Operational Risk And Data Robotics

May 2017OperationalRisk and DataRoboticsWHITE PAPEROperational riskand the Regulator’spoint of viewCosts of operationalrisk lossesData RoboticsSolutions

THE AUTHORSPaolo FabrisPaolo Fabris is a Partner at Avantage Reply. He has over 18 years of professional experience in FinancialServices. His focus has been on Regulatory projects and, in particular, on IAS/IFRS adoption in the bankingsystem, Individual and Group Financial Reporting, Risk Management on both market and credit risk, BaselII and Basel III adoption, and M&A.Paolo has focused his consulting skills in managing large and complex projects with particular attentionto the main regulatory aspects and requirements. He led important projects and programs taking careof business analysis, organizational change, regulatory compliance and internal auditing. His extensiveexperience in risk management spans all types of risks and a broad spectrum of financial products.Paolo Del GrandePaolo Del Grande is a manager at Avantage Reply Italy, specialized in financial services, risk managementregulation and accounting. He has ten years professional experience with the main Italian banks coveringcredit risk, operational risk, data governance, regulatory reporting and accounting standards. The mainactivities covered in theses years include: Statutory reporting audit, Italian and European regulatoryreporting (COREP/FINREP), RWA optimization, standard operational risk approach adoption, ECB AssetQuality Review 2014, Stress Test 2016, Data Governance projects and credit risk bank processes focusedon monitoring and NPL disposal.Francesca TerrizzanoFrancesca Terrizzano is a business analyst for Avantage Reply, with experience in Financial Services.Her focus is on risk management projects and specialising in Operations area and, in particular, businessprocesses mapping and optimization, especially in terms of automation, in order to increase the quality ofwork, reduce the processing times, decrease operational risk and reputational risk and limit the necessityof resources for the day-to-day operations. She takes part in the Reply Community of Practice on DataRobotics for the purpose of identifying new business solutions able to meet the needs of a constantlyevolving market. Francesca has a master’s degree in Economics and Finance from Bicocca University ofMilan.Roselisa GiulianoRoselisa Giuliano is a business analyst for Avantage Reply, with experience in the financialservices industry, in particular on European banking regulation and risk management. She isfocused in the ongoing transformation of banking business models, increasingly interconnectedwith the information technology. Roselisa participates actively in the Robotic Automation andMachine Learning project organized by Reply Group, such as Data Robotics Communityof Practice and Hackathon, with the objective of building real accelerators for the market.Roselisa has a master’s degree in Management and Business Innovation from Sapienza University ofRome.Avantage Reply, part of the Reply Group, is specialised in Financial Services with a focus on Risk, Compliance,Treasury and Capital and Financial Performance Management. With offices in Amsterdam, Brussels, Frankfurt,Hamburg, Lisbon, London (head office), Luxembourg, Milan, Munich, Paris and Rome, Avantage Reply counts someof the world’s most significant financial groups among its clients, including well-known and respected organisationsin the Banking, Insurance, Investment Management and Services, and Post Trading Services sectors. The firm’sdelivery capabilities cover advisory services (Risk/Finance/Treasury Subject Matter Expertise), Program and ProjectManagement, Business, Functional, and Data Analysis.

TABLE OF CONTENTSIntroduction2Operational Risk: The Regulator’s point of view3Costs of Operational Risk Losses6Data Robotics Solutions8What is Data Robotics?8What makes it work and what is the impact?9Where and when can it best be used?10Conclusion12Case Studies13Stress testing13Registry clean-up13Certified email management: foreclosures14Refund of employee loan repayments14Factoring: Credit acknowledgements141

Operational Risk and Data RoboticsIntroductionIntroductionData Robotics Solutions are emerging as a highlyeffective, yet practical approach for banks to reduceoperational risk, improve efficiency, reduce costsand derive additional value. From Robotic ProcessAutomation, which enables repetitive tasks to beautomated, to machine learning enabled IntelligentProcess Automation, which allows “robots” to take overcomplex and highly skilled tasks; banks that have startedimplementing these solutions are reaping the rewards,both from a financial and compliance perspective.This paper provides an overview of how Data RoboticsSolutions can help banks manage and reduceoperational risk, illustrated by case studies describingthe practical benefits of this powerful new technology,as implemented by Avantage Reply.2

