Henrik Nilsson 00:08

Welcome to the Buzz, a Bank Automation News podcast. My name is Henrik Nilsson, and I’m the Senior Associate Editor of Bank Automation News. In this podcast, we will hear from Hugh Shannon, head of sales and customer success at Oaknorth. He discussed how AI is improving the commercial loan process.

Hugh Shannon 00:28

Maybe the starting point is around what’s what’s the challenge with the traditional approach to commercial lending, where AI and you know, technology is not used in a large extent. So just just to walk through some of the elements there. So traditional commercial lending, you know, relies upon credit analysis and underwriting and the approach that banks take is often a fundamental approach. So banks capture historical financials from a borrower they ingest those, build them into financial model, usually using a program such as Excel, you know, and use that to count, calculate cash flow balance sheet income statement and associated financial and credit metrics, they then use that on a pro forma basis and on a sensitized basis to stress the borrower and understand, is this borrower going to be able to repay the loan? Are they gonna be able to service the interest, principal payments, etc. And that’s obviously important for banks, right? Because they’re earning single digit percentage interest on a loan. But if the loan defaults, and there’s no security there, they could lose all of the balance of the loan. So another thing that banks do as well as they look at peer analysis, and really benchmark, the borrower’s performance against peers to understand like, how is it performed historically? And does that provide an indication of how they look into the future? Now, there’s other aspects as well to underwriting but if you think about those two, very manual processes, very human capital intensive, and there arises limitations because of that, you know, there’s there’s human error that’s involved. There’s inconsistency and approach for analyzing or was, for example, that are in a similar industry, but because you’re going from a fundamental square, one analysis, maybe having an inconsistent approach to the analysis. Banks may not be capturing underlying shifts in the industry that they’re lending into, you know, so for example, one bank that we work with describes the automotive industry. And the automotive parts suppliers are part of that in manufacturing, they can look at a borrow and say, Look, fundamentally, this borrower looks great. It looks good against peers, it’s got good cash flows. But they’re ignoring the fact that there’s a marked shift in the automotive industry away from internal combustion to EVs. Right? So if you’re not capturing that underlying shift, there’s a challenge as well. And I think a final piece of the challenge with the current approach is risk models that banks are using, based on historical data, historical experience, historical, economic and credit cycles. But if you think back just over the past two years, as we’ve gone through COVID, that was really an unprecedented economic event. And if you actually start to pull apart the the macro variables at play, the massive spike in unemployment that we saw, really in all markets globally, historically, we saw that spike in unemployment, you would see a real dip in GDP, you’d see a rise in personal bankruptcies, you’d see a rise in, you know, businesses going into insolvency, but we didn’t see that play out. Right. So you need a model that’s much more adaptable. So those are some of the challenges that the current approach, thinking about how can AI improve that approach, just generally, and in terms of what we’re doing at oaknorth? Right. So the our approach that I know of is to say, we still want to maintain elements of that fundamental credit analysis. But also we need to be able to bring in elements of war sort of automation to that to drive more consistent outcomes to drive ultimately credited out comes to the considering the the economic environment. So the way that we do that is we analyze each borrower within the context of its geography and a very granular industry, and then monitor it relative to that industry performance and peer performance, right? That enables banks to have a independent approach to risk and their provisioning and really challenged the approach that they’re taking today. Ai comes into work. And you know, we can talk a bit more about that. In that it allows us to build out a much more granular approach to that financial, where credit analysis allows us to have a better sense of how businesses will potentially perform under different forward looking economic environments,

Henrik Nilsson 05:54

You mentioned that it allows you to go into the granular level. What do you mean by that? And how does it work?

