Predictive Analytics For Lending Book Of Business In light of stringent market condition and regulatory environment, banks are looking for new data-driven insights to grow, control cost, and a new way to manage accounts at risk.
Predictive Analytics For Lending Book Of BusinessIn light of stringent market condition and regulatory environment, banks are looking for new data-driven insights to grow, control cost, and a new way to manage accounts at risk.
2© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
Many of our larger clients request digital and analytics support in the context of desired outcomes. Regardless of how the client requests services – imperatives or outcomes – KPMG’s approach accommodates a discussion and is prepared to respond.
Digital & Analytic
Banks require substantive growth rates to compensate for lost revenues to new competition and an aging population.
Banks must reduce costs to compensate for higher regulatory and credit management costs and increased stakeholder expectations. Additionally, banks must reduce costs to operate in a more automated and competitive landscape .
Banks must increase the customer experience to retain existing customers and attract new millennial and high
value customers
Banks must automate largely manual risk and regulatory processes and implement digital solutions to enable,
coordinate and monitor compliance
Banks must adapt to a new business landscape with new millennial customers, untraditional competition, an
aging population and heightened regulatory demands.
KPMG enables bank leaders to solve complex business problems with digital strategies and advanced analytics. KPMG’s D&A solution is structured according to five fundamental value drivers.
3© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
Today’s disruptive technology (i.e., big data analytics) presents a paradigm shift to solve complex business problems, and has the ability to offer enhanced digital experience to the current traditional customers and emerging millennial generation.
Progression of Analytics
What happened?
Why did it happen?
What will Happen?
How can we make it Happen?
4© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
― Elements of market drivers such as low interest rates, cost of capital, funding liquidity volatility, high credit losses, and stringent regulatory environment have adversely impacted the growth and profitability of banks’ lending book of business.
― Impending FASB CECL could increase loss reserve from 30% to 50%.
― Emerging FinTech banking, with the use of disruptive technology and looking beyond conventional risk factors (e.g.; FICO score, Agency ratings, etc.), are rapidly capturing retail & small-business market share.
― Banks are looking for new data-driven insights to grow, control cost, and a new way to manage accounts at risk.
KPMG’s commercial credit experts, who specialize in complete loan life-cycle both operationally and analytically, will decompose each process component to pinpoint targeted use cases and employ predictive intelligence to:
― Enable P&L owners and RM with acquisition analytics to grow and manage origination pipeline
― Enable risk and portfolio managers with credit performance analytics to effectively manage or hedge credit quality
― Enable SAD and loss mitigation managers with collection & recovery analytics to manage or hedge SAD/REO book and control losses
An illustrative view of commercial loan life-cycle
5© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
Problem Statement:
― Elements of market drivers such as low
interest rates, cost of capital, funding liquidity
volatility, high credit losses, and stringent
regulatory environment have adversely
impacted the growth and profitability of banks’
lending book of business.
― Consume credit line of business, residential
mortgage in particular, witnessed steep losses
after the financial crises, and the rebound has
been relatively slow.
― Emerging FinTech banking, with the use of
disruptive technology and looking beyond
conventional risk factors, are rapidly capturing
retail & small-business market share.
― Banks are looking for new data-driven insights
to grow, control cost, and a new way to
manage accounts at risk.
