Over-indebtedness & responsible lending in the UK ... · Over-indebtedness & responsible lending in...
Transcript of Over-indebtedness & responsible lending in the UK ... · Over-indebtedness & responsible lending in...
© 2005, Experian-Scorex Proprietary and Confidential #1
Over-indebtedness & responsible lending in the UK
Dr Paul RussellDirector - Analytical Solutions
www.experian-scorex.com [email protected]
© 2005, Experian-Scorex Proprietary and Confidential #2
Agenda
• The headlines
• UK credit market and indebtedness
• Industry initiatives
• Measuring consumer indebtedness
• Affordability and customer value management
• Conclusions
© 2005, Experian-Scorex Proprietary and Confidential #4
October 2004
• £100k debts reported
• 16 credit cards
• 20 credit lines
• Nearly every account at max utilisation
• Weekly income £340
• No serious arrears
© 2005, Experian-Scorex Proprietary and Confidential #5
• 800 out of 60,000 had debts > £100,000
• 2,500 owed 66x monthly income
• 200 had > 15 credit cards
• One had 48!
CCCS report February 2005
CCCS – Consumer Credit Counselling Service
© 2005, Experian-Scorex Proprietary and Confidential #6
This House……is horrified to find that existing credit checking is not working;
...believes that there has to be full information sharing between all financial institutions, clear credit to income ratios set for maximum credit limits, agreed triggers for industry-funded independent debt counselling and that such systems must operate across all lenders;
...calls on a self-regulated system to be agreed before the House rises in the summer and to be seen to be working before the next Queen's speech; and resolves that, in the absence of such a system, Parliament must regulate the industry within the next year.
Early Day Motion 934 (30 March 2004))
Westminster concerns
© 2005, Experian-Scorex Proprietary and Confidential #7
Agenda
• The headlines
• UK credit market and indebtedness
• Industry initiatives
• Measuring consumer indebtedness
• Affordability and customer value management
• Conclusions
© 2005, Experian-Scorex Proprietary and Confidential #8
UK indebtedness
• Personal debt now exceeds £1Tr
• Consumer credit is low relative to annual incomes
• Personal debt is dominated by secured lending
Source: Bank of England, Experian Business Strategies
© 2005, Experian-Scorex Proprietary and Confidential #9
0.00%
1.4% 1.5%
2.3%
3.3%
1.5% 1.4%
1.8%
1.3%
2.1%
0.43%
0.26% 0.23%0.10% 0.09% 0.08% 0.07% 0.06% 0.04% 0.03% 0.01% 0.02%
2.7%
1.0%
1.3%1.2%
1.4%
1.8%
2.0%
3.0%
2.2%
1.8%1.8%
1.7%
1.1%
1.5% 1.2%
0.27% 0.29% 0.25% 0.32% 0.29%0.33%
0.38%0.29% 0.34%0.37% 0.29%
0.0%
0.5%
1.0%
1.5%
2.0%
2.5%
3.0%
3.5%
1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004
The UK retail credit marketBank of England Write-Off Rates (%)
Over the last 10 years, average write-off levels across the UK have increased on average by 5.3% p.a., with a marked decline in mortgage write-off levels. However, credit card write-off levels have surged over the last 5 years, reaching 3.3% by end 2004.
Source: Bank of England, Experian
Mortgages
Personal Loans
Credit Cards
TotalCAGR= 5.3%
© 2005, Experian-Scorex Proprietary and Confidential #10
18.8%
11.1% 11.4% 11.8%12.8% 12.9% 13.0%
14.1%15.4%
16.5%17.5%
18.3% 18.4%
17.9%
9.4% 9.3% 8.9%7.6% 7.2%
8.6% 8.2% 8.0% 7.9% 7.2% 7.3% 7.5%8.8%
11.6%13.1%
11.0%
27.6% 27.9% 27.7%
18.7%18.6%
20.4% 21.0% 20.9% 20.9% 21.2%22.7%
23.9%
26.3%
28.9%
31.5%29.9%
0%
5%
10%
15%
20%
25%
30%
35%
Mar
-96
Sep-
96
Mar
-97
Sep-
97
Mar
-98
Sep-
98
Mar
-99
Sep-
99
Mar
-00
Sep-
00
Mar
-01
Sep-
01
Mar
-02
Sep-
02
Mar
-03
Sep-
03
Mar
-04
Sep-
04
Mar
-05
Sep-
05
Mar
-06
Sep-
06
Mar
-07
Sep-
07
The UK retail credit marketDebt Servicing Payment* as a Percentage of Income - Total
Households now spend around 25% of their income to service debt (up from 20% 10 years ago), and this is predicted to rise to 28-32% over the next few years.
