Calculating Operational Risk Capital Craig Ivey Royal Bank of - - PowerPoint PPT Presentation

Document Classification: Public

Craig Ivey

Royal Bank of Scotland, Head of Operational Risk Appetite and Capital

Calculating Operational Risk Capital

Disclaimer: this presentation represents the speaker's opinions and does not necessarily represent the views of the RBS Group or its subsidiaries.

Document Classification: Public

2

Banks use a range of advanced tools, such as Economic Capital models, to enhance their management of risk and capital:

• This presentation provides an introduction into Operational Risk Capital

modelling and changes on course in light of regulatory developments.

• The following questions will be addressed:

1. How do we currently estimate our Operational Risk Capital requirements? 2. Why are the simpler approaches for Operational Risk Capital under review? 3. What drives the risk based Operational Risk Capital estimates?

OpRisk Capital: Executive Summary

Document Classification: Public

OpRisk Capital: Background Context

Operational Risk is evolving, quantification has proven difficult:

• Broad definition of Operational Risk (BCBS 2006); universally fundamental to a

bank’s risk management framework.

• Thought leadership to date highlights the on-going evolution of Operational Risk

Management and need for further research and critical evaluation.

• While some seek more advanced approaches, on going scrutiny of the internal

model approach known as AMA has also become a common theme: – Particularly due to challenges standardising this approach (Mr Coen, 2015).

3

Document Classification: Public

4

The management of Operational Risk is one of the components of the Bank’s risk management framework:

• We expect some level of Operational Risk losses as part of doing business,

such “expected” losses are generally absorbed by income.

• More severe losses (“unexpected” losses) are, however, to be absorbed by

capital reserves.

• Regulators set minimum levels of capital reserves to support financial stability.
• Additional Economic Capital is held where appropriate.
• Total capital requirements, including Economic Capital, are assessed through

the bank’s capital risk assessment known (in the UK) as the Internal Capital Adequacy Assessment Process (ICAAP).

OpRisk Capital: An Introduction

Document Classification: Public

5

Significant Operational Risk losses have arisen due to the following:

• Mis-selling of complex products:

– Mortgage Backed Securities fines and settlements e.g. JP Morgan’s $13bn. – Payment Protection Insurance mis-selling losses of £21.8bn have been

incurred since 2011 due to the length of time which mis-selling occurred.

– Use of deceptive marketing practices, involving breach of law and/or

regulations; e.g. Bank of America’s $8.4bn settlement. Ineffective controls within operating processes allowed for:

• Rogue trader losses on the back of market movements; e.g. Societe Generale

€4.9bn loss.

• Losses arising from the unsuitably leveraging and / or off-balance sheet

exposures; e.g. Bernard Madoff Ponzi scheme, which cost hedge fund investors $50bn.

OpRisk Capital: Loss Themes

Document Classification: Public

6

Banks are increasingly expected to self-assess their Operational Risk capital requirements during normal and ‘stress’ periods:

• There are two simpler approaches to calculate minimum Operational Risk

capital (BIA and the TSA).

• BIA and TSA are based on three years Gross Income multiplied by a set of

percentage rates (15% for BIA and 12% – 18% for TSA).

• A third approach uses a risk-based statistical method; it is known as the AMA.
• Operational Risk Economic Capital models are often built to AMA standard.

– Risk based approaches use inputs directly from the bank’s Operational Risk framework and external loss data. – In order to capture risks not fully identified or measured by these simpler approaches, bank’s can use Economic Capital models if not AMA.

• Bank’s separately consider risk during ‘stress conditions’ as part of formal

regulatory Stress Test and our financial planning process.

• Comparative analysis is performed as part of benchmarking process.

OpRisk Capital: How do we Calculate?

Document Classification: Public

7

The following criticisms of BIA / TSA have emerged:

• The use of Gross Income incorrectly assumes a direct relationship between

income and Operational Risk losses.

• Gross Income is historic and so fails to provide a forward looking view.

The Basel Committee intends to consult on a new standardised approach termed the Standardised Measurement Approach:

• This will be a new paper developing upon the current consultation paper which

was set out in October 2014.

• Capital may be flexed up or down based upon organisational size.
• Many banks already cover BIA / TSA limitations using the Economic Capital

model and / or other capital buffers.

OpRisk Capital: Simpler Approaches Under Review

Document Classification: Public

OpRisk Capital: Economic Capital Model

Brief overview of a ‘standard’ Operational Risk Economic Capital Model:

• An Operational Risk quantification system is normally built to AMA standards

(e.g to a 99.9th percentile soundness).