Hugh Shannon 06:02

Sure. And so again, if you think about a traditional approach to commercial lending, banks, at best, probably breaking down the universe into, you know, a dozen or maybe 20 different sectors, as they think about as they think about analyzing credits, right. So they would say, hospitality is one sector, right, which is covering everything from hotels, to full service restaurants, to fast food restaurants, right, that’s a manufacturing is another sector, which, you know, to come back to my previous example is going to capture auto parts manufacturing, but not differentiate between a manufacturer for internal combustion engine parts versus electric vehicle parts, our approach at North is to say, again, if we think about the business performance, and the credit performance of borrowers in those different sectors, across the economy, we need to be able to get more granular granular in that analysis, right. And if you come back to my first point around that fundamental human driven credit analysis, you know, the approach a human would say was, look, let’s start off with a general model. Let’s bring in the financials. But like, you know, I’m working in Excel here, right, but I’m going to start to make all these, you know, one time adjustments to capture the idiosyncrasies of an individual borrower, right? At oaknorth, we say, Well, look, that’s inherently an inefficient process. It’s an inconsistent process. It’s not scalable, right, but rather, we can, what we do is we say, look, let’s split the economy into a very granular view of industry by industry. So currently, we break that out into more than 270 industries. And for each of those, we create industry models, or forecast models that combine both bottom up industry forecasting, as well, as you know, the data set that we’re building out, you know, currently, we have over 400 billion of us commercial lending, going through our credit intelligence product, for example. Right. So what we then do at that granular industry level, is feed that data through our AI ml models, and use those models to identify for each individual sector, what our macro drivers that explain the performance of that sector, on a historical basis. And then we do an overlays on capsulate. You know, what are the current market dynamics, etc. So, you know, COVID and what have you, right? What that then means is that we can build out those forecast models driving from external data, but across those 270 industries, they’re each differentiated because they’re picking up different macro drivers and other drivers that that inform the performance. With that scenario, what we then do is we bring in borrower level data from the bank, combine it together with the you know, that industry forecast to come up with automated forecasts of potential borrower performance under different scenarios. Right. So, that means you know, whether a banks at the point of origination of your loans or monetary is existing loan book, or going through an annual review process, it can incorporate those industry level views on a forward looking basis to understand where are the borrowers within my portfolio that are potentially more at risk? Where are the borrowers that are less at risk, and then take a differentiated approach in terms of how they actually analyze, and ultimately spend time with those borrowers. And they can drive overall efficiency and better credit outcomes.

Henrik Nilsson 10:27

Can you talk about your efficiency ratio, as well?

Hugh Shannon 10:31

Yes, so OakNorth on a part of our software business at OakNorth, which we call OakNorth Credit Intelligence. However, at OakNorth, we actually started with our founders founding a bank, OakNorth Bank in the United Kingdom, so the bank is still still running. And it’s been a really successful story over the past seven years. So we have our bank, as our sister business to our software business, software business. And so the efficiency ratio, you know, it’s a typical, or common financial metric that banks look to measure as a measure of their performance. You know, in the US, we call it efficiency ratio in other markets, often called the cost to income ratio. And really, all it is, is a measure of the operating costs of running the business as the numerator, and the denominator is the revenue of the bank, right from lending and other fee income. Right. So the more that the lower that number is, the more efficient a bank is, right? So we’re the north bank, which is really the the first customer of ours North credit Intelligence Suite. And really, you know, we spun out with the bank itself. As I mentioned, the bank has been in operation for seven years. And it’s lent around $10 billion US equivalent. And it’s done that with a 26% efficiency ratio. So that means, you know, for every dollar or pound sterling of income or revenue that the bank achieves, it only costs us 26 cents or pence, to produce that dollar of revenue. Now, if you compare that to banks in the US, across the entire US banks have had an efficiency ratio in the low 60 percents and sort of bounce around between 61 and 63%. Over the past year, so there’s obviously a massive difference in efficiency that we’ve been able to achieve achieve without Northbank. Right. Yeah, and that’s Kapil, yeah. And that helps to drive, you know, a really strong return on assets return on equity and in a strong capital position for the bank, right. And the reason that we’ve been able to see that in the bank is through the application of our software, through the application of oaknorth credit intelligence, because by bringing in these sector level forecasts, by bringing in the automation, it really drives down the cost of underwriting these loans, initially, of monitoring the loans and going through an annual review process. You know, in for a bank, the human capital is a huge component of the costs, right? So if you can drive down, you know, the amount of human capital needed to service you can drive down costs and drive down the efficiency ratio. And, you know, beyond the bank, as we work with, you know, other partners, you know, predominantly in the US, that’s a key factor for them considering adopting, I’d love to credit intelligence is that ability to drive down their costs in their commercial lending portfolio?

