In the Vault: AI and Banking Innovation with Tim Karpoff of Citi

Episode Transcript

This transcript has been condensed and edited for readability and clarity.

David Haber: Tim, you’ve worked in the Treasury Department, had different policy roles, and worked for a number of CEOs of some of the largest financial institutions in the world. What have been some of the throughlines? What are some of the differences in those different roles?

Tim Karpoff: If you step back and look at it from a 30,000 foot level, there’s the work that is the: What are we trying to do? What are we trying to achieve? What’s substantive? And then there’s the: How you get it done. And the how you get it done varies a little bit from organization to organization, depending on its culture. But I would say one throughline that comes — whether working in the federal government, whether it’s the Treasury Department, the White House, the CFTC, at Barclays, or at Citigroup is that all of them are very very large bureaucracies. And part of the skill in working these kinds of jobs is figuring out how to navigate those bureaucracies and how to execute. Sometimes the what you need to do is a lot easier than how you get it done.

A lot of people can come up with the answer; it’s figuring out how to get the battleship to turn in the direction it needs to go that I think has been a consistent differentiator across all three organizations. And as I look at people who are really good in Treasury, really good at Barclays, at Citi, it was that regardless of the challenges that came from the size and complexity of the organizations, they figured out how to get it done, regardless.

David: Maybe we can shift gears to your current seat. You serve as the chief strategy officer of Citigroup, obviously. What does that role entail at such a large organization? How do you think about your role?

Tim: You know, it’s funny, the role seems to change on a day-to-day basis, depending on what needs to get done. And I’ve been privileged to do this at two different institutions, with a different executive team at each, obviously. Here at Citi, the mandate has really been around thinking about, okay. What is it that is core to what Citi does? And what is it that’s nice to do but not core? You know, for us, it was about identifying very clearly: international consumer banking.

A lot of these franchises are very good franchises, but they don’t actually make sense in the Citi umbrella. Consumer banking is not something that tends to cross geographic borders particularly well. It’s tough to get economies of scale, operational synergies in different markets. And so, let’s stop doing that.

At the same time, also, there was a variety of work that needed to happen with respect to Citi’s back- and middle-office functions, real investments that were required and financial risk infrastructure that had to be made if Citi was actually going to be scalable. So Citi had a good franchise, but growth was really challenging.

So, when you took that together, I think what you ended up with was a real emphasis on getting clarity around: Okay, Citi is this global transaction bank that serves the largest and most international clients in the world, corporates, regulated financial institutions, the buy side, increasingly, the middle market.

And in consumer, we still want to be really scaled, but we wanted to focus around consumer lending and payments with our cards franchise, right, to a retail franchise, as well, that, is increasingly shifting from a branch-led model to one that is much more reliant on digital capabilities, and for growth there. Still very much want to be a scale player, but saying, “this is going to be our niche going forward.”

David: For the audience that isn’t as familiar with Citi’s history, maybe you can walk us through the history of Citi and how that firm built such a global footprint. I’d love to better understand that history.

Tim: I think the real interesting answer to that question is: we follow the clients. You know, when we started to think about what our niche was going to be, we figured out what it is that our clients needed and then made sure that what we provided in the foundations that we had around the world were consistent with their needs. So, if you look to the late 19th century, early 20th century, as you start to see the first inklings of cross-border commerce in the world, Citi establishes offices everywhere from China to Russia, etc. It has the first global branch banking network that emerges from the United States, and that becomes Citi’s calling card that is different from everybody else.

What you then see is the next evolution, post World War II in the ‘50s. As you have industrial conglomerates start manufacturing all over the world, start distributing all over the world, Citi follows them even further, So General Electric, General Motors, the types of companies that start saying, “Hey, I need services in all of these markets in Europe,” and, increasingly, in parts of Asia. Citi sets up branches to service them in those locations. And so that’s wave number two.

Wave number three really happens in the ‘90s, because you see that same move that manufacturers and industrial conglomerates made in the ‘50s start to really shift to financial institutions, and you see more and more of them needing custody services, needing FX services, rate hedging and the like in jurisdictions all over the world, because suddenly they have interests in Hong Kong, Singapore, every European market, and increasingly in Africa. And not only did we follow them in terms of new geographic locations, but new capabilities that came up in the jurisdictions we may have already served so that we had a greater breadth for them.

And then I think the newest wave, which is most interesting, is this middle market wave. Now I think what you’re seeing, increasingly, is a lot of midsize companies actually go cross-border much, much earlier in their life cycle.

