Regulatory Reporting in the Digital Age

Introduction & Context

In an attempt to make our Banking and financial services Industry more stable and transparent, several regulations have been mandated with their own demanding reporting needs.  Some of the regulatory reporting has come a long way from being quarterly reports to monthly and even daily for some of the regulations.  Also, the enterprise-wide scope and nature of regulatory reporting cutting multiple lines of businesses have only added to the enormity of an already overwhelming activity for the regulatory reporting teams.  Though Banks and financial Institutions have a rich array of third-party reporting solutions available to them to be leveraged for their reporting needs, there are still a tremendous number of manual and tactical processes, which are performed as a prelude to and post generation of the reports.    It almost appears to be a herculean task to get the regulatory reports out on time with little or no scope for errors, to circumvent possibilities of a reputational and credibility loss for the Institutions.     

Potential USE cases for Digital Adoption in the Regulatory Reporting Process

1. Regulatory Interpretation – Interpretation of regulatory requirements and reporting guidelines is a very domain intensive and intellectually engaging task with financial Institutions spending a lot of time, effort and resources to ensure they are not being pulled up for noncompliance or MRAs (Matters requiring attention), post their regulatory submissions.  Though automation of regulatory interpretation features high on the business wish-list, it still has a long way to go before it becomes a reality.  With current advancements in Artificial Intelligence (AI) and Machine learning (ML), semi automation or accelerated regulatory interpretation is what financial institutions could practically aim for, in their pursuit to address this need.  

The recommended approach would be for banks and financial institutions to pilot with regulations specific domains of banking of finance, and extensively train the solution by using digital analytics techniques such as advanced Natural Language Processing (NLP).  For instance, one could start with the sub domain on mortgage banking, which could ensure that the training and orientation is practically manageable, and a level of predictable outcome could be achieved.  The approach if found successful and effective, can then be extended to other domains within banking and finance.   Boiling the ocean with a generic training of all banking and finance regulations is not only long drawn but is highly unlikely to result into a solution with a practical business utility. 

2. Regulatory Data exploration & analytics – Analyzing the data information needs of regulatory schedules and templates can be time-consuming and challenging process.  Most of these would be available in the banking relational data store, but there could be a few that may need exploration on a Data lake or upstream systems.  Banks currently do engage their business data analysts to do this, but this can be an intensely painful exercise, which could be more effective with some acceleration.

The recommendation here would be to leverage a Cognitive Data extraction tool that can read and analyze the metadata of data attributes in the upstream systems and recommend the likely mapping of the source element to the target fields of the Inbound Data model of the regulatory reporting solution.    Though this process cannot be fully automated at the outset, the machine learning algorithm over a period with feedback and validation will gain maturity in terms of accuracy by self-learning and can bring in a greater degree of acceleration to this process over a period. 

3. Data Preparation & Adjustments – Even with the advancements in IT and application technology, in most organizations we still have some of the non-systemized data inputs going into the final reports due to limitations of the business processes. Most of the non-automated tactical feeds are adjustments figures, which are sent to regulatory teams through spreadsheets.  Though over the years the controls established on these processes have matured, these non-automated feeds are still prevalent and are extremely time consuming to process. 

A Robotics Process Automation (RPA) Bot can be a very good digital lever that can perform these repetitive tasks, which are low on exceptions and variations and bring in efficiencies on speed and accuracy.  This will considerably fasten the process of data prep and adjustments and provision greater time to the regulatory team for analysis of the Reports.   RPA adoption will not only accelerate the process but also will be in greater levels of accuracy and traceability into it. 

4. Variance Analysis – Variance analysis is an important process of regulatory reporting to ensure that there are no major variances in reporting from one quarter to another and even if there are variations, the financial Institution is aware and can attribute it to a known business factor.   Even though the process of variance analysis by itself is not a very time consuming, it needs to be very performed diligently with little or no scope for errors.  

Hence, a recommended digital solution could be leveraging a Business Chatbot, which can be, trained for the key risk and regulatory data elements and plugged to the reporting results Datamart.  These can extensively slice and dice data across time horizons and across entities and geographies seamlessly with voice and text commands.  This will not only ensure an accelerated variance analysis but can also ensure that the variance is analyzed across multiple dimensions and granularity by business users from an analyst, all the way up to the senior management level.

5. Regulatory Reporting Infrastructure – Cost of compliance has been an ever-increasing problem for the regulatory institutions. Third party solutions come with their own licensing costs, and the cost of managing the associated infrastructure like servers and data marts for Integration.  This furthers the problems of regulatory compliance not only being demanding on the timelines but also increasing the overall cost of the implementation and ongoing regulatory compliance. 

The recommended approach is for financial institutions to evaluate and select regulatory reporting solutions, which are hosted on the cloud, along with the upstream data integration and analytical engines, which can cut down the cost of implementation and ongoing adherence to the regulatory compliance.  Depending on the sensitivity of the data, a public, private or hybrid cloud can used for hosting the regulatory reporting infrastructure. 

