Risk Analytics

Unit 3 – Credit Risk Analytics

Arup Duttaroy | Mar 2022

Agenda
§ Credit Risk – Introduction & Objectives
§ Understanding “Credit”, “Credit Worthiness” and other Terminologies
§ Pricing of Loans
§ Credit Risk & Basel Accords

§ Credit Risk Analytics – Analytics Implementation Considerations

§ Credit Risk Modelling - What does it entail?
§ Credit Scoring & Scorecard
§ Credit Risk Modelling - Some Terminologies
§ Credit Risk Scoring – Use Case
§ Credit Risk Scoring – Hypothesis vs. Data Needs
§ Customer onboarding for Credit Approval
§ Credit Risk Analytics – IPAD App
§ AI Platform for Credit Risk Model Building - Live Demo

2
 Risk Analytics
Unit 3 – Credit Risk Analytics
Topic 3.1 – Introduction & Objectives
Understanding the Word ‘Credit’
§ Credit is the trust which allows one party to provide resources to another party
§ Credit does not necessarily require money
§ Credit, in finance, is a term used to denote transactions involving the transfer of money or other
property on promise of repayment, usually at a fixed future date
§ Creditor
§ Debtor
§ Debt
§ Debt is given at certain rate of interest

This rate of interest is known as “Price of Loan”

4
Pricing of Loans

Interest Rate on any Borrowing = Base Rate + Spread

§ Type of Interest Rate

§ Fixed – The Spread and Base Rate both are fixed at the time of Borrowing
§ Floating – The Spread is fixed, and Base Rate keeps changing as per economic situations

§ Spread
§ Depends on the Credit Quality and Credit Worthiness of the Borrower
§ Depends on Collateral used for Borrowing

§ Earlier it was linked to BPLR (benchmark Prime Lending Rate)

§ Now it is Linked to Base Rate

5
What decides ‘Credit Worthiness’
§ 5’C of Credit
§ Capacity
§ Capital
§ Collateral
§ Conditions
§ Character
§ E.g. Tata’s vs Reliance, Kingfisher Airlines, Housing Loan, Greece, Dot Com Bubble
§ Who decides the Credit Worthiness?
§ Internal system of Banks
§ Credit Rating Agencies
§ CIBIL
§ Credit Union

6
Some Terminologies

Assets
Probability
Of Default
Liabilities
Cost of
Tenor Borrowing

Pricing of Loan Exposure at

Default
Credit Score

Non Performing
Assets Loss
Given Default
Net Interest
Margin

7
Credit Risk & Basel I
§ A committee was set up in year 1974 by central bank governors of G10 countries. It is to ensure that
banks have minimum enough capital to give back depositors’ funds. They meet regularly to discuss
banking supervisory matters at the Bank for International Settlements (BIS) in Basel, Switzerland
§ Basel I accord is the first official pact introduced in year 1988. It focused on credit risk and introduced
the idea of the Capital Adequacy Ratio which is also known as Capital to Risk Assets Ratio. It is the
ratio of a bank's capital to its risk. Banks needed to maintain ratio of at least 8%. It means capital
should be more than 8 percent of the risk-weighted assets. Capital is an aggregation of Tier 1 and Tier
2 capital
§ Tier 1 capital : Primary funding source of the bank. It includes shareholders' equity and retained earnings
§ Tier 2 capital : Subordinated loans, revaluation reserves, undisclosed reserves and general provisions
§ In Basel I, fixed risk weights were set based on the level of exposure. It was 50% for mortgages and
100% for non-mortgage exposures (like credit card, overdraft, auto loans, personal finance etc). See
the example shown below:
§ Mortgage $5,000
§ Risk Weight 50%
§ Risk Weighted Assets $2500 (Mortage * Risk Weight)
§ Minimum Capital Required $200 (8% * Risk Weighted Assets)

8
Credit Risk & Basel II
§ Basel II accord was introduced in June 2004 to eliminate the limitations of Basel I. For example, Basel
I focused only on credit risk whereas Basel II focused not only credit risk but also includes operational
and market risk. Operational Risk includes fraud and system failures. Market risk includes equity,
currency and commodity risk
§ In Basel II, there are following three ways to estimate credit risk:
§ Standardized Approach
§ Foundation Internal Rating Based (IRB) approach
§ Advanced Internal Rating Based (IRB) Approach