Henrik Nilsson 14:06

Is this a space where you see a lot of a lot of development in the future? And is there a lot of competition in this space?

Hugh Shannon 14:14

Look, a great question. If you take a step back, there is continued development of or application of software into the banking space, whether you think about that as transformation or whether you think about it as disruption. You know, it’s almost two sides of the coin, right? If you think about the, the timeline and approach to that, and where has that played out in the banking space, and where do you see most of the activity taking place from fintechs either as disruptors or as partners? It’s almost done in terms of sort of bottom up prior to what I mean by that is, you can think about consumer banking, right and really starting with the advent of credit cards, and in the US, Capital One was The big a big driver of that the application of technology and enabling automated decisioning in terms of, you know, granting credit for credit cards, and, you know, the monitoring and managing of a credit card portfolio. That sense shifted up over time into small business banking, right, where you see the application of technology and automation there. And again, that’s one where banks are adopting the technology. But there’s also a lot of non bank lenders that are beginning to come in and disrupt, right, and you see the application of players even like Amazon, or coming in and offering lending products directly to merchants through their existing relationship. As we get up into the commercial lending space. There’s a huge focus by banks on technology. But I think it’s probably fair to say that banks haven’t seen the level of disruption in that space at the moment. But as you talk to bank executives, banks, boards, I think there is an acceptance that that disruption will come right. So banks need to make the investment. So where do where do we see in the commercial lending space technology coming into play? I mean, banks have their core system, which, you know, which, again, there’s a whole separate conversation around legacy cause versus sort of newer generation or cause based in the cloud, they have a spreading system. But a lot of more recent focus, like banks has been implementing workflow systems, or loan origination systems in the commercial lending space, which is really taking a lot of their manual processes or processes driven by email, or shared drive, etc. And bringing that into a cloud based workflow tool. So you know, there’s players, such as Encino that, you know, it’s built on top of Salesforce that are playing a lot in that space. What we would say is, we’re really complementary to that, because if you think about a workflow tool, often what it is doing is really as the template is taking, taking a human driven process and digitizing it, but when I talk to banks about it, often they say, Well, look, if you take our existing credit, they’re all credit analysis, which exists in an Excel spreadsheet and a Word document that then gets turned into a PDF, right? If all you’re doing is taking the data from that, and putting it into a web based platform, but it’s the same analysis and the same data going in like, Yes, that makes my workflow a bit easier. Yes, that means I can find that information potentially bit easier. But it doesn’t fundamentally transform the way that of going about my underwriting or by monitoring, right? And again, like at best, what those systems are doing is, I’m inputting information into it, and I’m taking information back out, right, it’s not really telling me anything that I didn’t already know, just potentially organizing it a little better. When you think about growth and credit intelligence is really, with all that information, how can we actually derive better insights from it? So again, if you think back to what you know, what I was explaining a few minutes ago, by bringing in your one application that is by bringing in that industry level outlook, combining that with the data that the bank has today? Can that give you a differentiated view or score across your portfolio to say, Look, these are the borrowers that are at higher risk. We relate to you to that ahead of time, go and spend time with them. Now, these are borrowers that are at lower risk, you should feel comfortable, just continuing to sort of keep your active monitoring going on, but you don’t need to go and spend as much time with them. So that’s, that’s the approach that we’re taking there. And again, that, you know, that’s resonated with banks as they work through, you know, modernization or transformation of their commercial lending process and the associated technology that comes with that.

