An investment approach

to reducing long

term cost –

lessons from welfare

Hugh Miller

© Taylor Fry

What is a welfare system?

• A series of fortnightly payments to support cost of living

• Significant variation in duration, needs and outcomes

• The most common form of exit is finding employment

Has many fundamental parallels with injury schemes

How is a welfare system traditionally managed?

• Budget forward estimates

• Emphasis on ‘stock’, not flow

• Changes to policy conditions

with limited evidence base

• Supplementary benefits in

separate categories

• Movements across benefit

types not well understood 0

200

400

600

800

1,000

1,200

1,400

1,600

An

nu

al p

aym

en

ts (

NZ$

m)

Supported living payment Work ready Jobseeker

NZ govt, 2012 budget papers

Welfare Reform in New Zealand

"the scale and consequences of long-term benefit receipt are deeply concerning .... [welfare] is not sustainable, it does not provide equal and fair opportunities for those people on different benefit types and it is associated with poor social outcomes.“

Welfare Working Group (2011)

Key Recommendations:

• Up-front ‘investment approach’ to reduce benefit dependency

• Annual injury style valuations to make long-term costs of the benefit system transparent

• Accountabilities linked to managing long-term cost

Program of reform then developed by Ministry of Social Development (MSD) and NZ Treasury

Motivation

If we can

improve

employment

outcomes then

everybody wins

The Investment Approach

• Annual external valuation of lifetime welfare cost

• Incorporate macroeconomic trends, policy changes, trends in

experience

Level 1: Top level valuation

• Classify clients in segments to tailor interventions and monitor outcomes

• Set targets an accountability

• Allocate resources based on segment needs

Level 2: Segment lifetime costs

• Run small trials and evaluate impacts

• Translate small scale results into likely liability impact

Level 3: Client and cohort level initiatives

Taylor Fry

valuation

MSD internal

analysis

Not an ordinary valuation…

A sophisticated methodology that gives

more power to the user

• Individual based, not payment type

• Project various lifetime pathways

• Large number of predictive factors

incorporated

• Allows for macro-economic

conditions

An integrated model that runs from

aggregate estimates to individual level

detailed projections

JS

Project likelihood of each

individual’s pathways

JS SLP

Transition

probabilities

depend on

over 20

characteristics

No SLP

Cost models sit

on top of each

quarterly state

What the valuation provides

1. Top level attribution

2. Drill down attribution

3. Detailed impacts of legislative

changes

4. Detailed impacts of programme

changes

5. Better understanding of cost drivers

6. Basis for reform and intervention

1. Top line attribution

0

10

20

30

40

50

60

70

80

Cu

rre

nt

clie

nt

liab

ility

($

b)

Change

Expenses

Non-beneficiaries

Youth

Supported Living

Sole Parents

Jobseekers

76.5 73.9 71.7 69.5 69.0

-2.6 -2.2 -2.2 -0.5

Reconciling 2013 valuation to the 2014 valuation result

2. Drill down attribution

-800

-600

-400

-200

0

200

400

WR

<1

WR

>1

HC

D<1

HC

D>1

Ch

0-2

Ch

3-4

Ch

5-1

3 <1

Ch

5-1

3 >1

Car

er

Par

tner

HC

D YP

YPP

SUP

<1

SUP

>1 OB

Rec

ent

exit

s

Join

s

Oth

er

Jobseekers Sole parents Supp. living Youth Non-ben

Ch

ange

on

exp

ecte

d li

abili

ty f

rom

Ro

ll-f

orw

ard

(N

Z$m

)

Allocation of $2.2b experience related reduction amongst welfare segments

3. Detailed impacts of legislative changes

40%

50%

60%

70%

80%

90%

100%

Sep-11 Dec-11 Mar-12 Jun-12

Pro

po

rtio

n o

f co

ho

rt

2011 Valn, 11,450 clients in SPS>14

Jobseeker or Sole parent >14 Jobseeker - sick Supp. Living Sole parent < 14 Supp benefits only Not on benefit