• This includes four key data elements; internal & external loss event data,

scenario analysis and risk and control factors (BEICF).

• A range of capital measures (confidence intervals) and allocations (of differing

granularity and risk focus) can be devised in practice:

– A range of confidence intervals can provide a range of risk thresholds / limits.

• Assumptions are made regarding model granularity, diversification benefits and

correlation structures:

– Work best if meaningful and inform risk management; insight into risk

relationships provides insight into risk management.

• Such a design would facilitate a range of Risk Appetite measures such as risk

limits and capital allocation e.g. settling gross income to loss limits.

8

Document Classification: Public

9

OpRisk Capital: Loss Distribution Approach

Economic Capital models often use industry standard approach for modelling Operational Risk:

• The model provides a risk estimate for each risk unit.
• The severity and frequency of events is estimated for the next year based upon

Internal Loss Data (ILD) and External Loss Data (ELD).

• We can then read off the confidence level as desired, standards are very high so

we aim to be 99.9% confident that we will have sufficient capital.

Document Classification: Public

10

The following key inputs are required to run the model:

• ILD provides the starting risk profile reflecting the bank’s actual loss experience.
• ELD from the ORX database addresses the question ‘could it happen here’?
• Internally generated Scenario Analysis provides a forward looking risk view

which improves the model’s assumptions where appropriate e.g. by adjusting the severity estimates:

• Risk & control factors (BEICFs) are adjustments that reflects the state of the

control environment within the bank.

OpRisk Capital: Model Components

Document Classification: Public

OpRisk Capital: Forward Looking View

The forward looking view of risk:

• This is primarily done through scenario analysis and BEICFs.
• BEICFs can capture more granular risk and performance incentives.
• Scenario analysis is the use of expert opinion in conjunction with external data

to evaluate exposure to high-severity events (BCBS 2006).

• Scenarios can identify emerging tail end risk by allowing ‘what ifs’ analysis

which provides both risk management and measurement benefits.

11

Document Classification: Public

OpRisk Capital: Model Granularity and Allocations

ORX 2015: Capital is usually modelled at a combination of business line / event type level (83% of respondents) and then allocated to individual business units:

• Capital allocation varies across participants using

different factors smoothed over 3-5 years for stability.

• Allocation can be model based or use more

subjective approaches.

• Allocation metrics include; management accounting

parameters, loss frequency / severity, restricted to large events, income or revenue, FTE and assets.

• It’s desirable that an understandable risk sensitive

allocation approach is used.

12

Document Classification: Public

OpRisk Capital: Model Reliance?

Qualitative limits, top of house endorsement and the right risk taking culture are key enablers for any quantitative solution:

• The Operational Risk framework must be constructed with complementary

qualitative and quantitative elements.

• Model validation (Model Risk Management), verification (Audit) and governance

(Senior Management) are key elements to reduce model risk.

• Must also consider ‘use test’ and how this is demonstrated.
• Other benchmarks e.g. challenger models, scenario analysis, back-testing etc.

remain important.

13

Document Classification: Public

OpRisk Capital: AMA Context

The future of Operational Risk modelling:

• This is a divisive subject; concerns over uniformity across banks, differing views
• n where AMA rest, and nuances with modelling Operational Risk.
• The Basel Committee has recently noted it intends to consult on the possibility
• f removing AMA from the range of Pillar 1 approaches.
• Conversely international adoption of the AMA for Operational Risk by large,

complex, international financial institutions is now at an all-time high: – Eight US banks were AMA approved as of February 2014. – Ten European Banks have AMA approval as of December 2014. – The largest Swiss banks and largest Australian banks have AMA approval.

• It remains evident that internal models are increasingly relied upon for key risk

management and product pricing decisions for larger international banks.

14

Document Classification: Public

OpRisk Capital: Conclusion

Operational Risk Economic Capital Models likely to be part of the future for large, complex international banks:

• Operational Risk is not an easy area in which to quantify limits, but it can be

done: we now have over ten years of experience behind us.

• Operational Risk Economic Capital models can breach this gap if constructed

and used correctly.

• While the future of AMA is uncertain use of advanced risk based models for

Operational Risk has high value in the industry.

15

Document Classification: Public

16

Questions? Contact details: Email: craig.ivey@rbs.com

OpRisk Capital: Questions?