Data driven approach to KYC
Download as PPTX, PDF2 likes602 views
The document discusses a data analytics approach to Know Your Customer (KYC) processes in financial institutions, detailing various components such as customer identification, risk scoring, and ongoing transaction monitoring. It highlights challenges in customer due diligence, including data quality issues and regulatory compliance, while emphasizing the use of machine learning and automation to enhance these processes. The content also touches upon the importance of understanding the ultimate beneficial owner (UBO) and the implementation of advanced analytical solutions to improve operational efficiency and detection of financial crime.
Data driven approach to KYC
- 1. Data Analytics Approach to KYC Pankaj Baid 1
- 2. 2 KYC Value chain SOURCE: Mckinsey Customer Identification Program (CIP) Sanctions Screening Risk Scoring (Onboarding) Enhanced Due Diligence (EDD) Transaction Monitoring Risk Scoring (ongoing) Account Monitoring Case Management Documentary ID Validation Non- documentary ID Validation (e.g. Security Questions) Accounts with missing / mismatch info routed to case management Follow up with customers regarding missing/ mis- matched information (e.g. address mismatch) Tools – LexisNexis SBS, Equifax Cross Check against OFAC list, PEP list, negative news Tools: Manual search; or vendor apps (Filtering model) Reports & Notices (routine & annual blocked property reports; pre-penalty notices; hearing requests) Analytical engine segmenting customers into high, medium and low risk groups based on risk assessment factors & information collection during application process High risk customers routed to EDD; the rest are cleared Tools – Customer Risk rating (CRR) models In –depth review of high risk accounts (e.g. foreign, offshore) Follow up with customers if additional information is required for EDD Decisioning & customer communication Clear and release/ Restrict/ Close Ongoing screening of transactions (e.g. – Wire deposits, withdrawals) with an analytical engine Alert generation for suspicious activities Reporting (CTRs, SARs) Tools – AML Transaction Monitoring Models) Analytical engine re-scoring existing customers on a regular basis based on risk assessment factors and ongoing customer behavior (amount of transactions performed) High risk customers routed to EDD; the rest go through ED Tools – Customer Risk rating (CRR) models Regular review of the existing customer accounts based on risk score determine The Risk Rating determines the frequency of periodic KYC reviews Event-triggered sanctions re- screening Tools: Watch list filtering models Alert generation for suspicious transaction activities Reporting (CTRs, SARs) Alert intake and assessment False positive assessment Investigation management Disposition Periodic KYC review KYC remediation EDD of high risk customers Customer Communication KYC (onboarding of new customers) AML and periodic KYC (monitoring existing customers)
- 3. 3 Innovative analytical solutions to age-old problems Machine learning models can enrich transaction monitoring alerts and boost Suspicious Matter Report (SMR) conversion rates Analytics can also enable customer segmentation and profiling for various business purposes, including compliance and marketing. In addition to ensuring the screening engine is operating at peak performance with accurate data, emerging AI and analytical methods can also be used to address operational efficiency issues related to case investigation. Transaction Monitoring Know Your Customer Sanctions Screening SOURCE: EY Global - https://www.ey.com/en_gl/consulting/how-data-analytics-is-leading-the-fight-against-financial-crime Focus of the Presentation
- 4. 4 Challenges of Customer Due Diligence SOURCE: https://www.cognizant.com/whitepapers/digital-customer-due-diligence-leveraging-third-party-utilities-codex2386.pdf Data Quality Ascertaining the ultimate beneficial owner (UBO) Lengthy turnaround time ▪ Banks face numerous challenges when collecting required data for client due diligence (CDD). ▪ Information received from the public domain is often inaccurate, and clients often have privacy concerns when it comes to sharing personal information. ▪ Bank employees are hesitant to pressure clients for fear of antagonizing them. ▪ This is perhaps the most daunting CDD requirement, particularly with the increased pressures of FinCEN’s recently released Final Rule on Beneficial Ownership and the 4MLD regulation. ▪ In the European Union, banks are required to identify individuals with 25% or more equity interest in a legal entity. Most banks lack a well-established policy to determine the UBO for new accounts. ▪ The current CDD process is cumbersome and time-consuming, with bank employees relying heavily on complex spreadsheets and manual processes to assess AML risk for a client. ▪ The process is plagued by delays and manual workarounds. Siloed processes and lack of standardization ▪ In the absence of a standardized AML risk assessment template, the processes and rules for collecting, maintaining and updating client data differ vastly across most banking organizations. ▪ Multiple siloed systems and user interfaces are used for client onboarding and maintaining client data. ▪ Different lines of business use different search tools, and document storage methods Ever-changing regulatory requirements ▪ Because global banks come under the purview of multiple regulatory bodies, their internal systems need to be agile enough to accommodate these varied and often contradictory requirements across geographies.
