FNA roundtable event summary (10th November 2021)

This FNA roundtable, co-hosted with Planixs, offered an insight into the latest liquidity stress testing topics and innovations. 

The opening presentations discussed Intraday Liquidity Stress Testing from a technical and practitioner perspective. The technical view looked at the model components, construct and critical model drivers, with the practitioner’s perspective offering an insight into building and embedding an Intraday Liquidity Stress Testing Capability into a global business. The presentation also covered the practical problems, including data and regulatory issues that can arise. 

Why is Intraday Liquidity Stress Testing a critical and interesting topic? 

Firstly, Stress Models ultimately drive the liquidity requirements in a bank, for which regulators are accountable. Secondly, with the exception of a few, most regulators are inconsistent in terms of how prescriptive or non-prescriptive they are. Thirdly, the performance of these models is critical for internal alignment within a global bank, linking business units and legal entities with risk management, driving many discussions within large international banks. 

Presentation 1: Estimating the Intraday Liquidity Risk of Financial Institutions- A Monte-Carlo Simulation Approach 

The first presentation focused on a paper written in 2021 that examines risk from the perspective of a Central Bank. It introduces the Monte-Carlo Simulation Approach which provides a detailed measurement of intraday liquidity risk by answering the question: What is a financial institution’s maximum liquidity needs for a defined confidence level? 

The research identified the lack of synchrony between incoming and outgoing payments as the primary source of Intraday Liquidity Risk. Due to the interconnectedness of the financial system, monitoring synchrony is essential because a lack of synchrony can lead to broader disruption to the system. 

Measuring the circle of liquidity needs of an institution throughout the day offers a range of coordinated actions by settlement agents and gives an insight into the size and timing of received and executed payments. 

The presentation offered two examples, comparing two contrasting institutions: one a commercial bank, the other a broker-dealer. Commercial banks tend to have a sizable net balance at the beginning of the day due to reserve requirements; however, broker-dealers tend to put almost every dollar into securities, starting their day with a balance close to zero and borrowing from the commercial banks.

Demonstrating how much the synchrony of payments fluctuates throughout the day, enables Central Banks and regulators to gain a better understanding of an institution’s individual liquidity needs and the associated risks within large-value payment systems. 

The model also helps estimate the impact changes may have on the synchrony of payments as well as help financial institutions examine how they can avoid delaying payments. As the model also offers a view of the range of risk, supervisors and Central banks can potentially design liquidity buffers to help financial institutions further reduce liquidity risk. 

There are, however, areas where the model could improve and certainly challenges associated with its implementation in areas other than Central Banks. 

Although the model provides a complex view of risks within a system, additional methods using simple measurements should also supplement a regulator’s oversight. 

Secondly, it would be beneficial to model other sources of liquidity to show the number of securities a financial institution has in terms of cash equivalent securities, providing an improved understanding of whether or not a financial institution would need liquidity from a Central Bank. 

Thirdly, the model does not yet explore the impact on liquidity exchange if a crucial node is no longer in the network. 

As operators of RTGS systems, Central Banks can easily obtain the payments data the model requires to offer a detailed view of Intraday Liquidity risk. 

Presentation 2: Global Liquidity Risk Framework

The second presentation gave an insight into the strategic benefits, the implementation process and the lessons learned of implementing a Global Liquidity Risk Framework based on the Monte-Carlo Simulation Approach. 

The Bank chose to base the framework on the Monte-Carlo approach for its ability to capture behavioral assumptions rather than simply provide a historical look back at liquidity like the popular HLA method. 

However, the Monte-Carlo Simulation Approach was tailored towards the needs of Central Banks, so applying the approach to a commercial bank setting inevitably created some challenges. 

The Global Liquidity Risk Framework required the creation of a model to integrate with the Bank’s front-office platform, capable of monitoring liquidity risk on a daily basis. It was also essential that the model would work across the various entities within the global bank without the need to be redesigned. 

