Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted...

42
Credit Scoring Dr. Selim Seval Tehran, November 11 th , 2014

Transcript of Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted...

Page 1: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

Credit Scoring

Dr. Selim Seval

Tehran, November 11th, 2014

Page 2: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

Presentation agenda

2

How credit scores are developed - predictive modelling - validation The importance of scores - decision making - pricing

Credit scores and Basel II and III

Credit scores in an emerging market environment

Credit scores in credit insurance

The future in credit scoring

What is a credit score - advantages - types of scores

Page 3: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

Credit scores

3

What is a credit score?

Credit score is a

• A quantitative measure of failure or delinquency risk

• A predictive indicator: evaluates the likelihood of a future event

• Objective and consistent: statistically derived from actual

historical information

Probability of Default (%) Score (0 – 100) Rating (AAA, AA, AA)

Page 4: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

4

Origination

Approve / Decline

Initial credit limits

Risk-based pricing and collaterals

Cross-sell / Up-sell

Retention

Wallet share

Limit increase / decrease

Authorizations

Review

Prioritization

Resource allocation

Outsourcing

Acquisition Portfolio

Marketing

Risk

Management

Collections

Advantages of credit scoring

Page 5: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

5

Advantages of credit scoring

Speed • process a larger number of credit applications therefore cutting down the costs

Advanced risk management • know in advance the risks associated with any customer (predictiveness)

Consistency and objectivity • establish credit allocation and follow-up policies based on solid criteria that do not change within a

given time interval and applied consistently throughout the organization • decide objectively throughout the whole organization irrespective of the credit officer, branch and

regional peculiarities

Employee productivity • plan the work time of its credit officers; i.e. for top rated customers an automatic accept decision

may be taken while sparing more time on lower rated and problem cases. Lowest rated customers may also be automatically rejected

Portfolio management & monitoring • monitor customer and portfolio risks throughout time and take the necessary steps in advance to

avoid bad debts

Improved forecasting and strategy formulation • store credit data in a more organized fashion to enable further analysis including future runs and

tests of the scoring models

Page 6: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

6

How credit scores are developed – predictive modelling

Types of scorecards during the life-cycle of a credit

Debt Collection

Customer Management

Fraud Management

Origination

ANALYTICS

Application Score

Behavioral Score

Credit Bureau Score

Collection Score

Marketing Score

Attrition Score

Fraud Score

Page 7: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

7

Scorecard development

Credit scoring is a process whereby information available is converted into numbers that are added together to arrive at a score via a scorecard. • Generic Method (least advised method, may be used as a first step in consumer scoring) • Expert Method (judgmental, good for emerging market environments as a first step) • Statistical Method (well-proven methodology, may be strengthened with bootstrapping and reject inference for emerging market scorecards) • Hybrid Method (good to overcome deficiencies in electronic data storage)

How credit scores are developed – predictive modelling

Page 8: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

8

Scorecard development methodology

How credit scores are developed – predictive modelling

Scorecard development methodology has evolved for more than 50 years since it was first introduced to assist banks in making their credit lending decisions. Starting from the divergence based scorecard method which dominated the industry for the first a couple of decades, it has now diversified into a spectrum of methodologies: • logistic regression • decision trees • mathematical programming • neural network • genetic algorithm • survival analysis modelling • support vector machine • graphical models • double hurdle modelling Among all these, logistic regression is now the most commonly used method for scorecard building

Page 9: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

9

How credit scores are developed – predictive modelling

Most commonly used scorecards

• Application Scorecard (First-time Customer Scorecard) Derived with non-financial, financial, industry and external behavioral data

• Behavioral Scorecard Derived with internal and external behavioral data

• Existing Customer Scorecard Derived with full credit dataset (non-financial, financial, industry, external and internal behavioral data) • Credit bureau scorecard Provides a behavioral score of an entity pertaining to how its obligations were met to the entire universe of financial institutions

Page 10: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

10

How credit scores are developed – predictive modelling

Application Scorecard

• Application scorecards are used to analyze credit applications. Most application scorecards are custom, which means that they are based on the information collected from the applicant as well as external behavioral data from other sources.