Henrik Nilsson 19:23

So you have several partners that you’re working with PNC Capital One, Old National, just to name a few. Talk me through the process of integrating your software. You know, what’s the timeline and is it difficult?

Hugh Shannon 19:40

Yeah, good. Good question. So, our base product. We’ve consciously built that to be a standalone product, right a standalone platform. So the actual onboarding process or integration process can be very short, right? It is reliant on getting the relevant borrower data from the bank. But you know, we have the data spec, we work through the bank to understand the systems that needs to come out of your data and query those systems or their data, like get that data to us. And once we get that we can be running the analysis for them and giving them outputs in a matter of days or weeks, right is a very quick process, once we get the data that said, I say look at a challenge with getting customers on boarded. I say it’s a couple of things. Right? So one is just a question of prioritization. The bank’s decision making processes, the onboarding process in terms of you have third party vendor due diligence, going through procurement, can mean there’s a lot of hoops to get to jump through. You know, and again, from from me from a sales and customer success perspective, I’m very focused on that, which is like, how do we help to expedite that whole process and shorten the sales cycle, right, because again, like, it’s, it obviously benefits us. But it also benefits a bank, when you’re talking to, you know, leaders within the business, they’re excited about working with you. Like, they often you know, that they don’t want to go through a really lengthy process before they can get up and running, and people excited to want to get a kick to get going, you know, as soon as possible, obviously, understanding that we do Yeah, we do need to go through the requirements, required processes and checks, etc. Right. So I think that’s, that’s one challenge. I’d say the other piece is to actually get full value is not just a question of, okay, look, where we’ve stood up poked off credit intelligence, we’re now giving you these insights, you know, that’s it, right? You do need to integrate into bank workflows, you know, their processes. And this can take a longer period, you know, and potentially mean us that have deeper technical integration. So, for example, the signals that are software producers that can exist standalone in our open source application, or we can also feed them directly into existing bank systems or LLS is like Encino, so that out north insights show up on dashboards that the bankers or underwriters are already using. And, you know, so that’s an element of it. But then it’s also how then do their policies and processes change so that if they’re getting a Oaknorth, red signal or a green signal, like how does that actually change their processes to ultimately drive a more overall efficient approach, but also get better credit outcomes?

Henrik Nilsson 23:09

What’s next for Oaknorth and OakNorth Credit Intelligence?

Hugh Shannon 23:15

Look at that’s why they look, I mentioned earlier, you know, we’re now at a point where we have over 400 billion of commercial lending assets running through our software, which is like, a very big number, but it’s also only a very small fraction of the overall US lending market. And absolutely, look, as we continue to grow that data set, you know, that informs our approach and our models, right. If I think through, you know, what, you know, what’s our vision, or what’s our aim is really to continue to grow, penetrate into the US market, and really become a standard accepted way that banks think about, you know, their commercial lending. You know, much like the progress of other technologies in the commercial lending space, you know, the adoption of Spedding systems. More recently, the adoption of you know, LOS, you know, what’s the next step from that? And ultimately, what that will mean is as that technological transformation slash disruption moves up market, from small business banking into commercial lending, that the banks can really continue to compete and offer credit to commercial businesses to lower middle market businesses and you know, at its core, that’s why we’re here at North right is to enable the access to credit to funding for those businesses. that, you know, a core part of the overall economy, the growth engine, the employment engine, you know, whether that’s in us or other markets. So that’s, that’s how it all brings. That’s how it all comes together. I mean, that’s the end is really, you know, it’s a it’s not a zero sum game like it’s a net benefit to banks being able to operate more efficiently that they can turn that efficiency into more lending more growth for the banks, but that ultimately leads to more access to capital in the market and more economic growth for the economy as a whole.

Henrik Nilsson 25:37

You’ve been listening to the Buzz, a Bank Automation News podcast, please follow us on Twitter and LinkedIn. And as a reminder, you can rate this podcast on your platform of choice. Thank you for your time, and be sure to visit us at finainews.com for more automation news.