David: In your mind, what role does a bank like Citi play in the financial ecosystem, separate and apart from regional and community banks?

Tim: You know, the United States has got a banking system that is made up of, I think, roughly 5,000 or 6,000 different depositories. And I think the U.S. financial system is incredibly dynamic and works because of the combination of different types of institutions who play different roles in that system. You know, the Citis, the Bank of Americas, the J.P.s, the Wells of the world are able to provide global connectivity, resilience, breadth and depth of offering that you’re not going to find in some of your smaller institutions.

But then at the same time, we probably are going to be a little bit less nimble, a little bit less alert to the local conditions that might result in a smaller institution being able to provide different pricing than we are to a client that has unique needs that arise from the community in which they operate.

So, when you look at, more broadly, the regional financial bank stress that we saw last year, you had, I think, a set of institutions that were providing their clients with a lot of the very necessary services, and providing it well, But you ended up with some stresses in the market from the rapid rate increases that we saw over the course of the preceding 12 months that put some real challenges to the balance sheets of those institutions.

And there was a need, I think, for the larger institutions to step in, to provide some backstop. And that was the role that we played. And I think we want to make sure that we learn the lesson so it doesn’t happen again, but also recognize there’s a reason that we have the broadly diversified banking system that we do in the United States. And we don’t want to lose that.

David: And if you put your policy hat back on, what are some of the ramifications of that crisis a year ago? And how do you see the landscape shifting?

Tim: Look, I think something has to give when it comes to consolidation. The cost of compliance — whether it happens to be with prudential requirements, capital liquidity, AML onboarding, whatever it happens to be — has just gone up and up and up. And obviously, they are particularly high for the largest institutions in the world, and they are calibrated to a somewhat lower level for regional and smaller banks, but they’re still quite high for those institutions. And you have to be at a significant size really to amortize out those costs.

 And so, to some extent, perversely, the costs of compliance are driving a lot of institutions towards consolidation when the policymakers don’t necessarily want consolidation. But the cost has to come down, or you’re going to have to take advantage of economies of scale and get bigger.

Marc Andrusko: And Tim, one of the cornerstones of traditional corporate strategy is either building or reinforcing competitive moats. As you think about crafting a strategy for Citi, how much of the focus is on moats versus something else, like remaining true to the core? What always stays central in your and your team’s mind?

Tim: I don’t think that they’re totally distinct, but obviously there are different flavors associated with them. What is core to us is areas where I generally like to think of it as we have a right to win. Right? And not only that we have a right to win today, we’re going to have a right to win tomorrow, and potentially to win more tomorrow. So if you think about what home base is for Citi, which is its services franchise: our global TTS business, which provides liquidity, payment, trade, effectively working capital solutions for the largest multinationals and for the middle market anywhere in the world. We have branches in 96 different countries. The moats there are extremely deep and quite wide. And not only does our capabilities in these geographies mean that when somebody goes out and says, “Hey, I need service in 40 different markets,” we’re always going to be on the list for an RFP — and increasingly, we’re winning more and more of them — but it also means that it’s really hard for somebody else to enter.

So, that core for us very much is consistent with that moat thesis that you laid out. And as we think about the future, we don’t think this is getting any easier, We would expect that while the world will increasingly lend itself towards capitalism that will cross borders — I think that those trends will inexorably continue on towards more and more activity happening in a distributed manner over the world — central banks, banking regulators are getting more and more focused on what happens in their geographies and making sure their financial systems are protected, which is going to make it harder and harder for people to replicate the network that Citi has.

David: A few years ago, you announced a transformation of the company’s risk controls and data infrastructure. Maybe you can walk us through some specific examples of some of the work that you’re doing there and how much of an emphasis in that work is technology.

Tim: Absolutely. In 2020, just for those who don’t know, Citi received a little bit of a love tap from its regulators in the form of two consent orders: one from the Federal Reserve and one from the OCC. And it was related to a number of concerns around the risk compliance and financial infrastructure of the company. There were a variety of other components as well, but those were the big pillars.

And if I step back and I think about where that came from. It really was a function, I think, of a bunch of historical underinvestment in technology infrastructure in the firm. But this would extend from everything related to, loan processing, which was distributed across 30-some odd different systems, to underinvestment in general ledger systems, which had proliferated around the world, which really needed to get back to at least a single strategic ledger that actually bound everything together for the firm. And then you could generalize it to data concerns more broadly, where the governance around data was underdeveloped, so that you didn’t have consistent data definitions, consistent data logic, which resulted in people using different inputs in different parts of the firm and being very difficult to reconcile.

