
Settled Claims), for a given base table and CDD
• WP43 – Used to derive a set of „best fit‟ CI claim diagnosis rates
0
100
200
300
400
500
600
1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
Expected Settled Claims
0
100
200
300
400
500
600
1995 1996 1997 1998 1999 2000 2001 2002
Actual Settled Claims
CMI Critical Illness - WP33 Methodology
8
Critical Illness insurance:CMI 2003–2006 diagnosis rates
Agenda
• Summary of past work
• AC04 Diagnosis rates
• 2003-2006 Cause-specific rates
• Analysis of subsets of data
• Future work
9
Benefits of moving to 2003-2006 dataset
• More up-to-date
• Experience in 1999-2004 appears to have reduced in period
• Less affected by changes in the critical illness market?
• Shorter period (4 years v 6 years) ... But similar number of
settled claims
• Higher % of claims with date of diagnosis CDD more reliable
• Reduced dependency on off rates
• More stable contributing offices
• Analysis of two periods may show “real” features
10
Benefits of moving to 2003-2006 dataset
11
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
2003 2004 2005 2006
CMI accelerated critical illness exposure in 2003 to 2006, by office
Other
E
D
C
B
A
Figure 2.1 from Working Paper 50
Vulnerability of 2003-2006 dataset to market changes
12
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
1988 1990 1992 1994 1996 1998 2000 2002 2004 2006
CMI accelerated critical illness settled claims by policy commencement year
1999-2004
2003-2006
2003-2006 Claim Development Distribution (CDD)
Comparison of 2003-2006 CDD with the 1999-2004 CDD:
13
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 200 400 600 800 1,000
Pro
ba
bilit
y d
en
sit
y
Cu
mu
lati
ve
Pro
ba
bilit
y
Time interval from diagnosis and settlement (days)
1999-2004
2003-2006
Older age rates
• Need for full age-range table for the pricing, experience analysis
and valuation of whole-of-life policies
• But 2003-2006 dataset contained credible volumes of data for
only a limited age range
• Pragmatic approach used to extend age range
• A number of approaches to developing rates up to age 85 were
investigated
14
Older age rates: Approach
• Set target at age 85 for insured table as a percentage of population
table (CIBT02) based on ratio of insured to population mortality
• Extrapolate ultimate male and female rates at aggregate level to
progress smoothly from 60 to target at 85
• Derive ultimate smoker-segregated rates using:
– assumed proportion of smokers (observed to reduce with
increasing age) and
– assumed smoker to non-smoker differential at each age
(assumed to converge with increasing age)
• From age 86, project rates so that they increase at a lower rate at
each age, reaching unity at 110
15
Older age rates
16
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
70 75 80 85 90 95 100 105 110
Age
Progression of rates to unity at very old ages
ACMNL04
ACMSL04
ACFNL04
ACFSL04
Figure C.3 from Working Paper 50
2003-2006 All-causes Diagnosis Rates (AC04 rates)
Smoothed Annualised CI Diagnosis Rates by Gender and
Smoker Status; Accelerated CI; Ultimate; 2003-2006
17
0.000
0.005
0.010
0.015
0.020
0.025
0.030
0.035
25 30 35 40 45 50 55 60 65
Age exact
Male Non-Smokers
Male Smokers
Female Non-Smokers
Female Smokers
Figure 8.2 from Working Paper 50
2003-2006 All-causes Diagnosis Rates (AC04 rates)
Smoothed Annualised CI Diagnosis Rates by Gender and
Smoker Status; Accelerated CI; Ultimate; 2003-2006 as % of CIBT02
18
30%
40%
50%
60%
70%
80%
90%
100%
20 25 30 35 40 45 50 55 60 65 70 75 80 85
Age exact
Male Non-Smoker
Male Smoker
Female Non-Smoker
Female Smoker
Figure 8.5 from Working Paper 50
2003-2006 All-causes Diagnosis Rates (AC04 rates)
Durational pattern in Smoothed Annualised CI Diagnosis Rates
Accelerated CI; 2003-2006
19
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Male Non-Smokers
Male Smokers
Female Non-Smokers
Female Smokers
Dn0/Dn5+
Dn1/Dn5+
Dn2/Dn5+
Dn3/Dn5+
Dn4/Dn5+
Figure 8.4 from Working Paper 50
Key features of WP50 work
• Single CDD, based on data across 4 gender/smoker datasets
• Rates fitted by age only and by duration only, to broadly fit the
ESC‟s to the ASC‟s; each gender /smoker dataset is considered
independently
• Different selection patterns;
– Male Non-smoker 0, 1-4, 5+
– Male Smoker 0, 1-2, 3+
– Female Non-smoker 0, 1-4, 5+
– Female Smoker 0, 1, 2+
• Rates have been extended outside the age range where there
is credible volumes of data20
CMI Critical Illness: Scope of AC04 rates
• All-causes accelerated critical illness; Lives table only
• Based on claims settled in 2003-2006
• Four tables:
– ACMNL04
– ACMSL04
– ACFNL04 and
– ACFSL04.
