Printable versionSend by emailPDF version
March 27, 2017

The past decade has unarguably been one of the most revolutionary decades for the financial services industry. Starting with the financial market crisis during 2007-2009, the Libor manipulation scandal through 2009, the Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010, and Basel III by the Basel Committee on Banking Supervision established during 2010-2011, the financial services regulators internationally have stepped up their activity with sweeping regulatory reform to increase oversight, governance and enforcements of the participating banks. Financial institutions are experiencing an increased burden of transparency, internal controls and effective response more than ever before.

Additionally, the ever-increasing growth in the variety and volume of data makes it ever harder to comply with the mandate from regulators for more transparency and increased compliance analytics. Historically, transactional data analytics among financial services firms have taken the form of analysis on structured data derived from transactional applications, such as trading systems, settlement systems and general ledgers. And separately, the review of unstructured data has been focused on review of emails and other business documents, such as Word, Excel, PowerPoint, and PDF, through EDiscovery approaches. However, the majority of new data created now takes the form of semi-structured data, such as IM, chats, social media and audio/video.  

While the disparate analyses of structured data and unstructured data among financial institutions are comparatively mature, the analysis of the semi-structured data and perhaps more importantly the synthesis of these different data types is still largely unexercised. The failure to have a unified view of the information stored in structured, semi-structured, and unstructured data sets misses the opportunity to maximize the value of combined analyses and glean additional insights. To stay in compliance with increased regulatory requirements while overcoming the challenges posed by exponential data growth, investigators and compliance officers will need to utilize innovative analytics technologies.

Based upon traditional analytics techniques alone, the synthesis of the large volume of business records, communication data, and other third party data necessary in compliance reviews and investigations is impossible. The challenges commonly voiced by banks include:

  • Overwhelming Data Volume - Rate of new data creation makes volume of data too large for analysis
  • Unavailable Data - Data requested by regulators is not captured and unavailable
  • Disparate Data - Disparate data sets cannot be joined together for analysis
  • Wide Variety of Data - Variety of data types makes centralized archiving and analysis impossible
  • Disparate Systems - Multiple systems impede the ability to have a unified view

Many new technologies now exist which address the challenges described above. The advent of big data platforms such as Hadoop and Cassandra, enables systems to distribute processing across multiple servers which makes data processing performance highly scalable. Machine learning algorithms enable systems to quickly detect patterns and identify anomalies across multiple data points which may uncover noncompliance or potentially fraudulent activity buried deep within data. Natural language processing and entity extraction tools make analysis of data non-format-dependent by contextualizing information from unstructured data sets. Also, social data analytics tools make the collection, consolidation and analysis of social media data streamlined and relevant to corporate business intelligence gathering.

Below are a few examples of compliance and investigations objectives which benefit from advanced analytics techniques and technologies:

Trade Surveillance

Increase in computing power enables near real-time processing of trade data, chat data, emails and phone logs which may enable the identification of trades in breach of compliance sooner than previously possible. Also, the application of machine learning algorithms may detect patterns of deceptive market timing practices which were previously unknown. The review of social media data and instant messaging data may also provide additional insights to traders that may be colluding together for market manipulation.

Know-Your-Customer (KYC) / Enhanced Due Diligence

In addition to obtaining information directly related to the customer, enriching the customer information with social media data may provide additional insights regarding the customer’s questionable social networks. Also, marrying the customer’s ownership information from financial statement filings commonly available as unstructured data with watch-lists commonly available as structured data may uncover that a part owner of the customer company is under a watch-list.

Bank Secrecy Act / Anti-Money Laundering (BSA/AML)

The combination of big data platforms and use of multi-language entity extraction tools enables banks to gather information regarding the counterparties from a wide range of international information sources which may be used to verify the source of funds and any noncompliance of Office of Foreign Assets Control (OFAC) sanctions.

Fraud Investigations

The use of artificial intelligence natural language processing enables investigators to quickly sift through large volumes of documents and communication data to reduce false positives during investigations. Also, machine learning algorithms may detect patterns of internal and external fraud schemes, such as widespread creation of fake bank and credit card accounts inconsistent with the customer’s historical banking behaviors.

How A&M Can Help

Despite the development of artificial intelligence, machine learning, and other advanced analytics techniques, software systems alone cannot achieve meaningful analysis and results. The tools are only effective when implemented based upon the appropriate source data sets and well-designed reports which provide meaningful insights to support the compliance reviews and investigations. Below is a framework of necessary steps that A&M can walk your organization through to achieve optimal advanced analytics:

  • Workflow Efficiency - Establish efficient workflows to streamline data ingestion processes, minimize false positives, produce actionable findings and track resolution efforts
  • Technology Awareness - Conduct a market survey of the available technologies to determine the right tools for your needs and requirements
  • Source Data Quality Control - Assess the quality and completeness of source data sets to be ingested
  • Organization-Specific Analytics - Collaborate closely with the business divisions to develop analytics algorithms specific to your organization’s data and compliance policies 
  • Ongoing Calibration - Continuously reassess the analysis algorithms to fine-tune and re-affirm accuracy of results
  • Reporting - Design succinct and clear reports

The need for transparency and insight into the bank’s business operations is not optional but mandatory. Also, the use of advanced analytics is quickly becoming the norm for financial institutions to achieve such transparency. Financial institutions will need to quickly assess their compliance and investigations programs to determine how advanced analytics should be implemented to meet the ever-changing regulatory requirements.

For More Information

For more information, questions or to discuss the content in this newsletter, please contact Steven Lee, Managing Director with A&M’s Disputes and Investigations practice in New York.