- 5. 5 KYC Analytics revolves around multiple analytical activities across the value chain SOURCE: https://www.wns.com/insights/infographics/infographic-detail/653/navigating-the-kyc-aml-compliance-challenge-with-analytics • Automation saves cost and time (bulk of CDD tasks are logical and can be automated) • Advanced algorithms and statistical modeling reduce false positive rates Features across the analytical value chain • Customer Screening and due diligence • Nothing gets past without missing its rule-based detection • KYC document checks across multi-formats and multi-source and third party databases (OCR to extract document metadata) • AA enabled screening to check for any Politically exposed Person (PEP) and Special Designated Nationals (SDN) and OFAC (Office of Foreign Assets Control) and adverse news using NLP • Continuous updates as per global watchlists apart from facilitating near real-time investigation of negative news (ML algorithms) Banks leverage Automation, AI and Analytics Enhanced Due Diligence (EDD) Customer Due Diligence (CDD) Sanction Screening & Credit Checks Setting up a Watchlist that is continuously updated against various sanctions and non-sanctions databases like PEPs, SDNs, OFAC Sanctions etc Customer Identification Program Defining and implementing a name matching tolerance process Analytical Algorithms to implement various name checks Politically Exposed Person (PEP) checks Risk Identification and Scoring •Setting up exclusion criteria for the rules and scenarios implemented Determination of customer’s Risk in terms of propensity to commit money laundering, terrorist finance or identity theft Gather & Verify client Information
- 6. 6 Real-Time investigation of negative news involves significant advanced analytics SOURCE: • Negative-news screening (also known as adverse-media screening) has been recommended by regulatory authorities in high-risk as part of enhanced due-diligence procedures. • Many financial institutions still use manual approaches for negative- news screening. With many third- party solutions available, however, they can automate this process. • Investments in artificial intelligence (AI) and digital tools can dramatically improve the reach of the screening and the quality of insights it will surface. • Available solutions produce potential leads but also sets of insights to help analysts assess and prioritize information in the broader context of the case. Data acquisition by keyword search to retrieve articles Grouping of articles by subject or incident within each event type Natural-language processing to analyze language usage and extract a set of features (such as related people names) Integration with multiple sources across countries and languages Association model to relate searched entities and articles (from unassociated to highly associated) workflow functionality, including audit traceability, visualization, and integration with other tools Auto-adjudication to highlight potential false positives Event-Classification Model to organize by article topic (known as “event type”) Negative News Screening
- 7. 7 Name Screening and Matching Process • Defining the criteria • In-scope and out of scope entities • Country & relationship risk inclusion Exclusion Criterion • Defining the name screening process • In-scope and out of scope entities • Country & relationship risk inclusion Name Screening Definition • Deterministic vs. Probabilistic matching • Provision for phonetics, if available • Matching tolerance level setting Risk Based Name Matching Tolerance • Data cleansing • Extraction of additional date fields • Foreign language/numeral corrections Apply Data Standards • Source list to be used and filtering • PEP/Criminal/Notoriety lists (Local) • UN/OFAC/SDN lists (International) Watch List Selection SOURCE:
- 8. 8 Name - Matching Technology SOURCE: https://www.cognizant.com/InsightsWhitepapers/OFAC-Name-Matching-and-False-Positive-Reduction-Techniques-codex1016.pdf Useful algorithms have powerful routines that are specially designed to compare names, addresses, strings and partial strings, business names, spelling errors, postal codes, tax ID numbers, data that sounds similar (such as “John” and “Jon”) and more. There are two common types of matching technology on the market today: deterministic and probabilistic Direct Match or Exact Match (Deterministic) Partial Match (Probabilistic) Indirect Match Fuzzy Matching (Probabilistic)
