From chatbots to credit underwriting to stock market predictions, there is no shortage of use cases of machine learning in banking.

Despite the fact that risk management has always been at the top of banks’ agenda, many processes are still plagued with inefficiencies that are continuously draining bank resources. In this article, Andrey Koptelov discusses how banks can apply machine learning to streamline regulatory risk management and advance their fraud detection methods.

Streamlining regulatory change management

In the banking context, risk management and regulatory compliance are closely aligned. Banking employees have to manually monitor updates of thousands of regulatory documents, which involves visiting the websites of regulatory authorities and sifting through countless policy documents. This process is not only extremely resource-intensive but also error-prone and overall ineffective.

Machine learning can be used to automate a major part of the regulatory change management process. For example, Compliance.ai, a Silicon Valley startup founded in 2016, provides an ML-driven platform that helps banks keep up with regulatory changes. Using NLP and machine learning, the Compliance.ai platform automatically sources all relevant regulatory content from financial authorities, whitepapers, and news media. Importantly, when significant regulatory changes have taken place, the tool immediately alerts compliance officers.

Bank of Marin, a commercial bank that primarily operates in the Bay Area and has over $2 billion in assets, turned to compliance.ai to streamline its regulatory change management processes. Now Bank of Marin employees has access to all relevant regulatory content in one place, which significantly simplifies regulatory change management. As reported by compliance.ai, the Bank of Marin got returns on its investments in a short time by decreasing the number of resources for compliance.

Optimising stress testing

After the 2008 financial market crash, financial authorities substantially strengthened reporting requirements to ensure that banks can withstand significant economic downturns. This is why financial institutions need to routinely define and report their solvency to stay compliant, and banks with $50 billion or more in assets have to have their risk management teams conduct stress tests.

While undoubtedly important, these risk assessment processes often require hundreds of field experts and inordinate amounts of hours to complete. Moreover, finding relationships between various economic variables in relationship to banks’ performance is an increasingly complex and resource-intensive task.

Machine learning can help compliance experts to identify which exact combinations of values can cause risks, increasing reporting accuracy and decreasing the time it takes to complete these tests. This is exactly why Citi, one of the world's largest financial institutions that operate in 98 countries, joined forces with Symphony AyasdiAI to develop an ML-driven risk forecasting model.

Citi’s ML-driven stress testing solution

Prior to machine learning adoption, Citi had a hard time passing the annual Comprehensive Capital Analysis and Review (CCAR) conducted by the US Federal Reserve. CCAR requires a bank to submit its annual capital plans to the Fed to prove the bank's ability to deal with severe economic shock. Given the sheer number of economic variables at hand, Citi’s increasingly manual modelling approach took too long to complete, leaving business leads little time to understand the logic behind the models. As a result, Citi couldn't confidently defend its models in an annual submission to the Fed.

To create an accurate revenue forecast model, Ayasdi started with enriching the macroeconomic variables stipulated by the Federal Reserve. Then, Ayasdi used its proprietary machine-learning software to reveal how exactly these variables impact the revenues of each business unit. This allowed the company to define which unit-specific variables have the most impact on revenues. The detected variables were then used to create a custom ML-powered model that can accurately predict business units’ revenues under stressful economic conditions. As a result, Citi has managed to make this compliance process three times faster and less resource-consuming.

Enhancing fraud detection

Since the beginning of banking, fraud in its various shapes and forms has always been a persistent problem for many financial institutions. While many banks use sophisticated fraud detection systems, the rule-based nature of these solutions leads to a high probability of false positives. Importantly, fraudsters also keep innovating, which makes banks’ fraud detection systems grow obsolete.

Machine learning models, on the other hand, can learn and evolve together with the fraudsters. In very simple terms, a machine learning model can detect unfamiliar deviations from the normal pattern, notify the human employee about it, and, based on human feedback, learn if this kind of deviation is the new acceptable pattern or a case of fraud.

Just a few months ago, IBM launched a new generation mainframe that allows financial institutions to conduct fraud analysis of 100% of their transactions in real-time. To compare, around 10% of transactions could go through an AI-based fraud detection engine. In a nutshell, this means that banks that use the z16 mainframe can significantly reduce the number of false positives and increase customer satisfaction.

