Alan Greenspan:
President, Federal Reserve Board
May 1996
“… We should not forget that the basic economic
function of these regulated entities (banks) is to
take risk. If we eliminate risk taking in order to
reduce failure rates to zero, we will, by
definition, have eliminated the purpose of the
banking system.”
Everyday Questions
Balancing Marketing and Risk
Should I
target this
consumer?
… with what
message?
Will the
consumer
hear it?
Should I
approve?
… at what
credit level?
Will the
consumer
apply?
Will the
consumer
use it?
Increase value by improving
decisions
Use BI to optimize multiple
objectives
How will I
continue to
influence?
Will the consumer…
pay as agreed?
attrite too early?
build large balances?
repeat purchase?
buy add-on services?
be profitable?
Decision areas
• Solicitations
• New applications
• Account management
–
–
–
–
Credit line
Authorization
Collections
Reissue
• Cross-sell
• Keep / sell
Business Objectives
•
Increase consistency of lending decisions
– Consistent & unbiased treatment of applicant
• Customers with the same details get the same treatment
– Total management control over credit approval systems
• Allows for loosening or tightening of lending through credit cycles
• Potential increase in approvals
•
Reduce operating costs
– Increase in automated processing
•
Improve customer service
–
–
–
–
Fast and consistent decisions at application point
More appropriate limit and authorisation decisions
Reduction in collection actions on low risk accounts
Risk based allocation of credit limits and issue terms
10/18/2013
9
Business Objectives
•
Improved portfolio management
– Manage credit portfolios more effectively and
dynamically
•
•
•
•
Better prediction of credit losses
Management ability to react to changes fast & accurately
Ability to measure & forecast impact of policy decisions
Quick and uniform policy implementation
– Improved Information
• Permits information gathering to assist business needs and
marketing activities
• Information gathered can be fed back into future scoring
systems’ developments, collection activities and strategy
optimization
10/18/2013
10
The BI Solution: Scoring Models
DATA
MODEL
OUTCOME
BI in the Consumer Credit Industry
• Numerous quantitative advances have emerged in the
consumer credit risk area to support business strategy
throughout the customer life cycle - beyond simple
credit scores.
• At credit origination, analytical models are used to:
–
–
–
–
–
–
–
Identify likely consumers who are likely to be profitable
Predict propensity to respond to a credit offering
Align consumer preferences with products
Assess borrower credit worthiness
Determine line/loan authorization
Apply risk-based pricing
Evaluate relationship value of the customer
BI in the Consumer Credit Industry
• Throughout loan servicing, analytical methods are used to:
–
–
–
–
–
–
Anticipate consumer behaviour (risk) or payment patterns
Determine opportunities for cross-selling
Assess prepayment risk
Identify any fraudulent transactions
Optimize customer relationship management
Prioritize the collections effort to maximize recoveries in the event
of delinquency
• Analytical models are fast becoming the back-bone of
efficient consumer credit risk management.
• Consumer lending represents an analytically robust and
data-rich environment for credit risk and capital
measurement.
Account Life-cycle Scoring Progression
ACCOUNT STATUS
ON TIME
Credit risk in consumer banking has been (traditionally)
driven by 3 C’s of lending (based on judgment):
1.Character – “willingness to re-pay debt”
2.Collateral – “incentive to re-pay debt”
3.Capacity – “ability to re-pay debt”
•
•
•
The presence of a large number of consumers makes this
environment ideal for empirical modeling to predict
borrower behaviour as the basis for acquisition and
management of customers.
Markets with robust credit bureaus further provide the
impetus to use models to predict borrower behaviour.
Credit scores can summarize the details of credit report
and application data into a single actionable metric.
Basic Concept of Credit Scoring
A statistical means of providing a quantifiable risk factor for a
given customer or applicant.
• Credit scoring is a process whereby information provided is
converted into numbers that are added together (hence it is an
example of ‘Generalized Additive Models’) to arrive at a score
(using a “Scorecard”).