The presentation then gave an overview of the nine-step journey that resulted in the creation of a prototype in March 2020. The bank: 

• Gathered historical data to determine the b
• Removed double counts from the data
• Assessed the data, considering behavioural patterns between inflows and out
• Gathered historical stress evidence from academics and regulators 
• Set model parameters
• Undertook the simulation, in this case performing 75,000 to gain a minute by minute view of liquidity 
• Aligned simulations with scenarios 
• Set requirements by taking the 95% percentile target of cumulative net payments 
• Tested the sensitivity of the model’s parameters 

The simulations help set the daily buffer on the resulting front-end platform, reviewing the buffer each day once the actual data comes in. 

For simulations to be effective and valuable, they must represent banking patterns, but that’s not the only challenge. How can banks effectively scale buffer requirements to market stress over time, and how can the Bank reduce its overestimations when setting liquidity buffers? 

Despite different perspectives on the market, both presenters were in agreement that due to the interconnectedness of the financial system, if liquidity risk is not effectively measured, the system as a whole could be at risk. Problems must be dealt with as early as possible to avoid propagating across the system and having a knock-on effect on other financial institutions. 

Roundtable Discussion

Today’s session had already provided one example of applying the Monte-Carlo Approach but do other methods exist to monitor liquidity risk? 

As mentioned in the earlier presentations, many banks still take a historical approach to measuring liquidity risk while often applying qualitative overlays and stress testing looking back at a concentration of scenarios. However, there is a desire to move towards the more quantitative approach the Monte-Carlo Model offers. 

There are concerns around the issue of lost or recycled money, prompting the question, to what extent does the model capture money outside of the system’s measure. It could though be argued that

Over-buffering:

Another question concerned the issue of over-buffering. Does every bank use a substantial buffer? Does this in itself pose a risk? Should there be more of a centre point for which institutions should be aiming? 

While this may be an issue for commercial banks, central banks on the other hand see over-buffering as a positive as their main priority remains the efficiency and safety of the large value payment systems they oversee, preferring a ‘safety-first’ approach. 

To preserve liquidity, banks sometimes turn to diversification, though from a Central Bank point of view, if every bank diversified in the same way, the market is not diversified and can’t be trusted as a buffer source. 

The 2008 financial crisis highlighted the need to better understand financial networks and the interconnectedness of the system. Problems can quickly spread through, so over-buffering or having a large buffer may help protect a bank from systemic risks. 

Pattern Analysis

The discussion then turned to look at patterns asking if a detailed analysis of patterns could offer additional benefits when looking at ways to optimise liquidity.  

While patterns could be observed, particularly around the Trump transition and the Covid-19 crisis, conclusions were largely hypothetical. However, patterns could potentially help when considering optimising liquidity or testing specific scenarios, such as changing the timings of payments. 

Operational Frameworks:

The roundtable discussion also covered operational frameworks and tools used to help flag instances where banks may be behaving abnormally. Large banks often monitor their largest counterparts on a minute by minute basis with a significant degree of confidence; systems can flag up instances where a payment has been made. Operators can then make the call to ensure everything is okay. The challenge lies when looking at the broader network. However, holding such information on each counterpart could be used against individuals though it is necessary when looking at the most prominent banks, due to the knock-on effects should something go wrong. 

Liquidity Saving Mechanisms:

Participants also briefly discussed Liquidity saving mechanisms and their role in terms of future real-time growth globally while reducing risk. In one example, liquidity saving mechanisms reduced liquidity requirements by 20% in a high-value payment system. 

Although Liquidity saving mechanisms can help optimise liquidity to a certain degree, liquidity is a network problem. Increasing collaboration within the industry is essential, as optimisation requires behavioural change within individual banks. 

Looking ahead: 24/7 Payments and beyond

The roundtable concluded by exploring what impact 24/7 payments would have on liquidity and future challenges likely to arise.