6. Management and Supervisory Reviews – Business users in regulatory reporting teams are required to create management summary reports, both for review by internal management team and for the review by the regulators.  This is currently a manually intensive and time-consuming process, which comes towards the end of an already exhausting reporting cycle further compounding the problem.  Most of these reports contains summary of the key variances, the reasons for the variances, dashboards of capital and exposure across various dimensions.  

A recommended remedy for management and supervisory review reports would be a cognitive Natural Language Generation (NLG) solution based on narrative science, which can read, analyze and narrate the variations in business metrics.   Users would need to create the standard narration templates as a one-time exercise for most likely or frequent sentences to summarize problems, the bot is then subsequently expected to pull out specific data points and matrices with variations and pick out the templatized sentences most appropriate for a given data movement.   Though this cannot be leveraged to churn out an entire management summary report, it is likely to accelerate the overall creation of such a report to a large extent.  

Conclusion

The Digital and cognitive Solutions recommended above for their potential leverage in the regulatory reporting process is to enable greater time for the Regulatory reporting team on business and regulatory analysis as against being inundated with the operational aspects.  Regulatory reporting is a highly skilled and responsible business function of a financial Institution and it is essential to ensure that there is adequate amount of digital automation that could be provisioned, to aid and assist the regulatory reporting teams to interpret, analyze and report the regulatory requirements in a timely and effective manner. 

Disclaimer: The views and opinions expressed in this article are authors own and do not represent the opinions of any entity/employer whatsoever with which they have been, now, or will be affiliated.

About the Authors

Deependra Kushwaha is the Head of US Basel III Reporting for a Global Bank.  He has more than 17 years of experience in Banking and Finance Services Industry working on a critical role with some of the biggest financial institutions across the globe (US, Europe, Asia pacific and India). He has led multiple credit risk and regulatory capital related engagements and proven track record of success in working cross functional teams including the risk, finance, regulatory reporting and IT functions.

Manoj Reddy is the Head of BFSI Risk & Treasury Practice at TATA Consultancy Services with an experience of more than 18 years in the areas of financial services, IT, and business consulting. Reddy has led several risk & regulatory consulting and implementation engagements for financial firms globally. He has provided both Regulatory and Strategic Business solution to his customers over the last decade primarily in CCAR, Basel, Liquidity Risk and Enterprise Risk management and is currently leading TCS efforts in North America with respect to providing Business & technology solutions to BFSI customers in the Treasury & Risk Management space. 

Introduction & Context

In an attempt to make our Banking and financial services Industry more stable and transparent, several regulations have been mandated with their own demanding reporting needs.  Some of the regulatory reporting has come a long way from being quarterly reports to monthly and even daily for some of the regulations.  Also, the enterprise-wide scope and nature of regulatory reporting cutting multiple lines of businesses have only added to the enormity of an already overwhelming activity for the regulatory reporting teams.  Though Banks and financial Institutions have a rich array of third-party reporting solutions available to them to be leveraged for their reporting needs, there are still a tremendous number of manual and tactical processes, which are performed as a prelude to and post generation of the reports.    It almost appears to be a herculean task to get the regulatory reports out on time with little or no scope for errors, to circumvent possibilities of a reputational and credibility loss for the Institutions.     

Potential USE cases for Digital Adoption in the Regulatory Reporting Process

1. Regulatory Interpretation – Interpretation of regulatory requirements and reporting guidelines is a very domain intensive and intellectually engaging task with financial Institutions spending a lot of time, effort and resources to ensure they are not being pulled up for noncompliance or MRAs (Matters requiring attention), post their regulatory submissions.  Though automation of regulatory interpretation features high on the business wish-list, it still has a long way to go before it becomes a reality.  With current advancements in Artificial Intelligence (AI) and Machine learning (ML), semi automation or accelerated regulatory interpretation is what financial institutions could practically aim for, in their pursuit to address this need.  

The recommended approach would be for banks and financial institutions to pilot with regulations specific domains of banking of finance, and extensively train the solution by using digital analytics techniques such as advanced Natural Language Processing (NLP).  For instance, one could start with the sub domain on mortgage banking, which could ensure that the training and orientation is practically manageable, and a level of predictable outcome could be achieved.  The approach if found successful and effective, can then be extended to other domains within banking and finance.   Boiling the ocean with a generic training of all banking and finance regulations is not only long drawn but is highly unlikely to result into a solution with a practical business utility. 

2. Regulatory Data exploration & analytics – Analyzing the data information needs of regulatory schedules and templates can be time-consuming and challenging process.  Most of these would be available in the banking relational data store, but there could be a few that may need exploration on a Data lake or upstream systems.  Banks currently do engage their business data analysts to do this, but this can be an intensely painful exercise, which could be more effective with some acceleration.

