Credit Risk Modeling for Alternative Lending

Keith Shields: Chief Analytics Officer - Magnify
Bruce Lund, Ph.D.: Director, Thought Leadership - Magnify
Credit Risk Modeling for Alternative Lending
Copyright ©2012 Marketing Associates LLC. All rights reserved. Magnify is a division of Marketing Associates, LLC. 1

magnifyAnalytic Solutions
magnify • simplify • amplify
Speaker Intro
Keith Shields
 20+ years in financial services, developing and deploying statistical models to predict
credit risk across multiple asset classes
 CAO Magnify & SVP Loan Science: 7 years of Analytic Services and Solutions
 14 years at Ford Motor Credit, Director of Global Analytics
 2+ years at Credit Acceptance Corp, VP Portfolio Management
Dr. Bruce Lund
 22 years at Ford Marketing: created the Customer Knowledge System, a DW that
houses customer data, predicts customer behavior, and steers 1-1 marketing efforts for
30+ million Ford customers
 Former Math professor
 Analytics “Godfather” of Magnify

Tying to the Abstract…
Key Elements:
Credit Risk Modeling
Scorecards and Cutoff Scores
Alternative Lending, Marketplace Lending, and Fintech
Common predictive modeling conventions
Regulatory Pressure
The Demand for Creativity and Disruption
What Makes “Alternative Lending Analytics” Different from “Banking
Analytics”?
The Continued Relevance of Logistic Regression
“It is not the model fitting techniques that need to change, as much as it
is the treatment of samples, potential predictors, and model refits”.
3Copyright ©2012 Marketing Associates LLC. All rights reserved. Magnify is a division of Marketing Associates, LLC.

Definition and Function
 A credit risk model is, usually, a linear statistical model that uses combination
of credit, contract, and personal attributes to predict the likelihood that a
loan applicant will default (fail to pay back the loan).
 “Attributes” = “Characteristics” = Independent variables
 Credit Score
 Loan-to-Value Ratio (LTV)…insert diatribe on 2008 mortgage disaster
 Payment-to-Income Ratio (PTI)
 Revolving Utilization
 Recent Credit Inquiries
 Integral part of how lending is done.
 PD’s, Basel III, SR 11-7
 Is lending ever done without models? The Big Short…
Copyright ©2012 Marketing Associates LLC. All rights reserved. Magnify is a division of Marketing Associates, LLC.
4

History
 Bill Fair, Earl Isaac…”FICO”
 Credit Scoring System: Credit Risk Models are converted to
“scorecards” by creating a point system whereby the parameter
estimates of the credit risk model are multiplied by the possible
values of the independent variables to create points.
 Example: Y = b0 + b1*x1. Suppose x1 can take on a value of 1, 2, or 3.
Scorecard:
X1 Points
1 B1 * 1 * 100
2 B1 * 2 * 100
3 B1 * 3 * 100
Note: Typically we
transform X1 first
5

Key Elements:
Regulatory Pressure
Analytics”?

Example: Assume we make $1,000 profit on “good” loans and suffer a
$10,000 loss on “bad” loans:
Credit Score
% of
Applicants Good Rate Loss (Bad) Rate
Expected
Net Profit
Cumulative
Approval
Rate
Cumulative
Loss Rate
500 40% 99.5% 0.5% $945 40% 0.5%
400 20% 96.0% 4.0% $560 60% 1.7%
300 15% 92.0% 8.0% $120 75% 2.9%
200 15% 87.0% 13.0% -$430 90% 4.6%
100 10% 75.0% 25.0% -$1,750 100% 6.7%
• Cutoff score = 300. We lose money on borrowers who score below 300.
• We will decline roughly 25% of our applicants.
• Our loss rate will be 2.9% (as opposed to 6.7% if we approved everyone).
• Our profit per contract is $677 (as opposed to $268 if we approved everything).
Some simple math on each credit tranche leads us to very useful conclusions:
(40%*.5%+20%*4%+15%*8%)/75%
7

Key Elements:
Regulatory Pressure
Analytics”?

Definition and Players
 The simplest definition is that an alternative lender is any lender that
isn’t a bank (or captive, in my opinion) .
 Historically the customers of alternative lenders are people, or small
businesses, who are unable to secure a bank loan. This is still largely
the case, but not nearly to the extent it once was.
Alternative Lending
9

Modern Nuances and New Entrants
 Recently many start-up alternative lenders have distinguished themselves
from banks based on heavier use of technology, removal of traditional loan
structures, and tapping into overlooked sources of collateral and non-
traditional funding sources:
 Technology: Fintech platforms
 Removal of traditional repayment plans: Merchant Cash Advance, Payday
Loans, and Uber
 Overlooked sources of collateral: merchant invoices
 Non-traditional funding sources: Peer-to-Peer (P2P) or “Marketplace”
Lending
Alternative Lending
10

Is there a place for us (statisticians, data scientists, analysts)?
 In a big way yes:
 Huge expenditures on alternative data (Yodlee, Clarity, LendDo, Yelp)
 100% technology-enabled, model-driven, loan adjudication and pricing
 Loan acquisition strategies driven completely from customer behavior and
“in-market” models
 But there are also minefields:
 Small samples and the “overfitting contest”
 Biased samples
 Relentless pressure to grow (before regulation) and strong appetite from
secondary markets. The Big Short?
Alternative Lending
11

Key Elements:
Regulatory Pressure
Analytics”?