- 9. 9 Risk Assessment Parameters SOURCE: The following table defines the risk assessment parameters and provides details irrespective of the customer type Definition Risk Assessment Parameter Detail Identification Verification , IDV for Interested Parties Proof of Identity Geography Risk - Country of Citizenship Risk associated with the country of citizenship of a customer. Geography Risk - Country of Residence Risk associated with the country of residence of a customer. Geography Risk - Country of Taxation Risk associated with a country where the customer pays tax. Source of Wealth Risk associated with the source of wealth defined by the customer in relation to the account being opened. Occupation Risk associated with the occupation a customer performs. Length of Relationship Risk Risk associated with the length of relationship a customer has with the bank or FI. Watch List Risk Risk associated with the customers being listed in the watch list maintained by the bank. The score assigned would be based on the list where the customer is matched. Negative News Risk Risk associated with a customer who is available in the negative news search. Risk Assessment Parameter Detail Geography Risk - Countries of Operations Risk associated with the country where the customer's business is being operated. Geography Risk - Country of Headquarters Risk associated with a country where the headquarters of the customer is located. Industry Risk Risk associated with the Industry where the customer is employed. Legal Structure & Ownership Risk Risk associated with the legal structure (Trust) of a customer based on whether it is publicly or privately held. Corporation Age Risk Risk associated with the age of the corporation in the industry Risk Associated with the Markets Served Risk associated with difference markets served as stated by the customer for its operations. Risk Associated to Public Company Risk associated with type of the company, public or private. Risk Associated with the Products Offered Risk associated to the different products served as stated by the customer. Risk associated with Method of Account Opening Risk associated with different method of account opening. For example Online/Walk in/Phone
- 10. 10 Risk Identification and scoring approach using a Customer Risk Rating (CRR) Model SOURCE: Objective Client Type Dimension Dimension Weight Factor Factor Weight Standardized Value Factor Score Dimension Score CRR Score Individual Client 25% Citizenship 100% NO 5 5 4.2 Geography 30% Country of Primary Address 100% HR 5 5 (Max of Country Scores) Country of Mailing Address MR 3 Country of Statement Mailing MR 3 Product 25% Product 100% HR 5 5 Delivery Channel 10% Delivery Channel 100% FACE-TO-FACE 1 1 Length of Relationship 10% Length of Relationship 100% >5 YEARS 1 1 Build a comprehensive risk assessment model using data analytics techniques to: • Carry out detailed risk assessment of customers by identifying factors like customer profile information, transaction/behaviour attributes, delivery channels, etc. • Calculate a customer risk rating score (normally a low, medium or high score) by applying specific tolerance limit on the expected incoming and outgoing volumes for profile monitoring purposes. CRR Process Data Collection Data Standardization Account Analysis Customer Aggregation Customer Analysis CRR Score Identification of data sources and data extraction Reviewing the data quality Removing inconsistencies such as white spaces, dashes etc. Calculation of CRR Factor scores at account level Computation of individual factor scores Aggregation of account level data Defaulting the MSB, PEP and SAR customers as identified by the client to a high risk score Final CRR score as a weighted sum of individual scores CRR Calculation Risk Score = (Factor Weight * 100) / Sum of weights of all factors
- 11. 11 Potential applications of AI in the Automated Review System of AML processes SOURCE: White Paper on AML and fundamental rights
- 12. 12 Discovery of Transactional Behavior using Advanced Analytics algorithms SOURCE: https://www.incubegroup.com/blog/know-your-client-better-discover-transactional-behaviour/ Step 1 – Feature Engineering ▪ Design characteristics that describes an account’s transactional behavior ▪ Aggregate Transactions into a Vector Step 2 – Discover Cluster Structure ▪ Use a clustering algorithm on these vectors. Choices include K- means, hierarchical clustering, density-based clustering, etc. Objective Build a comprehensive model using data analytics techniques to: • Detect cases of potential fraud and accounts misuse for money laundering or tax evasion • Increase the efficiency of transaction analysis and automatic pattern discovery ▪ A typical inheritance account experiences one or two huge inflows, after which it is slowly drained with relatively infrequent outflows. (#3333333 in this case) ▪ Savings accounts usually have regular, but infrequent inflows of about the same volume. (#111111 in this case) ▪ Commercial accounts, which perform many transactions, have a huge turnover and wide geography. (#222222 in this case) ▪ Two outliers are seen instantly: those are a savings and an inheritance account that were allocated to the cluster of commercial accounts by the algorithm. ▪ The cluster detection algorithm monitors transactional behavior and immediately notices deviations and anomalies. The accounts with detected inconsistencies are brought higher up in the periodic review priority and submitted to a compliance officer.
