• No results found

Basel II: Operational Risk Implementation based on Risk Framework

N/A
N/A
Protected

Academic year: 2021

Share "Basel II: Operational Risk Implementation based on Risk Framework"

Copied!
61
0
0

Loading.... (view fulltext now)

Full text

(1)

BG-9002 Varna

Tel. +359 52 612 367

Fax +359 52 612 371

eMail

[email protected]

WEB:

www.eurorisksystems.com

Basel II:

Operational Risk Implementation

based on Risk Framework

(2)

Presentation Agenda

Overview of Approaches

Basic Indicator Approach

Standard Approach

OR Loss Database and Loss Event Data Entry

Self Assessment

Advanced Measurement Approach (AMA)

Requirements of AMA

Internal Operational Risk Data Model

-

Stochastic Models for Severity

-

Stochastic Models for Frequency

-

Correlation of Key Risk Indicators

Business Structure and Aggregation

(3)

Main Risk Types

Capital allocation

Market Risk ca.

15 %

Credit Risk ca.

50 %

Operational Risk ca.

35 %

(4)

Operational Risk Regulatory Capital

8 Business Lines

Investment Banking

Corporate Finance

Trading & Sales

Banking

Retail Banking

Commercial Banking

Payment & Settlement

Others

Retail Brokerage

Agency Services & Custody

(5)

Operational Risk Regulatory Capital

3 Methods

 Basic Indicator Approach

BIA

 Standard Approach

STA

(6)

Risk Framework Modules

English, German

Databases Oracle, MS SQL, …

Import Interface, Excel Export

Crystal Reports, XML/XSL

COREP / XBRL Reporting

Batch Processing, Internet GUI

Model&rule-based Modules

ALM&Liquidity

Asset /Liability Mgmt. Cash Flow(GAP) Fund Transfer, LVaR

Liquidity Scenarios ……… English, German ……… English, German English, German ……… English, German ……… English, German ……… English, German ……… Turkish

Core System

(7)

Main Features

Encloses all regulatory risk types: credit, operational, market, rating&scoring

Registering of main objects: Exposures, Instruments, Customers, Collateral’s

Simple Integration to existing portfolio management or core banking systems

Support of different data bases without changes: Oracle, MS SQL

WEB Browser User interface and server mode

Supporting Functionalities

Multi-User Feature, Roles and Rights: show, create, modify, confirm, delete

Support of historic storing/accessing for all Object persisted into data base

Rule based expert system

Rule based business Logic: model scripts controls fully Risk Framework

Definition of GUI, automatic structuring of the data base, business logic

Automatic adjustment of GUI and data base at model changes

Interfaces and Reporting

Importer to import data from external sources using flat files

Risk Framework

System Features

(8)

Risk Framework

Product Architecture

Data base

Busines Logic

Importer

Risk Framework

Windows

desktop

Model-

Script

CLIPS Expert

System Core

Core DB

Tables

Risk Framework Tables

Import

Files

Host

Data Base

RFW

Tables

Rule based Engine

WEB

Browser

(9)

Basic Indicator Approach

Capital Requirement Calculation

Operational Risk Requirement = ∑ (α x indicator)/n

α = 15%

n = max 3, average over 3 years

Indicator = gross income

Gross income = yield income + non-yield income

(10)

Basic Indicator Approach

Income Example

Yield incomes

100

100

Yield expenses

(70)

(70)

Pure yield incomes

30

30

Provisions

(3)

Net fees / commissions

5

5

Realized gains / losses from sale

of securities in the banking book

5

Income from trading

5

5

Other non-interest income

5

5

Non-interest income

20

15

Non-interest expenses

(25)

(25)

Taxes

3

Extraordinary income

6

(11)

Basis Indicator Approach

Input Data and Capital Requirements in Risk Framework

Evaluation Currency

Incomes for 3 years

(12)

Standard Approach

Capital Requirements Calculation

Operational Risk Requirement =

∑ {years

(1..3)

max [∑(GI

(1..8)

x β

(1..8)

),0]}/3

GI

(1..8)

= Gross annual income for each of the 8 business lines

β

(1..8)

=

fixed percentage, set for each business line

The activities of the bank is distributed in the 8 business lines: GI

(1..8)

Indicator: Gross income averaged over 3 years: years

(1..3)

Capital requirements for each business line is calculated by multiplying

its gross income at a certain Beta percentage:

β

(1..8)

Negative income is included in the calculations

Total capital requirement is the sum of capital requirements on

business lines

(13)

Standard Approach

(14)

Standard Approach

Income Example

Year 1

Gross income

* „Beta”

Total

Corporate Banking

60

15%

9

Retail Banking

-10

12%

-1.2

Marketing and sales

30

18%

5.4

TOTAL

80

13.2

Year 2

Gross income

* „Beta”

