Financial Markets Infrastructures (FMIs) generate immense amounts of data that is valuable for many purposes ranging from optimising liquidity to enabling better risk management and generating insights about the economy.

Dr. Kimmo Soramaki and Dave Sissens of FNA have recently published a paper on Future proofing the next generation of financial market infrastructures: The devil is in the data where they discuss various practical outcomes that can be achieved with advanced analytics using artificial intelligence (AI) and machine learning (ML) algorithms, how to get there and some potential pitfalls along the way.

The paper aims to provide sufficient insight for readers to consider new technologies and approaches for meeting industry guidelines. Additionally, it shares best practices on how as a result of increasing interconnectedness in the global financial markets, FMIs and their regulators, need to help one another improve their Anti-Money Laundering (AML) posture and systemic risk exposure.

Below is a short summary of the paper’s key insights:  

Only a small number of organisations are using low-level or transactional data to improve the stability, quality or robustness of their services to their clients.

From the large datasets observed at a global and national level, one thing is for certain: Financial Services organisations are far more interconnected than we would have ever thought. From a system perspective, from a process-interaction perspective, across asset classes, across infrastructures and across jurisdictions, models demonstrate that a ‘ripple’ in one area of the Financial Services industry can send waves across another.

Despite many organisations now collecting or generating large datasets, there are very few infrastructures using this low-level or transactional data to improve the stability, quality or robustness of their services to their clients.

Advanced analytics allow for the use of graph theory to create ‘networks’ which provide a deeper understanding of a financial organisation’s or even an individual’s behaviour.

By using transactional data, FMIs are now able to analyse and understand the interconnectedness of their clients and participants. This understanding is extremely valuable and can be used to improve not only the robustness of the FMI itself along with its clients, but to also improve the overall robustness of the industry. This is exactly what the CPMI is championing in its 2018 strategy paper.

Sample use of advanced analytics on transactional data include:

  • Designing payment systems of the future – By using advanced analytical simulations and transactional data, it is possible to design the ‘optimal’ version of a payment system.
  • Monitoring data-driven systems – Performing sophisticated analysis on in-flight transactional data can provide tangible business benefits. For example, a number of FMIs now use historic transactional data combined with the knowledge of liquidity and solvency issues, to predict the likelihood of future occurrences.  Similar advanced institutions will soon be using Artificial Intelligence approaches to better manage their liquidity needs. Payments departments can enhance their value to the overall business through analytics over granular transactional data.
  • Advancing supervisory functions – One South American Central Bank is now using its RTGS System’s transactional data on a daily basis to gain greater risk insights: By observing the payment behaviour of its participants vis-a-vis historical behavioural experience, it gleans important “early warning” information about the liquidity and solvency of its participants.
  • Using transactional data for war gaming – By using real data, simulated failures can be interjected during particular times of certain business days to play out realistic scenarios. What would happen if our biggest participant had a solvency issue on an IMM quarter date? Or, what would happen if a natural disaster took out a key region in the middle of a settlement cycle? All of these answers increase the preparedness of the FMI and in turn, the global financial industry.
  • Reducing the level of cyber risks – So far, this article has focussed on transactions. These can generally be thought of as payments or trades. But also important is the rapidly increasing interconnectedness of the technology upon which we run our services. Some sophisticated FMIs are now developing an understanding of the networks of technology which their participants and clients use. There are very few institutions that can understand and simulate these sorts of networks, but the number is growing. The FMI of tomorrow will have a complete picture of interconnectedness, from a technical and economic perspective, ready to hand in a financial crisis.
  • Improving fraud detection – The next generation FMI can consider providing its clients with a real-time confidence rating of the transactions it is processing, based upon its assessment of the transaction’s alignment to normal behaviour. The client could then be presented with the opportunity to rethink whether the transaction should be cleared or failed. FMIs that don’t want to go this far could simply provide features of the data for machine learning models that their participants can run. Network features of the data are particularly valuable as these are not available to any single institution, only to the FMI.  It is possible that some of these network features of transactional data can be shared without the fear of compromising proprietary data.

Here are some key steps to future proof your FMI:

  1. Make sure the data is available – Many legacy systems have in fact a very hard time providing the transaction-level data required for analytics. It may not contain all the necessary fields and the field values may not be clear – or not reliably and consistently populated. You can increase the value of the data significantly if there are good Application Programming Interfaces (APIs) to access historical data. So, when you are building or choosing vendors for transactional systems, this should be a key question from an analytical point of view.
  2. Establish whether it is legally possible to use the data for the type of analytics that you wish to build – It is much easier to establish data availability and readiness when the system and the rules are being set up, rather than going back to the members and possibly their clients to get these permissions retrospectively.
  3. Get the right people and tools – People usually means data scientists or analysts who may come from a variety of quantitative fields in economics, physics, mathematics, etc. These data analytics teams can help an FMI in many areas ranging from regulatory engagement to developing new businesses. As data challenges continue to prevail, inventive ways of developing, attracting and retaining these skills will become more critical for an FMI. Having a plan to address this increasing demand will become paramount.
  4. Create an appetite for change and culture of innovation – The journey starts with each FMI and its determination to move up the analytics maturity curve, both for its own development and benefit and in order to improve the risk profile of the industry. An FMI’s top management must get behind this as a priority and allocate resources accordingly.

Click here to purchase the complete paper. 

More News

What is Suptech Analytics?

Author: Will Towning, Central Banks and Academia Programme Manager   What is Suptech Analytics?   Suptech is the application of innovative supervisory technology by central banks and financial authorities. It has traditionally focused on helping authorities collect and manage data more effectively, as well as digitise and automate certain processes. Such early iterations mostly support […]

FNA Talks Data Science in Economics and Finance with the Bank of England

FNA Talks Data Science in Economics and Finance with the Bank of England    Adrian Waddy, Data Consultant at Australian Prudential Regulation Authority and Developer at the Bank of England, joins host Adam Csabay to discuss his contribution to the Risk books publication, Data Science in Economics and Finance for Decision-makers. Adrian’s chapter, Implementing Big […]

Reconstructing and Stress Testing Credit Networks

By Amanah Ramadiah, Fabio Caccioli, Daniel Fricke Financial networks are an essential source of systemic risk. Unfortunately, detailed data on (direct and indirect) interactions between individual financial institutions is often unavailable, and the only the total aggregate position is known. To conduct a stress test, one must resort to network reconstruction methods to infer the […]
Copyright FNA © 2021 | Privacy Policy