sunz 2011 - acc case study

Projection of ACC Long Term Claim Numbers

weekly compensation

Todd NicholsonBee Wong Sim

24 February 2011

Section 1: BackgroundWeekly Compensation, Long-term claims, RIS

• Weekly compensation (WC) claims are those that receive “income replacement” payments after their accident

• Long-term WC claims are those that have received more than 365 days of weekly compensation

• Outstanding claims liability provision for WC at June 2011 is> $6 billion – 90% due to claims active one year post accident

• Recover Independence Service (RIS) set up in July 2009

Section 1: BackgroundNumber of Weekly Compensation Long-Term Claims

0

2,500

5,000

7,500

10,000

12,500

15,000

17,500

20,000

Jul-0

0

Jan-

01

Jul-0

1

Jan-

02

Jul-0

2

Jan-

03

Jul-0

3

Jan-

04

Jul-0

4

Jan-

05

Jul-0

5

Jan-

06

Jul-0

6

Jan-

07

Jul-0

7

Jan-

08

Jul-0

8

Jan-

09

Jul-0

9

Jan-

10

Section 1: BackgroundNet Change in Long-Term Claim Numbers (Rolling 12 Months) by Duration

-3,000

-2,500

-2,000

-1,500

-1,000

-500

0

500

1,000

Jun-

00

Dec

-00

Jun-

01

Dec

-01

Jun-

02

Dec

-02

Jun-

03

Dec

-03

Jun-

04

Dec

-04

Jun-

05

Dec

-05

Jun-

06

Dec

-06

Jun-

07

Dec

-07

Jun-

08

Dec

-08

Jun-

09

Dec

-09

Jun-

10

1 to 2.5 WC years 2.5+ WC years

RIS established

Section 2: Purpose• Project the optimal size of the long-term claims pool

• Gain an insight into the characteristics of long-term claims (what is the case-mix and is it likely to change over time?)

• Allow testing under different scenarios

• Assist in claims liability estimates

• Provide a better understanding of the impact of targeted intervention analysis and performance measures

Section 3: Modelling1. Segment the existing long-term WC claims pool to

understand claim-mix

2. Construct a survival analysis model to determine which factors influence claim duration

3. Construct a simulation model to project future long-term claim numbers

Section 3a: Segmentation1A

11%

1B13%

416%

55%

613%

PP2%

SI5%

26%

39%

720%

Section 3a: SegmentationSegment Claim

durationFund

(predominant) Gender Age at accident

1A short Earners more males more younger and older people

1B medium Earners more females fewer young people

2 long Treatment Injury more females more 50+, a lot more 60+

3 short Work more males more older people (particularly 60+)

4 short Work more males middle to late middle age

5 long Non-Earners more females more young people

6 long Motor Vehicle fewer females a lot more young people

7 very long Residual Claims mixed less older people

Section 3b: Survival Analysis• Want to model claim duration at the individual

claim level• So take into account all the claim characteristics

such as diagnosis, claimant age and claim management applied to that claim

• For example:– younger claimants should generally heel faster– seriously injured claimants should have longer claim

durations– claimants near to 65 years old have an upper bound on

duration

Section 3b: Survival AnalysisBut we have a problem with our training data• Some claims are still open so we don’t know their

final duration

Date of first payment Date of last payment Duration

1-Jan-11 15-Jan-11 151-Jan-11 10-Feb-11 411-Jan-11 23/02/2011 and still open ?