Total

Corporate Banking

30

15%

4.5

Retail Banking

-40

12%

-4.8

Marketing and sales

-10

18%

-1.8

TOTAL

-20

-2.1

Year 3

Gross income

* „Beta”

Total

Corporate Banking

80

15%

12

Retail Banking

10

12%

1.2

Marketing and sales

30

18%

5.4

TOTAL

120

18.6

13.2 + 0 + 18.6

3

Requirements

=

10.6

(15)

Standard Approach

Input Data in Risk Framework

Incomes for 3 years for

marketing and sales

Incomes for 3 years for

retail banking

(16)

Standard Approach

Capital Requirement Calculation in Risk Framework

Capital Requirement

fixed percentage for each

(17)

Basic Indicator & Standard Approach

Operational Risk COREP Reporting in Risk Framework

Preparation of COREP

Reproting Data

Gross income of

business lines for

last 3 years

(18)

Basic Indicator & Standard Approach

COREP and Crystal Reporting (Example in Cirillic)

Show COREP Reporting

XBRL on Internet Explorer

Generate Reports by

Crystal Reproter

(19)

Basic Indicator & Standard Approach

COREP Reporting in Excel forms (Example in Cirillic)

Generate COREP

Reports in Excel forms

(20)

OR Loss Data Base

Loss Event Data Entry in Risk Framework

Dates:

Occurence,

Encovering,

Accounting

Loss Event Description

Loss Amounts:

Gross Amount,

Recovery Amount,

Net amount

(21)

OR Loss Data Base

Loss Event Data Entry in Risk Framework

Gross Loss Amount

Distribution of Loss

over business lines

(22)

OR Loss Data Base

Loss Event Data Entry in Risk Framework

Basel II Categorisation

of Loss Event:

(23)

OR Loss Data Base

Loss Event WEB Protocol in Risk Framework

Loss Events sessions are stored

by time stamp.

Every session registers a Loss

Event Loss.

A standard session protocol can

be generated and printed out.

(24)

OR Loss Data Base

Calculation of Risk Matrix for COREP in Risk Framework

Preparation of COREP

Reporting Data

Matrix: Busines Lines

x Key Factors

Total Loss Amount

Max Loss Amount

Number of Losses

(25)

OR Loss Data Base

COREP and Crystal Reporting (Example in Cirillic)

Show COREP Reporting

XBRL on Internet Explorer

Generate Reports by

Crystal Reproter

(26)

OR Loss Data Base

COREP Reporting in Excel forms (Example in Cirillic)

Generate COREP

(27)

OR Loss Data Base

Single Losses and COREP Reporting in Risk Framework

Preparation of COREP

Reporting Data for single

Losses

Show COREP Reporting

XBRL on Internet Explorer

for single Losses

(28)

OR Loss Data Base

COREP Reporting in Excel forms (Example in Cirillic)

Generate COREP

Reports in Excel forms

(29)

Self Assessment

Example for Internal Frauds in Risk Framework

Questions about

potential occurence of

Internal Frauds

Question Replay giving

Severity (0..4) and

Probability (0..4)

(30)

Self Assessment

Risk Matrix for Internal Frauds in Risk Framework

Categorization within risk

matrix:

Loss Volume x Probability

Categorization within

matrix of remaining risk:

Risk Reduction x Risk

Distribution of Losses

over business lines

(31)

Management of Internal historic Events and Loss Data

Fitting historic Series to theoretical Distribution Models

Fitting to Severity Models (Loss Probability Distribution)

Fitting to Frequency Models (Probability of Event Frequency)

Numerical Distributions Construction by Self Assessment

Mapping Events and Loss Data from external Sources

Use Network Representation of Bank Processes based on:

Nodes, Rules and Distribution Aggregation Expressions

Script Language to ensure high level of Flexibility

Connection of Distributions to leave Nodes of Network

Perform Monte Carlo Simulation over the Network

Aggregate total OR Loss Distribution and obtain OR VaR

Advanced Measurement Approach (AMA)

Approach Requirements and Steps

(32)

Interviewing experts in business lines

Identify OR types (e.g. Basel II definition)

Mapping OR types to business lines (Basel II)

Definition of loss event data base

Import the own data in data base

Using and mapping of external OR data bases

Advanced Measurement Approach (AMA)

Data Modeling for Operational Risk

(33)

Advanced Measurement Approach (AMA)

Stochastic Data Modeling

No requirements for the selection of a particular model or

distribution function of parameters such as:

Frequency

- Poisson, Negative binomial, Binomial

Severity

- Lognormal, Weibull, Pareto, Gamma, Inverse Gaussian

Banks should demonstrate that their approach meets the

standard of comparability to internal models for assessing

credit risk (1-year horizon and 99.9% confidence)