* Debt servicing payment is the sum of capital and interest payments for mortgages, credit cards and other unsecured lending.Source: Bank of England, Experian Business Strategies
Total
Capital
Interest
Actual Projected
Downside ScenarioOptimistic Scenario
© 2005, Experian-Scorex Proprietary and Confidential #11
0.38%
0.43%
0.29%
0.29%
0.25%
0.32%
0.29%
0.33%
0.34%
0.47%
0.50%
0.54%
0.20%
0.25%
0.30%
0.35%
0.40%
0.45%
0.50%
0.55%
0.60%
15% 17% 19% 21% 23% 25% 27% 29%
19961997199819992000200120022003200420052006 Optimistic2006 DownsideLinear (Trendline)
The UK retail credit marketWrite-Off Rate by Debt Servicing (12 Month Lag)
Write-off levels have traditionally been closely related to debt servicing to income ratios, and further increases in the ratio could see write-off levels reach 0.50% in the next few years (from 0.43% today).
Note: The graphs shows write-off % in a given year and debt servicing ratio in the previous year.Source: Bank of England, Experian
Using the trend line, a prediction of write-off can be estimated for 2005 and for both scenarios in 2006
© 2005, Experian-Scorex Proprietary and Confidential #12
UK indebtedness: survey data
Source : Bank of England and NMG Research. Figures up to 2002 are calculated using the BHPS, 2003 and 2004 figures are from the NMG Research surveys.
• Significant differences between owners and renters
• Renters are more likely to have problems but their share of total debt is small
• The vast majority of debt is attributable to homeowners but veryfew have current problems
• 40% of total household debt is owed by those spending over 25% of gross income servicing debts
• The share of debt owed by those with current debt problems is lower than a decade ago
• Generally, UK credit industry is very healthy
• Isolated cases of irresponsible borrowing exist
• Detection of those cases can be improved
© 2005, Experian-Scorex Proprietary and Confidential #13
Agenda
• The headlines
• UK credit market and indebtedness
• Industry initiatives
• Measuring consumer indebtedness
• Affordability and customer value management
• Conclusions
© 2005, Experian-Scorex Proprietary and Confidential #14
Industry initiatives
• Department for Trade and Industry (DTI) Task Force on Indebtedness
• Financial Services Authority Financial Capability Initiative
• APACS agreement to share all available data
• Industry/DTI/Department for Constitutional Affairs – historical consent
• Steering Committee on Reciprocity working group – new data sources
Student loansUtilities
© 2005, Experian-Scorex Proprietary and Confidential #15
Supporting industry initiatives – The Debt Test
• Simple questions about personal financial circumstances
• Answers ‘scored’ to determine likelihood of indebtedness problems
• Score drives ‘risk messages’:Suggestions on how to improve the situationSources of help – Citizen’s Advice etc
• Supports ‘What-if’ scenarios
• Initial launch on BBC website
© 2005, Experian-Scorex Proprietary and Confidential #16
Agenda
• The headlines
• UK credit market and indebtedness
• Industry initiatives
• Measuring consumer indebtedness
• Affordability and customer value management
• Conclusions
© 2005, Experian-Scorex Proprietary and Confidential #17
Measuring consumer indebtedness
Full data sharing will help
UK credit industry has committed to share full data on most products by 2006
Approximately 40 credit accounts that cannot be shared due to lack of consent
Need to properly quantify individual indebtedness
A more refined view of credit bureau data has led to better understanding of individual commitments
Developed the Consumer Indebtedness Index (CII) as a predictive measure of risk based upon a more detailed view of indebtedness
True affordability requires some knowledge of income
Developed Estimated Disposable Income models to assist in the assessment of individual affordability.
Developed an Affordability Index (AI) as a predictive measure of risk based upon a good understanding of individual affordability
© 2005, Experian-Scorex Proprietary and Confidential #18
Research - existing credit card customers
G:B Odds
• Highly-committed customers
identified via current Delphi block 10.4:1
now identified 2.6:1
• Customers with current payment problems 2.4:1
© 2005, Experian-Scorex Proprietary and Confidential #19
Segmentation
Total Population
No Trace of Address
Valid Address
Very Low Balances
All other balances
Very High Balances
Recent Derogs (Payment problems L3Y)
Uns Balance <= £2,000
Uns Balance > £2,000
Uns Balance <= £2,000
Uns Balance > £2,000
No Recent Derogs & No Derogs
Low Balance
High Balance
Low Balance
High Balance
© 2005, Experian-Scorex Proprietary and Confidential #20
Consumer Indebtedness Index (CII)
• Number of active accounts in use
• Number of revolving accounts in use
• Limit utilisation across revolving accounts
• Type of neighbourhood1.921-30
10.661-70
16.371-80
30.481-90
57.791-100
4.341-50
6.551-60
0.71-10
1.311-20
3.431-40
Bad Rate (%)Index
Key Predictors Strong Prediction
© 2005, Experian-Scorex Proprietary and Confidential #21
Results
Credit applicants with high balances (>£2k)
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80 90 100
% Goods
% Ba
ds
CII
Traditional Delphi Score
© 2005, Experian-Scorex Proprietary and Confidential #22
Consumer Credit Counselling Service (CCCS)
• Consumers approaching CCCS in 1 month analysed (~3,000)
• Credit bureau data (including CII) attached retrospectively at 3 monthly intervals over previous year
• AnalysisCompare CII profile of indebted cases with general populationCould lenders have made better decisions?