Conclusion

Risk management is a natural playground for machine learning. With the amount of data that banks have accumulated in the past decades, it’s only right to use the technology for its analysis. Machine learning models can drastically increase the accuracy of forecasts, decrease operational costs, and make risk management more effective overall.

Yet the use of such emerging technologies also continues to grow exponentially across a host of industries, according to McKinsey. However, financial services organisations are likely to be the biggest beneficiaries with the adoption of such technologies in terms of actual bottom-line business value.

A critical reason for this is that these companies already have access to extremely complex and large datasets, which are a requirement to create different predictive models from this type of technology. Such models are also hugely powerful. They can be applied across a wide variety of financial products and situations, helping these organisations to better understand everything from the possibility and probability of defaults on loans, to customer purchasing intention to detecting fraudulent transactions.

Even the BoE and FCA believe that ML technologies can “make financial services and markets more efficient, accessible and tailored to consumer needs”.

Despite such advantages, to really harness the power of AI and ML and put in place for projects that make an impact with everyday consumers, organisations within the sector do still face some very specific and sizable challenges.

Firstly, financial services is a highly regulated industry whereby personally identifiable information (PII) is required to be protected. Yet this does hinder collaboration as a huge amount of time is necessary to clean and compliance check this information. Due to this, a project’s timescale can really lengthen.

Secondly, despite collecting and managing such a wealth of data, financial organisations face some limitations with the information they have. Even when data has been prepared to develop AI solutions, the actual dataset itself may be under-representative and such limitations are the most cited major barriers that prevent finance organisations from utilising their data assets. up to three quarters (73%) of data actually goes unused for analytics by companies, according to Forrester.

In fact, it is actually quite common that the most valuable information for an organisation is hidden in an under-representative customer category. A biased dataset means the insights gleaned will also be biased. The knock-on effect of this can be quite damaging. It can lead to false assumptions about customer segmentation that leads to higher costs for the acquisition of customers (banks already spend over £279 each year on acquisition per bank account), inappropriate offers being made to customers, which ultimately makes them less likely to purchase, or worse.

Its biggest impact could come in the area of personalisation of financial products to everyday consumers.

Advanced approaches using AI and ML are helping to tackle these challenges. Synthetic data generation technologies have emerged as a highly credible method of protecting PII, while also eliminating the limitations that organisations are facing with their data. The technology, underpinned by AI and ML, constructs a new, entirely synthetic dataset from the original information, one that is highly statistically accurate (up to 95%) but crucially does not reveal individuals’ PII.

It could be transformative for the financial services industry, with organisations like JP Morgan already touting its potential.

Its biggest impact could come in the area of personalisation of financial products to everyday consumers. Undoubtedly such personalisation has improved from tactics such as the ‘Fresno Drop’ which saw over 60,000 pre-approved credit cards mass-mailed to consumers in the Californian city in 1958. However, AI and ML technologies are built specifically to extract insights from data which encapsulates consumers' preferences, interaction, behaviour, lifestyle details and interests. Not only this, but the technology is also developing to such a stage that it can spot and, in effect, ‘rebalance’ biased datasets.

When this approach is implemented accurately, research has found that synthetic data can give the same results as real data. Yet crucially the key benefits include full data privacy compliance and a major reduction in the time needed for product development and testing.

While the successful personalisation of offers, policies and pricing makes a large contribution to the revenues of the business, it also keeps customers happy as they aren’t being bombarded by irrelevant information. This matters hugely as McKinsey found that highly satisfied customers are two and a half times more likely to open new accounts and products with their existing bank.

Having access to such deep insights into all segments is not something that can be put off much longer as consumer behaviour, across generations, is undergoing radical changes already. Research from PwC found that half of younger consumers (those under 35) will open a primary bank account based on a trusted referral from friends or family, by contrast, however, one out of two consumers over 35 will choose a primary bank based on the local presence of a branch or ATM. Such generational differences need to be spotted quickly to keep a financial organisation in step with rapidly changing preferences.

While it is encouraging that more and more financial organisations are using AI and ML technologies, any approach to maximise data’s value must have a coherent strategy behind it. Used in the right way and with the right strategy in place, the opportunity from these technologies offers unlimited potential to financial services organisations.