• Credit bureau scorecards are useful for institutions without historical data or the resources to develop custom application scorecards. They can also be used in conjunction with application scorecards to increase overall predictiveness.

Page 11: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

11

How credit scores are developed – predictive modelling

Behavioral Scorecard

• Behavioral scorecards are used for ongoing management of existing accounts. These scorecards are based on the customer’s actual payment and credit usage behavior.

• Credit bureau scores are also very useful for the management of existing accounts because they consider a customer’s credit risk by analyzing their payment behavior across all trade lines reported to the credit bureau at that point in time.

Page 12: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

12

How credit scores are developed – predictive modelling

Developing a scorecard - data vs information

• Data = information lacking structure (for instance the alphabet is data)

• Information = data with structure (a word, a sentence etc.)

• Information consists of facts and data organized to describe a particular situation or condition. Information can be judgmental. Data is objective, impartial.

• Data is an electronically readable piece of alphabetical, numerical or alphanumerical information stored according to a given layout.

• Data is often meaningless unless one or several of them are used to derive an information with the help of a formula or an algorithm.

i.e. A balance sheet does not tell whether the company will fail unless analytics is applied on it.

Page 13: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

13

How credit scores are developed – predictive modelling

Types of data elements (predictive variables)

According to sources • Application data: Data that the applicant provides to the financial

institution at the time of the application • Behavioral data: How the credit customer behaves with regard to

meeting its obligations Internal behavioral data: Behavioral data of the customer that

the financial institution captures External behavioral data: Behavioral data of the customer that a

credit bureau captures from all financial institutions

According to characteristics • Financial information • Non-Financial data and information • Industry and economic information

Page 14: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

14

How credit scores are developed – predictive modelling

Variable selection

Starting from the Long List of available variables, these steps lead to a Short List of candidates to became part of the Final Model.

1. Exclusion of variables with a high percentage of missing values

2. Exclusion of variables with a low predictive power

3. Exclusion of incoherent variables

4. Exclusion of correlated variables

Identification of the Short List, constituted by variables

• Statistically highly predictive

• Well interpretable from a business and operational perspective

• Uncorrelated

That are candidates to be entered the statistical model

Available Variables

75

61

40

29

16

LONG LIST

MEDIUM LIST

SHORT LIST

Data consistency

Data analysis

Correlation Analysis

Page 15: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

15

How credit scores are developed – predictive modelling

Steps in scorecard development

Project Objectives Definitions

• Define objectives for the project • Organizational objectives and scorecard role • Review credit policy (past, present, and future)

Model Design • Definition of the sample and performance windows • Review credit policy (past, present, and future) • Target measure definition

Sample Definition

• Sizing • Data requirements specifications • Data Extractions

Data Organization

• Data Aggregation, merging and manipulation • Normalization of the sources attributes • Data Integration

Project Objectives Definitions

• Define objectives for the project • Organizational objectives and scorecard role • Review credit policy (past, present and future)

Model Design • Definition of the sample and performance windows • Review credit policy (past, present, and future) • Target measure definition

Sample Definition

Page 16: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

16

How credit scores are developed – predictive modelling

Steps in scorecard development

Exploratory Analysis

• Data Quality Check, in terms of readability, correctness and completeness

• Variable understanding

Segmentation Analysis

• Segmentation of the known population, with the purpose to augment the predictive power of the model.

• Business driven segmentation

Reject Inference

• It allows to recover the target measure (e.g. Bads

and Goods) for the data records that do not have one (these records are typically the rejected cases).

Page 17: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

17

How credit scores are developed – predictive modelling

Steps in scorecard development

Data Analysis

• Purpose of this analysis is the selection of predictive data variables, and group the data values into robust categories

• Selection of the best predictive and “business meaning” variables

Preliminary Model Development

• Process a multivariate model based on the target

variable and on the set of predictive variables; • Verify the quality and performance of the model

with Statistical Measures of discrimination

Fine Tuning and delivery of the model

• Refinements with the feedback from users • Adjusting of the model and identification of

anomalies or concentrations in the distribution of the score

• Eventual intervention in the grouping categories of the variables

Page 18: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

18

How credit scores are developed – validation

Validation tools – Perfect Model, Gini Coefficient and ROC

There are several statistical methods used to measure the scorecard power, among which two are the most commonly used.

