critical illness insurance - institute and faculty of …...critical illness insurance: cmi...

© 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


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


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


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


8


Agenda


• AC04 Diagnosis rates



• 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



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



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+


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


Agenda





• 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


CI Diagnosis Rates by Cause as % of All-causes Rates


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


27

Smoothed Annualised CI Diagnosis Rates by Cause


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


28

CI Diagnosis Rates by Cause as % of All-causes Rates


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


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


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


Agenda





• 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


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


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



Agenda





• 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

critical illness insurance - institute and faculty of …...critical illness insurance: cmi...

Documents