So this firm started on a major journey, which was announced in 2020, and it’s the most important thing the company is doing. In the early days, I think a lot of it was focused on investments in people to make sure that we really understood the problems and had good plans to fix them. But increasingly, that investment has shifted overwhelmingly towards technology. And it’s investment in those things like strategic ledger and risk systems, but also more foundational components like data. So it’s going to take time. But technology, I think, is probably the foundational component. And I think, quarter to quarter, the benefits of that investment are starting to translate into dollars and cents.

Marc: Yeah, and that’s actually a perfect segue way into what I want us to talk about next, which is: How does Citi work with startups? How do you decide what gets built internally versus possibly working with a best-in-class company to solve a problem together? Walk us through that, in your capacity as Chief Strategy Officer.

Tim: To be perfectly honest, I think that it really is company-by-company and people-specific at this point. I think the area where we’re improving is around how we make decisions internally about where we should build and where we should partner.

If somebody comes with an idea and says “I want to do this, and this is the strategic need I’m filling,” we try to make sure that early on in that conversation, we get to that question, like, “Great. Does that mean we buy it? Does that mean we build it? Have you scanned the market for who can do X, Y, Z, etc. etc. Not perfect yet, but we’re getting a little bit better.

Marc: And out of curiosity, what would a good example be of the gold standard of a partnership that you felt really good about how it played out?

Tim: So one that I’m excited about is our partnership with Navan, formerly TripAction. We have engaged with them for provision of commercial card expense management, for our commercial banking segment. We see that a lot of companies regularly treat commercial card and expense management as their first financial product. You know, what is the way that they actually think about their budgeting, how they actually run on a day-to-day basis? And we had a gap.

And it was pretty clear that if we wanted to get deeper with that client segment, which was very core to our strategy, it’s really important to catch them at the front end — and we didn’t have the capability. In our discussions internally, we evaluated: Did we want to build this ourselves or were we better off working with somebody who is best-in-class in the market? And it became pretty clear that there were just a couple of players we potentially wanted to do business with. So we’re really excited about what we’re seeing there.

David: I want to shift gears a bit and imagine the future. Maybe you can paint a picture for what Citi looks like in 10 years. How might technology reshape the day-to-day workflows of different personas within an organization like Citibank?

Tim: It’s a really good question, but it is a super hard question as well. I think the most interesting thing is probably what role generative AI is going to play over that medium-term to longer-term. And what you see from all of our peers is everybody is investing in the same general capabilities to see what plays out, whether it happens to do with technology development, application development, whether it happens to do with fraud scoring, client engagement and the like. All the right things: anywhere large language models can play a role, predictive analytics can play a role. So put money out there, see what we can do. You know, let’s try a bunch of small things. Let’s try and move really quickly. It doesn’t work, then let’s turn it off and try something else, which I think is really healthy. We’re being far more agile in how we think about it.

But back to your question: longer-term, if you move away from the specific use cases and you think thematically, what does the work of a big financial institution look like over time? The role of banks and financial services really boils down to three big things: it’s payments; it’s capital formation — so, when I say capital, I don’t only mean equity, but lending debt, capital markets, equity investment, IPOs and the like; and risk transfer, so, individuals who potentially may be longer-risk wanting to lay it off, speculators who have capital they want to put to work and are willing to take a risk for a price. Everything that we do boils down to some version of: payments, capital formation, risk transfer in some way, shape, or form.

And if you think about all of those things, I can take out a significant amount of the work associated with every step in the value chain. So in capital formation, identifying and helping you figure out, where are my origination opportunities? How do I use the data they have around payment flows for my clients to understand when they may actually have a need? How do I get in front of them at the right time? How do I price differently? I think that’s likely to happen. I think you’ve already seen the move in trading, in markets. Equity markets at this point, particularly cash equity markets, are dominated by electronic players, which I think will only get more and more sophisticated over time. While you will still have major roles for human beings, it’s going to be more around the negotiation, the financing, the structuring side. And that role will be there, but increasingly it will be informed by data.