Cu

mu

lati

ve %

of

defa

ult

s

Cumulative % of total sample

Perfect model

Random model

(No differentiation)

0% 100%

Area A

Area B

Best scores Worst scores

Area A

Potential model

Page 19: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

19

How credit scores are developed – predictive modelling

Validation tools –Gini Coefficient and ROC

Gini Coefficient = A / (A+B)

ROC = A + (A+B) / 2*(A+B)

Cu

mu

lati

ve %

of

defa

ult

s

Cumulative % of total sample

Random model

(No differentiation)

0% 100%

Area A

Area B

Best scores Worst scores

Area A

Gini Coefficient = A / (A+B)

Page 20: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

20

How credit scores are developed – validation

Validation tools

• Test the statistic stability, i.e. the model capability to replicate on one or more samples performances coherent to the ones observed during the development phase.

• The validation analysis involves comparing results from the validation samples against results from the original development database.

• In particular, comparisons are made for the bad rate against score,

BOOTSTRAP: sub-sample iterating generation starting from 25% of the population

OUT OF SAMPLE: 10-25% holdout sample from the development database kept separate from all analyses

OUT OF TIME: more recent population than the development

Page 21: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

21

How credit scores are developed – predictive modelling

Focus on the Overrides Analysis

• Verify the stability of the decision scoring tool compared to a discretional usage

• It allows to know the utilization of the decision scoring tool and to evaluate the impact in terms of cut-off strategy

• The override is a decision that is in contrast with the decision suggested by the score

• The override is a physiologic element in the application scoring process. It is very important for the financial institutions to have consistent policies that define some thresholds in the usage of the overrides and tools dedicated to codify, measure and store the overrides and the reasons.

• The override could be permitted in relation to the commercial strategy of the institution (or if the model doesn’t cover the variables that are considered particularly relevant for the decision related to the credit request.

• The quantitative measure between the override rate and the Reject / Accept rate should be always kept under control

“Disciplined” utilization of the overrides

“Randomized” utilization of the overrides

Page 22: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

External

22

The importance of scores - decision making

Application for Credit

Application Process

External Information

Decision Rules

Automatic Approval

Automatic Reject

Gray Area

Review or Adjust Terms

Credit Allocation Process

Page 23: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

23

The importance of scores - decision making

Scores/ratings form the basis of a creditor’s line allocation, increase / decrease and termination decisions. Scorecards facilitate the implementation of: • risk-based pricing and limit allocation

