Jan. 10 2023 11:22 AM

Banking and financial institutions are looking to automate compliance and auditing for their lending practices. But why the urgency?

    The cost of compliance has skyrocketed in recent years. According to Forbes, banking regulatory costs will rise from 4% to 10% of banking revenue by 2021. The Forbes article went on to say, “… like Moore’s law in the field of computing, there is a ‘Regulatory Law’ that the operational burden of controlling regulations will double every few years.”

    Bain & Co estimated that governance, risk and compliance (GRC) costs account for 15% to 20% of the total “run the bank” cost base. Other research cited in the Forbes article indicates that lenders annually spend over $100 billion on regulatory compliance and this cost continues to rise. Compliance with Dodd Frank alone has cost the industry $36 billion to date.

    Banks have been reducing their operational cost base for several years now by deploying process automation solutions. But many of these efficiency gains have been more than offset by the increasing costs and resources required to meet fast-expanding regulatory requirements and to settle fines.

    To cope, financial firms must accelerate the move from the human-centric processing model using Microsoft Office tools and paper to an intelligent automation framework.

    Technologies for Process Automation
    Intelligent process automation is all about the use of digital workers (or “bots”) that mimic human interactions with a computer. The bots can automate manual, repetitive and rule-based processes, freeing the staff to work on more complex tasks. This automation improves process speed and brings data to disparate systems across the enterprise.

    To deploy bots, companies purchase Robotic Process Automation (RPA) software. While banking RPA applications have been beneficial to replace the simple process tasks, they alone cannot handle more complex workflows. Financial companies are eager to expand the use of RPA to automate more complex tasks that currently require humans to open and read documents.

    Compliance adds a much higher level of complexity because of lending or onboarding processes where document certification, KYC (Know Your Customer) and CDD (Customer Due Diligence) functions are critical to staying within the regulatory guardrails. For example, a PDF arrives via email and a human must open it, validate the document’s authenticity and data fitness, and then enter selected data into the underwriting system. While a lot of interest is focused on loan underwriting, there are a number of post-close processes where dozens of different documents in various formats and the data within must be verified and validated.

    At this point, RPA leaves the safe and happy shire of data predictability and reliability and enters the Mordor of computer vision and text analysis. For bots to achieve an acceptable level of automation requires specialized software with the ability to read documents like a human.

    Anyone with experience in document capture or document imaging knows how hard it can be to achieve, let alone maintain, an acceptable rate of automated extraction of data from diverse and unpredictably formatted documents.

    This specialized software is called Intelligent Document Processing (IDP). It is indispensable to the lender’s automation technology stack. IDP uses advanced computer vision and machine learning algorithms to perform document classification, identification, validation, data extraction and data verification. IDP ensures that the data is purified before the bot passes the data onto the next stage of a lending compliance process.

    IDP also provides higher reliability to reduce the false positive exceptions that require human intervention. This can be a game changer as there are many possible scenarios where humans are still integral to the process: fraud detection, exceptions handling, dispute management, alerts management and credit approvals.

    Compliance Use Cases in Lending

    There are two specific processes found in lending operations that can benefit from more automation.

    Customer onboarding — Customer due diligence (CDD) and know your customer (KYC) are highly regulated functions. These processes are essential to help lenders assess the risk of new customers and to determine the products and relationship that the bank can develop with them.

    The most important step for KYC is identity and income verification, which requires the accurate validation of verification documents. In the USA, there is not one single, standard identity document. There are an infinite variety of documents submitted for income verification.

    Oftentimes, the customer submits a poor-quality image taken with a smartphone that is difficult for a computer to read.

    By shifting from largely manual processes to automated processes, lenders are looking to automate CDD and KYC processes to get better control of the costs and to improve the quality of the due diligence. But because of the volume and diversity of unstructured data and documents involved, end-to-end automation has been an aspiration until now.

    Mortgage remediation — There are several remediation scenarios that could benefit from automation. Here are a few examples.
    • Customer Complaint Handling
    • Post-Closing Loan Auditing
    • Loan File Certification

    Preparing and validating the data contained in documents is an important first step in the remediation process. RPA with IDP can be used to automate these processes.

    As you can see, there is a lot of work to be done to apply process automation for lending compliance. RPA and IDP software tools are capable today of a fairly high degree of automation for complex workflows and definitely worth the investment to reap the reward.

    Where Is the Puck Headed?

    Wayne Gretzky, the Canadian ice hockey legend, said the secret of his scoring prowess was he never skated to where the puck was, but to where it was going next. Let’s now lift our vision up from the processes of today and imagine where process automation is going next.

    One of IDP’s current challenges is the large dataset required to train the software how to read accurately. It can take a company weeks or months to collect enough documents to reach an acceptable exception rate level. If you are paying consultants by the hour, the training step can also become the most expensive phase.

    Leading-edge IDP software companies are continually applying data science and machine learning to dramatically reduce the time and number of documents for training.

    Another encouraging trend is the integration of RPA with best-of-breed IDP that complements and extends the basic OCR capabilities of some RPA software. Companies such as UiPath, Automation Anywhere and Blue Prism have formed partnerships with the leading IDP companies, especially those that focus on document automation for specific processes. Expect these alliances to produce even more tightly integrated solutions in the near future.

    Dan Lucarini is an AIIM Fellow and independent consultant to Parascript with over 25 years’ experience in capture and content management technologies. Learn more at www.linkedin.com/in/danlucarini.

    Most Read  

    This section does not contain Content.
    0