- 13. 13 Client Profiling Increase scope to include additional data fields while client profiling Retail – KYC (Know Your Customer) Corporate – KYB (Know Your Business) Implement enhanced due diligence (EDD) for clients belonging to high risk geographies to ensure that you’re capturing all the customer information needed to assess risk Implement enhanced due diligence (EDD) for entities operating in high risk zones and in businesses with significant cash component to ensure that you’re capturing all the company information needed to assess risk Additional fields like – educational background and institution, occupation, transaction value/frequency, past 3 residential addresses, past 3 employers, criminal history, check on politically exposed persons (PEPs) should be collected while creating a client profile Additional fields like –ownership structure, ultimate beneficial owners (UBO), criminal history of promoter, past corporate addresses, transaction value/frequency, source of funds, photo IDs for their employees should be collected while creating a business profile Use this to build a robust client profile which captures all business relationships, transaction behavioral information Advanced analytics algorithms to profile and screen all transactions Transaction Profiling • Based on historical data, use ML algorithms to classify a transaction as either a fraudulent, clean or a doubtful transaction • Automate the process to map these transactions to the existing and new client profiles • The client profile is also updated on a continuous basis and matched against sanctions and PEP-lists • Use this mapping for risk classification to identify the verification mode and strictness of verification for these customers Use ongoing due diligence, analytics based KYC norms for different client profiles Ongoing Analytics led KYC • Identify new cases/avenues of money laundering and update client profiles and transaction profiles as an ongoing process in compliance systems hereby improving data quality • Ensure in-person verification happens for the fraudulent one and ensure simple app verification for clean cases and periodical reviews for the doubtful cases Data driven approach for enabling a robust KYC (Know Your Customer) SOURCE:
- 14. 14 What is an Ultimate Beneficial Owner (UBO)? SOURCE: UBO refers to the person or entity ultimately benefiting from the business relationship Definition • The reason why UBO has grown in importance is because malicious actors, like terrorists, narcotics dealers and others increasingly try to obfuscate their identity through the use of legal entities and even other individuals. • In the case of legal entities, the Ultimate Beneficial Owner: • Holds at least 25% of the entity’s capital OR • Holds at least 25% of entity’s voting rights at the AGM of stakeholders OR • Is the beneficiary of at least 25% of the entity’s capital Steps in UBO Screening UBO Screening • Step 1 – Verification is usually done with a government issued document such as a passport. • If the identity of a company needs to be verified, then Articles of Association will be checked. • Step 2 – Perform a compliance check • Essentially this involves estimating the risk of the UBO. This might include checking to see if the UBO is on politically exposed person (PEP) lists, sanction lists, freeze lists or is under investigation. Negative publicity should also be taken into account. • Step 3 – Prepare a risk estimate • An organization must use objective indicators to place their clients into different risk categories that range from low to high. The nature and scale of the client screening may be adjusted based on this risk assessment. Essentially, is a client is lower risk, less effort is needed to reduce potential risk. For low risk clients, simplified screening suffices. For mid to high risk clients, stricter screening is needed. • Step 4 – Filing of Identification Data • Organizations need to file and update their data for a minimum of five years. Information also needs to filed efficiently and consistently in order to keep data up to date.