• auto-reject and auto-accept rules

• cut-off policies

• customers’ credit performance evaluation

• collateral policy

• early warnings during credit follow-up

From scores to decisions

Page 24: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

24

The importance of scores - decision making

SME

•Reject / Very

Low Limit

•Strong Collateral

•High Price

•Low Limit

•Strong / Medium Collateral

•Medium Price

•Medium Limit

•Weak Collateral

•Low Price

MID-MARKET

•Very Low Limit

•Strong Collateral

•High Price

•Medium Limit

•Medium Collateral

•Medium Price

•High Limit

•No or Weak Collateral

•Low Price

CORPORATE

•Low / Medium Limit

•Strong Collateral

•Relatively High Price

•Medium / High Limit

•Medium Collateral

•Medium Price

•Very High Limit

•No or Weak Collateral

•Low Collateral

LOW SCORE

MEDIUM SCORE

HIGH SCORE

Application Score above cut-off and Credit Decision

Page 25: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

25

The importance of scores - decision making

POOR

•Terminate Limit

•Keep or Reduce Limit

•Keep or Increase Collateral

•Keep or Reduce Limit

•Keep or Increase Collateral

MEDIUM

•Keep or Reduce Limit

•Increase collateral

•Keep Actual Limit

•Keep Actual Collateral

•Marginal Limit Increase

•Keep Actual Collateral

HIGH

•Keep or Reduce Limit

•Keep or Increase Collateral

•Marginal Limit Increase

•Keep Actual Collateral

•Big Limit Increase

•Reduce Collateral

POOR

MEDIUM

HIGH

Internal Behavioral Score & Credit Bureau Score

I N T E R N A L B E H A V I O R A L S C O R E

B U

R E

A U

S

C O

R E

Page 26: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

26

The importance of scores - decision making

HIGH

Early legal or

collection agency

Accelerate effort or

early legal

Standard collection

practice

MEDIUM

Accelerate effort

or early legal

Standard collection

practice

Delay work queue

or low key follow-up

LOW

Standard collection

practice

Delay work queue

or standard practice

Very low-key

follow-up

LOW SCORE

MEDIUM SCORE

HIGH SCORE

Early Past Due Accounts and Collection Score

P A S T D U E A M O U N T

Page 27: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

27

The importance of scores - decision making

From behavioral scorecard to an action model

Action models are advised primarily for micro and SME credits

An action model uses the outcomes of the behavior

scorecard and produces decisions like: • Continue as is • Increase collateral margin • Change collateral type • Change price • Decrease limit • Terminate credit contract • Start legal action

Page 28: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

28

Decision models - pricing

Scores and pricing

Price is a function of: 1. PD for each rating group 2. Collateral recovery ratio 3. Targeted base interest rate / premium rate

Page 29: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

29

Decision models - pricing

Expected Collection

(within related rating

group)

Defaults Live Credits Defaults Live Credits

0 (1-PD) × radj × P PD × RR × P (1-PD) × P (1 + rbase) × P

Interest Collection

(within related rating

group)

Principal Collection

(within related rating group)

PD = PD related to a rating group

RR = Collateral recovery ratio (as a percentage)

P = Principal of credit at a rating group

rbase = Base interest rate

radj = Risk-adjusted interest rate for a rating group

Scores and pricing

For any rating category:

Page 30: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

30

Decision models - pricing

PD

RRPDrr baseadj

1

)1(

Risk-adjusted interest rate for a rating group

A 24,95% 0,67% 20,64%

B 23,90% 1,20% 21,17%

C 16,17% 2,18% 22,32%

D 14,57% 3,50% 23,82%

E 14,17% 7,33% 28,38%

Risk-based interest rate

considering

loss given default (radj)

Default Probability

(PD)

Rating

Category

Collateral

recovery ratio

from defaults

(RR)

When rbase = 20%

Page 31: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

31

Developing scorecards in an emerging market

Developed markets: • Full disclosure • no secrets, SarOx • well-maintained public records • efficient databases Emerging markets: • Confidentiality • reliance on company-sourced data • several sets of accounting books • limited accessibility to public information • Banks may not have written or well-established credit assessment

and decision policies i.e. No clear formulation of bad customers, good customers, auto-reject rules etc.

Differences in market environments

Page 32: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

32

Developing scorecards in an emerging market

• In an economic environment where volatility is an everyday phenomenon, finding an observation window may be difficult

• The model, whether expert or statistical, should be based on a time period free from extraordinary events

• We have to consider a 12 or 18-month maturity period (i.e. sufficient time period for a credit to mature and allow us to assess accurately its performance)

12 / 18 months

Outcome Period Observation Window

Observation Point Today

Differences in market environments

Page 33: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

33

Developing scorecards in an emerging market

• Financial institutions in emerging countries usually do not store the contents of their credit files (information) in their computer systems/databases. Information stored in Word or Excel files are not usable for statistical methods

• External data is not readily available in electronic format. Exclusion of such data elements may significantly alter the predictiveness of models.

• It is difficult to change old ways of doing business. Credit officers (underwriters) who do not have prior credit scoring experience are usually prejudiced against scorecards. They fear the scorecard will replace their jobs and duties.

• The scorecard is like a living creature. It must have an owner within an organization. This is highly important for the success of scorecard management.

From a financial institution’s perspective

Page 34: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

Place - 01/01/2014 Name Surname 34

Credits to Individuals

Consumer credits

Credit cards

Mortgage

Vehicle credits

Small Business Credits

Micro Companies

SME Credits Mid-market Credits

Large Corporate Credits

• A person or a family

• Structured on fixed income

• A company or group of

companies

• Commercial activity

• Structured on a future

variable income

Developing scorecards in an emerging market

Typical segmentation of credits

Page 35: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

Place - 01/01/2014 Name Surname 35

Why company size matters?

