Transcription:
Penny Crosman (00:03):
Welcome to the American Banker Podcast. I'm Penny Crosman. Risk management in banking has never felt more important in the wake of the failures of Silvergate Capital, Silicon Valley Bank and Signature Bank. An estimated $1.7 trillion in unrealized losses still sits on U.S. banks' balance sheets today, according to analysis conducted by researchers at New York University. What should banks be thinking and doing about this? We're here today with Sumeet Chabria, founder and CEO of Advisory Firm ThoughtLinks. He was formerly Chief Operating Officer of Global Technology and Operations and Head of Global Business Services at Bank of America. Welcome Sumeet.
Sumeet Chabria (00:41):
Thank you, Penny. I'm very glad to join you here.
Penny Crosman (00:44):
Glad to have you. So some analysis of call report data has shown that one of the problems that drove Silvergate, Silicon Valley Bank and Signature Bank out of business is shared by many banks that a lot of banks have underwater bond portfolios and would be in trouble if they were forced to sell off a lot of their holdings. How do you think this problem became so widespread?
Sumeet Chabria (01:09):
We've had, Penny, one of the fastest rises in interest rates in recent history to curb inflation just in 2022 and six months. The Fed funds rate rose by more than two percentage points, after almost 15 years of mostly near zero rates coming out of the last financial crisis. Now the Fed funds rate stands at 4.75%. So as the economy grew, helped by stimulus recently, banks built up excess deposits that they invested in long-dated securities and much of it in treasuries and treasuries are almost risk free. They're as safe an investment as you can have backed by the full faith and credit of the United States. But prices of long-dated treasuries, just as any long-dated bonds, are sensitive to changes in interest rates. So as interest rates went up so quickly these bonds lost value and banks were carrying unaligned losses sizeable, to the point you made, which they do not have to realize if they hold these securities to maturity. But for some banks, as they start losing deposits for various reasons, got into a liquidity crunch and were forced to sell some of these bonds. And once that happened with a few banks, it creates a contagion risk, a more systemic risk with more banks risk losing deposits as consumers lose confidence. So the Fed, the FDIC had to step in to reassure consumers, businesses, the deposit and provide a backdoor. So that's how it's become more widespread. Penny.
Penny Crosman (02:43):
And is there any kind of risk management technology that could have caught some of this earlier or could have maybe provided some alerts or some kind of warning that Silvergate and Silicon Valley Bank and Signature maybe didn't have?
Sumeet Chabria (03:02):
Yeah, there's a strong case to be made first that all the banks that failed, and this could be argued, had some unique challenges with their business model, non diversified client base and lack of strong risk management, oversight and governance. But the next banking crisis may have different factors and we are not done yet with this crisis. So I do think that there's an opportunity now to refresh and rethink risk management capabilities enabled by technologies. Some should that should have already helped prevent this from happening, but also some that might prevent the future crisis from happening. And the technology should be available across the bank to all lines of defenses in the bank. Every bank should have multiple lines of defenses. The first line being the business unit itself that should manage their risk and own the risk, followed by bank-wide risk and compliance departments. And then ultimately internal external audit.
(04:05)
And then the bank supervisors and these units should all follow a strong risk management framework and culture and have risk metrics, limits, thresholds, tolerances part of their risk framework that should be monitored through tech daily intraday. And I would even argue real time covering all types of risks, whether they're financial, credit rate, risk, interest rate, operational. So in terms of these failures, one of the key issues we discussed earlier was liquidity risk, the ability to kind of fund increases in assets and meet obligations without incurring losses. So technology should provide you all the information about capital liquidity and funding. The bank funding concentration, contractual maturity mismatches, tech should allow you to stress test your cash flows during or before you get into a stressful situation or at times of stress. Just think about for a second sort of technology that again that provides you sort of all your inflows and outflow real time.
(05:21)
Think about a metric as an example called a liquidity trigger that measures from say 1% to gradually 5% if your flows out of the bank are outside your normal floors for a particular type of deposit base. And if it gets beyond a certain number that you've set as an organization or a bank, it's red, it's a red flag and it should raise the right in alarms and sirens and cause you to have the right intervention. So there's a lot of tech that could be in place that could have helped, should help and should help in the future, test out different scenarios. The other thing I would just add here is the speed at which things happened surprised me just in a day or so. Silicon Valley Bank lost 42 billion of deposits. That's very quickly. Customers that were armed with digital tools, I took action even when they were in transit potentially fueled by social media. So I think there's an opportunity to rethink risk management technology and tools to be more agile, more real time and factor for all this learnings coming out of it, this.