Section 3b: Survival Analysis

Reference Characteristics:40-50 year oldMaleReceiving $400-$520 per weekIn medium intensity employmentSoft tissue injuryUpper limb

Section 3b: Survival Analysis• lag between injury and lodgement of claim• multiple injury indicator• injury diagnosis• injury site• scene of injury• serious injury indicator• at work injury indicator• occupation• pre-injury work strenuousness• hours at weekend indicator• WC rate per week• gender• age at start of WC payment


-40% -20% 0% 20% 40% 60% 80% 100% 120%

less than 50

50-90

90-150

150-250

250-400

400-520

520-670

670-850

850-1100

1100-1400

Wee

kly

com

pens

atio

n ra

te ($

)

Change in odds of recovery compared with reference WC rate

Reference

Bad Good


-30% -20% -10% 0% 10% 20% 30% 40% 50% 60% 70%

less than 2020-3030-4040-5050-5555-6060-6161-6262-6363-6464-65

65+

Age

at th

e tim

e of

inju

ry

Changes in odds of recovery compared with reference age

Reference

Bad Good


Segment RIS hazard ratio

Claim duration

Fund (predominant) Gender Age at accident

1A 1.5 short Earners more males more younger and older people

1B 1.6 medium Earners more females fewer young people

2 2.0 long Treatment Injury more females more 50+, a lot more 60+

3 1.5 short Work more males more older people (particularly 60+)

4 1.4 short Work more males middle to late middle age

5 1.5 long Non-Earners more females more young people

6 2.0 long Motor Vehicle fewer females a lot more young people

7 1.5 very long Residual Claims mixed less older people

How effective is RIS?


Factor Change in chances of exit

Serious Injury -84%

Multiple Injury -16%

Female -16%

Injury at work -6%

Long lag between injury and claim -6%


Factor Influence on chances of exit Area of influence

Injury site best upper limb, lower limb

worst head/face, back/spine

Work type best very heavy, heavy

worst sedentary, light

Fund best Earners, Self-Employed Work

worst Residual Claims, Treatment Injury, Motor Vehicle


Injury type is an excellent predictor of claim duration for short term claims

Injury type is a poor predictor for long term claims

Therefore:For long term claims its less about the type of injury and more about other factors

Section 3b: Survival AnalysisTo get the predicted survival curve for each claim we:

• Take the baseline survival curve

• Adjust for the claim characteristics

• Adjust for how long they have survived so far

• Adjust for the introduction of the service delivery model

• Adjust for the claim being transferred to the RIS team


Reference Characteristics:40-50 year oldMaleReceiving $400-$520 per weekIn medium intensity employmentSoft tissue injuryUpper limb

Section 3b: Survival Analysis• So we now have a survival curve for each claim,

which takes into account that claim’s characteristics.

• We also know the date of birth so we can calculate when they will retire.

• So we can calculate a predicted duration for each claim

Section 3c: Simulation

60%

70%

80%

90%

100%

110%

Jan-09 Jul-09 Jan-10 Jul-10 Jan-11 Jul-11 Jan-12 Jul-12 Jan-13 Jul-13 Jan-14 Jul-14 Jan-15 Jul-15

Entry into long-term claims pool

Perc

enta

ge o

f cur

rent

num

ber

• We also need to take into account new claims• Simulate new claims using the most recent

year’s claims

Per

cent

age

of c

urre

nt n

umbe

r

Section 3c: Simulation• We test three scenarios to give a best case, a

most likely case and a worst case

Section 3c: SimulationTotal Number of Long-Term Claims

0

2,000

4,000

6,000

8,000

10,000

12,000

14,000

16,000

18,000

Jun-

01

Jun-

02

Jun-

03

Jun-

04

Jun-

05

Jun-

06

Jun-

07

Jun-

08

Jun-

09

Jun-

10

Jun-

11

Jun-

12

Jun-

13

Jun-

14

Jun-

15

Jun-

16

Jun-

17

Jun-

18

Jun-

19

Jun-

20

Scenario 1

Actual Projected

Scenario 3

Scenario 2

Section 3c: Simulation

0

2000

4000

6000

8000

10000

12000

14000

Jun-10 Jun-11 Jun-12 Jun-13 Jun-14 Jun-15 Jun-16 Jun-17 Jun-18 Jun-19 Jun-20

Scenario 2 - Total Long-Term Claims by Segment

Permanent Pension Serious Injury Segment 1A Segment 1B Segment 2Segment 3 Segment 4 Segment 5 Segment 6 Segment 7

Questions…

sunz 2011 - acc case study

Economy & Finance