Richard Harmon, Managing Director of Financial Services at Cloudera, discusses the importance of relevant machine learning models in today's age, and how the financial sector can prepare for future changes.

The past six months have been turbulent. Business disruptions and closures are happening at an unprecedented scale and impacting the economy in a profound way. In the financial services sector, S&P Global estimates that this year could quadruple UK bank credit losses. The economic uncertainty in the UK is heightened by Brexit, which will see the UK leave the European Union in 2021. In isolation, Brexit would be a monumentally disruptive event, but when this is conjoined with the COVID-19 crisis, we have a classic double shock wave. The duration of this pandemic is yet to be known, as is the likely future status of society and the global economy.  What the ‘new normal’ will be once the pandemic has been controlled is a key topic of discussion and analysis.

It’s not easy to predict the unpredictable 

In these circumstances, concerns arise about the accuracy of machine learning (ML) models, with questions flying around regarding the speed at which the UK and EU will recover relative to the rest of the world, and what financial institutions should do to address this. ML models have become essential tools for financial institutions, as the technology has the potential to improve financial outcomes for both businesses and consumers based on data. However, the majority of ML models in production today have been estimated using large volumes and deep histories of granular data. It will take some time for existing models to be re-estimated to adjust to the new reality we are finding ourselves in.

The most recent example of such complications and abnormalities, at a global scale, was the impact on risk and forecasting models during the 2008 financial crisis. Re-adjusting these models is by no means a simple task and there are a number of questions to be taken into consideration when trying to navigate this uncertainty.

ML models have become essential tools for financial institutions, as the technology has the potential to improve financial outcomes for both businesses and consumers based on data.

Firstly, it will need to be determined whether the current situation is a ‘structural change’ or a once in a hundred years ‘tail risk’ event. If the COVID-19 pandemic is considered a one-off tail risk event, then when the world recovers, the global economy, the markets, and businesses will operate in a similar environment to the pre-COVID-19 crisis. The ML challenge, in this case, is to avoid models from becoming biased due to the once-in-a-lifetime COVID-19 event. On the other hand, a ‘structural change’ represents the situation where the pandemic abates, and the world settles into a ‘new normal’ environment that is fundamentally different from the pre-COVID-19 world.  This requires institutions to develop entirely new ML models that require sufficient data to capture this new and evolving environment.

There isn’t one right answer that fits every business, but there are a few steps financial services institutions can take to help them navigate this scenario.

How to navigate uncertainty with accurate machine learning

• Modify existing models: This will be the starting point for all data science teams and can involve using the latest data elements to modify current models while also creating scenario-based projections adjusted for various levels of model bias.
• Stress testing: A practice that helps businesses gain a clearer understanding of their hidden vulnerabilities. Finance leaders should work alongside Chief Risk Officers to set up multiple dynamic stress testing scenarios.
• Industrialisation of ML: This is the ideal time to invest in a platform that supports the entire ML life cycle to build, validate, productionise, manage and monitor all their models across the entire enterprise. Data about customers is coming from different places and is stored in different places, whether that’s the public cloud or on-premise. For ML models to be effective, they need to be taking all these streams of data into account, with the ability to understand what the different data is saying quickly. This can only be done effectively in a unified enterprise data cloud platform.
• Prescriptive analytics: This approach is complementary to ML and uses simulations for more accurate decision-making for different scenarios, brought on by shocks or market changes. One common approach is Agent-Based Modeling (ABM), a bottom-up simulation for modelling of complex and adaptive systems. ABMs help business project thousands of future scenarios without having to depend upon the limitations of historical data.

[ymal]

When facing a crisis of unprecedented size such as this one, it’s time to look inwards and review the technology investments in place and whether crucial tools such as ML models are being deployed in the best way possible. Financial institutions should face this issue not as responding to a one-off crisis, but as a chance to implement a longer-term strategy that enables a set of expanded capabilities to help prepare them for the next crisis. Businesses that put in time and effort to re-evaluate their machine learning models now will be setting themselves up for success.

Of course, the rise of Human Factors Analysis Tools (HFAT) has forced financial services firms to push the envelope, but AI is gradually beginning to be integrated into the operations of firms across other industries. Perhaps, one sector which lags behind is insurance. However, according to Nikolas Kairinos, CEO and Founder of Fountech, attitudes are definitely shifting and in large part, this is due to the possibilities presented by AI toolsets.