© 2005, Experian-Scorex Proprietary and Confidential #23
CII Comparison
SP Cumulative % Population
-
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.80
0.90
1.00
0 2 4 6 8 10 13 15 17 19 21 24 27 31 33 38 40 42 46 48 52 56 62 64 68 70 78 86 91 93 95 97 99
CII Value
Perc
enta
ge
CCCS Cumulative % CIB Cumulative %
~26%
~95%
© 2005, Experian-Scorex Proprietary and Confidential #24
Affordability-based methodology
• Standard Good/Bad DefinitionBad - current delinquency 3+
current delinquency = 2 and overlimitIndet - current delinquency = 2
current delinquency = 1 and worst delinq L24> 1overlimit
Good - current delinquency = 0current delinquency = 1 and worst delinq L24 = 1
• Affordability-based Good/Bad DefinitionBad - currently classified as 'bad' or CII at outcome = 50+Indet - currently classified as 'indet' or CII at outcome = 30 - 49Good - all other accounts
© 2005, Experian-Scorex Proprietary and Confidential #25
Affordability-based methodology
• Applicants very highly indebted at outcome (CII 50+)
247
378
659
# Accepts
-62.5%9.7Affordability-based methodology
-42.6%14.8Standard methodology
_25.9Current scorecard
ChangeAccept Rate
© 2005, Experian-Scorex Proprietary and Confidential #26
Agenda
• The headlines
• UK credit market and indebtedness
• Industry initiatives
• Measuring consumer indebtedness
• Affordability and customer value management
• Conclusions
© 2005, Experian-Scorex Proprietary and Confidential #27
Estimated Disposable Income
Estimated Disposable
Income
Net Monthly Income
Monthly Expenditure
MonthlyCredit
Commitments
Taken from application form
Derived from credit bureau data
Models based on application form information
© 2005, Experian-Scorex Proprietary and Confidential #28
Estimating Disposable Income
Estimated Disposable
Income
Net Monthly Income
Monthly Expenditure
MonthlyCredit
Commitments
Supplied Gross
Income
Tax& NI
Rules
MaximumIncomeModel
© 2005, Experian-Scorex Proprietary and Confidential #29
Estimating Disposable Income
Estimated Disposable
Income
Net Monthly Income
Monthly Expenditure
MonthlyCredit
Commitments
Personal Loan (CAIS) monthly
repayment
Proportion of Credit Card
(CAIS) balance
OtherAccounts
New Account(including
consolidation)
© 2005, Experian-Scorex Proprietary and Confidential #30
Estimating Disposable Income
Estimated Disposable
Income
Net Monthly Income
Monthly Expenditure
MonthlyCredit
Commitments
Mortgage CAISor estimated
Mortgage/Rent
Estimated essential
expenditure
© 2005, Experian-Scorex Proprietary and Confidential #31
Expenditure Survey Data
• Random sample of 12,096 Households with responses from 6,809 Households
• Response rate of 62%• Weekly expenditure amounts• Diary entries for 2 weeks
AdultsChildren
• Household and personal level values converted into totals for "Financial Units"
© 2005, Experian-Scorex Proprietary and Confidential #32
Expenditure Models
Models
Financial Units
Sub-pops
Mortgage
Rent
Expenditure
Grown-up
Children
-
Owners Pensioners
Renters
Owners Singles
Renters
Owners Working Families
Renters
Owners Low Income
Households Renters
Key PredictorsAgeIncomeUK regionCouncil Tax bandTime at address# ChildrenSingle parent
© 2005, Experian-Scorex Proprietary and Confidential #33
Affordability Index
• Builds on CII using income and other personal data
• AI is score-based using:• Estimated disposable income
• Consumer Indebtedness Index
• Residential/Marital Status
• Net Monthly Income (NMI)
• Monthly Credit Commitments as a % NMI
• Monthly Mortgage Payment as a % NMI
• Applicant’s Age
• AI is tuned to ‘potential accepts’ – not the through-the-door population
1
100
0 Unclassified
Very Low Affordability
Very High Affordability
.
