Follow the money: How analytics can aid the fight against financial crime

In 2020, the FinCEN file leak rocked the financial community and beyond. Published by Buzzfeed News, the files consist of 2,500 suspicious activity reports sent to the Financial Crimes Enforcement Network by financial institutions around the world between 1999 and 2017. The International Consortium of Investigative Journalists, which analysed the files, said that the leak revealed that financial institutions had “defied money laundering crackdowns by moving staggering sums of illicit cash for shadowy characters”.

In 2019, 12 of the world’s top 50 banks were fined for non-compliance with anti-money laundering and Know Your Customer regulations.

More recently, Capital One was fined $390m fine for its “egregious” response to money laundering.

In this context, banks and other financial institutions are under growing pressure to step up their response to financial crime. However, doing this requires them to make sense of a complex web of financial data, sifting out the illicit activity from the millions of legitimate transactions and contending with the sophisticated techniques criminals use to cover their tracks.

That’s where Quantexa comes in. Founded in 2016, the UK-based startup uses AI to help organisations understand the vast volumes of data they have accumulated. Through “Contextual Decision Intelligence”, Quantexa is aiding banks and other financial institutions in uncovering evidence of financial crime from within their data, identifying hidden risks and detecting criminal activity faster and more accurately.

Clients include HSBC, Standard Chartered Bank and OFX, with Danske Bank added to the list in December. Last year, Quantexa raised $64.7m in series C funding to support expansion in North America, Asia-Pacific and European territories and the deployment of its Contextual Decision Intelligence technology in new sectors.

Verdict Magazine heard from Quantexa CPO Alexon Bell on how the startup is helping banks follow the money behind illicit activities.

Over the past decade, organisations across numerous industries have woken up to the benefits of utilising the vast volumes of data they generate. But according to data from market research company Forrester, between 60 and 73% of data within organisations that could be analysed goes unused. Siloed data makes connecting the dots between pieces of information even more of a challenge.

For financial institutions, effectively analysing the reams of data they generate is not only valuable for improving customer experience, but also for tackling crime.

In the UK, banks are required comply with anti-money laundering (AML) laws and Know your Customer (KYC) laws to ensure criminals are not using their services to carry out illegal activity.

Money laundering is an issue that banks are under increasing regulatory pressure to address. According to the United Nations Office on Drugs and Crime, an estimated 2-5% of global GDP is laundered each year. Creating a more complete picture of the parties involved in money laundering activity and how they might be connected is key to stamping it out, with many turning to tech to aid in this.

“[Understanding data] is probably one of the most critical elements in a successful AML programme” says Bell. “If you look at the volumes at the big banks, they might process, you know, 30, 40, 50 million transactions a day. They've got tens if not hundreds of millions of customers spread around the world in many cases for the big universal banks.

“So you can't have people solve that problem. You need technology to do that. And that technology has to assist them in a number of ways. The most critical one, I suppose, is bringing and joining together that data across multiple internal systems in one geography and then across geographies.”

According to EY, over the last decade $26bn in fines has been imposed by global regulators for non-compliance with AML, KYC and Sanctions regulations.

To tackle the issue of money laundering, Bell explains that financial institutions utilise both internal and external data, but many rely on outdated or manual approaches that use only a small number of data points and that sophisticated criminal operations can easily evade.

However, for Quantexa, constructing a more complete picture of a transaction is key to determining whether it is cause for concern. The company’s platform uses advances in AI to provide a contextual view of data, asking in-depth questions about who is sending and receiving money to gain a better understanding of the networks that exist.

“We're trying to understand the context of this particular individual business, whatever it is,” says Bell. “And that's really important. If we can give a human being or a system 50, 100, 200 additional relevant data points, they can really make a different decision. What type of business are you trading with? And many of the things that I look at is the kind of hidden risk.”

This includes identifying “heavily leveraged” shell companies that hide a “whole wealth of crimes from human trafficking to corruption to drugs” says Bell.

“And they're typically funnelled through businesses, so getting a really detailed understanding of the businesses is becoming a critical facet in solving the AML problem.”