§ Standardized Approach - For wholesale banking, the banks relies on ratings from certified credit
rating agencies (CRAs) like S&P, Moody etc. to quantify required capital for credit risk. Risk weight is
20% for high rated exposures and goes up to 150 percent for low rated exposures. For retail banking,
risk weight is 35% for mortgage exposures and 75% for non-mortgage exposures (no rating by credit
rating agencies required for retail). See the example shown below:
§ Corporate Exposure $5,00,000 | Credit Assessment AAA | Risk Weights 20%
§ Risk Weighted Assets $1,00,000
§ Minimum Capital Required $8,000

9
Credit Risk & Basel II contd…
§ Internal Rating Based (IRB) Approach – It has four Credit Risk Component:
§ Probability of Default (PD)
§ Exposure at Default (EAD)
§ Loss given Default (LGD)
§ Effective Maturity (M)

§ Probability of Default (PD) – It means the likelihood that a borrower will default on debt (credit card, mortgage
or non-mortgage loan) over a one-year period. In simple words, it returns the expected probability of customers fail
to repay the loan. Probability is expressed in the form of percentage, lies between 0% and 100%. Higher the
probability, higher the chance of default
§ Exposure at Default (EAD) - It means how much should we expect the amount outstanding to be in the case of
default. It is the amount that the borrower has to pay the bank at the time of default.
§ Loss given Default (LGD) - It means how much of the amount outstanding we expect to lose. It is a proportion of
the total exposure when borrower defaults. It is calculated by (1 - Recovery Rate). See calculation below:
§ LGD = (EAD – PV(recovery) – PV(cost)) / EAD
§ PV (recovery) = Present value of recovery discounted till time of default.
§ PV (cost) = Present value of cost discounted till time of default.

10
Credit Risk & Basel II contd…
§ Internal Rating Based (IRB) Approach – Someone takes $100,000 home loan from bank for purchase
of flat. At the time of default, loan has an outstanding balance of $70,000. Bank foreclosed flat and sold
it for $60,000.
§ EAD is $70,000
§ LGD is calculated by dividing ($70,000 - $60,000)/$70,000 i.e. 14.3%
§ Suppose Probability of Default (PD) = 5%

§ Expected Loss – is calculated by (PD * LGD * EAD)

= 5% x 14.3% x $70000
= $500.50

§ Effective Maturity (M) - It is a duration that reflects standard bank practice is used. For Foundation
IRB, the effective maturity is 2.5 years (exception is repo style transactions where it is 6 months). For
Advanced IRB, M is the greater of 1 year or the effective maturity of the specific instrument.

11
Credit Risk & Basel III
§ Basel III accord was scheduled to be implemented effective March 2019. In view of the coronavirus
pandemic, the implementation has been postponed by a year till January 1, 2023.
§ Basel III has incorporated several risk measures to counter issues which were identified and
highlighted in 2008 financial crisis.
§ It emphasis on revised capital standards (such as leverage ratios), stress testing and tangible equity
capital which is the component with the greatest loss-absorbing capacity
§ In Basel III, the concept of building internal models and external ratings for estimating PD, LGD and
EAD remains same as it was in Basel II. However, there are some changes introduced in Basel III. It is
shown in the table below:

Basel II Basel III

Common Tier 1 capital ratio(shareholders’ equity + retained earnings) 2% * RWA 4.5% * RWA
Tier 1 capital ratio 4% * RWA 6% * RWA
Tier 2 capital ratio 4% * RWA 2% * RWA
Capital conservation buffer(common equity) - 2.5% * RWA

12
 Risk Analytics
Unit 3 – Credit Risk Analytics
Topic 3.2 – Credit Risk Analytics Implementation Considerations
Credit Risk Modelling – What does it entail?
§ Credit risk modelling refers to data driven risk models which calculates the chances of a borrower
defaults on loan (or credit card)
§ We also need to determine, if a borrower fails to repay loan, how much amount he/she owes at the
time of default and how much lender would lose from the outstanding amount
§ In other words, we need to build probability of default, loss given default and exposure at default
models as per advanced IRB approach under Basel norms