How we can help?KPMG’s consumer credit experts, who specialize
in complete loan life-cycle both operationally and
analytically, will decompose each process
component to pinpoint targeted use cases and
employ predictive intelligence to:
― Enable P&L owners and RM with acquisition
analytics to grow and manage origination
pipeline
― Enable risk and portfolio managers with credit
performance analytics to effectively manage
credit quality
― Enable loss mitigation managers with
collection & recovery analytics to control
losses
An illustrative view of Consumer loan life-cycle
Fraud
Legend
Parties
Borrower
Co-Borrower
Agent/Officer
Exposure
Portfolio
Risk Exposure
EAD
Expected Loss
Risk Rating
Portfolio
Risk Rating
PD/LGD
Mortgage
Account
Fraud
Attributes
Mortgage
Account
Fraud
Profile
Collateral
Valuation
Event
Fraud DB, MSP, Mortgage C/O SS
Financial
Account
Repurchase
Mortgage
Account
Repurchase
Attributes
Collateral
Valuation
Event
Mortgage
Account
Repurchase
Profile
Collateral
Valuation
Event
Repurchase DB, Fraud DB, MSP
Collection
Mortgage
Account
Collection
Attributes
Mortgage
Account
Collection
Event
MSP and other collection sources
Scorecard
Broker
Attributes
Scorecard
Profile
Scorecard Application
QC/QA
Mortgage
Account
QC/QA
Attributes
Mortgage
Account
QC/QA
Profile
QC/QA Database
Recovery
Recovery Sources
Mortgage
Account
Recovery
Attributes
Mortgage
Account
Recovery
Profile
Mortgage
Account
Recovery
Event
Mortgage
Account
Credit
Application
Mortgage
Account
Mortgage
Account
Term Profile
Mortgage
Account to
Interest
Rate
Mortgage
Account
Balance
Profile
Mortgage
Account
Term
Structure
Collateral
Real
Property
Core Portfolio Structure
Ref. Data
Reference Data
Income
Revenue,
Interest,
Fees and
Pricing
Loss Mitigation
Mortgage
Account
Loss
Mitigation
Attributes
Mortgage
Account
Loss
Mitigation
Profile
Collateral
Valuation
Event
MSP, Loss Mitigation Resources
Default
Mortgage
Account Default
Attributes
Mortgage
Account Default
Profile
MSP, Repurchase DB, Mortgage C/O SS
Proposed Source Systems
Delinquency
Mortgage
Account
Delinquency
Profile
MSP and other servicing sources
Real Estate Owned
Real
Estate
Owned
Item
Attributes
Real
Estate
Owned
Offer
Event
Real
Estate
Owned
Write
Down
Event
RES.NET, MSP and REO SS
Real
Estate
Owned
Expense
Event
Collateral
Valuation
Event
Slowly
Changing
Dynamic/
Profile
Transaction
s/
Events
Core
Portfolio
Structure
CDC
6© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
Lending Book Predictive Analytics Must be Business DrivenA combination of bank’s existing data and industry’s emerging data encapsulates information value of multidimensional propensity that can, with the help of advanced technology, unleash anunparalleled predictive power to catalyze a bank’s ability to achieve growth profitably, manage credit performance optimally, and control cost by managing credit losses effectively.
— Achieve longitudinal and latitudinal growth — Expand into emerging FinTech segments — Design new channel and product offering to tap into millennial consumers — Acquire portfolio through strategic merger and acquisition— Optimize deposit/FTP margin spread, generate fee, and control customer & balance attrition
— Optimize credit performance to retain lifetime economic value of loans — Target top P&L movers to optimize risk-return goals and improve credit quality— Prevent large prepayment shift with optimal loan restructuring— Detect borrowers at risk to minimize credit cost by understanding predicted vs. actual KRIs — Control default rate/charge-offs to optimize loan loss reserve and unleash trapped capital
— Detect early signs of problem loans and accounts to optimize loss mitigation strategy — Optimize time to recovery and reduce recovery cost— Optimize collection strategy and reduce resources for collection and cost— Prevent impaired real estate loans leading to foreclosure and reduce REO book inventory — Improve SAD book asset disposition rate and increase return on SAD asset
External Industry
Data
Credit Bureau Data
Emerging Social
Media & News
Media Data
Internal Portfolio & Customer
Data
— Information Retrieval — Natural Language Processing— Network Theory, Signal Processing, and Diffusion Model— Machine Learning and Decision Science— Behavioral Science and Propensity Modelling — Knowledge Representation and Reasoning— Hypothesis and Evidence Scoring
7© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
― Identifying Potential Growth Areas― M&A Due-diligence (valuation, elasticity, churn)― Proactive targeting of loan commitment completion― Making offers and customer communication
relevant and timely― Targeting ‘risk of attrition’ customers with save
strategies― Identifying price-sensitive customers for retention
strategies― Predict profitability (i.e., RAROC) and optimal
pricing decision
― Ability to identify loans at risk of default and employ strategies for loss mitigation
― Identifying downward migration of risk rating for potential loan restructuring
― Proactive strategies on legally binding exposures for loss mitigation
― Payment adjustments to prevent risky loans becoming delinquent
― Optimized TDR and SAD disposition― Optimized allocation of collection resources― Reduced collection cost― Increased recovery― Reduces delinquency― Enhanced collection efficiency
Lending book predictive analytics use case
— Emerging Sector/New Footprint
— M&A Analytics
— Total Borrower/Relationship Exposure
— Account Renewal
— Limit Enhancements
— 360 Customer view and Customer Lifetime Value— Cross-sell/Upsell
— High Value Customers
— FinTech for Small Business Banking
— Omni channel design
— EWI and Covenant Breach
— Propensity of Delinquency
— Rating Migration
— Non-accrual Likelihood
— Time to Delinquency and Default
— Bankruptcy Likelihood
— Drawdown (speed and amount) Likelihood in case of default
— Gross and Net Charge Offs
— Obligation Pay-offs
— Line Utilization
— Deep-dive Analytics for Top 20 Accounts
— Potential to Repay— Change in Ability to Repay— Cure Rate— Recovery Rate— Time to Recovery— Return on Recovery— Likelihood of Recovery from Bankruptcy— SAD Restructuring Likelihood and Performance
— Customer Segmentation
— Campaign Analytics
— Advanced Credit Approval Scorecards
— Credit Card Activation Rate
— Credit Card Risk Profile and Limit Enhancements
— Credit Card Dormancy and Re-activation Analytics
— Balance and Account Attrition
— Cross-sell/Upsell/Next Best Product Offer
— Price Elasticity and Propensity to Buy
— Millennial consumer acquisition strategy
— FinTech for Consumer Banking
— First Payment Default— Never-Pay Analytics (Credit Card) — Propensity of Delinquency— Roll-rate Prediction— Time to Default— Gross and Net Charge offs— Foreclosure Likelihood— Credit Card High Usage— Prepay Propensity— Time to Prepay and Amount
— Potential to Repay— Change in Ability to Repay— Potential to Self-cure— Short-term default outlook (3-6 month) — Prioritizing Accounts or Segments Outreach— Agent (pricing and call volume)— Call Analytics— REO book disposition analytics
8© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
Target potential new customers at various risk profile levels to maximize return potentialAcquiring new customers, based on customer-specific risk profiles
Purpose: to identify potential new
customers, then assess their
likelihood to join USAA and the
expected return
Purpose: to determine acceptable
levels of risk/return tradeoffs, and the
cost associated with acquisition; place
potential new customers into
segments that will be targeted the
same
Purpose: to achieve a targeted
acquisition approach for potential new
customers that maximizes expected
return at lowest cost with a balance
across the risk spectrum
Segment
Risk/R
eturn
Marketing
campaign
S1 L/L Secure Checking
S2 L/M Auto Insurance
S3 M/MPreferred Cash
Awards Visa
S4 H/H VA Loan
Identification & assessment
Network Modeling
Risk Profile
Propensity to Buy
Potential Lifetime Value
Marketing optimization
— “How do I effectively target
different segments of potential
new customers?
— What campaigns do I run?”
Risk management
risk
retu
rn
Metric Score
Risk M
Return M
Cost to Acq. $K
Prop to Buy 42%
LV $2.7M
9© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
Expand product market share from existing customers through target marketing
Purpose: to differentiate banks’
customers across multiple dimensions
that impact pricing decisions
Purpose: to rollup customers into
“Pricing Segments” by grouping “like”
customers that ultimately will be
priced/marketed-to the same
Purpose: to maximize growth while
minimizing cost and managing for
risk; growth/cost/risk are all
trade-offs, where do I get the most
return on investment?