For example, through its partnership with HSBC, the Quantexa automatically screens more than 5.8 million trade finance transactions each year against over 50 different scenarios that indicate money laundering.

“That's really the fundamental difference between Quantexa and a legacy generation system,” says Bell. “The Quantexa platform is preassembling and understanding and asking that question of every transaction. Who did it come from? Where do we find them on other corporate information registries? Do they connect and have they sent transactions to any other existing customer within the bank? And we are then pre analysing those connections for risk in any dimension.”

As well as helping organisations keep pace with high volumes of data, data analytics can also help reduce the problem of false positives. According to Mckinsey, typically banks will use a customer’s profile and transaction history to generate risk ratings and flag suspicious behaviour. However, this method generates a very high level of false positives – up to 95% according to Bell.

This means that organisations must dedicate a considerable amount of time and resources to identifying genuine cases and can easily become overwhelmed with data. Quantexa’s anti-money laundering solutions, on the other hand, reduce the rate of false positives by 75%, says Bell.

“In some banks there can be many thousands of people going through the alerts and only finding five interesting alerts out of every 100. And that's because of a lack of information. What we have observed at Quantexa is that these older generation platforms look at these limited datasets, and then the human investigation process assembles the data that they need to make the right decision.

“For example, where has the money come from? What type of business is it? Are they connected to anything that might be risky, like a politically exposed person, or, heaven forbid, a sanctioned entity or a terrorist entity in the background? They do that research manually.”

He explained that this process can be improved by enriched data:

“we are looking at networks. Who are they connected to? We provide a couple of extra capabilities into that process and what we've seen is that when we do this, a legacy system may generate more than 7,000 alerts for something like trade-based money laundering or trade finance AML. A Quantexa platform will generate only 200. And out of those 200, 35% of them will get up to the level that would require a disclosure to the regulator. So 200, that’s about 70 alerts or so, which is completely turning the alert problem on its head. So previously 7000 false positives. Now we have 200 out of which 35% are very interesting and require disclosures.”

As well as money laundering, platforms like Quantexa’s can be used to uncover the financial trails left behind by other serious crimes, such as human trafficking.

Initiatives such as IBM’s Traffik Analysis Hub, which helps financial organisations disrupt human trafficking networks, have demonstrated the value of analytics in monitoring the flow of transactions and identifying anomalies that may point to criminal activity.

Just like money laundering, contextual decision intelligence can help uncover hidden links between individuals, which can lead to the discovery of covert criminal networks.

“We found a human trafficker who had been arrested and gone to jail and we were able to uncover the companies that he was using to traffic, unfortunately, women from Eastern Europe to London,” says Bell. “He had logistics companies and he had, ‘entertainment companies’ in London. We found all these businesses that he and his associates were connected to.”

Quantexa was able to follow the digital breadcrumbs, establishing that he was reopening the same type of businesses that he’d previously spent five years in jail for.

"This is the connected network context that we're generating of who are the directors and associates of particular businesses,” says Bell. 

However, evidence of human trafficking often spans multiple financial organisations. Bell says that Quantexa works with the likes of Dunn and Bradstreet, Dow Jones and Liberty to connect the dots between data held by different entities.

“And actually that can be expanded to everything, and so one of the core things that we talk about is that while people perpetrate the crimes, they are typically facilitated by businesses, and all the money laundering is done through businesses,” says Bell. “So that's why we partnered with Dun and Bradstreet, because their data gives us that insight, and it is Liberty and Dow Jones that provide us with the bad people. We use our software to make those connections to Dunn and Bradstreet, and then identify new businesses that these people have set up.”

With modern slavery and human trafficking generating approximately $150bn a year, banks have an important role to play in uncovering criminals and the financial footprint they leave behind, using technology unlock insights from the swathes of data they handle.

However, while analytics certainly has an important role to play in addressing such issues, Bell emphasised that it is crucial that it does not create issues of its own. The European Banking Authority said that the "use of AI in financial services will raise questions about whether it is socially beneficial [and] whether it creates or reinforces bias", meaning that banks must be acutely aware of algorithmic bias. As such, tools like Quantexa’s must be deployed with care.

“There are many programs within the bank that want to make sure that the AI is not biased by previous decisions that it's learning from. There are some certain high-risk jurisdictions around the world for this kind of corruption, such as Somalia or Sudan. What they don't want to do is have their AI and machine learning discriminate against Sudanese or Somalian people based on that piece of information. It's only the transactions going to and from Somalia that needs to be looked at with scrutiny. Just because you're a Somali national that's come over here means that you should not be tainted with that.”

Over the past year, another area that banks have had to contend with is criminal activity connected to Covid-19. It comes as no surprise that many fraudsters have capitalised on the pandemic as a means of carrying out scams. According to Experian, as of June 2020 fraud rates in the UK rose 33% during Covid-19 lockdown.

Bell said that the pandemic has not drastically altered Quantexa’s way of working:

“From a Quantexa perspective, we're always asking the same questions. Who is sending the money? We've seen some of the cases where they're sending 30,000,000 government contracts to a company that has only just been set up and has got no financials at all. For us we would flag that and say ‘look at this’. It makes no sense for you to send this amount of money to a company that has no trading history. Whereas if you look at some of the traditional value volume systems and some of the fraud applications, they just look at it and say, well, the governments made the purchase so it's fine. And that's not really good enough in the current environment. You need to ask the questions.”

Bell also explains that the pandemic has created further opportunities for corruption. According to Quantexa, federal and state agencies have been flooded with reports of potential fraud in many of the coronavirus relief programmes:

“As the aid packages start to be distributed, there's huge amounts of corruption that happens over the course of that. And again, the value and volume of these things are typically not uncovered by the types of businesses that are involved. It is the questions about where the money is going to, and if they're connected. And especially in the kind of more corrupt countries, to businesses that are linked to politicians. And you can't tell that by the name of the country or the value of the volume. You need to do a lot more research into that. And that's really where Quantexa is focusing its efforts.”

The pandemic has meant that for many legitimate businesses, their operations have been considerably disrupted or altered. For banks, this has meant that their legacy systems have struggled to make sense of the rapidly changing business landscape, leading to an even greater number of false positives. This is another area in which Quantexa’s products can be put to use:

“Outside of fraud, we've heard that many customers have been struggling with massive amounts of false positives from their legacy systems because essentially what's happened because legitimate trade has morphed into something completely different. Lots of online stuff, supply chains heavily disrupted. Businesses are behaving completely differently to how they would behave or have behave normally.

“One of the approaches that is consistent for legacy systems is they build a profile of activity over the previous 12 months and they use that as the basis to determine what normality is. So as you can see, Covid is a completely abnormal set of behaviours, so customers are firing left, right, and centre because they're just looking at those basic elements around the value and the volume and the type of transaction and country, which is just not giving the answer that the banks need at the moment. It's causing more problems than solving.”

“We're also seeing that outside of the AML world it's very relevant in things like credit risk,” Bell added. “And especially now in Covid times you can't assess a business based upon their balance sheet. You have to really assess a business on who they're trading with. So a bank could look at the business and say who’s their customer? And they could see transactions going in and out of their accounts. What Quantexa’s is providing is that visibility of who's making those payments. So if you know you make meals, you might be doing really well if you're selling them to supermarkets. But if the mainstay of your customers that you supply to are airline companies or hospitality, then you're in serious trouble and your credit rating should be degraded accordingly. And that's really the insights that we're providing. And this is what we mean about relevant and meaningful information. It's transforming the view of data from the customer to your customer’s customer. It's extending the visibility from just what's within the realm of the bank to that touch point that's outside of the bank.”

The financial services sector is thought to be one of the most data-intensive industries in the world. The likelihood that AML regulation will be tightened following the FinCEN leaks means deploying technological solutions to generate comprehensive insights from the information banks have hold is more important than ever.

Looking to the future, Bell believes that significant change is on the horizon:

“I think it's going to change dramatically out of sight from where it has been previously. I think that is through a combination of regulatory pressure and frustration from the banks themselves. We talked about 95% false positive rates. And that's just not sustainable. It's not sustainable to have very large teams that are very expensive wasting effort on this. The banks want to do something differently.”