The model should cover the main risk factors, including

potential extreme losses

Correlation can be used if the system exposes stable

calculation, fully encompasses the activities which take

(34)

4 Parameter

3 Parameter

2 Parameter

1 Parameter

Generalized Beta 1

Generalized Beta 2

Beta 1

Log-t

Generalized

Gamma

Beta 2

Burr 12

Power

Uniform

Log-Cauchy

Lognormal

Weibull Gamma

Pareto

Normal (Gaus)

Rayleigh

Exponential

Model Quality Test by Kolmogorov-Smirnov Method

Advanced Measurement Approach (AMA)

(35)

Poisson

Binomial

Negative Binomial

Geometric

Hypergeometric

Polia-Appli (Poisson-Geometric)

Model Quality Test by Chi-Squared Method

Advanced Measurement Approach (AMA)

(36)

Aggregation of Severity and Frequency Distribution Models

1 Step: Sample Frequency Distribution (for example Poison)

2 Step: Sample Severity Distribution (for example Lognormal) many

times as given in Step 1 by Poison Distribution

3 Step: Sample possible Recovery Distribution and reduce Losses

4 Step: Cumulate Losses for Step2 accounting for Recovery of Step3

5 Step: Repeat 10 000 times Steps 1,2,3,4 and build Loss Distribution

6 Step: Obtain OR VaR at 99,9 % Confidence Level

Testing the Prediction Quality

Back Testing using historic loss event series

Use Kupiec Test or Q-Test (Crnkovic-Dracham)

Advanced Measurement Approach (AMA)

(37)

Class

Definition

Loss Volume (EUR)

A

Catastrophic

>= 30 Mio

B

Large

5 Mio < B < 30 Mio

C

Middle

0.5 Mio < C < 5 Mio

D

Small

50 000 < D < 500 000

E

Minor

< 50 000

Advanced Measurement Approach (AMA)

Severity of Loss Events – Example Figures

(38)

Class

Definition

Expected Event Frequency

A

Extremely high

< 50 times in year

B

High

10 < B < 50

C

Middle

2 < C < 10

D

Low

1 < D < 2

E

Extremely Low

Very rarely,

once every 20 - 30 years

Advanced Measurement Approach (AMA)

Frequency of Loss Events – Example Figures

(39)

Advanced Measurement Approach (AMA)

List of Basic Key Indicators for Operational Risk

1.

Internal fraud: Unauthorized Activity, Theft and Fraud

2.

External fraud: Theft and Fraud, Systems Security

3.

Employment Practices and Workplace Safety : Employee Relations, Safe

Environment, Diversity & Discrimination

4.

Clients, Products & Business Practices: Suitability, Disclosure & Fiduciary,

Improper Business or Market Practices, Product Flaws, Selection,

Sponsorship & Exposure, Advisory Activities

5.

Damage to Physical Assets: Disasters and other events

6.

Damage to Physical Assets: Disasters and other events

7.

Execution, Delivery & Process Management: Transaction Capture, Execution

&Maintenance, Monitoring and Reporting, Customer Intake and

Documentation, Customer / Client Account Management, Trade

Counterparties, Vendors & Suppliers

(40)

Business Units

Business Lines

Bank Processes

OR VaR

BL3

BL1

BL2

Basic Key Indicators

VaR

Monte

Carlo

Simulation

Distributions

Script for

Bank

Structure

VaR Result

Advanced Measurement Approach (AMA)

Hierarchic Network for OR Simulation and Aggregation

(41)

The model should cover the level of expected and unexpected losses. The bank may

Expected Loss (ЕL)

Unexpected Loss (UL)

OR VaR

Stress (catastrophic) Loss (CL)

Pr

o

b

ab

il

it

y

OR Losses

Capital

Requirements

Can’t be

assumed by

bank

Recognition

by Provisions

and prices

50% of distribution square

99,9% of distribution square

Advanced Measurement Approach (AMA)

Operational Risk Distribution an Confidence Levels

(42)

Corp.

Finance

OR Total VaR

BU1

BL1

BL3

Basic Indicator

Distributions

VaR

Basel II

Oper. Risk

Event

Types

Business

Units

VaR Result

External Fraud

BU2

BL2

BL4

Basel II

Business

Lines

INVESTMENT

BANKING

BANKING

Trading&

Sales

Retail

Banking

Comm.

Banking

Clients, Products, Bus. Services

Execution, Delivery, Processes

BANK

—>

Integration, Mapping Business Process Model to Basel Bank Model?