• Corporate scorecards focus more on a company’s commercial and financial performance while a small business scorecard tries to predict a company’s future payment record

• In small business scorecards, the main shareholder’s payment behavior, his/her socio-economic status and demographic traits are also major determinants

• In small business credits, the scorecard should be able to process large volume of applications in a short period of time and help credit officers to assess majority of applications automatically

Developing scorecards in an emerging market

Page 36: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

Place - 01/01/2014 Name Surname 36

• Lack of sufficient information and data accuracy are major concerns with small businesses

• Small business variety is vast to enable them consolidate under meaningful groups

• There are no benchmark segmental ratios for small businesses

• Statistical demographic risk-related information is not available

• Generic models implemented in developed countries will not work effectively in emerging countries

Difficulties in small business scoring

Developing scorecards in an emerging market

Page 37: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

37

• Establish an international benchmark for banks’ credit evaluation

• Create awareness for data quality, transparency, credit scoring and analytical risk management systems

• Separation of default risk and loss risk

Expected Loss (EL)=Probability of Default (PD) x Loss Given Default x Exposure At Default (EAD) Borrower Rating (What is the likelihood that the customer will default on an obligation?) Facility Rating (In the event of a default, how much does the creditor expect to lose?)

How the Basel Accord helps?

Credit Scores and Basel II and III

Page 38: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

38

Credit insurance = Non-cash credit

Credit Scores in Credit Insurance

• The task of an underwriter in a credit insurance company is not any different than a credit allocation officer in a bank or a lending institution

• The same credit allocation and risk assessment processes should apply

• Credit insurance is akin issuing a bank letter of guarantee (a non-cash credit product)

• A major difference is that an insurer obtains the credit information report from external sources. A credit insurer should also have access to bank credit bureau databases and scores

• The insurer should insist that the credit reports contain credit scores developed and maintained with appropriate methodolies explained in this presentation. The insurer should have periodical access to the informaton provider’s score development and validation processes.

Page 39: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

39

Importance of scores in credit insurance – future trends

Credit Scores in Credit Insurance

• As economic and market conditions change rapidly, credit insurers will need «smart scores» that quickly adopt to current market conditions in an effort to avoid increases in claims.

• Credit insurers will continue to obtain scores from external competent score providers. Therefore, score providers should rely more and more on analytics to calculate more sophisticated scores which will have the capacity to warn the insurers more accurately and timely.

• Platforms developed and operated by information or score providers will gain importance as these platforms will enable the users to integrate any type of credit data from any source to integrate and come up with «smart decisions».

• As volumes increase when insurers expand their portfolios, there will be

more reliance on scores for small ticket policyholders. Scores will be «the most important factor» in making a decision to issue a policy to cover a small scale debtor.

Page 40: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

40

The Future in Credit Scoring - Yesterday and Today

Yesterday • limits only to well-known names • highly collateralized transactions • more large scale customers than SMEs • personal judgments of the credit officer as key to credit decisions • no mathematical modeling (scoring and rating) • No well-defined risk management processes Today • micro and SME customers replacing big ticket customers (the world

is becoming one marketplace) • IT developments facilitate accumulation of credit data • credit risk management analytics (objective methodology – scoring

and rating for default prediction) • universal banking regulations (Basel Accord) • reputation collateral

The credit environment – Yesterday and today

Page 41: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

41

The Future in Credit Scoring - Yesterday and Today

Tomorrow

• Automated self-service credits from the internet

• Credits will go mobile – accessible from mobile devices

• Centralized world-wide databases will gain importance (KYC and anti-money laundering)

• Scoring and decisioning models will replace personal decisions, databases and analytics will dominate the financial businesses

• No limits will be given without a credit score

• Universal financial practices will be more enforceable

The credit environment – Tomorrow

Page 42: Credit Scoring - Aman UnionCredit scoring is a process whereby information available is converted ... customer’s credit risk by analyzing their payment behavior ... candidates to

© 2

014

Thank you

Dr. Selim Seval