Penny Crosman (06:36):
It seems like there could be a number of challenges to what you just described. For instance, one could be having all the right data, timely data to feed this kind of risk management framework. Another thing I would think might be determining what the right triggers are for those red flags. How do you know when this underwater bond situation is truly dangerous versus something that you could live with? Because if you're not having a run on the bank, you can live with that situation for a while. And it seems like another one could be what you were describing as having the same kind of risk management dashboard among many different parts of the business, the compliance and the business lines and risk managers and other people. How difficult is it to create this kind of framework and is does it actually exist in some banks today?
Sumeet Chabria (07:43):
Yeah, I think broadly speaking, that's exactly right. Penny, broadly speaking, the ability to run these reporting exists in almost all banks that I know. The question is, is it sophisticated? Is it real time? Is it comprehensive with the metrics? And to your point, you need a very good data management layer and capability to make sure that you have a data lake, a data and a big data capabilities that can aggregate all this information in real time. You need to have the right compute capability in compute farms that can run scenarios and analytics quickly. Normal fast stress testing in large organizations can take weeks to put together. I think we are in a day and age where money we've just seen moves so quickly fed is also going to move to real time payments now in 2023. So I think these capabilities need to be built, data compute capabilities, analytics, and exactly right.
(08:54)
The discussion on metrics limits threshold tolerance is not an easy one. You need to have many of them that are right for your business. You need to have them populated real time and you need to have the ability and the culture and the governance to take action from it. So if you see couples, of these metrics yellow or trending red, but they're yellow right now or yellow trending green, what does that mean? What action do you take and do you have the next level of actions in the playbook already rehearsed? So you don't sort of look at the metric and say, I don't know what I'm going to do next. Let's have a meeting. So I think this needs to become, needs to become sort of a way to run the business real time rather than a way to look back on what just happened.
Penny Crosman (09:47):
That makes sense. So you mentioned stress testing and I think a lot of people have suggested that more stress testing, if the regulators required more stress testing of these mid-size banks and smaller banks, that this crisis could have been prevented. I've always been a little skeptical of stress test because I feel like somebody's coming up with some kind of outlandish or an invented scenario that might never happen in real life. How do you feel about it and what's the key to doing stress testing in a really realistic and helpful way?
Sumeet Chabria (10:25):
See, it's a great point. Stress testing does definitely help if what you stress, the scenario stress tested play out? If you look at the last stress test this happened, the Fed just did 2022. The scenario as laid out, especially for not the baseline scenario, but they have baseline and a severely adverse scenario. The severely adverse scenario as tested did not happen in reality, if you look at sort of FEB 2022, when FED issued the numbers, it had actually predicted inflation dropping to little over 2% now in Q1 2023. And inflation is much higher right now. As now it predicted that the three month treasury would be yielding shy of 1% now, and the three month treasury I looked this morning was over 4.6%. So almost five times the numbers. So the stress test do test a certain set of scenarios in shocks and market shocks.
(11:27)
And if they kind of play out that way, you get the result that you tested for. I think it's just one, it should be one part of a risk management framework for any bank. The other parts have to be, as we talked about, real time information and metrics and dashboard and the ability to scenario, test a number of scenarios that you as an organization believe may happen. And they don't have to be totally onerous and test 16 different variables. You can test a few variables at a time, but more frequent testing of more scenarios, more realtime of quick testing results based on the information you see. Even using, let's say some AI artificial intelligence to spot anomalies or to predict something that may happen even if you get some false positive is a better way to holistically manage the risk. And also having the right culture, the risk culture. So it's just not up to the chief risk officer or the chief financial officer and their officers to aggregate these numbers and figure out what may or may not happen. The risk should be a responsibility for every employee in a bank almost. And every department head to be able to provide insight on this.