I think the regulatory change, though, that is going to matter more than anything else is on the treatment of funding and liquidity and, we’re all waiting for the resolution of Basel III endgame. I think what’s likely to come is greater regulation and focus around liquidity and funding management. It has become pretty clear, I think, that supervisors are increasingly worried that this legal fiction — this idea that you take a deposit, the deposit is money good, but then can be lent out to somebody else — is something that they are increasingly worried about. And you haven’t yet really seen it on the regulatory side, but you’ve definitely seen it on the supervisory side. And the result has been really putting a crimp on the amount of money that banks and financial institutions can make from maturity transformation. That’s where I think there’s some real unknowns over the next 10 years. But I don’t think it’s ever going back: with the changes in technology, the ability of people to move money super quickly from one institution to another, I think it has freaked out banking regulators all over the world. They’re like, “We have to figure out some other way to mitigate these risks.”

David: I imagine having to hold more capital against these different risk assets reduces the leverage you have on your balance sheet, but also reduces the velocity of money in the system, right? Does that effectively increase cost of capital for all of your counterparties? What are the downstream implications of Basel III endgame? I think you described the positive potential of reducing systemic risk, but I’m curious to better understand the negative implications of this too.

Tim: So it’s a really good question. And at its most simplistic, the higher the regulatory costs are in the regulated banking system, the higher the cost of financing — whether it’s debt or equity financing — from the regulated financial system is going to be. It has to work out that way in some way, shape, or form, which is why I also think it’s really important to make sure that you get the nature of the regulation right. And so, what’s the most efficient way to address that risk? And this is why I distinguish  that increased capital is not, at least in my mind, the most efficient way to address liquidity risk. I don’t know exactly what it is, at this point. Sometimes I think it could be prearranged liquidity facilities; who knows? I know that the supervisors are very, very focused on figuring this out and you got a bunch of smart people there who are working on it. But you have to get the right tool for the right risk. Otherwise, you’re going to raise the cost too much.

Marc: What do you think everyone is getting wrong when it comes to AI In financial services? Where do you disagree with the zeitgeist?

Tim: I think that likely the impact of AI is going to be super profound in a couple of specific places. So, technology development: my sense is that the cost of app development is going to see a step change and that really will be transformative, which is impressive. I see in other areas: fraud screening, I think we’ll continue to get better and operational losses from fraud will come down further and further as a result of the deployment of AI, machine learning, etc. I think some basic account servicing through AI power chats, which are far more sophisticated than the asynchronous chats that you see today, will lead to much better customer experiences, but probably not materially lower costs.

And then, I think, you get further out. There are other things that I think are unlikely to be quite as much of a step change. To give you an example, I’ve seen some capabilities that some non-U.S. banks have rolled out on things like mortgage origination, which are increasingly AI-enabled. Really, really interesting, but it’s not clear how much take-up you’re getting from customers. Maybe that’s a function of time. But it feels to me that there are certain components of the financial services ecosystem, particularly in the consumer space, where for the largest, thorniest transactions, people still want a human being and that they’re really dependent on other human beings’ willingness to accept a machine-provided solution. The limiting factor on the speed of adoption isn’t actually technology, it’s people. And those are the ones where I think probably may not be quite as quick as people expect.

Marc: As you discuss the implications of generative AI at the board level and with the very senior leadership of the firm, I’d imagine on the one hand, there’s a very real fiduciary responsibility to be focused on the near-term and the low-hanging fruit and what I would describe as the sort of “cost out” of the equation. But I’m curious: Are you also being pushed to really think about incremental revenue generating opportunities? Which of those voices tends to win out in the boardroom, at least right now?

Tim: I think that the near-term opportunities for generative AI probably show up first and foremost in the expense line. We have to be a little bit careful. So there’s a speed component that we have to be cognizant of — can’t move too fast, need to make sure we understand the potential customer impact. But I think your point about the tension between revenue generating opportunities and expense generating opportunities is the right question, in part because it really comes back to something that I think is more fundamental about banks and the value of big banks longer-term and how they generate increased value for their shareholders.

Because if you think about corporate finance 101, the value of the company is its current earnings above its cost of capital plus the present value of its future earnings over its cost of capital, with the second part of that equation being much, much more important to its current valuation. And how do you get to those future earnings? It really is from growth, right? There’s only so much you can do on the expense line. You’ve got to focus it. But if you really want to grow your shareholder value, your stock price, it requires having a clear path to: How am I going to make more revenue? How am I going to generate more earnings? And as such, you know, longer term AI has got to be deployed in a manner that’s going to generate that. We’re trying to remind people growth is where it’s at. Growth is where you’re going to find increased value for your shareholders. And, luckily, I think we’ve gotten that mindset firmly embedded here.

David: Tim, thank you so much for joining us today.

Tim: I’ve really enjoyed it, guys. I’ve learned so much from you guys and from your portfolio companies.

Marc: Absolutely. Thank you, Tim.