Liquidity risk:
How are management techniques evolving?
With Chandni Bhan,
Global Chief Risk Officer, Wise
Following bank failures and volatile market movements, liquidity risk is on everyone’s mind. One of the reasons for such scrutiny is due to the surprising behaviour of customers. Are yesterday’s liquidity risk assumptions still correct today?
Chandni Bhan, Global Chief Risk Officer, Wise, explores how liquidity risk and its management is changing today, and how to be proactive about upcoming risks by employing a multi-model approach in early warning dashboards.
I think liquidity risk is front and centre of a lot of things that have happened in the recent past. That's been the story for eons, but it's funny that we never seem to learn a lesson from it, right? Pretty much all bank failures have happened because of liquidity. The triggers must have been other things, but eventually, the reason banks fail is because they're not able to hand out the cash that they ought to hand out, so they end up becoming liquidity risk failures.
During Covid-19, what we considered to be highly liquid assets, like the money market funds, which are basically super liquid very short-term credit notes with enterprises, or even treasury funds, literally became illiquid! Money market funds were struggling to meet the requirements of liquidity at that point, so it really makes you want to test and makes you wonder how we used to think about liquidity before. Are they truly actually as liquid as they used to be in the past?
One of the reasons why liquidity if different today versus ten years ago is because the velocity of reaction has sped up so much. First, I think there's a lot more headline driven decision-making and connectivity to information. Investors are more prompt and smarter than before, meaning that the panic of sellouts is happening a lot faster and at a far greater volume.
The second thing is that you're also trying to remove friction from the system. For example, you're trying to make payments faster. The more friction you remove, married with faster reaction times, creates the perfect symphony of what would make liquidity get pressure tested as a concept. We saw that with SVB; we saw that with money market funds; we saw that with exchanges. And it’s the same story with crypto exchanges.
These inherent changes in the environment have really changed the dynamics of how we ought to think about liquidity.
Keeping an eye on everything that can drive liquidity is becoming important.
Liquidity is usually an aftereffect of a bunch of other events. One of the reasons why liquidity risk is a lot more pertinent is that the markets have become much more correlated than they were before. In the past, we used to keep bonds and stocks as uncorrelated instruments in our portfolios for diversification. When we look at the recent history of the markets though, it’s slightly different. They have become correlated, even within the same segment. For example equity markets in the US, China, Germany, and the UK are more correlated. This creates more liquidity stress. In this environment, anything can be triggering and the cascading effect is quick, and will impact liquidity one way or another. For example, an issue can start in credit, but lands as a problem in equities or commodities, creating more liquidity demand.
The second thing that makes liquidity a lot more interesting is headlines because it prompts faster reactions to events. And the third thing that makes liquidity even more interesting is the frictionless nature of moving towards T+1 or T0 supplements. But by reducing friction in the markets, you create more stresses in the markets – that’s a very different environment.
The people that don’t react to news fast can end up taking the worst brunt of it all. It’s not the institutions; it’s not the banks; it’s not the hedge funds. It’s actually the retail customer. The regulations, which are developed with a clear and good ethos, make it clear that you should have access to your money now, but making processes faster and offering swifter benefits can have a significantly bigger impact on the markets. I think that’s something that we haven’t really considered much in the industry, but the way we’re going, the disparity will become more apparent. Keeping an eye on everything that can drive liquidity is becoming important.
Well, I’ve always looked at markets and across assets, but never as a single asset class (due to the correlated nature of markets). Looking forward, it’s important to do that. We’ve developed a tool called the Early Warning Dashboard to help keeping an eye on risks and see where issues might be bubbling up from and cascading down, causing a spike in liquidity.
I worry that we're going to sleepwalk into a crisis if we don't think about liquidity proactively
Another thing we’re doing to understand liquidity better is to study behaviour patterns and customers. We’re trying to develop a much more systemic way of managing liquidity and to become smarter at what you need to hold inside funds and banks. There are a lot of great tools, especially with machine learning and artificial intelligence, that can give you a much more adept and nimble sense of temperatures and evolving patterns. I think this will be a meaningful place of innovation in the industry.
We also have structural changes in the industry, like regulations on stress testing and pressure testing, where a lot of work has already been done. There’s also a lot of focus on fund managers to do the same. For example in the money market funds, there’s a lot more emphasis on liquidity management now. Same can be said about usage funds. Beyond regulation though, when we’re going to T+1 or T0 redemptions, frictionless payments, and blockchain technologies, we’re not that long out ahead, but close enough to change the dynamics of liquidity risk. I worry that we're going to sleepwalk into a crisis if we don't think about liquidity proactively, and we don't put in the right sort of thinking as an industry about how we're going to manage velocity in those scenarios.