§ Probability of Default Modelling

§ This can be done using a “Supervised Model”
§ It needs a Dependant Variable which will have binary values 0 and 1 to indicate “Good Customer” vs.
“Bad Customer” respectively
§ Bad Customers Customers who defaulted in payment. By 'default', it means if either or all of the
following scenarios have taken place:
§ Payment due more than 90 days. In some countries, it is 120 or 180 days.
§ Borrower has filed for bankruptcy
§ Loan is partially or fully written off

14
Credit Risk Modelling – What does it entail? contd…
§ Indeterminates or rollovers - These customers fall into these 2 categories :
§ Payment due 30 or max 60 days but paid after that. They are regular late payers.
§ Inactive accounts
§ All the other customers are good customers
§ Indeterminates should not be included as it would reduce the discrimination ability to distinguish
between good and bad. It is important to note that we include these customers at the time of scoring.

§ We consider 12 months as performance window to flag defaults which means if a customer has
defaulted any time in next 12 months, it would be flagged as 'Bad’

§ Statistical Method using Machine Learning algorithms are best to calculate Probability of Default as
it can achieve higher accuracy while working with hundreds of data points or independent variables to
identify defaulters
§ The models once deployed, can offer automated and faster solution for making credit decisions, while
being unbiased and free from dishonest or fraudulent conduct by loan approval officers or managers

15
Credit Risk Modelling – What does it entail? contd…
§ Data Sources for PD Modelling - Following are the data needed:
§ Demographic Data : Applicant's age, income, employment status, marital status, no. of years at current
address, no. of years at job, postal code
§ Existing Relationship : Tenure, number of products, payment performance, previous claims
§ Credit Bureau Variables : Default or Delinquency history, Bureau score, Amount of credits, Inquiries etc.

§ Steps of PD Modelling
§ Data Preparation
§ Variable Selection
§ Model Development
§ Model Validation
§ Calibration
§ Independent Validation
§ Supervisory Approval
§ Model Implementation : Roll out to users
§ Periodic Monitoring
§ Post Implementation Validation : Backtesting and Benchmarking
§ Model Refinement (if any issue)

16
Credit Risk Modelling – What does it entail? contd…
§ Statistical Techniques / Algorithms used for Model Development:
§ Logistic Regression is most widely used technique for estimation of PD
§ Gradient Boosting
§ Markov chain Modelling
§ Random Forest
§ Neural Network
§ Survival Analysis is generally used to compute lifetime PD (required for IFRS 9)

§ Model Performance in PD Model - There are main 2 levels of performance testing:

§ Discrimination : Ability to differentiate between good (non-defaulters) and bad (defaulters) customers.
Following are used for model testing: Area under Curve, Gini coefficient, Kolmogorov-Smirnoff (KS) Statistics
§ Calibration : Check whether the actual default rate is close to predicted PD values. Following are used for
model testing: Hosmer and Lemeshow Test, Binomial Test

17
Credit Scoring & Scorecard
§ Probability of Default model is used to score each customer to assess his/her likelihood of default.
When you go to Bank for loan, they check your credit score. This credit score can be built internally by
bank or Bank can use score of credit bureaus
§ Credit Bureaus collect individuals' credit information from various banks and sell it in the form of a
credit report. They also release credit scores.
§ In US, FICO score is very popular credit score ranging between 300 and 850.
§ In India, CIBIL score is used for the same and lie between 300 and 900.

§ Scorecards - There are 2 types of scorecards:

§ Application Scorecard : It applies to new (first time) customers applying for loan or credit card. It estimate
probability of default at time applicant applies for loan
§ Behaviour Scorecard : It applies to existing customers to assess whether customer will default in loan payment.
Performance window is generally 6 to 18 months

18
Application Scorecard & Behavior Scorecard
§ Scorecard Points - Application scorecard is used majorly for
the following tasks:
• To determine whether or not to approve a
customer for a loan.
• To assist in 'due diligence'. Suppose an
applicant scoring very high or very low can
be declined or approved outright without
asking for further information.
§ Suppose Cut-off for granting Loan is 350

§ Profile of a New Customer:

Customer 1 Customer 2 Behavior scorecard is used majorly for
Gender – Male Gender – Female the following tasks:
Age – 30 Age – 30 • To set credit limit i.e. increase or decrease
credit limit
Salary – 15000 Salary – 15000
• Debt provisioning and profit scoring.
Total Points = (85+100+120) = 305 Total Points = (170+100+120) = 390 • Renewals
Decision – Refuse Loan Decision – Grant Loan

