Instant Cash Advance & Payday Loan Information - Financial Help And Advice
Friday, 1 May 2015
AFFORDABILITY AND CREDITWORTHINESS
RESPONSIBLE LENDING
LENDING DECISIONS
RECORD KEEPING
PUBLIC INFORMATION
Call Credit Scoring
Overview
As part of
Callcredit’s standard data return, a generic credit risk score called Gauge is
displayed. All bureaus have a score of this nature – it’s typically what people
mean when they think of their credit score. Scores like Gauge are generally
built using a ‘twelve-month outcome’, meaning the score is intended to predict
which potential customers will have missed multiple payments in a year’s time.
In short-term
lending, money is loaned to a customer for a much shorter amount of time, and
as a result of this, traditional credit risk scores are not successful in
predicting which customers will and won’t repay. This has created the need for
an alternative method of deciding which applicants to accept.
To enable our
clients in short-term lending to make informed decisions that use the full
depth of our data, our consultancy team has created a score. Unlike Gauge, this
score is specifically designed to be used in short-term lending and has been
built using only single-payment loans of approximately one month in length.
The score build
A sample of 625,155 short term loans was sourced, using Good/Bad
outcomes supplied by clients.
The following
exclusions were applied to the file:
Exclusion
|
Removed records
|
Applicants not found on Callcredit's database
|
4,026
|
Applicants currently restricted
|
96
|
Applicants with a CCJ in last 12 months
|
23,307
|
Remaining population
|
597,725
|
The remaining 597,725 records had a bad
rate of 13.4%:
Outcome
|
Records
|
Bad
|
80,018
|
Good
|
517,707
|
Total
|
597,725
|
These records were then divided up into a Build and Holdout samples of
75% and 25% respectively. The score build took place on the Build sample, and
was later validated against the Holdout sample.
This build population then entered our score-building process, which
involves analysing each concept in isolation to evaluate how predictive they
are, before passing through coarse classing, variable clustering, and logistic
regression. A presentation describing this process in more detail is available
upon request.
Concepts in the
scorecard
Here is a list of the concepts contained within the scorecard. With
the exception of Age, all of these are sourced from Callcredit’s BSB data:
Age - Age of customer at Application (this is
derived from the application form)
- BCC - Number of months since most recent 1+ cycle excluding historic defaults
- BN - Total limits now as a % of total limits 12 months ago
- IZB - Value of Cash Advances in the Current Month as a % of Total Credit Card Repayment Amount in the current month
- JIC - Worst Payment Status On A Pay Day Loan Last 3 Months
- ND - Months since last CCJ
- PN - Total monthly repayments on any fixed term accounts which are settled
- RT - Number of defaults with a balance of >£100 but <£250
- SAC - Age of most recent active Home Credit account
- SEB - Number of all credit application searches in last 12 month
- YGC - Number of all credit application searches in last month
- VM - Total balances now as a % of total balances 3 months ago
- VR - Number of accounts defaulted in the last 12 months
- XHC - Number of Short Term Loan accounts
Understanding the
scorecard
Each concept in the scorecard contributes points to the score. The amount
of points contributed per concept varies depending on the result returned for
that concept.
These points are
added to the intercept (the base figure), which is 383. Therefore the minimum
possible score is 383, and the maximum achievable (by scoring top marks on each
concept) is 661. In practice, the vast majority of scores will sit somewhere
between 475 and 575.
Here are some
pointers on how to interpret and code the scorecard:
Example Concept
Age of customer
Class
|
Min
|
Max
|
Score
|
|
9
|
18
|
21
|
0
|
|
8
|
22
|
22
|
2
|
|
7
|
23
|
26
|
7
|
|
6
|
27
|
34
|
8
|
|
5
|
35
|
36
|
13
|
|
4
|
37
|
38
|
15
|
|
3
|
39
|
42
|
23
|
|
2
|
43
|
50
|
26
|
|
1
|
51+
|
34
|
||
- Customers aged between 18 and 21 score 0 points
- Customers who are 22 years old score 2 points
- Customers who are aged 51 or over score 34 points
- Therefore, in terms of credit risk, the older the applicant the better
{ND} = “Not
derivable” – this occurs if it is not possible to calculate a value for the
individual on the concept in question. For example, concept BN compares limits
now to twelve months ago. If the customer has not had any credit limits in this
period, this cannot be calculated and {ND} will be returned.
{OB} = “Out of
bounds” – this means that a number too big to fit in the field has been
returned. For example, concept XHC (Number of short term loans) has a length of
two digits. If a customer has had more than 100 short term loans, {OB} will be
returned.
Consultancy
A generic scorecard
is a great way of quickly setting up a robust strategy using an
“out-of-the-box” solution.
Over time, as your
volumes increase, you may wish to create a strategy that is more bespoke to
your customer base. Our consultancy team is able to use your historic good,
bad, and declined customers to create a scorecard specific to you. This ensures
that the scorecard you are using contains the concepts most appropriate to your
data, and that it is as predictive as possible for the applicants you are
attracting.
We are also capable
of working with you on other projects that may be of interest, for example:
- Rejects analysis - Examine what happens to your rejected applicants in the weeks after you decline them. Are they able to take out credit elsewhere, and if so, how do they perform on that account? This insight can be useful in helping to tweak your acceptance criteria.
- Process optimisation - We can conduct analysis to ensure that you are conducting your checks in the most cost-effective manner. If this turns out to not be the case, we can tell you the optimal order and the estimated savings.
- New products - A lot of lenders in alternative finance are starting to explore different product offerings. Traditionally this can be quite a slow process, involving a soft rollout while the lender works on their strategy as they go. We can expedite this transition by allowing you access to bureau data and performance history for accounts opened in the sector you are entering. This allows you to set up policy rules and scores in advance, enabling you to quickly launch a new product with a strategy already in place.
To discuss any of these offerings, or for any further questions you
might have regarding this scorecard, please get in touch with our Consultancy
team.
Subscribe to:
Posts (Atom)