Is Automated Reconciliation Impossible in Complex Financial World?

4 mins

The 2008 global financial crisis and fairly new regulatory requirements like Basel III asked financial institutions (FIs) to cut operating costs and adopt a lean operations structure, pushing FIs to rethink on current reconciliation measures. Most of the banks are in the process of adopting an advanced reconciliation process that is fully compliant with applicable regulations in various regions. Previously, the reconciliation function across banks was only meant to prevent control failures that affect reputation. Banks had independent reconciliation teams for each line of business, and all these teams were reliant on manual workflows and spreadsheets, making the process labor-intensive and error-prone. Due to a combination of factors like increasing process complexity, the emergence of new asset types and structured deals, high transaction volumes, a large number of data sources, and ever-changing regulatory requirements, the process became more challenging and unsustainable. Result: compliance lapses and higher operating costs. The introduction of Basel III with its stringent requirements with regard to minimum capital, supervisory review process and market discipline created the urgency to rethink on technology advancements and implement new measures to cut costs and improve efficiency in the process.

The traditional manual reconciliation approach is no longer feasible in today’s scenario due to the ever-increasing data volumes. The same is the case with spreadsheets which are still used by a large number of organizations. Spreadsheets cannot manage the rapid handling of data as demanded by the regulations today. Spreadsheet reconciliation can consume up to four hours of an accountant’s time every day as he/she has to manually sum up the numbers and spend additional time in the mechanics of reconciliation. Manually reconciling thousands of items makes it challenging to meet deadlines and poses the risk of more human errors. Further, spreadsheets cannot provide real-time information with regard to transparency in financial reporting. In a nutshell, Excel-based reconciliation would lead to an overall lack of consistency in procedures and disjointed systems creating audit concerns.

The accurate aggregation, analysis and reporting of data cannot be done without process-focused solutions in today’s scenario. In order to develop software solutions that can effectively address the reconciliation problem, it is important to realize and accept the challenges including the segmented nature of the reconciliation process across the organization, identifying the true cost of reconciliation and its associated risks, data availability and data quality issues.

Software and Hosted Reconciliation Solutions

As part of the structural changes in the aftermath of the global financial crisis, many banks and their in-house teams started adopting software and hosted solutions to cut down costs, bring in operational efficiency and mitigate compliance risk. Through partially automating reconciliation processes, these software solutions could greatly reduce errors that came via manual processing. They could address matching of transactions more effectively with pre-set business matching rules and create cases around exceptions/breaks which need human intelligence to reconcile. Some of these solutions enabled real-time processing, which immensely helped reconcilers with shortened settlement cycles. In addition, they helped reconcilers with flexible and easy-to-use interfaces – key assistance as reconciliation moved beyond traditional accounting to new areas such as risk management, cash management and trading. These solutions increased efficiency to a large extent as compared to manual reconciliation.

With banks having entirely unrelated reconciliation areas such as GL, FOBO, Nostro and Vostro that have diverse data formats and sources, these solutions came as scattered packages, catering to only a particular need and with manual or semi-automated workflows. The rules-based nature these solutions came as another hurdle as the complexity of cases the volume of data to be reconciled kept increasing. Processing errors (reference id mismatch or missing, incorrect amount entry, duplicates, etc.) and system limitations result in data mismatch. Rules-based applications fail to tackle such mismatch and generate breaks, which are to be manually investigated and reconciled. In most cases, new rules are created to handle them. The process of updating current systems with granular rules every time a new scenario is detected is cumbersome and resource-intensive. Another problem with these solutions is the lack of a proper workflow to handle break or exceptions. A higher number of breaks would mean adding more to the already tedious work of break investigators. Naturally, the mounting work pile would impact the reconciliation quality, leading to compliance lapses. These solutions also lack when it comes to reliable regulatory or management reporting as they don’t have an audit trail to support.

AI/Machine Learning-based Solutions

These solutions came into play to address the drawbacks of rules-based solutions. Mixing and matching certain attributes of data across multiple files will help match records. It is not manually possible to figure out attribute-mix and create that many rules. AI/Machine Learning (ML) can automatically identify attribute-mix/pattern and create rules for matching. Exception handling, a key reconciliation process, is completely manual today. AI/ML-based solutions such as Tookitaki’s Reconciliation Suite can make a paradigm shift here as they are able to learn patterns from historical manual interventions and help detect breaks/exceptions automatically and resolve them in a faster manner. They can streamline and automate reconciliation processes across any line of business, dramatically enhance internal controls while enforcing standardization to improve the quality and accuracy of financial data. In addition, these solutions can help increase transparency in financial reporting.

How does machine learning help in reconciliation?

  • Connects to multiple data sources and bringing standardization in data requirement and quality
  • Automatic pattern detection and matching
  • Automatic break detection & resolution
  • Records all activities for audit purpose
  • Provides scalability and helps to streamline the reconciliation process

Given the novelty in the approach and the multitude of operational benefits, AI/ML-powered solutions will undoubtedly define the future reconciliation processes in the financial services sector. By employing these solutions financial services can achieve unmatched operational efficiency improvements while ensuring compliance with the toughest of the regulations.