19
Credit Risk Modelling – Some Important Terminologies
§ Stressed PD vs. Unstressed PD
§ Stressed PD: A stressed PD depends on the risk attributes of borrower but is not highly affected by
macroeconomic factors as adverse economic conditions are already factored into it
§ Unstressed PD: An unstressed PD depends on both current macroeconomic and risk attributes of borrower. It
moves up or down depending on the economic conditions

§ Downturn LGD and EAD

§ Under Basel II and III, financial institutions need to estimate downturn LGD and EAD.
§ By 'downturn', it means adverse economic conditions.
§ We need to select the month with highest default rate and then take two consecutive quarters (6-month)
window on both sides of this point and consider it as downturn period and then take maximum of EAD and LGD
which provides the downturn estimates.
§ It is required because LGD and EAD can be affected by downturn economic conditions

§ Conditional PD
§ It is the probability of default during the second year given that it does not default during the first year.
§ To calculate conditional PD, we need probability of not defaulting by the end of year 1 (P0) and unconditional
probability of defaulting during the second year (P1).
§ If P0=0.5 and P1=0.1 so Conditional PD i.e. Prob(default | Survival) would be 0.1/0.5 = 20%
20
Credit Risk Modelling – Some Important Terminologies contd…
§ Lifetime PD vs 12 month PD
§ As per IFRS 9, we require two types of PDs for calculating Expected Credit Losses (ECL)
§ 12-month PDs for stage 1 assets - Chances of default within the next 12 months
§ Lifetime PDs for stage 2 and 3 assets - Chances of default over the remaining life of the financial instrument.

§ Calculation Logic:
§ Suppose 12-month PD is 3% which means survival rate is 97% (1 - PD).
§ 2nd and 3rd year conditional PD is 4% and 5%.
§ 1st year cumulative survival rate (CSR) is same as first year survival rate (SR).
§ 2nd year cumulative survival rate = 1st year CSR * SR of 2nd year = 97% * 96% = 93%
§ 3rd year cumulative survival rate = 2nd year CSR * SR of 3rd year = 93% * 95% = 88%
§ Lifetime PD = 1 - 88% = 12%

§ Macroeconomic factors to consider to estimate ECL

§ GDP, Unemployment Rate, Index of Industrial Production, House price index
§ Imports & Exports
§ Interest rate, Inflation rate, Exchange rate

21
 Credit Risk Scoring – Use Case
Description : Customer Credit Risk Scoring to identify the possibility of loan default
• Customer walks into the bank and applies for a loan providing his KYC documents.
• Once the user information is keyed into the system – The risk associated with the application must be calculated.
• To make the process of risk calculation more intelligent, bank will need to make use of the historical datasets and
data from external agencies like credit bureaus
• Bank’s Credit risk department have the data scientists and business analysts who will require data to be collated
from across various systems of the bank.
• The Data scientists are required to build a credit risk model showing “Probability of Default (PD)” & “Loss given
Default (LGD)” using the historical dataset and measure the model accuracy
• The model developed must be deployed and integrated into the Loan origination application via API.
• Each loan application should be inferenced with the risk model to identify the risk. Model outcomes should be
fair and hence should also be explainable as per the compliance standards
• Data scientists needs to monitor the behavior of the model deployed to look for model drift and re-train the
model as and when needed based on the real time feedback

22
 Credit Risk Scoring – Use Case

Problem Statement
• Data Analysts / Data Scientist / Business Analyst requests for summarized and collated information from external
and internal systems of the bank.
• Credit risk impacting attributes, required for analytical consumption, which are derived from –
• Credit bureau data for other loans from different financial institutes,
• Extract meaningful information from KYC documents, which are mostly unstructured data sources
• Current loan application details,
• Previous loan repayment history and credit card repayment history details of a customer with the bank
• Ensuring fairness in Credit Risk Rating i.e., explainability of Model outcomes

23
Credit Risk Scoring – Different Hypothesis vs. Data Needs
SL # Hypothesis Data Need
1 Those who provide false or wrong information during Unstructured KYC Documents – ID Proofs,
the KYC process are most likely to default on their Salary Slips, Application Form
loan repayment This will need AI/ML based Intelligent
Document Processing capability

2 Those having low credit rating are more likely to External Data from Credit Bureaus.
default on their loan repayment This data needs to be married with internal
data, which can be challenging. Identity
Resolution can be a potent need.