Operational Risk and Data RoboticsOperational Risk: The Regulator’s Point of ViewOperational Risk: TheRegulator’s Point of ViewLosses attributed to operational risk in recent years haveresulted in increased focus by banks’ risk managementfunctions and heightened attention from regulators. Anumber of regulatory changes have been implementedin recent years, and there is more to come, including: A requirement to include and project operational riskinformation in regulatory stress testing processes(both the EBA and Bank of England exercises); andProposed changes to operational risk capitalrequirements, which include removing the AdvancedMeasurement Approach (‘AMA’) and using its basiccomponent (loss data collection) in the StandardizedMeasurement Approach (‘SMA’), a methodologythat implicitly values the management approach tooperational risk and involves two main components:1) A Business Indicator Component (‘BI’), repre-senting the operational risk associated with thebank’s business model; and2) A Loss Component, representing loss eventsover the last ten years.While the latter change remains hotly debated amongbanks and even national regulators, regardless of theoutcome, it serves to remind us of the importance ofoperational risk to a bank profitability and financialresilience.In order to reduce operational risk capital requirements(approximately 10% of total Risk Weighted Assets (‘RWA’) forEuropean commercial banks, as shown in Figure 2: Exampleof an EBA Risk Dashboard 3Q2016), and to thereforeimprove capital ratios, banks are incentivised to reduce thevolume and magnitude of operational risk losses.Figure 1: Drivers of new regulatory requirementsCurrentSimplicity Excessive complexity of AMA modelingPotential increased complexity for bankswith BIA and TSA related to data collectionand data quality processReduced computation complexity due to: the use of a “closed” algorithm for calculatingregulatory capital the absence of scenario analysis andexternal data use Difficulty in comparing the capitalrequirements for operation risks dueto the lack of homogeneity of the AMAapproaches used Little sensitivity to the actual exposureto operational risks for banks that usesimplified methods BIA and TSA Comparability Risk SensitivitySma Greater comparability in view of theapplication of the same algorithm to allbanks, even with simplified methodsGreater ability of the regulator to identifyand respond to potential systemic issuesIncrease in risk sensitivity by introducingspecific Business Indicator Componentfocused on the business model of thebank and Loss Data3

Operational Risk and Data RoboticsOperational Risk: The Regulator’s Point of ViewFigure 2: Example of an EBA Risk Dashboard 3Q20161RWA composition% of BE85.7%86.5%0.2%82.3%82.4% al riskcapital 1.2%11.5%11.6%83.1%82.6% 2%8.9%ES87.0%86.8% 8.5%8.6%FI82.7%82.4% .5%13.4%13.8%15.3%15.3%IE90.4%88.5% 91.0%90.8% 7.6%7.8%LV86.2%86.2% %86.9% 0%11.6%12.0%EE*n.a89.8% 0%10.1%10.1%10.0%10.2%79.0%79.0%Jun-16market risk equirementsCredit risk capitalrequirements %6.4%6.3%6.3%1 European Banking Authority, risk dashboard - data as of Q3 2016, p. 27, available 9/EBA Dashboard - Q3 6Sep-16

Operational Risk and Data RoboticsOperational Risk: The Regulator’s Point of ViewAccording to a recent paper from Harvard Business School on Operational Risk2: For G-SIBs – the average percentage of RWA for operational risk is 15% (higher than the EBA statistic of 10%); The range of these percentages is about 45% to 5%; The proportion of operational RWA to total RWA has risen 50% from 2008 to 2015; Of the types of operational risks, 75% are ‘regulatory’ related.European Systemic Risk Board (‘ESRB’) research on misconduct risk in the banking sector, conducted from 2009 to2014, shows that regulatory bodies have imposed fines of 200 billion euros, both in the form of sanctions and businessrestrictions. These fines are in relation to bank misconduct and its effects on financial stability.Figure 3: Cumulative misconduct costs3cumulative misconduct costs worldwidecumulative misconduct costs EU banks200150100500Dec 2009Dec 2011Dec 2010Dec 2012Dec 2013Dec 2014Furthermore, to take an example of one sizeable banking market within the EU, a recent Bank of Italy survey hasshown that approximately 50% of all banks perceive an increase in operational risk in their institution, up 10% fromonly six months prior.Figure 4: Operational risk across the Italian Banking Sector410%20%30%40%You see an increase inoperational risk in yourbank0%50%December 2016June 2016December 20152 Harvard Business School, Rethinking Operational Risk Capital Requirements, available equirements.pdf3 European Systemic Risk Board, Report on misconduct risk in the banking sector, p. 12, available at:http://www.esrb.europa.eu/pub/pdf/other/150625 report misconduct risk.en.pdf4 European Banking Authority, Risk Assessment Report December 2016, p.53, figure 59, available 97/EBA Risk Assessment Report December 2016.pdf

Operational Risk and Data RoboticsCosts of Operational Risk LossesCosts of Operational Risk LossesDue to the broad and typically sensitive or confidential nature of operational risk losses, it is challenging to achievefull visibility of the distribution of losses. However, staying with the Italian example, “DIPO”, the Italian Operational RiskLosses Database, contributed to by 33 banking groups, gives a good indication, as shown below.Figure 5: Distribution of operational risk losses in Italian banks over the course of 20165Percentage of loss amountPercentage of loss numberEvent type23.6%727.4%61.2%0.9%Losses due to shortcomings inthe completion of transactionsor handling of processes, andin relations with commercialcounterparties, vendors and suppliers2Losses due to malfunction/unavailability of IT systems30.5%5Losses due to natural disasters or otherevents such as terrorism or vandalism1.2%Losses due to default, relating toprofessional obligations to customersor to the nature or configuration ofthe product/service provided44.2%439.4%3Losses due to action not compliantwith laws or agreements on usage,health and safety in the workplace,compensation for personal damagesor discrimination or failure in applyingequal treatment terms4.9%4.6%Losses due to fraud, embezzlement orinfringement of laws by external people12.4%223.9%123.2%Losses due to unauthorized activities,fraud, embezzlement, infringementof laws, regulations or corporatedirectives involving at least oneinternal person2.5%0%610%20%30%40%50%5 Database Italiano Perdite Operative, Standard Report, p.2, available at: t%2002SE2014%20ENG.pdf1