Indeed, the venture capital community considers the insurance industry to be so ripe for disruption that Lemonade, a US InsurTech company, managed to raise $300 million in seed funding earlier this year. As an AI developer myself, I believe that the technology can drastically improve insurers at all levels, but only if industry leaders understand what AI actually offers and how to effectively integrate it into their organisations.

AI in InsurTech

The first, and arguably, most important part of this process, is having a sophisticated awareness of what AI in insurance actually means. For most firms, the benefits of AI actually come through robotic process automation (RPA); in other words, automating existing processes to save time and resources. For example, insurance AI exists which could remove the need for firms to manually classify documents, write contracts or process claims.

However, the most significant advantage that AI offers to insurance firms specifically stems from the way in which sophisticated algorithms can use vast datasets in order to predict and monitor risk. This would have many applications across the crucial functions of underwriting, pricing and risk management. Going further, the technology could even be used to prevent fraud by detecting tiny inconsistencies in either publicly available data or a client’s financial history.

However, AI doesn’t simply provide a competitive advantage for the forward-thinking firms who employ it, it also benefits policyholders who would enjoy cheaper premiums as a result of lower overheads and reductions in the amount of fraud.

Managing the transition

Still, some within the industry remain apprehensive about the impact of AI on either the employees or customer base of an insurance firm. The first thing is to say that many of these concerns, particularly around data security, are legitimate but it’s important that industry leaders do not see these apprehensions as an insurmountable obstacle. Integrating AI is not about saving resources for the sake of it but rather adopting new tools with the potential to improve the industry as a whole.

I’ve been developing software for professional services companies for years and based on what I have seen, I believe that successfully integrating AI into your services boils down to three things. Understanding the limitations of both the technology and your organisation, working with developers as much as budget and time constraints allow and being critical about where and why you’re integrating AI into your company’s operations.

At Fountech, we think it’s important for firms to understand what AI has to offer the insurance industry, and so we recently released a new white paper which explores how insurers might integrate AI into their business. Ultimately, with a proper understanding of AI’s strengths and limitations, industry leaders can begin adapting their firms to the rigours of the new data-driven landscape.

Towards a more intelligent future

As AI begins to play a central role in the functioning of insurance firms, it’s important that industry leaders remain invested in the technology’s potential to change insurance for the better. At root, this means having a sophisticated understanding of how AI can benefit your organisation but also remaining vigilant to any problems that might arise as a result.

Finally, as we move towards a more data-driven insurance industry, it’s essential that insurance firms begin playing a more active role in the development of new AI either through investment, active feedback, or by providing a breeding ground for new tools to be refined. Now is the time for insurance firms to begin playing a more active role in the development of the tools that are going to fundamentally reshape the industry over the next few years.

Salvatore LaScala is a Managing Director at Navigant Consulting, where he is Co-Lead of the Global Investigation and Compliance and Anti-money Laundering (AML) Practices. Mr. LaScala has over 20 years experience conducing AML and Sanctions compliance programme reviews, Risk Assessments, Monitorships and Remediations and regularly assists his financial services clients with Navigating regulatory or law enforcement actions. Mr. LaScala also applies his expertise by assisting clients with AML & Sanctions optimisation services that increases the breadth and scope of risk coverage while making the programme more efficient. He oversees Navigant’s AML Technology Team and has helped develop STAR, Navigant’s proprietary Case Management System and Rules engine regularly utilised for AML Look-Backs, Sanctions Look-Backs, CDD Remediations and other compliance and investigative projects. Additionally, Mr. LaScala provides his clients with outsourced Financial Investigation Unit (FIU) teams to both augment existing FIUs on a permanent basis or by providing FIU Surge protection services whereby the Navigant team is deployed to handle an increase in investigative or compliance activity pursuant to a compliance technology transformation or acquisition of another institution or large scale customer on-boarding.

Mr. LaScala began his career as an accountant, attorney and Special Agent with the IRS Criminal Investigations Division of the Treasury Department and thereafter spent over 20 years providing AML compliance and investigative services. He has been with Navigant since 2010.