Segment Auto loan Checking
S1 Rate (S1, A1) Rate (S1, C1)
S2 Rate (S2, A2) Rate (S2, C2)
S3 Rate (S3, A3) Rate (S3, C3)
S4 Rate (S4, A4) Rate (S4, C4)
360° View of customer
Customer Lifetime Value
Attrition
Risk Profile
Rate Sensitivity
Pricing optimization
Where do I spend dollars and
accept varying levels of risk to get
biggest bang for my buck?
customer segmentation
Combining predictive modeling with optimization will enable banks to grow in a cost effectively manner while managing risk.
S1 S2
S3S4
Use existing customers base for target marketing and pricing
10© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
Optimizing product pricing through risk profiling and customer-centered ROI What is Pricing Optimization?
Pricing Optimization applying analytics to predict how customer behavior can change under different rate pricing scenarios at the
micro-market level, then optimally price products to maximize revenue, while adhering to defined business rules and constraints.
Our Approach
KPMG built a closed-loop pricing optimization solution that combines predictive modeling and optimization. The solution optimally
assigns rate to maximize balance growth and minimize interest expense, while not adversely impacting attrition and CLV.
The solution is automated and includes an iterative feedback loop to continuously improve pricing and result in higher and higher
return for the bank.
Results
Since implementation in November 2015, the bank has seen growth of 5%, reduced interest expense by 8%, and maintained
attrition.Client-Centric Pricing Optimization Solution
Model Inputs
Internal Data: Customer,
Product, Markets
External Data: Customer
Potential (IXI) and Competitor
Pricing (Informa)
Predictive Models: Customer
Lifetime Value (CLV), Rate
Sensitivity, Balance Growth,
and Attrition
Customer Segments
Pricing Business Rules
Model Outputs
Client-Centric
Pricing
Optimization Model
Segment
Product P
Tier 1 Tier 2 Tier N
S1 S1T1 S1T2 S1TN
S2 S2T1 S2T2 S2TN
SN SNT1 SNT2 SNTN
Implement prices, gather data on customer response, provide feedback to models and clusters, continuously improve pricing to maximize returns
What-if Scenario Analysis
Product Profitability
#BPS
# Exception Codes
Rate Sensitivity
CLV
# Products per Customer
Product Profitability
11© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
― Classes of data stemming from banks internal activities, internal activities, digital interaction, news media, credit bureau, and capital market information ingested into an analytical data lake using Big-Data technology. Data is mined to determine abstract patterns.
― A range of data science techniques are employed to formulate mathematical algorithm that would analyze current and historical facts to make predictions about future or otherwise unknown events.
― A set of advanced credit profile are developed using data science that is predicated upon identifying plausible relationship hidden in data ( i.e., data association) looking beyond conventional credit scoring mechanism.
― Models will capture relationships among many factors to allow assessment of risk potentially associated with a particular set of conditions, guiding decision making for candidate credit transactions.
― Mathematical algorithms are built to self-learn the probability of credit events and subsequently, determine the velocity of the events to culminate in default, enabling credit managers with early detection of problem accounts.
Predictive analytics ecosystem
CUSTOMER ACCOUNT DATA LOAN ACCOUNTING DATA
CREDIT BUREAU/AGENCY DATA
CRMLOAN/ACCOUNTORIGINATION DATA
MARKETING & CAMPAIGNDATABASES
ONLINE ADDS & COMMERCIALS
CONSUMER FEEDBACK
STOCK MARKET DATA
MOBILE BANKING STREAMS
FINTECH DATA
CAPITAL MARKET DATA
SOCIAL MEDIADATA
REGULATORY DISCLOSURES
NEWS FEEDS
Data Mining
Data Relation &
Association
Diffusion Model & Signal
Processing
Natural Language Processing
Machine Learning
Propensity Modeling
13© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
Commercial lending book executive dashboard
— Create Credit Package
— Credit Approval
— Closing Preparation
— Document Preparation
— Closing
— Funding & Booking
— Monitor Portfolio
— Manage Exposures
— Monitor & Manage Credit Quality