Penny Crosman (12:57):
Well, to that point, I've heard that some of these problems, for instance with Silicon Valley Bank, were clear if you looked at its December call reports from 2022 and yet the bank wasn't really making changes accordingly, do you think technology could somehow bring more of a sense of urgency to balance sheet challenges or is that more of a, as you were saying, sort of a cultural policy issue?
Sumeet Chabria (13:32):
Yeah, I actually think it, it's both. I think it is, there's an oversight and governance issue because some information, to your point, was available readily available. It's not that they were not, Silicon Valley bank was not seeing loss of deposits. At some point they had built off significant amount of deposits that we all know massive numbers very quickly, but they were seeing some loss of deposits. They were obviously seeing more unrealized losses on their bond portfolios as interest rates were rising. So it wasn't, and that's why I think a number of institutions, including the Fed basically said it's a management and supervision failure. But when you see these trends happen and they can happen very quickly and then ultimately the big draw happened in a matter of days. But once these things do happen, I do think the technology can provide the right information and reporting and help the risk team build a trend report.
(14:36)
So that could forecast what would happen under different scenarios. If silicon valley never thought that they could lose 10 billion in a day, they could have run a report and say if that would have happened, they could be in the zone of becoming liquidating or becoming in solvent. So I do think the technology can build intelligent reporting around forecasting and trends when different levers change, levers change. And in this case, the levers were deposits, the levers were a treasury being repriced, but tomorrow could be something else that could change. I mean, banks deal with a lot of risk, interest rate credit, operational risk technology, risk, cyber. There's a host of risks, fraud. And I do think that the tech can enable intelligent action and make information visually available visually in charts and graphs and easily digestible both for management but also all the way after the board.
Penny Crosman (15:33):
So I know after some of our past banking crises, there's been talk about giving the government an industry wide view of risk, especially through the Office of Financial Research that I think was created through Dodd-Frank, if I remember right. But the idea was to give the government a dashboard of all those different kinds of risks you talked about and where they are prevalent throughout the industry where there are a lot of banks of a certain size or in certain sectors or certain counterparties that are all subject to related or similar risks. And I think that that effort has kind of fallen by the wayside due to a variety of reasons. But do you think that still might happen? And would that be helpful?
Sumeet Chabria (16:28):
I think it could happen, and I think it was a good idea. But since that point in time, our economy has increased the size of bank balance sheets, just the amount of intra transactions that happen, new asset classes. It's all get gone to a point where I think there's always a risk that if you don't have intelligence or insights that get created, that sort of point to where things may go wrong versus the amount of data that gets generated today and the amount of reporting that gets generated and the amount of metrics people have to monitor. Sometimes you miss the forest from the trees, then they miss the big picture. So the question, when all this is the information is already out there, the amount of revenue reporting that happens, the amount of reporting happens in each individual company, the amount of credit and counterparty reporting that's out there.
(17:31)
And this, I do think the information's out there, the question is how do you get aggregated and how do you make intelligent, so how do you point to intelligent... intelligently point to places where the next big event may happen versus a report of 30 to 50 or 70 different areas of concern a day and then nobody looks at the report anymore, which is the risk we al always have. So I do think that generative AI even, or AI generally going forward could play a role there. And you have to experiment on some of these things, but I don't think we have a choice not to do so given what happened and the level of intervention that's happened and the concerns around ensuring all uninsured deposits and the pros and cons of any such major action. I don't think we have there. There's too much to lose for all of us without having some level of intelligence cross-industry.
(18:30)
And we should all look back and say, is that the right department still to do this? Have, do they have the right capability? Do they have the right team? Do they have the right technologists and the financial expertise? So I do think coming out of this, there have to be a deeper discussion and dialogue on it. But I do think that that more has to happen than less here. But without stifling innovation and without stifling and creating sort of regulation that burdens ability for banks to lend responsibly, which many of them do and create in the cycle in this economy of our businesses, fund themselves and we run our lives and get credit.
Penny Crosman (19:17):
Sure. That makes sense. Well, Sumeet, this has been very interesting. Thank you for sharing your insights and to all of you, thank you for listening to the American Banker Podcast. I produced this episode with audio production by Kevin Parise. Special thanks this week to Sumeet Chabria at ThoughtLinks Rate us, review us and subscribe to our content at www.americanbanker.com/subscribe. For American Banker, I'm Penny Crosman and thanks for listening.