Logistic regression offers the right construct for estimating risk…
 Let PD = “Probability of Default”. Let Y=1 be default and Y=0 be
paid as agreed
 PD is related to attributes Xk’s by taking a weighted sum of the
attributes called “xbeta” where xbeta = β0 + β1X1 + … + βnXn and
substituting xbeta into the logistic function
PD = P(Y=1|Xk) =
exp(xbeta)
1+exp(xbeta)
… the logistic function
We need to find the βk’s which best connect the Xk’s to the Y’s
 If Y=1, want PD big If Y=0, want 1-PD big
 Done by maximizing the likelihood function “L” as a function
of βk’s
Max [ L(β | X) = PDY (1 - PD)1-Y ]
13

Logistic regression models are at the root of most credit scores
 Per Wiki: Although logistic (or non-linear) probability
modelling is still the most popular means by which to
develop scorecards, various other methods offer powerful
alternatives, including MARS, CART, CHAID, and random
forests.
Machine learning has become really popular lately, particularly
among alternative lenders. The overfitting contest continues
(next slide)…
The Continued Relevance of Logistic Regression … But
14

Comparing a Logistic Regression Model with SVM Model
The sample was split 50/50 (sample1 / sample2).
The following steps were performed for both the logistic regression model and SVM
model:
 Sample1 was used to fit a model that was subsequently used to score sample2
 Sample2 was used to fit a model that was subsequently used to score sample1
38.1% 38.5% 40.3%
58.3%
66.7%
73.1% 70.3%
0%
5%
10%
15%
20%
25%
30%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
00%-37%37%-45%45%-52%52%-60%60%-67%67%-72% 72%+
%ofFundedContracts
RenewalLikelihood
SVM Prediction
Validation of Machine Learning Model: Test Samples
% of Funded Actual Predicted
37.3% 38.1%
66.7%
47.4%
68.9% 67.9% 68.5%
0%
5%
10%
15%
20%
25%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
00%-37%37%-45%45%-52%52%-60%60%-67%67%-72% 72%+
%ofActiveContracts
RenewalLikelihood
Logistic Regression Prediction
Validation of Logistic Regression Model: Test Samples
% of Funded Actual Predicted
KS=28.0
KS=24.4

Key Elements:
Regulatory Pressure
Analytics”?

Building predictive models has become a canned process for many
banks – why?
 Reg B: Empirically-derived, statistically sound
 ECOA
 OCC 11-7: Supervisory Guidelines on Model Risk Management
 CFPB and lenders’ lack of transparency
 Credit Ratings and Secondary Markets
 Banks shrinking desire to do retail banking. The alternative?
Commercial lines-of-credit for alternative lenders
 Millennials, millennials, millennials…
Regulatory and Societal Pressures
17

But we still have to find ways to accurately assess risk and price
loans…
 Big Data can’t be just a buzzword
 Unstructured data has to be structured in an intelligent
way
 Personally-verifiable, or volunteered, data
 Technology devoted to collecting unstructured data and
structuring it
The Demand For Creativity and Disruption
18

Analytics: Banking vs. Alternative Lending
Issue Banking Alternative
Lenders
How do alternative lenders handle the
modeling problems?
Sample
Sizes
Large Small Strong priors
Fix parameter estimates
“Wider” samples, not longer
History Spans
multiple
economic
cycles
1-3 years,
these are
mostly start-
up companies
“Look-alike samples”: BUY DATA
Refit and relearn
Introduce products that control risk without
having to model it (e.g. SoFi)
Change the target event to something that
happens quickly (e.g. Loan Science)
Regulation Onerous Light Actively explore social media (Yelp)
Huge investments in data that banks and
captives don’t think they can deploy
Legacy
code
Lots of old
SAS
processes
No legacy
processes
Technology “free for all”: R processes are daisy-
chained with SAS, then connected to Python.
“The IT nightmare”: 100% deployment flexibility
19

“It is not the model fitting techniques that need to change, as much as it is
the treatment of samples, potential predictors, and model refits”.
 The dependent variable can be redefined to something that is more
frequently occurring and ultimately worth predicting.
 Unstructured data can be structured as x’s and multiplied by betas to
create expanded credit scores. For example:
 Yelp: # times the word “terrible” is mentioned
 Student lending: # times “broke” is in the same tweet as “student
loans”
 Rapid refit and relearn processes are relatively easy to run routinely
when data are appropriately structured.
How Does Alternative Lending Change What We Do?
20

Thank you for your time.
Follow-up questions and requests:
 kshields@magnifyas.com; kshields@loanscience.com
 blund@magnifyas.com
Questions
21

Credit Risk Modeling for Alternative Lending

More Related Content

What's hot (20)

Viewers also liked (7)

Similar to Credit Risk Modeling for Alternative Lending (20)

More from Magnify Analytic Solutions (8)

Recently uploaded (20)

Credit Risk Modeling for Alternative Lending