Advanced Measurement Approach (AMA)

Example Network for OR Simulation and Aggregation

(43)

Advanced Measurement Approach (AMA)

Theoretic Distributions for Basic Key Factors

Possible assumptions of example theoretic distributions for Basic Key Factors:

 Normal

(44)

Possible assumptions of example theoretic distributions for Basic Key Factors:

 Beta

 Rayleigh

Advanced Measurement Approach (AMA)

Theoretic Distributions for Basic Key Factors

(45)

Possible assumptions of example theoretic distributions for Basic Key Factors:

 Weibull

 Inverse Normal

Advanced Measurement Approach (AMA)

Theoretic Distributions for Basic Key Factors

(46)

Advanced Measurement Approach (AMA)

Editing and Aggregation of Distributions for Basic Key Factors

Left and right min and max

values can be set to adjust

the distribution X - axis

The distribution profile

can be edited by mouse

allowing for user defined

distribution shape

(47)

Advanced Measurement Approach (AMA)

OR VaR from Result Distribution at Confidence Level

Distribution mean =

Expected Loss = 25,30

Value at confidence

= 39,28

Unexpected Loss

OR VaR = 13,98

(48)

Algebraic Distribution Operators

Add two Distributions

Add Constant to Distribution

Multiply two Distributions

Multiply Constant to Distribution

Scale Distribution Value

Scale Distribution Probability

Normalize Distribution

Advanced Measurement Approach (AMA)

(49)

Advanced Measurement Approach (AMA)

Severity, Frequency and Recovery Input Distributions

Internal Frauds

External Frauds

Result

Distribution

Expected Loss

and OR VaR

(50)

Advanced Measurement Approach (AMA)

(51)

Advanced Measurement Approach (AMA)

Correlation Matrix (default values)

Correlations between

each two OR Indicators

The upper triangle matrix

is editable

(52)

Advanced Measurement Approach (AMA)

Correlation Matrix (loaded from Data Base)

(53)

Advanced Measurement Approach (AMA)

Loss Distribution Reporting on Business Units Level

(54)

Equally Distributed

And Correlated

Random

Samples (0...1)

Loss Event

Correlation Matrix

Normal Distributed

Correlated Random

Samples

Aggregated

Values

Cumulative

Distribution

Cumulative

Distribution

Beta

Distribution

Lognormal

Distribution

Aggregated

Distribution

OR Net

Work

Advanced Measurement Approach (AMA)

(55)

Advanced Measurement Approach (AMA)

Monte Carlo Simulation Approach – Data Flow

(56)

0,45

0,25

0,30

From Loss Data Base

Potential Losses

From Self Assessment

Monte Carlo

Simulation

Advanced Measurement Approach (AMA)

Monte Carlo Aggregation Approach – Different OR Sources

(57)

 Severity

Frequency

Gross economic capital

by loss event type

Monte Carlo

Simulation

1.

Internal fraud

2.

External fraud

3.

Employment Practices and Workplace Safety

4.

Clients, Products & Business Practices

5.

Damage to Physical Assets

6.

Damage to Physical Assets

7.

Execution, Delivery & Process Management

Total gross economic capital

Severity

Extern

Loss Data

Database for

internal losses

5 years

Correction of probability

Advanced Measurement Approach (AMA)

(58)

Advanced Measurement Approach (AMA)

Consolidation along the Bank Structure in Risk Framework

Hierarchy of Finance

Centers and Divisions

Finance Centers and

Divisions of a Bank

Concern

(59)

Data Sourcing

Internal historic Events and Loss Data

Mapping Events and Loss Data from external Sources

Fitting internal historic Series to theoretical Distribution

Models

Numerical Distributions Construction by Self Assessment

Aggregation of Key Factor Distributions:

From Loss Data Base

From Potential Loss

From Self Assessment

Advanced Measurement Approach (AMA)

Current steps in AMA Development

(60)

Risk Framework using WEB Browser Interface

Examples of OR Models - BIA

(61)

Risk Framework using WEB Browser Interface

Examples of OR Models – Self Assessment

Internal Fraud Model

Questionnaire

References

Related documents

* Population : Theorists of sustainable development have generally rejected the concept of unlimited growth, whether of population or of economic production.. Even if a

In fact, in the case of heating, power and transport in the second-half of the twentieth century, the average energy price was rising, while the price of the service was rising

The TV advertisement affects positively and significantly to build brand awareness, the TV advertisement affects positively and significantly to the purchase

The impacts of standards are also visible in the total annual number of diagnoses, and discharges, the satisfaction rate of out-patients, the satisfaction rate of discharged

(d) co-operation of sports physicians, physicians in general practice, coaches, physical education instructors and teachers, for the mutual exchange of information on basic

We tested if a permanent freshwater (FW) surface layer in snorkel sea-cages would lower AGD and salmon lice levels of stock relative to snorkel cages with seawater (SW) only and

The four step process for chronic disease home care management process, organised through the TDP funding initiatives, enables a very high degree of integration

Because a competitive bank chooses the level of asset risk to maximize the welfare of its creditors, the effect of changes in the level of transparency on the bank’s asset risk