• Durations 0,1,2,3,4 and 5+ for ages 18 to 65; ultimate only for
ages 66+
• Age exact basis
21
Critical Illness insurance:CMI 2003–2006 diagnosis rates
Agenda
• Summary of past work
• AC04 Diagnosis rates
• 2003-2006 Cause-specific rates
• Analysis of subsets of data
• Future work
22
2003-2006 Cause-specific rates: Scope
• Restricted dataset
• Main causes only (200+ claims):
– M Non-Smoker Cancer, HA, Death, Stroke, CABG, TPD
– M Smoker Cancer, HA, Death
– F Non-Smoker Cancer, Death, Stroke, MS
– F Smoker Cancer, Death
• Cause-specific CDDs again combine data across 4
gender/smoker datasets (except Cancer)
23
2003-2006 Cause-specific CDDs
24
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 100 200 300 400 500 600 700 800 900 1,000
Cu
mu
lati
ve
pro
ba
bilit
y
Interval (days)
Central
Cancer (male only)
Cancer (female only)
Heart Attack
Deaths
Stroke
CABG
TPD
Multiple Sclerosis
2003-2006 DRAFT Cause-specific Diagnosis Rates
Smoothed Annualised CI Diagnosis Rates by Cause
Accelerated CI; Durations 5+; 2003-2006
0.000
0.002
0.004
0.006
0.008
0.010
0.012
30 35 40 45 50 55 60
Age
Male Non-Smokers
Residual
TPD
CABG
Stroke
Deaths
Heart attack
Cancer
25
0.000
0.005
0.010
0.015
0.020
0.025
30 35 40 45 50 55 60
Age
Male Smokers
Residual
Deaths
Heart attack
Cancer
2003-2006 DRAFT Cause-specific Diagnosis Rates
CI Diagnosis Rates by Cause as % of All-causes Rates
Accelerated CI; Durations 5+; 2003-2006
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
30 35 40 45 50 55 60
Age
Male Non-smokers
Residual
TPD
CABG
Stroke
Deaths
Heart attack
Cancer
26
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
30 35 40 45 50 55 60
Age
Male smokers
Residual
Deaths
Heart attackCancer
2003-2006 DRAFT Cause-specific Diagnosis Rates
27
Smoothed Annualised CI Diagnosis Rates by Cause
Accelerated CI; Durations 5+; 2003-2006
0.000
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008
0.009
30 35 40 45 50 55 60
Age
Female Non-smokers
Residual
MS
Stroke
Deaths
Cancer
0.000
0.002
0.004
0.006
0.008
0.010
0.012
0.014
30 35 40 45 50 55 60
Age
Female smokers
Residual
Deaths
Cancer
2003-2006 DRAFT Cause-specific Diagnosis Rates
28
CI Diagnosis Rates by Cause as % of All-causes Rates
Accelerated CI; Durations 5+; 2003-2006
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
30 35 40 45 50 55 60
Age
Female Non-smokers
Residual
MS
Stroke
Deaths
Cancer
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
30 35 40 45 50 55 60
Age
Female smokers
Residual
Deaths
Cancer
Comparison of DRAFT cause-specific rates
29
0
0.001
0.002
0.003
0.004
0.005
0.006
0.007
30 35 40 45 50 55 60
Age
Ultimate cancer rates by age
Male non-smokers
Male smokers
Female non-smokers
Female smokers
0
0.001
0.002
0.003
0.004
0.005
0.006
0.007
30 35 40 45 50 55 60
Age
Ultimate death rates by age
Male non-smokers
Male smokers
Female non-smokers
Female smokers
Comparison of DRAFT cause-specific rates
30
0
0.001
0.002
0.003
0.004
0.005
30 35 40 45 50 55 60
Age
Ultimate heart attack rates by age
Male non-smokers
Male smokers
0.0000
0.0005
0.0010
0.0015
0.0020
30 35 40 45 50 55 60
Age
Ultimate stroke rates by age
Male non-smokers