40%

50%

60%

70%

80%

90%

100%

Sep-11 Dec-11 Mar-12 Jun-12

Pro

po

rtio

n o

f co

ho

rt

2011 Valn, 11,450 clients in SPS>14

Sep-12 Dec-12 Mar-13 Jun-13

2012 Valn, 10,690 clients in SPS > 14

40%

50%

60%

70%

80%

90%

100%

Sep-13 Dec-13 Mar-14 Jun-14

2013 Valn, 8,850 clients in SPS >14

Oct-12 Jul-13

Transition pathways for clients who were Sole Parent Support recipients at 30 June each year

4. Detailed impacts of program changes

0%

20%

40%

60%

80%

100%

Stat

us

afte

r 2

ye

ars

Quarter of 17 year old IYB/YP snapshot

JS SPS/Other Not on benefits

0%

20%

40%

60%

80%

100%

Quarter of 18 year old DPB/YPP snapshot

SPS JS/Other Not on benefits

Youth Payt Young Parent PaytProgram rollout Program rollout Full Youth serviceFull Youth service

Two year welfare status for youth programme clients

5. Better understanding of cost drivers

0 100 200 300 400

Re

lati

ve n

um

be

rs

Inidividual liability ($000s)

Non-beneficiary

Supp. living

Sole parent

Jobseeker (sick)

Jobseeker

Distribution of individual level liabilities, clients aged 18-24

5. Better understanding of cost drivers Usefulness of predictors in estimating the lifetime welfare cost of current 18-24 year old clients

Current benefit type

Age of entry

Duration - any benefit

Inter-generational info

# qtrs Sole Parent

Gender

# qtrs Jobseeker

Ethnicity

Prev benefit

Education

District

Duration current benefit

Relative variable importanceSole parent and Supp.

living highest

Entry before 19 very predictive

>5 quarters continuously on benefits high risk

Increase for intensive family benefit history

These reflect Sole parent re-entry risk

Maori particularly highHeightened risk if Sole

parent or Supported inving in the past

5. Better understanding of cost drivers

60

70

80

90

100

110

120

130

140

Auckland Benefittype

History Region Ethnicity Age Gender Education Other Northland

Ave

rage

clie

nt

liab

ility

($

k)

Comparison of average client lifetime cost for Auckland and Northland regions

$10k of the difference

is due to a differing

distribution of benefit

types received

6. Basis for reform and intervention

0.0

0.5

1.0

1.5

2.0

2.5

16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64

Cu

rre

nt

clie

nt

liab

ility

($

b)

Age at valuation date

Entry before age 20 Entries after age 20

Age of entry estimated

Actual age of entry

75% of the liability is attributable to those who enter before age 20, even though these represent less than 5% of the current caseload

Lifetime welfare cost, split by current age and age of entry into welfare system

Putting it all

together –

crucial elements

Stakeholder buy-in

Interventions

Data

Modelling

Communication

What next?

Political success has broadened

the horizons of the investment

approach:

• Impact investing

• Other sectors

• Australian and international

movements

Technical success has

broadened the horizons of

individual based multivariate

modelling

• Social sector outcomes

• Injured worker outcomes

• Commercial sector

applications

Completing the circle –

What can injury schemes learn form welfare?

• Are you getting full value from your valuation?

– Betting linking operations and top level

control

– Longitudinal view of cost

• How effectively are initiatives measured and

targeted?

• How much can be gained by more detailed

insights?

Injury

schemes

Social

Sector

Learn

Learn

Further reading

• NZ Welfare 2014 Valuation report: https://www.msd.govt.nz/about-msd-and-our-

work/newsroom/media-releases/2015/reforms-succeed.html

• Welfare Working Group http://igps.victoria.ac.nz/WelfareWorkingGroup/Index.html

• More effective social services, NZ Productivity commission

http://www.productivity.govt.nz/inquiry-content/2032?stage=4

• NSW Government social impact investment hub

http://www.dpc.nsw.gov.au/programs_and_services/social_impact_investment