- 15. 15 Information collected for Customer Due Diligence (CDD) and Enhanced Due Diligence (EDD) SOURCE: https://justcoded.com/blog/what-is-the-difference-between-kyc-and-aml/ % of respondents % of respondents
- 16. 16 Latest Trends in this space.. Behavioral Biometrics is the next generation of security Behavioral Biometrics • An innovative approach to user authentication based on creating a unique profile for every customer. • The idea of behavioral biometrics is to essentially monitor user activity over time, establish their regular behavior – behind the scenes. • Typically the most commonly used type of behavioral biometrics include: Automatic recognition of patterns such as how keystrokes are made on a phone or tablet, how a mouse is used, IP addresses and geo-location • NatWest plans to replace passwords with behavioral biometrics and is in testing phase with Visa as their technology partner Network Analysis is the future of AML Network Analysis • In Practice, statistics from a network (for e.g. How closely it resembles a known money laundering topology) would be incorporated into existing customer risk rating and transaction monitoring models as inputs to improve model accuracy. New capabilities such as community detection would help accelerate investigations and identify hidden risks • Start by building a network of existing customer links by using account transfers, shared account ownership, and payments to build linkages both internally and to external institutions using the destination account number. Then create inferred links between customers by looking at shared addresses, employer, or social media data Block chain Technology in KYC Block Chain • Current KYC processes also entail substantial duplication of effort between banks (and other 3P institutions). While annual compliance costs are high, there are also large penalties for failing to follow KYC guidelines properly. Rabobank proposed the use case that “KYC statements can be stored on the Blockchain. • Once a bank has KYC’d a new customer they can then put a summary of the KYC documents on Blockchain which can then be used by other banks and other accredited organizations (such as insurers, car rental firms, etc.) without the need to ask the customer to start the KYC process all over again. • Overall operational cost savings are estimated to be around $2.5bn dollars. AML penalties will also be reduced by $0.5bn to $2bn dollars. SOURCE: Multiple Sources from Internet
- 17. 17 Gemalto IdCloud – A Thames Group Banking Solution using Behavioral Biometrics
- 18. 18 Making your KYC remediation efforts risk and value-based (Mckinsey recommendations) SOURCE: https://www.mckinsey.com/industries/financial-services/our-insights/banking-matters/making-your-kyc-remediation-efforts-risk-and-value-based Most banks expend disproportionate effort on customers who pose very little or no risk A model that segments customers more finely – perhaps into as many as 10 to 30 categories – can ensure remediation efforts are aligned with the level of risk. To manage both risk and value, segment customers more finely Self-service should be the default option for customers providing KYC information. By automatically posing more questions to customers whose responses suggest higher risk, the burden on less-risky customers is kept to a minimum Deploy self-service solutions that are risk-sensitive and carry minimal execution costs Remediation efforts will be more powerful if teams follow the approach used by digital marketers. Banks can track each customer’s progress though the KYC and due diligence process, determining appropriate actions at each stage depending on the customer’s preferences, behavioral profile and risk categorization. Tailor and track remediation efforts at the individual customer level There are plenty of off-the-shelf solutions and data providers that can help quickly stitch together an integrated solution. AI can then accelerate learnings from these outputs To quicken progress, make use of third-party data, external providers and artificial intelligence (AI)