This month, Finance Monthly had the pleasure to connect with Mr. LaScala and discuss AML in the US and the impact that AI, Machine Learning and Robotics Process Automation have had on the sector.

What drew to the AML field? What excites you about the sector you work within?

My background initially drew me in. As an accountant, attorney and former law enforcement officer, it all came together initially with a consulting job in 1997 with a Big 4 firm specialising in AML and Forensic Accounting. I enjoyed both but spent far more time in AML. I loved developing and dispositioning AML and Sanctions alerts and constantly found ways to make the process more comprehensive and efficient. Eventually I developed ways to make large scale AML remediations, including Look-Backs more efficient by building rules engines, false positive review platforms and custom case management systems. My perspectives as an accountant, attorney and former law enforcement officer helped make these technologies, auditable, regulatorily responsive and feature rich for investigators, respectively. These days I am still excited to be involved because I like working with clients, and because the regulations, financial institutions, and money launderers constantly change. It’s constant learning, which works for me - otherwise I’d be bored.

What is the current state of AML in the US?

This is a very important time for AML compliance - regulators, examiners and law enforcement now know more about AML programmes, compliance technology and payment platforms than ever before, and as such, the stakes for financial institutions regarding compliance become increasingly higher. Financial institutions are quickly adapting and upgrading their technology and overall programmes to maintain compliance and prevent and detect money laundering, terrorist financing and fraud. The ‘bad guys’ however, seem to have far more payment methods and venues at their disposal to commit crimes than ever before in history.

What are some of the key challenges you face on a daily basis and how do you overcome them?

The key challenges include finding innovative and cost effective ways to serve our clients, who are often faced with fines and expensive remediations. Providing the right breadth and depth of services to them in a cost effective way is critical. We also work for financial institutions of all different shapes and sizes, some have been through enforcement actions two or more times and are in a position to better plan their way through those actions with a great appreciation of the effort it takes. Others have either not been through too many regulatory or law enforcement actions, or are unable to communicate to a home office in a foreign country the gravity of a US regulator or law enforcement action, and don’t get the financial support they need to get through it. The challenge in both instances still becomes handling ongoing work or “business as usual work” (BAU) along with regulatory action or some compliance technology transformation. Without consultants helping, there are just not enough hours in the day. Regardless of a financial institution’s capacity to respond to a regulatory action, it’s often best if we get in there early and get them off to a timely start so they don’t also fall behind on BAU, or react to regulators too slowly, which can lead to additional issues.

What are the current AML issues and solutions affecting American businesses?

AML is constantly undergoing transformations. Some of these are based on new and emerging AML and Fraud schemes that the industry has to respond to, other transformations are due to new regulations, such as NYSDFS Part 504 regulations, which add additional layers of accountability on AML programme owners. Still, other transformations are the result of enhancing the regulations and the technology behind it because every time we close a door on money launderers and fraudsters, they both seek out institutions without robust compliance and find new venues through which to launder money. The US and several other markets are attractive to money launderers, fraudsters and terrorists because the financial services industry is vast and because these markets are segmented. This means that some players in capital markets or money service business spaces are very technologically savvy with respect to compliance, while other smaller players in the same segment are not. In fact, we often see challenges where the larger and more sophisticated financial institutions de-market or close customer/client accounts which later pop up at smaller or less sophisticated financial institutions.

How has the introduction of Artificial Intelligence, Machine Learning and Robotics Process Automation impacted compliance and investigative solutions?

Navigant is highly focused on applying Artificial Intelligence (AI), a form of Machine Learning (ML) and Robotics Process Automation (RPA) to our clients in many different areas, including AML and Sanctions. For AML example, we believe that AI/ ML can help existing AML Transaction Monitoring Systems deliver enhanced detection scenario parameters by grouping behavioural patterning to cover more risk and produce fewer false positives. Concurrently, we are applying RPA to the expedite portions of the dispositions of such alerts by removing mundane rote tasks from the analysts purview so that he/she is spending more time on considering the facts, CDD, news and current transactions to determine whether the transaction should be filed on, and less time hunting for data and writing the disposition. Specifically, AI/ML, which helps increase coverage and reduce false positives, and RPA which provides the Investigator more time to analyse the actionable items, are remarkably powerful together.That said, there is a fair amount of work to do, and in the beginning, we need to focus AI/ML only on matters for which the data feed is clean and comprehensive and apply it in a way that is transparent and can readily be described to regulators, examiners and internal audit. The AI/ML revolution won’t survive if the providers that developed it and the financial institutions that use it are not completely transparent. Moreover, even RPA will be better received if it is introduced in stages and when implementations are accompanied by statistically valid data showing that it is more accurate, and saves time such that the ultimate work product contains more thoughtful analyses and is generating comprehensive filings useful to law enforcement.