3
Those habitually delay their credit card payment or Internal Structured Data about Credit Card
does not pay credit card balance in full (carries over Payment History
debts) are more likely to default on their loan This will need Descriptive & Diagnostic
repayment Analytics on Aggregated Data at the
Customer Level

24
Customer On-boarding for Credit approval in a retail banking
scenario : Challenges Artificial Intelligence:
Automated Data
Automated cross-
verification to avoid
Extraction errors

Customer Bank

Standard KYC processes and checks Manual checks –

Formalities and Data Collection
approvals/rejections
Loan Form
Document Cross-

On-Boarding Channels
Mobile Self On-boarding Client Identification Verification

Pay-Slip Web Self on-boarding Data Controls-External

Tax Status Verification
Customer requests for Back office
account opening/loan Manual/Walking-in on-
application boarding Credit Request Check KYC Status
Passport

+more
Continuous request API based
processing: faster Verification with
Challenges SLA external data sources

• Multiple interactions with • Manual Processes – High error rate

bank • High Operational Cost – Back-office/FTE base in large banks
• Account Opening takes time • Incorrect and poor KYC status
• Unsatisfactory experience • Insufficient background checks
• Lack of data controls effecting credit terms

25
 Customer On-boarding for Credit approval in a retail banking
scenario : AI/ML Powered Solution
Customer Bank

Automated KYC processes and checks Human-in-the-loop

Formalities and Data Collection
checks –
Loan Form Client Identification Tax Status Credit Request Check approvals/rejections

On-Boarding Channels
Mobile Self On-boarding
Refract By Fosfor *
Pay-Slip Web Self on-boarding Continuous
Customer requests for batch Document Cross- Data Controls-External
KYC Status Back office
processing Verification Verification
account opening/loan Manual/Walking-in on-
application
Passport boarding Spectra By Fosfor *
+more Scanner

Benefits

1. Better experience with the bank 1. Better customer service with improved SLAs in customer on-boarding
2. Faster KYC status feedback and 2. Less load on back-office teams with automated digitization of documents and KYC process
loan approval status 3. Better compliance in Customer KYC process: Matching Name-Entities and Photo/signature
across customer documents

* These are products by LTI 26

Credit Risk Analytics Platform – Live Demo

* Fosfor is an Advanced Analytics Platform build by LTI 27

 Lorem ipsum dolor sit amet

Thank
Header 3
Lorem ipsum
Check out01
Header 1
dolor sit amet video tutorials atdolor
Lorem ipsum … Coming
sit amet soon…
Header 1
Lorem ipsum
dolor sit amet

You
Your headline will go here 02 Header 1
Lorem ipsum dolor sit amet

Header 2
Lorem ipsum For any queries please write to us at
dolor sit amet
[email protected]
03 Header 1
Lorem ipsum dolor sit amet
 Credit Risk Analysis iPad App
Highlights the KPIs for the bank Delinquency view shows the Business Challenge
Heat map highlighting
along with the 12 month trend bank’s 12 month performance
risk analysis by product A global financial bank’s Chief Risk Officer wished to
for each key risk indicator based on its non-performing roles develop an application which would help him
visualize consolidated exposures across sectors and
credit ratings

Solution
We helped the client build an intuitive app using
MicroStrategy mobile application to help them keep
a track of the key performance indicators which
revolved around delinquency ratio and exposure
within different assets and liabilities

How they use the system

The application is used by the C-level executives of
the risk team to monitor KPIs for different assets and
changing exposures accordingly

ROI to Customer/ Business Benefits

With the help of the application the client was able to
effectively measure loss due to defaults, delinquency
rates and return on risk weighted assets and also
anticipate the expected losses

Summary view shows the bank Asset Liability Management Graph highlighting credit “Unique” or “cool” features
performance versus key risk indicators View shows the bank rating wise exposure for The app includes many features for representation
and regulatory parameters such as heat maps, drill through and mouse hover
performance based on gap indirect loans
between asset and liabilities

29