- 19. 19 To adopt the new generation of customer risk-rating models, financial institutions are applying five best practices SOURCE: https://www.mckinsey.com/industries/financial-services/our-insights/banking-matters/making-your-kyc-remediation-efforts-risk-and-value-based Leading institutions examine their AML programs holistically, first aligning all models to a consistent set of risk factors, then determining the specific inputs that are relevant for each line of business. The approach not only identifies risk more effectively but does so more efficiently, as different businesses can share the investments needed to develop tools, approaches, standards, and data pipelines. Simplify the model architecture Machine-learning algorithms can search exhaustively through subsegments of the data to identify where quality issues are concentrated, helping investigators identify and resolve them. Sometimes, natural-language processing (NLP) can help Improve data quality Connecting the insights from transaction-monitoring models with customer risk-rating models can significantly improve the effectiveness of the latter. A more effective risk-rating model updates customer information continuously, flagging a change of address to a high-risk country, for example. Continuously update customer profiles while also considering behavior Statistically calibrated models tend to be simpler. And, importantly, they are more accurate, generating significantly fewer false-positive high-risk cases. Experts still play an important role in model development. They are best qualified to identify the risk factors that a model requires as a starting point. And they can spot spurious inputs that might result from statistical analysis alone. Complement expert judgment with statistical analysis Feature-selection algorithms that are assumption-free can review thousands of potential model inputs to help identify the most relevant features, while variable clustering can remove redundant model inputs. Predictive algorithms (decision trees and adaptive boosting, for example) can help reveal the most predictive risk factors and combined indicators of high-risk customers Deploy machine learning and network science tools
- 20. 20 Effective, efficient risk-rating models use a consistent set of risk factors, though inputs will vary by business line SOURCE: McKinsey
- 21. Recommended FATF1 Guidelines for an effective KYC – Know Your Customer SOURCE: 1 Financial Action Task Force Recommendations •Establishing business relations •There is suspicion of money laundering or terrorist financing •There are doubts about the veracity or adequacy of previously obtained customer identification data •Carrying out occasional transactions above a threshold Customer Due Diligence (CDD) measures will be undertaken when: •Identify the customer and verify that customer's identity using reliable, independent source documents, data or information •Identify the beneficial owner, and take reasonable measures to verify the identity of the beneficial owner •Obtain information on purpose and intended nature of business relationship •Conduct on-going due diligence on the business relationship and scrutiny of transactions undertaken thorough the course of that relationship CDD Measures to be taken are as follows:
- 22. Appendix
- 23. 23 23 23 23 23 23 23 23 23 23 SOURCE: https://www.mckinsey.com/business-functions/risk/our-insights/the-investigator-centered-approach-to-financial-crime-doing-what-matters Case Study – Low performance risk rating models without advanced analytics should not be allowed into production • This represents a typical multifactor customer risk-rating model for the retail business of a large North American universal bank • A manually conducted expert review of the results revealed that for every 100 customers rated high risk, 72 were actually medium to low risk; furthermore, 57 of every 100 customers rated medium to low risk by the model proved on review to have a high-risk profile • To put this into perspective, a credit- risk model with this kind of performance would never be allowed into production 100 72 57 85 High risk customers according to customer risk rating model "Low risk" cases removed (false positives) "High risk" cases added (false negatives) High risk customers after expert review High risk customers sent to enhanced due- diligence units (disguised real data example), indexed to 100
- 24. 24 24 24 24 24 24 24 24 24 24 1 Suspicious Activity Report SOURCE: https://www.mckinsey.com/business-functions/risk/our-insights/the-neglected-art-of-risk-detection Case Study – Bank used enhanced data and analytics to dramatically reduce the money laundering activities • At one large US bank, the false-positive rate in anti–money laundering (AML) alerts was very high. The remedial process involved a two-stage investigation. One team would determine whether an alert was truly triggered by suspicious activity. It would eliminate clearly false positives and pass on the remainder to experts for further investigation. Very few suspicious-activity-report filings resulted. • The bank rightly felt that this elaborate procedure and