— Manage Payments & Advances
— Manage Exceptions
— Manage Payouts & Release Collateral
— New Business
— Increased Exposures
— Renewals
— Underwriting Policy Exceptions
— Exposure/Outstanding
— Criticized/Classified
— NPLs, ORE, NPA
— Charge-offs
— Delinquencies
— Rating Migration, WARR
— SAD Managed
— Total Borrower Exposure Relationship
— Top Customers/Exposure & Changes
— Trading Exposure
— Top Criticized, Classified, NPAs & Changes to each
— Real Estate Collateral Concentration
— Document Exceptions
— Large Payoffs
— Pending Maturity
— Active loans with expired maturity date
— ACH Penetration
— Exceptions: Stale/Missing/Mismatch Ratings
— Weekly Credit Production Report
— New and Increased Loan Commitments
— Renewed Facilities
— Trackable Policy Exceptions (New and Renewed)
— Group/Region Line of Business Asset Quality Summaries
— Largest CRS Relationships (Exposure & Outstanding)
— NPL and ORE
— Criticized and Classified Exposures
— Risk Rating Migration
— Delinquencies (30-89, 90+)
— Credit Relationship Groups Report
— Daily Trading Exposure (Counterparty Limit) Reports
— Industry Concentrations
— Top Lists
— Past Due Reports
— Maturing Loans
— Payoffs
— Past Maturity
— Data Quality Scorecard
— Data Exceptions
— Document Exceptions
— Unassigned Loans
— Credit Quality Dashboard
— Interactive Credit Quality Reports –Customizable credit measures and time periods
— Rating Migration Report
— Customer Search – Total Borrower Exposure
— Interactive Detail Reports – Top Lists & Changes for a selected credit measure
14© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
are designed to reflect the key business drivers, metrics and risk budget components impacting organizational decision making at the executive level.
are the first layer of drill-down into specific portfolios bringing tabular and graphical data together around key performance and risk indicators, as well as risk budget components to focus attention on relevant issues.
are designed to facilitate analysis into more granular levels to better diagnose and evaluate the drivers of trends.
Consumer lending book executive dashboard
Executive Dashboard Rolling 9-12 Quarters Credit Quality SummaryCredit Performance Summary & U.S. Heat MapRepurchase SummaryFraud SummaryOrigination Quality Summary Post/Pre-FundingIndustry comparison & External Indicator Summary
Run-on and Run-offsDelinquency & Roll rate ReportingNon-Performing Assets ReportingNon-AccrualsREO90+ to 120+ DPDTDRCharge-Off ReportingFFIEC Write-downsIndustry Comparison
Production & Servicing:Units, $Amount By Geo, Property Type, Loan Type, Loan Purpose, Occupancy Type, Doc Type, LTV/FICO/DTI/DpD/Interest Rate/Maturity Term/UPB/Reg. Classification bucketsPortfolio Stratification with US Geo Map
Weighted Average SpreadsLTVFICODTIInterest Rate
Vintage TrendsDelinquencyNon-AccrualCharge-offsLoss
16© 2016 KPMG LLP, a Delaware limited liability partnership and the U.S. member firm of the KPMG network of independent member firms affiliated with KPMG International
Cooperative (“KPMG International”), a Swiss entity. All rights reserved. NDPPS 559933
Leveraging Big Data Analytics, Artificial Intelligence and Machine Learning Across FinTech
Analytical rigor in FinTech landscape
Credit scoring
Marketing
Risk management Investment management
Customer
acquisition
Customer
retention and
loyalty
— Gather customer data from
multiple available sources
— Quantify qualitative aspects
— Customize scoring models
iteratively
— Customer acquisition: Focus on
digital channels
— Improvising digital touch points
to engage consumers
— Creating complete customer
preference profiles by going
beyond transactional data
— Personalized, contextual
offerings
— Enhanced fraud & authentication
solutions
— Eradicate vulnerable access
points
— Device identification, biometrics,
behavior analysis
— Automated advisory solutions
— Combine multiple data points
(social media, search data, etc.)
and provide visual insights
— Identifying anomalies
Note: Company list is not exhaustive and is focused on startups
Source: LTP, Powered by MEDICI
Thank you
ABCD
Saroj Das
Managing Director
Credit Risk Consulting
303 Peachtree Street Tel. +1 703 343 2336
Atlanta, GA 30308 Mobile +1 404 242 6852