Females non-smokers
2003-2006 DRAFT Cause-specific Diagnosis Rates
31
Durational pattern in CI Diagnosis Rates at durations 0,1,2,3 and 4 as
a percentage of duration 5+ rates; 2003-2006
0%
20%
40%
60%
80%
100%
120%
Male non-smokers
Male smokers
Female non-smokers
Female smokers
Cancer rates
Dn0
Dn1
Dn2
Dn3
Dn4
0%
20%
40%
60%
80%
100%
120%
Male non-smokers
Male smokers
Female non-smokers
Female smokers
Death rates
Dn0
Dn1
Dn2
Dn3
Dn4
2003-2006 DRAFT Cause-specific Diagnosis Rates
32
0%
20%
40%
60%
80%
100%
120%
140%
Male non-smokers Male smokers
Heart attack rates
Dn0
Dn1
Dn2
Dn3
Dn4
0%
20%
40%
60%
80%
100%
120%
Male non-smokers Female non-smokers
Stroke rates
Dn0
Dn1
Dn2
Dn3
Dn4
Durational pattern in CI Diagnosis Rates at durations 0,1,2,3 and 4 as
a percentage of duration 5+ rates; 2003-2006
Critical Illness insurance:CMI 2003–2006 diagnosis rates
Agenda
• Summary of past work
• AC04 Diagnosis rates
• 2003-2006 Cause-specific rates
• Analysis of subsets of data
• Future work
33
CMI Critical Illness: Scope of other analyses
• Experience split by
– Office
– Sum assured band
– Distribution channel
– Year of commencement
• Central CDD used except for analysis by office
• Published in Working Paper in 2011
34
2003-2006 Office-specific CDDs
35
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
0 100 200 300 400 500 600 700 800 900 1,000
Cu
mu
lati
ve
pro
ba
bilit
y
Interval (days)
Central
Office A
Office B
Office C
Office D
Office E
Office F
Office G
Office H
Office I
Experience for 5 selected offices
36
70
80
90
100
110
120
130
Office B Office D Office E Office G Office H
Male Non-smoker
Male Smoker
Female Non-smoker
Female Smoker
Results are expressed as a percentage of the AC04 diagnosis rates
Experience by sum assured
37
80
85
90
95
100
105
110
£0 - £40,000 £40,001 - £80,000 £80,000+
Benefit Amount Band
Male Non-smoker
Male Smoker
Female Non-smoker
Female Smoker
Results are expressed as a percentage of the AC04 diagnosis rates
Experience by distribution channel
38
80
85
90
95
100
105
110
Bancassurer Direct Sales IFA
Distribution Channel
Male Non-smoker
Male Smoker
Female Non-smoker
Female Smoker
Results are expressed as a percentage of the AC04 diagnosis rates
Experience by year of commencement
39
80
85
90
95
100
105
110
Pre - 2000 Post - 1999
Year of Commencement
Male Non-smoker
Male Smoker
Female Non-smoker
Female Smoker
Results are expressed as a percentage of the AC04 diagnosis rates
Critical Illness insurance:CMI 2003–2006 diagnosis rates
Agenda
• Summary of past work
• AC04 Diagnosis rates
• 2003-2006 Cause-specific rates
• Analysis of subsets of data
• Future work
40
CMI Critical Illness: Future work
• Cause-specific rates (expected to be published May 2011)
• Analysis by benefit amount, distribution channel, year of
commencement, office and product type
• Sensitivity of final rates, e.g. to CDD
• Imputed stand-alone rates + comparison of experience
• Published in separate Working Paper later in 2011
41
Critical Illness:Learning from experience
Questions??
Slides on website next week:
www.actuaries.org.uk/research-and-resources
Any queries to [email protected]
42