The recommendation here would be to leverage a Cognitive Data extraction tool that can read and analyze the metadata of data attributes in the upstream systems and recommend the likely mapping of the source element to the target fields of the Inbound Data model of the regulatory reporting solution.    Though this process cannot be fully automated at the outset, the machine learning algorithm over a period with feedback and validation will gain maturity in terms of accuracy by self-learning and can bring in a greater degree of acceleration to this process over a period. 

3. Data Preparation & Adjustments – Even with the advancements in IT and application technology, in most organizations we still have some of the non-systemized data inputs going into the final reports due to limitations of the business processes. Most of the non-automated tactical feeds are adjustments figures, which are sent to regulatory teams through spreadsheets.  Though over the years the controls established on these processes have matured, these non-automated feeds are still prevalent and are extremely time consuming to process. 

A Robotics Process Automation (RPA) Bot can be a very good digital lever that can perform these repetitive tasks, which are low on exceptions and variations and bring in efficiencies on speed and accuracy.  This will considerably fasten the process of data prep and adjustments and provision greater time to the regulatory team for analysis of the Reports.   RPA adoption will not only accelerate the process but also will be in greater levels of accuracy and traceability into it. 

4. Variance Analysis – Variance analysis is an important process of regulatory reporting to ensure that there are no major variances in reporting from one quarter to another and even if there are variations, the financial Institution is aware and can attribute it to a known business factor.   Even though the process of variance analysis by itself is not a very time consuming, it needs to be very performed diligently with little or no scope for errors.  

Hence, a recommended digital solution could be leveraging a Business Chatbot, which can be, trained for the key risk and regulatory data elements and plugged to the reporting results Datamart.  These can extensively slice and dice data across time horizons and across entities and geographies seamlessly with voice and text commands.  This will not only ensure an accelerated variance analysis but can also ensure that the variance is analyzed across multiple dimensions and granularity by business users from an analyst, all the way up to the senior management level.

5. Regulatory Reporting Infrastructure – Cost of compliance has been an ever-increasing problem for the regulatory institutions. Third party solutions come with their own licensing costs, and the cost of managing the associated infrastructure like servers and data marts for Integration.  This furthers the problems of regulatory compliance not only being demanding on the timelines but also increasing the overall cost of the implementation and ongoing regulatory compliance. 

The recommended approach is for financial institutions to evaluate and select regulatory reporting solutions, which are hosted on the cloud, along with the upstream data integration and analytical engines, which can cut down the cost of implementation and ongoing adherence to the regulatory compliance.  Depending on the sensitivity of the data, a public, private or hybrid cloud can used for hosting the regulatory reporting infrastructure. 

6. Management and Supervisory Reviews – Business users in regulatory reporting teams are required to create management summary reports, both for review by internal management team and for the review by the regulators.  This is currently a manually intensive and time-consuming process, which comes towards the end of an already exhausting reporting cycle further compounding the problem.  Most of these reports contains summary of the key variances, the reasons for the variances, dashboards of capital and exposure across various dimensions.  

A recommended remedy for management and supervisory review reports would be a cognitive Natural Language Generation (NLG) solution based on narrative science, which can read, analyze and narrate the variations in business metrics.   Users would need to create the standard narration templates as a one-time exercise for most likely or frequent sentences to summarize problems, the bot is then subsequently expected to pull out specific data points and matrices with variations and pick out the templatized sentences most appropriate for a given data movement.   Though this cannot be leveraged to churn out an entire management summary report, it is likely to accelerate the overall creation of such a report to a large extent.  

Conclusion

The Digital and cognitive Solutions recommended above for their potential leverage in the regulatory reporting process is to enable greater time for the Regulatory reporting team on business and regulatory analysis as against being inundated with the operational aspects.  Regulatory reporting is a highly skilled and responsible business function of a financial Institution and it is essential to ensure that there is adequate amount of digital automation that could be provisioned, to aid and assist the regulatory reporting teams to interpret, analyze and report the regulatory requirements in a timely and effective manner. 

Disclaimer: The views and opinions expressed in this article are authors own and do not represent the opinions of any entity/employer whatsoever with which they have been, now, or will be affiliated.

About the Authors

Deependra Kushwaha is the Head of US Basel III Reporting for a Global Bank.  He has more than 17 years of experience in Banking and Finance Services Industry working on a critical role with some of the biggest financial institutions across the globe (US, Europe, Asia pacific and India). He has led multiple credit risk and regulatory capital related engagements and proven track record of success in working cross functional teams including the risk, finance, regulatory reporting and IT functions.

Manoj Reddy is the Head of BFSI Risk & Treasury Practice at TATA Consultancy Services with an experience of more than 18 years in the areas of financial services, IT, and business consulting. Reddy has led several risk & regulatory consulting and implementation engagements for financial firms globally. He has provided both Regulatory and Strategic Business solution to his customers over the last decade primarily in CCAR, Basel, Liquidity Risk and Enterprise Risk management and is currently leading TCS efforts in North America with respect to providing Business & technology solutions to BFSI customers in the Treasury & Risk Management space.