meager result was overtaxing resources. To improve the specificity of its tests so that AML expertise could be better utilized, the bank looked at the underlying data and algorithms. It discovered that the databases incompletely identified customers and transactions. By adding more data elements and linking systems through machine-learning techniques, the bank achieved a more complete understanding of the transactions being monitored. • It turned out that more than half of the cases alerted for investigation were perfectly innocuous intracompany transactions. With their more sensitive database, the bank was able to keep the process from issuing alerts for these transactions, which substantially freed resources for allocation to more complex cases Before enhanced data and analytics,% After enhanced data and analytics,% 0 90 10 8 2 100 50 45 5 3 2 100 Total alerts Known intra- company transfers Reviewed by primary team and closed Reviewed by secondary team Closed by secondary team Filed as SAR 1
- 25. 25 25 25 25 25 25 25 25 25 25 • It is estimated that the combined revenue lost due to financial crime is $1.45tn annually. • This is despite the collective billions of dollars that are spent in an effort to prevent crimes like money laundering, fraud, theft, and corruption. Combined revenue lost in 2018 ($bn) SOURCE: Refinitiv, “Revealing the True Cost of Financial Crime,” 2018. The True Cost of Financial Crime 188 209 239 241 267 309 0 50 100 150 200 250 300 350 SLAVE LABOR & HUMAN TRAFFICKING THEFT FRAUD CYBERCRIME MONEY LAUNDERING BRIBERY & CORRUPTION Recent studies have surveyed the impact of financial crime on business, society, and people, highlighting the high cost and difficulty in overcoming it.
- 26. 26 A financial crime analytics framework must have these essential components SOURCE: Deloitte Financial Crime Analytics Business Rules • Logic rules (if…then..) to select cases with a high financial crime risk directly Exploratory Analyses • Profiling and outlier detection based on statistics or visualizations Predictive Modelling • Self learning risk classification based on advanced analytics techniques Social network analysis (SNA) • Focused analyses on relations of a particular set of customers Alert generation, risk score models Outlier detection, cluster analysis, data visualization Machine learning, regression analysis, decision trees, neural networks Regular SNA, Graph Databases Technique Description Example Suitable for well known financial crime patterns Suitable for unknown financial crime patterns Suitable for complex financial crime patterns Suitable for large financial crime investigations Application
- 27. 27 Financial crime and fraud are common in numerous industries
- 28. 28 Key Categories of BSA/AML Risk for Banks – Risk Assessment Approach SOURCE: Products and Services • Do we have significant volumes of electronic payments, such as wire transfers, prepaid cards, and remittances? • Do our customers actively engage in, or have we recently implemented, electronic banking services, such as remote deposit capture, online account opening, and/or Internet transactions? • Do we provide services to third- party payment processors or senders? Customers and Entities • Do we have a significant portfolio of cash-intensive business customers, such as privately owned ATMs or convenience, liquor, or retail stores? • Does our customer base include foreign entities, such as financial institutions corporations, and/or individuals or governments? • Do we have a significant number of professional service provider customers, including attorneys, accountants, real estate brokers, etc.? •Does our customer base include a significant number of politically exposed persons? Geographic Location •Do our customers engage in or process transactions involving international locations identified by the U.S. Treasury Departments, the FATF, or other international bodies as having strategic deficiencies in their countries' AML frameworks? • Are any of our customers located in, or do they conduct transactions with, offshore financial centers? • Do we maintain branches in or have significant customer populations located within domestic locales designated as High Intensity Drug Trafficking Areas and/or High Intensity Financial Crimes Areas? Banks should ask itself several questions to help identify some of these areas of heightened BSA/AML risk
- 29. 29 List of High Risk business activities that are considered potential source of money laundering SOURCE:
- 30. 30 Model of an Integrated Architecture supporting Financial Crime Compliance for KYC, AML, fraud SOURCE: