Episode 18: From Back Office to Frontline - Evolving Fraud Prevention in the Financial Sector

This episode of the Fraud Technology Podcast discusses the evolving approach to fraud detection and prevention in the financial sector. Traditionally viewed as a back-office function, fraud management has gained prominence due to rising consumer scams and increasing financial losses. The conversation highlights the shift towards proactive measures, drawing parallels with Anti-Money Laundering (AML) practices. Our guest, Nisan Bangiev shares insights from their experience in implementing fraud monitoring systems, emphasizing the importance of model validation, risk assessments, and leveraging advanced technologies like AI. Additionally, they discuss the persistent issue of check fraud, noting its surprising resurgence despite a decline in check usage.

Transcript

0:00

Hi, welcome back listeners. This is Ravi from Fraud Technology Podcast. And today we have another esteemed guest, Nisan Bangaev. He has over two decades of experience across multiple financial crime rules, especially around AML and fraud. And it's a pleasure to have him here and would love to know more about his journey and his insights into the fraud itself. Welcome Nisan to the podcast. Thanks, Ravi. Great to be here. Awesome. So we'd love to have a brief overview of what you've been working on. What's your journey been like? Sure, sure. So like you mentioned, I've spent a considerable amount of time in my career in the financial crime space, really sort of starting in 2008 post the financial crisis where you saw a lot of banks hit with large fines for email violations and Things of that nature. So, you know, I found starting in the AML transaction monitoring space, the whole investigative process to be a great way to sort of understand financial crime, sort of from a bank perspective. So was able to really sort of start with, like, understanding how correspondent banking works. And then from there, I moved on into working at a large institution at Bank of New York Mellon, where. Continue to learn about the correspondent banking space as well as all the different sort of products offered by banking York Mellon, including broker dealer treasury and a whole bunch of other sort of products that they offered. So, was able to do a lot of email on fraud investigations there. That's where I really sort of got my chops understanding how financial crime works ways that banks could be susceptible to fraud. And money laundering. And then from there, as I sort of built my knowledge, I moved over to Morgan Stanley, where I spent time sort of understanding sort of the broker dealer space flow of funds there, how potentially criminals could use other financial instruments, stocks, bonds, things of that nature to potentially layer funds. And then I continued my journey on to Oppenheimer. Similarly, It's a broker dealer. Similarly, looking at how things like penny stocks and other sort of instruments such as that could be used in potential money laundering or other sort of crimes and other sort of ways to manipulate the market and defraud customers and things of that nature. And then I moved over to. Bank Leumi, where I did private banking and commercial banking, and that was acquired by Valley Bank a few years ago. And then recently I was moved into the fraud space where, um, the last year or so I've been heavily involved in bringing the fraud program up to speed. Wonderful. I mean, quite a bit of experiences in different domains, even though all of it sounds like AML, but I'm sure given the different nature of business that each of the banks was going through and the roles that you're doing, I'm guessing that there was significant difference in the roles that you did. There's been a sort of a, uh, recently, you know, AML and fraud have, have start to come together and Valley Bank fraud and AML sit under one umbrella. So there's a lot of information sharing, a lot of utilizing data and resources, shared environment where we can sort of allow us to get sort of a sort of shared ability to combat financial crime. Wonderful. Actually, my question was around that as actually traditionally what I've seen is that AML is And fraud are traditionally two separate departments in most of the banks. And I see that you've done a lot of, almost 90% of your experience has been to, uh, a ML and recently into fraud. Yeah. So one, I wanted to understand the reasons of how and why you ended up. being moving into fraud and also wanted to understand your perspective of what you learn from AML. And I also see there is a synergy between these two. So I wanted to understand more about how has your journey from AML to fraud has been. I know you just mentioned Valley for both of them are same. I also would love to know more about that as well. Sure. So my journey, essentially, I see fraud as the new frontier. Traditionally, I think fraud has always been looked at as a back office operations. And I think that has changed recently. I think with the attention that has been paid to consumers, scams, things of that nature, really affecting day to day consumers. And I think more attention has been paid. To protecting individuals in the insert of the financial world. So I think that has probably precipitated a big push of looking at fraud. Similarly to AML, where it requires the same sort of rigor and attention, having that sort of the models in place, model validation, risk assessments, tuning of your rules, understanding. Your customers and all of those components really dovetail pretty well into fraud as well. And I think personally for me, as I've been sort of working in the email space, some of my skills around tuning and setting up rules and parameters around transaction monitoring. Those are parlayed into looking at fraud systems. So even at Leumi, I was tasked with implementing the fraud monitoring system because the CCO at the time thought, Hey, who better to implement a fraud system than the, the AML transaction monitoring guy. So he got me involved. I started to learn all the aspects of. Fraud monitoring, and then as we were acquired by Valley Bank, that need became even greater because, you know, Valley being a retail and commercial bank, the amount of fraud they see is significantly more than what you may would see. So it was an opportunity. For me to continue to apply sort of the skills I've learned, um, from a transaction monitoring and investigation standpoint into fraud. So that has sort of precipitated into taking this director of a fraud role. Wonderful. So you talked a little bit about how fraud has transitioned from being a back office operation to something more prominent, right? So can I understand a little bit more about what do you mean by that? Yeah. I'm assuming when you say back office, more customer service operations where a customer complains about fraud. So it is handled in the back office process, right? So can I understand more about what changed? Sure. So I think what has changed is banks are starting to pay a little more attention to the losses they're sustaining as a result of fraud. And well, you see all the numbers of how fraud has continually increased as far as, you know, losses that banks have sustained. The financial institutions in general, the industry in general has sustained a lot of losses due to fraud. And I think what that has happened is banks have started to put more resources into. Hey, how could we detect fraud? How could we get better detecting fraud? How could we introduce things like a fraud risk assessment? While it's not a requirement, like a risk assessment is on the AML side, regulators are starting to ask for that. Having the rigor around model risk And model validation at Valley Bank. We have a model validation team that validates all the models that we use on the fraud side. So all of the, all the systems, the models, the rules, they go through model validation and we are also investing in. In our resources, especially our investigative staff on the digital side, we're seeing an increase on the digital side of of fraud reported and losses sustained. So investigating reports of fraud. Getting better at detecting the fraud, investing into systems that are using the latest technology like AI and machine learning, validation of documents, you know, so the process, they're just getting better and better. And if we don't keep pace, we're going to continue to sustain heavy losses. Okay, so if I understand correctly, so what you're saying is instead of being reactive to, if let's say a fraud happens to a customer and then they report to or raise a ticket to the customer service and then there is an investigation, which typically is a back of the process and has always existed, what's happening now is there is a proactive approach to mitigate the losses that we are getting, instead of like budgeting some loss for it and saying, you know, that's the budget for customer service for them to reimburse the money to the customer if their investigation proves to be so. Keep that as a budget and move on and keep it as a back office budget. Now you're like proactively looking at how do I reduce that process? And also I'm assuming that's all the customer experience part to the fraud, uh, itself. Uh, that's why there's a productiveness as well. Yeah. Yeah. So fraud is very, you know, when customer goes through fraud, that could be a extremely negative customer experience. Correct. So we are starting to get attuned to that. And part of that is being able to respond to the customer. Tell them that, you know, someone is investigating, uh, manage their expectations because especially, you know, we're seeing a lot of check fraud. So, you know, you see a lot of issues where, let's say, at the bank of first deposit, there was an issue with the endorsement and the bank of first deposit was supposed to detect that check was fraudulently endorsed and deposited. So, but our customers saying, Hey, where's my money? And we're reaching out to the bank at first deposit. We're trying to get those funds back. So, and we're really sort of under their mercy. A lot of the banks, they, they take months, if not years to respond. A lot of times I'll deny the claim. So it's all about. Understanding that this can have a very negative impact on your customer experience And being able to position yourself in a way where you can manage the customer's expectations You can give them a little some comfort in that someone at the bank is prioritizing and handling their claim And that essentially there'll be light at the end of the tunnel for them. It's interesting that you talked about check fraud. I always assumed the check fraud, and I also even remember, I don't remember the exact statistics, but I remember that the number of checks. is going down significantly. But you mentioned that being a significant problem. Could you probably share a little more about what is the key challenge around check fraud itself? Yeah. So, I mean, for us, it's been the opposite. While I've heard reports recently that check fraud has gone down, it hasn't for us. And you know, and what we've been, you know, I think people have been talking about checks going away for, for decades. And, you know, as we introduce new, new, new payment rails, there's always the talk about, well, you know, checks are going to finally go away. We haven't seen that. And I think the big reason why is I think it's so easy for all of these transnational organizations, these criminal organizations to conduct this fraud that I don't think it's going to go anytime soon. And with the prevalence of dark web and information sharing on the dark web, and it's quite easy for. These criminals to steal checks out of the mails to intimidate or steal keys from post office workers, recruit post office workers, and then get these checks, sell the information online. And then that ends up, you know, turning into potentially a big problem for our customers. I think with a lot of the controls we put in place many times on the digital side, and maybe a lot of customers are a little bit more aware of, of potentially being able to be scammed by links and stuff. I don't know, but it just seems like check fraud has really made a return really post COVID and it, for us, it hasn't really gone away. Oh, wow. So two parts to this, right? One is the number of checks as a quantum being issued. I'm assuming it's going down still. The quantum of check fraud is constant. Are we saying that it's growing? Which means the percentage of frauds past overall checks is probably going crazy at the moment. Yeah. I mean, we advise our customers whenever we can, don't use checks, don't mail checks, secure your checkbooks. You know, we're trying to institute campaigns to continue to, to educate our customers because. You know, 80 percent of our losses are coming from check fraud and we've been trying to benchmark that's, you know, I would say many banks are in our situation where they're seeing, you know, still check check fraud is probably the, their biggest driver of losses. And, you know, at this point, it's all about educating our clients, getting better at detection, using the latest tools and technology, because it continues to be a problem. Wow. So is that also an incidence of age with the checks? Because I think The latest generation, the younger crowd would probably be attuned to using, would that mean the older generation, there's more higher incidence of older generation to use checks because that that's the more that they have been used to for like decades. And so for them to change that behavior is very hard. And so, and also it's easy to target them, uh, as well, basically what I'm getting to is check fraud and, uh. Older age fraud. This is a particular term for this. I'm assuming there's a huge overlap between these elder elder. Yeah, yeah, yeah. You'd be correct in saying that. And that is true. Usually we see a correlation with well, not just the ages is the susceptibility of the customer to being, you know, there's. You know, you also have younger customers that are also just not very savvy with technology and, uh, very much still dependent on their sort of their old old way of sort of writing checks. So, yeah, it's probably exactly what you mentioned. It's probably, you know, still our elderly customers that are doing this. Got it. Got it. Got it. I understand. I understand. And this is also typically physical, right? Because when, when I hear fraud, you hear scams and people sending you romance scams and like links and account takeovers and all the, all the fancy words that people talk about, right? But I'm assuming this is actual physical because the examples that you gave is stealing checks from mails. Right. Uh, stealing keys from post office workers or recruiting. These are traditional way of doing things, right? And 80 percent of your problem is on that side of things, which is very surprising, uh, to me, actually, because I've been hearing it's overwhelming the kind of information that's available. Yeah. So there is a, there's very much a physical element. And also I think with the prevalence of things like telegram, right, where, um, now it's much easier for criminals. To be able to buy and sell this information. They also can recruit, they recruit bank employees, they recruit postal workers. So the possibilities, it really opens up a lot of possibilities. You know, we do dark web monitoring for our customers, checks our customers, credit card information. We do scrub. The dark web to see if we could, there's any chatter about any latest broad schemes that we could get ahead of, but there's a whole underworld there and they seem to really like tech flow. Wow. I have never thought about this use case because I've always been targeted with people randomly reaching out and adding me to telegram groups, Whatsapp. Whatsapp is not prevalent in the US, but the rest of the world does use quite a bit of Whatsapp. So random people add you to a Whatsapp group and they say, you know, we have a small task for you. And I've never thought about it as recruiting people to aid some sort of fraud. I've never thought about that use case on which makes sense because I keep getting quite a bit by the way. So I'm sure banking employees are also being targeted. You're looking for a mule, Ravi. Yeah, they're looking for a mule. And yeah, so you see that a lot, right? They'll and that's like a primary vehicle. Like you even talking about, you know, we're talking about, um, you know, all these scams, right? Uh, you wonder, well, you know, it's not like these scammers are opening accounts in their own name. No. I mean, it kind of goes back to they get mules, whether willing or unwilling synthetic IDs. You know, we've seen a lot of that too. Basically we have, or just our customers information is just compromised and they they'll slap their face on our customer, like on a fake ID. And they'll open up an account. So we we've seen a lot of that type of mule type activity in the scans. And yeah, that's just a part of it, right? They just, they reach out, they try to find. People that are willing to make a few quick bucks just so that they can pass though that fraudulent money through their account. And what we've seen is really like we, we do monitor our first party fraud. We definitely see a skewing towards the younger, the younger people. So the older people are being defrauded. The younger people are doing the fraud, uh, the, the defrauding. So you see a lot of the mule activity and the scam activity going through our younger customers. Yeah. I've not heard about this particular insight that the susceptibility of younger people for being mules is higher. Interesting. I'm like now thinking about what would be the motivations of the younger people itself would be. I'm assuming it's to do with gig work and gig business. mentality that probably says I'm just doing something smaller. Yeah. Yeah. Think, think of yourself when you were 19 and, uh, you were a college student or, or just out there working and trying to making minimum wage. And, and then all of a sudden you say, Hey, here's a quick way to make 500. So yeah, there's, there's obviously also, you know, young people are also susceptible to being influenced and, you know, so it's not, you know, obviously, you know, The elderly are as well, but, um, young people also, they're, you know, they, they're also susceptible to, uh, that type of influence. Okay. Okay. I've never heard of that particular segment being also being susceptible, but when you're explaining it, it kind of sounds obvious that they would be targeted. Um, I guess scammers have to teach us how different people are susceptible, right? Unless they teach us, then probably we've never think about it unless we see start seeing holes out. Why did that whole happen? Uh, that's when, uh, it strikes probably to us. Cool. Wonderful. I mean, a lot of great insights around. The check fraud and the frauds that you're, uh, essentially seeing. So, some points that you touched upon a little bit about is. You said there is parallels between transaction monitoring that you've predominantly done, uh, in your prior career to fraud monitoring, um, itself. So we'd love to know a little bit. I'm assuming the, the way both the workflows work is probably very similar. I would love to know one, uh, what are the similarities that you found, which is, which is very helpful. And what are the key differences itself between a transaction monitoring versus a fraud monitoring? Sure. I mean, obviously the biggest key difference is, you know, you're doing transaction monitoring on a look back basis, whereas fraud, you're doing it on a, on a real time basis. Right. So let's say something like wire monitoring. You could have monitoring rules in place to pick up. Certain wires that could be higher degree susceptibility to some sort of scam or some sort of potential business email compromise fraud, but really, because the rules are fairly, it's still fairly limited. Like you're looking at, well, you know, was there a change of instruction? Is this a new beneficiary? Things that, so you can, it's similar to how you can monitor wires. On the AML side, but with AML, you're looking at a sort of wider range of potential activity. Whereas. We, you know, on the, on the fraud side, but although we're sort of looking to, to continue to, to add to our cross channel functionality and being able to monitor the session information, that's very important, right? When a customer logs in to their online account to monitor that session information, and then when they try to conduct a transaction, see, Hey, is there some risk indicator there that, that makes the, the wire, you know, You know, more susceptible to being fraud. So you kind of don't have those elements. In AML, but within AML, you have the ability to look at a historical look back, assess the wire, see if it's outside of the parameters of what's traditionally done to a slightly greater extent than on the fraud side. Um, this is sort of what I, what I found, you know, when you're doing check fraud monitoring, I mean, that's a completely different animal altogether because really there you're looking, you're doing image analytics. You're looking at, you know, looking at the signature. You're not doing the type of AML type monitoring where you're potentially looking at why the funds being paid to this individual. What is what is the what is the nature and purpose of this? Um, uh, of this payment, so on the on the check side, it's predominantly looking at the face of the check and saying, hey, is this altered? Is this a fictitious check? Is this potentially is it have had been somehow compromised and changed in any way? And you're looking at. Check, uh, you know, images in your the previous checks that were presented against the count to see, hey, does this look different? So the monitoring is definitely more real time, a little different, but ultimately, you know, you know, we're starting to see a lot of the parallels, especially in some of the more real time payments. Hmm, I understand. So you talked a little bit about. Validation of the models itself, right? And I'm assuming with fraud, because in AML, you are expected to respond to a regulator in very detail because it's been a very mature practice. In fraud, it's still, I mean, I'm sure regulators are asking, but it's not very clearly laid out regulation that you need to do SARS, you need to do this, you need to do, you need to do this audits, you need to do blah, blah, blah. Right. And also, I also see that the amount of data or the kind of data that is being used, for example, device analytics to how the person is using the mobile app, et cetera, which is not typically a data that is used in AML transaction monitoring, not typically. And in fraud, these are heavy usage of such data. Right. How do you do model validation in a scenario where one, you have a lot more data and two, you are moving also a lot faster and you're not required necessarily to be that thorough, for lack of a better word. So I wanted to understand how would you do model validation in these two scenarios. So I can't probably go into too many specifics with the model validation because we have a dedicated team that essentially, I mean, really the model validation is, look, you're right. The rigor around model validation on the fraud side, isn't the same as on the AML side. I know on the AML side, our regulators spend a lot of time looking at the model validation reports. Now on the, on the fraud side, I think they are starting to have that. expectation that there's some level of model validation, but a lot of the model validation is, uh, still, I mean, it's still very nascent and, uh, you know, it's still fairly basic to make sure, you know, um, the data is flowing in, they can replicate the rule. It's picking up what's intended, doing any sort of data validation. Is there any missing data? Is there any incomplete data? So a lot of those like similar elements, but not to the rigor and level of on the AML side. But I do think, and then the same thing with the fraud risk assessment, I do think there's going to continue to be that push to have fraud, um, have that same sort of level of rigor as, as AML. And like I mentioned, I think it's, it's the next frontier, which is why I was sort of excited to kind of jump to the fraud side because of this recent trend. And, and, and what I will say is. I do think now, um, certain regulators are using the AML exam to take a good look into your fraud, uh, exam as well. How are you filing SARS? What are your, you know, how are you detecting the, the fraud? How are you reporting it? How are you filing SARS? How are you, you know, so all those things. How are you keeping metrics? Like, how are you keeping track of the type of losses? Where it is at the bank, which customers is it affecting? So we're starting to see that expectation and that's where we're starting to, we are trying to invest very heavily in is getting, you know, Power BI dashboards to visualize our, our metrics being able to pull, you know, cause we're decentralized and a lot of banks are probably decentralized in their fraud monitoring while, you know, we might be doing some of the monitoring. We might have a wire team that's doing some of it. We might have the ACH So we're trying to get that visibility across the organization. And that's where really having a good data team helps. So we have a dedicated data scientist within financial crimes compliance. He works with us to help us get, pull that data, get it into a visually, you know, usable format so that we can then make informed decisions. We can, first of all, report up to our board, to our senior executives, the type of losses we're sustaining, the type of fraud we're seeing. And then also be able to drive our fraud program from a training perspective, from a model, a tuning perspective from, uh, just resource and technology perspective. So we're starting to see the benefits of basically. having that type of data analytics on the fraud side as well. Wonderful. One last question around AML. I mean, because you've done a lot of AML and come to fraud. So for me, that difference is what I'm trying to learn more from you. I have not seen any other profile who had such a long experience in AML moving into fraud. So I have a lot of questions around it. One is on the overall AML operations, right? Majority of your cost is on Certifying them, training them. Probably 60 percent of your cost is on people, right? The operations team and analysts. Obviously tech is 20 percent and the rest of it will be 20%, right? So in fraud, technique traditionally has been a more predominant technology operation, not a very human centric operation. I know you mentioned back office. Back office is the only example that I know where. Uh, for teams and also you're doing real time. You don't necessarily have a chance to do, uh, human intervention, right? So I want to understand and again, um, I asked this question because one is when you have models and when you're doing an email and you have a person behind, you can say that, okay, my threshold setting is done so and so, so that the reviewers could catch some of that as well. But in fraud, you don't have that safety net. In real time, at least when you're making decisions in real time. So how is the operation? So do you think we need more people involvement like in AML, or do you think it will predominantly be a tech oriented operation? I see it more gravitating towards it being a tech oriented, but I will say the heavy manpower is actually on the investigation side. And That sort of is almost like a sort of a feedback loop. We have more investigators than we have analysts reviewing the sort of the first level. And, you know, the, the, the, really sort of our human capital is predominantly, uh, you know, after the fraud occurs. So, because It's technology really is, we're so dependent on the technology, especially when it comes to check fraud, right? I mean, we do have certain human review of checks, but really, we're heavily, heavily reliant on our check fraud detection systems to, to basically do that image analytics and, and give us a pop an alert and have us reviewed. So, you know, whereas, you know, I've always felt on the AML side, it's almost like, um, fraud is yes or no. Was there fraud? Yes or no. Right. On the AML side, it's always, you're dealing in the gray area, right? You're painting in the gray area, whereas fraud is fairly black and white. So, you know, You could see how AML would be a lot more human, you know, resource intensive because you're constantly trying to potentially pick up a, some sort of pattern of activity that you might subjectively. Think is potentially suspicious, right? I mean, we do get some feedback from regulators when we've reported something that is, uh, is that, that is actual money laundering. But just think about the percentage of those to the number of stars we file. It's still, you know, a drop in the bucket. Whereas fraud is finite. You have a better sort of set of, of sort of data to work with. I think on the fraud side, the unfortunate thing is. It's very hard to replicate rules and build detection rules around your fraud cases. So that's why I'm always looking for better technology that could do that sort of detection. I do think that on the fraud side, technology is the key. Hmm. Okay. Wonderful. Again, uh, I, I, in that note, I uh, probably would ask you a final question, which is around what are the main, uh, things that you think in the front. You talked about technologies. What are the key things that you are excited about and looking to explore or trying to solve for the next, uh, one year? What are the key things that are on the, on, on the top of your mind? If I had my wishlist, I would love to have to continue to explore. Technology sending around validation of identification sort of, you know, for us, there's a getting good right there. Now they're using AI to create these fake documents, right? And not just ideas. I'm talking about business or incorporation documents, E I N letters. Bills, uh, we get fake utility bills. We are expending a lot of effort on, especially on our digital side to review all these manuals and it's very prone to human error. So would love to explore and continue to explore technology. That could assist us in sort of detecting all of these fake documents and, and be able to sort of use AI to start learning how to spot those type of customers that are opening accounts, potentially to defraud the bank. Right? So that's where my, my focus is. And I think potentially there's a lot of opportunity in that space for to introduce all these various platforms that could assist in that. Wonderful. Document verification, document forgery, and technologies to detect document forgery itself. Wonderful. Okay. Thank you so much. Listen, that was really insightful and really, really, I mean, a lot of things that you talked about around, uh, Younger people being suspect susceptible as well as some of the things around AML and fraud has been extremely insightful and I'm sure a lot of our listeners would love this insights. I know most of these are probably when you say it, it seems. Okay, that makes sense, right? But I am. Like, just like me, most of the people would probably never have thought about it in that particular way. And so thanks to guiding us into some of these key insights, which would help us do some of our day to day tasks easier as well. Thank you, Nisan, for your time, and we'd love to host you again soon. Great. Thank you, Ravi. I appreciate it and enjoyed my time speaking with you. Thank you, Nisan.