.
.
.
.
.
© 2005, Experian-Scorex Proprietary and Confidential #34
Results - standard cut off
1.32.07.223.542.651.661.168.776.884.6
5.57.29.78.911.213.424.641.080.5
239.5
Decile 1Decile 2Decile 3Decile 4Decile 5Decile 6Decile 7Decile 8Decile 9
Decile 10
Accept RateG:B OddsApp. Score
Gini Coefficient = 49.4
Affordability Index case study
Traditionally a cut-off decision has been driven by a requirement to accept a certain quality of applicant, as defined by the GB odds, or have an accept rate in line with the company’s growth plans.
© 2005, Experian-Scorex Proprietary and Confidential #35
Results - Affordability Index
10.024.633.555.579.9
5.510.718.540.4112.1
Lo – 6869 – 7677 – 8990 – 94
95+
Accept RateG:B OddsAI
Gini Coefficient = 46.7
Affordability Index case study
The Affordability Index is a highly predictive characteristic in its own right.
© 2005, Experian-Scorex Proprietary and Confidential #36
Results – G:B Odds misalignment
AIApp. Score
26.1112.140.418.510.75.5Total
5.57.29.78.9
11.213.424.641.080.5239.
5
-4.4-
30.726.590.883.368.4
138.5252.2
1.55.86.6
21.119.921.332.954.499.4617.4
6.910.79.08.1
11.611.823.541.148.093.1
8.04.59.97.28.48.8
12.415.924.4
-
3.24.49.94.94.35.55.08.0--
Decile 1Decile 2Decile 3Decile 4Decile 5Decile 6Decile 7Decile 8Decile 9
Decile 10
Total95+90 – 9477 – 8969 – 76Lo – 68
Affordability Index case study
Using the Affordability Index, it can be shown that at the given cut-off there are large differences in the quality of the population.
© 2005, Experian-Scorex Proprietary and Confidential #37
Affordability Index case study
Results - potential improvements
Through the use of the Affordability Index, we can significantly reduce the bad rate or increase the accept rate.
3.0
3.2
3.4
3.6
3.8
4.0
40 42 44 46 48 50
Accept rate
Bad
Rat
e
CurrentSame Bad rate -
Improved accept rate by 7%
Same Accept rate -reduced bad rate by 11%
© 2005, Experian-Scorex Proprietary and Confidential #38
Estimated Disposable Income as a % of Net Monthly Income
1%14%10%15%20%40%
< - 50%-49 to - 1%
0%1 - 10%11 - 20%
>20%
% New A/csEDI / NMI %
Affordability Index case study
EDI/NMI % is not predictive of good/bad, but is another measure of affordability.
Small group with very low affordability. Large group with high affordability.
© 2005, Experian-Scorex Proprietary and Confidential #39
New credit application process
Delphi for New BusinessModule
CVM ModuleRisk Models
Application Input
Decline - Risk
Potential Accepts
Credit Decision
© 2005, Experian-Scorex Proprietary and Confidential #40
New credit application process
Delphi for New BusinessModule
DetectModule
CVM ModuleRisk Models
Application Input
Decline - Risk
Decline - Fraud
Potential Accepts
Accepts
FraudDecision
© 2005, Experian-Scorex Proprietary and Confidential #41
Delphi for New BusinessModule
CVM ModuleReward Models
DetectModule
CVM ModuleRisk Models
Application Input
Take most appropriate up-sell/ cross sell action
Decline - Risk
Decline - Fraud
Potential Accepts
Accepts
Up-Sell / Cross-SellDecision
New credit application process
© 2005, Experian-Scorex Proprietary and Confidential #42
Using the AI with the credit risk score
High Credit Score
(Low Risk)
Low Credit Score
(High Risk)
Risk Score Cut-Off
SCORE ACCEPTS
SCORE REJECTS
AFFORDABILITY INDEX
Previous Accepts now declined for low
affordability
Previous Refers with high AI's can be auto accepted
CREDIT SCORE
© 2005, Experian-Scorex Proprietary and Confidential #43
Agenda
• The headlines
• UK credit market and indebtedness
• Industry initiatives
• Measuring consumer indebtedness
• Affordability and customer value management
• Conclusions
© 2005, Experian-Scorex Proprietary and Confidential #44
Conclusions
• Sharper focus on indebtedness
• Increased pressure to demonstrate responsible lending
• Credit bureau data provides a more complete picture
• More than just data sharing
• Analytics will be vital in making sense of the data
© 2005, Experian-Scorex Proprietary and Confidential #45
Over-indebtedness & responsible lending in the UK
Dr Paul RussellDirector - Analytical Solutions
www.experian-scorex.com [email protected]