Ensuring compliance through automation.

Watch now
Nordea Logo

April 15th traditionally marks Tax Day in the United States (July 15th in 2020) and global tax evasion is big business. 

Between 2008-2010, the U.S. Internal Revenue Service (IRS) identified a total tax gap of $458,000,000,000. That’s $458 Billion USD! Eighty-five percent (85%), or $387B, is due to underreporting and six percent (6%), $32B, is due to non-filing.

It’s no wonder that, in an effort to combat offshore tax evasion in 2010, the United States enacted the Foreign Account Tax Compliance Act (FATCA).

This federal law requires financial institutions outside of the United States to identify customers with a connection to the United States, including birth and current or prior residency, and report the identities and their assets to the United States Department of the Treasury. In 2014, further financial regulations saw the Common Reporting Standard extended to bank accounts on a global level. To date, more than 100 countries have agreed to these laws, mitigating tax evasion on a global scale.

Banks have no choice but to comply, risking being frozen out of global markets. Yet many financial institutions are struggling to scale conformity with constantly evolving global reporting and compliance laws.

Nordea, a globally oriented multinational bank with strong Nordic roots, understands that you cannot apply a linear solution to an exponential problem. Classical financial solutions of people and paper can no longer keep pace. Instead, Nordea applies cognitive automation to keep pace with the increasing regulations and consumer expectations. Cognitive automation is the application of artificial intelligence and machine learning to automate repetitive tasks that require increasing cognition beyond RPA (Robotic Process Automation), relying on high human involvement.

Nordea Bank
by the numbers

countries with Nordea presence
David Rasool

David Rasool Head of Automation

David Rassool is head of automation for finance and treasury reporting at Nordea Sweden. Having joined Nordea in 2010, prior roles at Nordea include head of business data science and head of data architecture. His background includes advanced analytics and enterprise information systems.

Nordea sees cognitive automation, using Teradata Vantage™, as the solution to exponential problems banks are experiencing.

Vantage’s built-in Machine Learning Engine allows Nordea to:

1.  Identify and automate key features of data that accurately predict reportable customers across their 11M customer base.
2. Discover and learn about new, reportable customers or fraud schemes.
3.  Enable a broader community within the bank to contribute by integrating into the current tax reporting ecosystem.


Cognitive automation for regulatory compliance

“Automation is about using data to get to the goal with as little human intervention as possible.”

David Rasool, head of automation

The heaviest burden in operating a compliance process, where bank employees check every reportable customer, is to confirm the reporting classification is correct—to the tune of many thousands of customers each day.

These bank employees also conduct spot-check on non-reportable customers for due diligence. These spot-checks result in several hundred customers out of millions and millions that need to be efficiently filtered. The equivalent to finding a needle in a haystack.

Operating manual processes at such scale is neither efficient, strategic, nor faultless.

Human error increases the probability for false positives and false negatives. Compound this human error with increasing customer expectations and increasing requirements on privacy, regulation, tax reporting, and financial crimes and it is hard to expect banks to keep track of, and maintain high accuracy against, all of these new requirements.

The reputational and operational risk for improperly classifying a customer to the IRS is significant.

A customer who is audited at the fault of their bank’s own human error is a costly consequence; leading to eroding customer trust, closure of accounts, and losses in total assets for the bank. Alternatively, the bank must identify potential fraud for reportable customers that are deliberately not reporting.

Nordea realizes its customer data, over 10TB, is an asset towards automating foreign tax reporting processes that accelerate and improve accuracy, minimize non-strategic personnel efforts, and maximize customer satisfaction related to financial regulations and compliance management.

Therefore, placing rules into analytical models in the form of automation is an easier approach.

manual process burden

“A significant portion of processes that we run today involve some kind of manual step in them. The business case for automation is if we can look at a process and remove that burden from the people working at Nordea, then we have an opportunity to do it quicker, to do it more accurately, and we free up time for our workers to be able to actually contribute in more cognitively challenging areas.”

With 11M customers in 20 countries, Nordea uses the Machine Learning Engine in Vantage to create a machine learning-based classification that predicts customers as reportable (the dependent variable).

The ML classification compliments a traditional rules-based classification which relies on approximately 70 unique features (independent variables) to classify and flag reportable customers. Typically, a quantity of number from 0 to 1 (positive one) and 0 to -1 (negative one) shows the importance of a variable.

Common examples of the features may include:   

Tax IDs and tax forms

Social security numbers and passports

FICO scores and credit forms

Place of birth

Current or prior country of residency

Volume and frequency of transactions

Transaction origination and destination

Deposits and withdrawals

This means that no data is left behind or considered garbage prior to ingestion and reading. 

The benefits are that the more data that may be ingested, the more accurate predictions and answers may be, as well as only discarding data when deemed invaluable.

Nordea uses native Vantage filtering and sorting techniques to remove the superfluous data and only make the valued data available for machine learning analysis. Nordea data scientists continue to do analysis, visualization, and operationalization all in one platform.

Cognitive automation reduces Nordea’s back-office processes costs by approximately $5.6M USD.

A cost-savings that may be reinvested into other areas of Nordea’s automation efforts.

“Before implementing machine learning, Nordea had to organize a lot of people and prepare for performing manual validation of the FATCA and CRS reports. It's quite a labor intensive task.

If we can prioritize the list of customers where we believe there may be an error in the classification then that makes a much lighter-weight back-office process to validate and know the report is correct. We also have a secondary benefit that we can actually scan the millions of customers that we have and make sure that we also haven't missed someone.”

Beyond tax evasion tactics, other areas where artificial intelligence for automation may prevent financial crimes include bust-out fraud, money laundering, and electronic fraud.

As financial institutions face diverse challenges when it comes to successfully combating fraud, money laundering, and cybercrime, Nordea realizes that the only way to combat the rise in sophisticated financial crimes is to leverage advanced analytics techniques and emerging practices.

Simply put, the fallout is too costly with hefty regulatory fines, loss in consumer trust, and stolen assets at stake.

Data as an asset

Nordea’s cognitive automation efforts provide a deeper understanding on where to focus data quality efforts – on data that matters most.

By running predictions on all of its customers, Nordea has become a bank built on data. With a complete picture of its customers, the broader community within the bank contributes to the overall tax reporting system by meeting increasing financial regulations, avoiding costly fines, and exceeding consumer expectations.

Turning its data into an asset, cognitive automation using Teradata Vantage provides Nordea with the critical business processes for fast, accurate, and cost-effective solutions to their toughest business challenges.

Other customer stories