I think there’s a lot more to be done in terms of connecting the dots. Liquidity is a consequence of dots that can cascade, so figuring these out can help understanding how the domino effect works. An organisation or our risks should be seen as a network effect rather than single factor effects (which is what banks and the industry have done traditionally). As a bank, you have to stress test across ICAAP, ICARA, CCAR, but you’d want to get smart about the correlations between these. How? That’s a major area of development. The smarter you get at this, the better you can position yourself and make yourself far more resilient to your customers, as well as the system.
I also see a lot of change in the risk management space where risk will have to move from exposed to a preventative dynamic in the organisation, while staying systematic. Risk must be done through much larger models than humans, where the human expertise is overlaid. The data and system should be able to synthesise information and alert to future bubble-up events and provide answers to questions like “what’s changed?” or “how has the pattern shifted that causes us to have a look?” This is where the subject matter expert comes in to give context and connect with the right people to pre-emptively plan.
As an example, you can build live runbooks. When the Russian war on Ukraine started, overnight, all systems and all companies had to shift gear and figure out how to deal with a live, large-scale sanctions segregation of accounts. Some of us foresaw something like this, so we did table-top exercises and made sure that a runbook is ready, tested, and pressure tested, which was helpful. It reduced the organisation’s risk exposure, but more importantly, it reduced risk for the customers and markets.
Yes, it’s about having a playbook that sits in the back. It’s almost like having a manual for your computer or like doing fire drills. We do fire drills on a regular basis. We know when the bell goes off what we need to do because we built up muscle memory. In a crisis, muscle memory can be impactful, and these exercises should be done.
I can’t really speak for all, just my own organisation, but I think we’ve gone through a great evolution in risk management to become a data driven and signal driven function. The emphasis on harnessing the power of data and signals is starting to see a lot more focus, which is great because we need that to be able to manage future risks as the world is becoming more interconnected. It’s a network of risks, rather than isolated risks.
But you’re talking about tons and tons of data that need to be synthesised in real time and processed to give you the signals. Data and harnessing the power of data will be the biggest differentiators for the risk management function. But if you have good data and you also know how to harness the power of that data, you should start thinking about a parallel model.
This is what I do when thinking about early warning indicators. I never rely on a single model because I can never have 100% confidence that the model world work and serve every single scenario. Markets are changing; environments are changing. I usually have two or three models aiming to do the same outcome check. I build parallel models, some which leverage machine learning, some which leverage traditional sort of SME rule-based logic, some that are based on simplified versions of models based on regressions. They are helpful because when you have three or four models coherently signalling towards something, you’ll be far more confident that you’re right.
When you have models competing with each other, you are sometimes generating interesting information to go and look at. Why is something telling me one thing while the other isn’t? That can bring in new insights as well. Finding more novel ways of investigating the right things would be a very good place for risk management to evolve to.
Yes, fact checking. You can almost give it a weighted probabilistic outcome plan. If all of them say yes, you say yes. If three out of four say yes, you still say yes. But if it’s 50-50, you have to validate. You have to go through with judgment. It makes your process a little bit more thoughtful.
We are living in a period where we’re just starting to see some course correction from the nonstop trajectory of inflation. I think even central banks have a lot less ammo to support a downturn. We’re coming from a place of vulnerability because we don’t have that many tools in our arsenal to support the next big catastrophe.
We also have so much geopolitical dynamics going on behind the scenes. There are so many other elements at play which impact markets and could impair markets’ ability to operate as effectively as we are. The threats are on the rise. The mitigants in our toolkits are low. And again, the infrastructural changes we see – instant reactions and more concentrated markets – make the velocity of the next crisis a lot more unpredictable, and therefore potentially a lot more impactful. We could have sharp dips and sharp recoveries (like at Covid-19), but for a lot of places that sharp dip could be the end of the story. If you’re not fast enough or proactive enough to foresee them and position yourself for them, the existential questions are becoming much more real.
I’m really excited to see all the innovation that’s happening across the board. There are so many amazing speakers with a wealth of knowledge. What I love about conferences like QuantMinds it that even though it may not directly impact my line of work, I can see the idea and it’ll plant something in my brain for me to figure out how to leverage it. I’m really excited about hearing what everyone’s been up to.