• The objective is to forecast future performance from past
behaviour.
• Credit scoring developed by Fair & Isaac in early 1960s
– Widespread acceptance in the US in early 80s and UK
early 90s
– FICO scores make 75% of US Mortgage loan decisions
– Behavioural scoring, introduced later, has been accepted as
more predictive than application scoring
•
National (US) Distribution of
FICO Scores
30%
28%
% of Population
25%
19%
20%
16%
15%
12%
10%
11%
8%
5%
5%
1%
0%
Up to 499
500-549
550-599
600-649
650-699
FICO Score Range
700-749
750-799
800+
Bad Rates of Major FICO Score
Ranges
Score Range
Bad %
300-500
500-599
600-699
700-850
48%
30%
11%
1.50%
Evaluating the credit applicants: Judgment
Versus Scoring
CHARACTERISTICS
Time at present address
Time at present job
Residential status
Debt ratio
Bank reference
Age
Income
•
••
# of Recent inquiries
% of Balance to avail. lines
# of Major derogs.
Overall
Decision
Odds of repayment
JUDGMENT
CREDIT SCORING
+
+
+
+
N/A
-
12
20
5
21
28
15
5
+
+
+
-7
10
35
212
•
••
Accept
?
•
••
Accept
46:1
Risk Scorecard - Example
Own
Rent
15
25
Years at
<.5
.5-2.49
address
12
10
Prof
Semi-Prf
Occupation
50
44
<.5
.5-1.49
Years on job
2
8
Dept St /
None
Dept-St
Major CC
0
11
Bank
Chq A/c Sav A/c
reference
5
10
<15
15-25
Debt ratio
22
15
No. of recent
0
1
inquiries
3
11
<.5
1-2
Years in file
0
5
# of Loan
0
1-2
Accounts
5
12
% Credit line 0-15%
16-30%
utilization
15
5
Worst
No Rcrd Maj. Def.
reference
0
-59
Behavioural scoring is a statistical means of
assessing risk for existing customers through
internal behavioural data
– Customers/accounts scored repeatedly
The objective of a scorecard is to use characteristics that
discriminate between Good and Bad accounts with sufficiently
high accuracy.
The score is a measure of the probability of being a Good or
Bad performer.
If the scorecard is a good one then the mean score of ‘Bads’ is
lower than the mean score of the ‘Goods’.
Good/Bad Odds (probability)
•
•
•
•
•
A scoring system does not individually identify a good
performer from a bad performer, it classifies an
applicant in a particular “Good/Bad odds” group.
An applicant belonging to a 200 to 1 group, appears
pretty safe and profitable.
If the applicant belongs to a 4 to 1 risk group, we would
no doubt find the risk unacceptable.
There is a “cut-off” point where it is not profitable for
the bank to accept a certain Good to Bad ratio
Based on the above, it is accepted that there will be
some “bads” above the cut-off level set, and some
“goods” below the cut-off level set.
Credit score = odds (risk)
SCORE
ODDS
220
=
64 / 1
180
=
16 / 1
=
4/1
=
1/1
•
•
•
•
•
•
140
•
•
•
100
•
•
•
•
•
•
•
•
•
Scores are Calibrated (Aligned) to Odds
(on a Log Scale)
1
0
Ln(Odds)
-1
Actual Data
-2
Std Line
-3
-4
-5
505
515
525
535
545
555
565
575
585
595
Score
For any given score the probability of ‘Bad’ can be found
using the equation of the Log-odds (straight) line.
How Is a Credit Scoring Model
Developed ?
• Analysis of a large set of consumers (>= 1
Million)
• Identification of common variables that
define behavior
• Statistical models are then built that assign
weights to each variable
• Adding all variables combines to make an
individual score
Scorecard Construction
Data Gathering
•Population Identification
•Data Availability
•Data Extraction
•Sampling