Far from taking human jobs in future, Artificial Intelligence (AI) and Machine Learning (ML) technologies are going to free up finance professionals from spending too much time on monotonous tasks and allow them to focus on more strategic tasks of higher value to the business. Does this mean that finance roles will mostly be driven by robots? Below Tim Wakeford, VP of financials product strategy EMEA at Workday, discusses with Finance Monthly.

A recent EY study revealed that the majority (65%) of finance leaders said that having standardised and automated processes—with agility and quality built into those processes—was a significant priority when it came to investing in emerging AI and other technologies. And, following on from this, 67% of finance leaders said that improving the relationship between finance and the wider business strategy was also a key priority.

Again, this is an area where automation and AI technologies are helping free up time for finance to spend more time working with other teams within the business. This enables them to figure out where to go next as opposed to looking backwards and dealing with unproductive and time-consuming legacy finance systems.

Freeing up talent to focus on high-value tasks

Freeing people up from repetitive jobs to enable them to focus on high-value tasks is the opposite of the oft-cited “robots putting people out of work” narrative.

Indeed, automation is a huge opportunity to reduce the unnecessary burden and pressure that’s put on finance professionals, particularly around traditional tasks such as transaction processing, and audit and compliance.

The adoption of AI applications within finance enables forward-thinking executives to move info far more strategic business advisory roles. This means that they can focus less on number crunching and more on financial analytics and forecasting, strategic risk and resilience, and compliance and control. This shift to data-driven financial management delivers a much wider benefit across the business.

The Rise of the robots: AI in finance

Computer systems performing tasks that previously required human intelligence is the definition of AI, with experts viewing AI and automation as viable solutions to efficiently deal with compliance and risk challenges across different sectors.

With the rise of the ‘big data’ era comes a parallel growth in the need to analyse data for financial executives to be able to properly manage compliance and risk.

This is another reason why finance teams cannot ignore the opportunities that embracing AI technologies offers them. It allows them to process vast amounts of data faster and easier than large teams of humans can.

Individuals are then able to make better strategic decisions based on the information that AI is able to rapidly extract from what were previously time-consuming and repetitive and monotonous tasks such as transaction processing.

Jobs least likely to go to robots

Forward-thinking and highly-skilled financial executives are happily embracing AI, as they see the clear opportunity it presents to play a more valuable and strategic role within their organisation.

“The challenge for managers will be to identify where automation could transform their organisations, and then figure out where to unlock value, given the cost of replacing human labour with machines and the complexity of adapting business processes to a changed workplace.” This is how writers James Manyika, Michael Chui and Mehdi Miremadi so fittingly describe the process in their book These Are the Jobs Least Likely to Go to Robots.

“Most benefits may come not from reducing labour costs but from raising productivity through fewer errors, higher output, and improved quality, safety, and speed.”

AI and automation in finance has to be about reducing repetitive manual tasks and raising overall productivity through data-driven business strategy. The bottom line is this: any technology that can reduce manual input and the associated human errors for transaction processing and governance, risk, and control (GRC) will free up finance professionals for more strategic work.

Any organisation’s most important asset is its people. And finding out which emergent AI technologies and applications are the best for a business and its people is going to be key for the future of finance.

Giving skilled finance staff the autonomy and opportunity to move into far more strategic data interpretation roles and letting the machines take on the grunt work is a necessary shift in the finance function.

As well as automating a large part of the finance function, AI technology will also help skilled finance executives to make a far more sophisticated analysis of complex data sets and to provide genuinely valuable insight to drive the business forward.

There is very little doubt that the future of finance will be one that embraces technological innovations to improve effectiveness, increase efficiency, and enhance insight.