By Pascal vander Straeten January 02, 2023
Modern organizations face several challenges when it comes to risk management. Risks can crop up insidiously and swiftly around the globe due to the interconnectedness of the markets and other factors. This increases volatility and uncertainty and puts more pressure on risk managers.
Currently, the financial system relies on about thirty-three Global Systemically Important Banks (G-SIBs) (not to mention Central Clearing Parties and Insurance firms), all of which are too big to fail.
In recent years, central banks around the globe have been printing money in record numbers, pretending that the economy was rebounding by simply lending more money to households/corporations/firms/governments to maintain the illusion of solvency (think about the Eurozone debt crisis, but also the overall amount of debt in the U.S., Japan, U.K., and even China). But shielding all economies from shocks harms them all; avoiding small errors magnifies the big ones.
As atoms and bytes interact, new projections and narratives shape our world every minute. There has been an increase in the number of Black Swan effects because of complexity, interdependence between parts, globalization, and ‘efficiency’. This allows people to sail close to the wind. The proliferation of uncertainty and volatility poses many dangers. How can banks deal with them?
In order to make informed decisions, they need accurate data due to uncertainty and volatility. In addition, they need to consider various issues and alternatives when making decisions. Their decisions will be informed as a result of this. According to practice, the following six factors have become increasingly relevant since the GFC: operational risk, models and their role, systemic risk caused by Black Swans, capital charges, corporate governance, and the role of the Board of Directors.
This article aims to discuss one major regulatory issue that dominates the global regulatory discussion in terms of systemic risk. Several issues take up a considerable amount of time for risk managers and raise complex questions about risk. Is it necessary for global systemically important banks’ balance sheets, or environmental risks, like sovereign risks, to be considered in a firm’s models if they want to model the future? In particular, I am interested in understanding how each of these factors impacts the firm and the risks associated with the system as a whole.
Since the Global Financial Crisis (GFC), financial risk management has faced many challenges, including Black Swans, a concept introduced by Nassim Taleb in 2007. According to Taleb, a black swan is a highly improbable event with three principal characteristics: 1) It is highly unpredictable; 2) It carries massive severity; and 3) People contend that they expected the event to occur. We don’t acknowledge Black Swans until after they have occurred. Why is that?
Dr Taleb says humans are hardwired to learn specifics when they should focus on generalities. It is common for us to concentrate on what we already know and ignore what we don’t know. As a result, we are incapable of estimating opportunities, too prone to simplify, narrate, and categorize, and not open enough to reward those who can imagine the impossible.
We fool ourselves into thinking we know more than we do. Significant events continue to surprise us and shape our world while we limit our thinking to the irrelevant and inconsequential. Black Swans and blind spots can be prevented by properly exploiting organizational data.
Managing risks with data involves data acquisition and merging, risk analytics, and risk management decision support systems. To generate integrated risk scores, qualitative and unstructured data, such as text and voice recordings, can be combined with quantitative data, such as balance sheets and technical performance. Consequently, “data-based risk analysis” complements and reinforces the more traditional subjective scoring methods used in classical risk management.
Financial services companies increasingly realize the importance of leveraging and mutualizing their massive data sets. Identifying relationships, correlations, and early warning signs can increase a firm’s competitive advantage.
Quantitative models and algorithms cannot determine investment/business decisions solely. In order to make wise decisions, decision-makers need to use quantitative tools exclusively.
In isolation, these data outcomes will not be helpful to a firm in predicting the future. However, when combined with qualitative advice from experts, they can assist you in mapping risks more holistically. Because the historical sample is so small, tail events are particularly interesting. In light of this, risk managers will need assistance explaining why statistical predictions of future events are better than rolling the dice.
Risk managers may need more historical records if the extreme event they are trying to predict is rare to overcome the statistical biases inherent in small samples. They may be able to explore the deep past in those cases, but the events they are looking for are as sparse as Earth-like exoplanets in space. There is no way of knowing when or if they will happen again.
A fundamental limitation of data science is that it is based on models and assumptions closer to theory than reality. In most problem spaces, it is assumed that steady equilibrium will not exist (hence, consideration of entropy). It may be okay for some physical subsystems (including computing systems) to make this assumption. A reasonable assumption about biological substances (including social networks) is rarely valid.
It is often flawed to assign probabilities to events based on the assumption of a closed (and steady) set of outcomes. If the results are generated on the fly, we may not be able to model them (a kind of second-degree “statistics”). Most practical systems become nonlinear at some point, sometimes violently and strongly.
A black swan’s behaviour is usually nonlinear as well. Black Swans, however, are not always unknown unknowns. A million-dollar question when discussing Black Swans is: what triggers extreme events? It is pertinent to remember that The Black Swan emphasizes that the most significant social changes come from unknown unknowns. Because extreme events are impossible to predict, they are called Black Swans.
The most spectacular failures of financial institutions, however, deserve closer examination. Often, risk managers will be surprised to find that the causes of extreme events are quite obvious, such as a reckless M&A decision or an inexcusable business strategy error.
Thus, most tail events aren’t always caused by unknowns but by known unknowns. For example, it was mismanagement by bankers and regulators that caused the latest Global Financial Crisis. The GFC was not a tail event, which should result from something exceptional.
Financial risk indicators (ratings, VaR, spreads, etc.) were not used to predict the financial meltdown because the tools were inadequate.
People still believe that tail risk events only occur due to prime and unforeseen causes, i.e. something out of the ordinary. In reality, however, things are quite different. People often create extreme risk events due to basic and preventable actions, such as stupidity or recklessness.
A common cause of extreme events in the financial sector is strategic mistakes. Several notable bank failures were caused by strategic mistakes by their bosses (e.g. UBS, AIG, Northern Rock, HBOS, RBS, Dexia, and Wachovia).
Is big data capable of mitigating black swan effects?
Modern information systems collect data at an unprecedented rate. Risk management and governance can be enhanced by incorporating such data. Information of any type is considered data. In a call centre, for example, operators assist customers in entering written text. Alternatively, they can use web logs or data warehouses where all transactions and tasks are tracked.
Unstructured semantic data must be organized using data extraction tools and ontologies. Experts have developed the fields of natural language processing and artificial intelligence. In order to develop a robust risk management infrastructure, quantitative and qualitative data must be combined. Combining nonlinear analysis with Bayesian networks is required.
This vast amount of data needs to be mined, collected, analyzed, and translated into information that has distinct added value for the organization. These data will not help you avoid extreme events. Instead, it will help you understand what is likely to happen, what could happen, and how it will impact the firm. That approach relies on large amounts of data, and global regulators are preparing to oversee this brave new world.
Increasing amounts of data are mined, collected, analyzed, and stored. Data science is the multidisciplinary evolution of business intelligence and analytics. As well as having a solid foundation in statistics, math, data engineering, and computer science, data scientists must also be industry experts. This is to correctly identify the fundamental problems and the kinds of answers that should be sought in any given area.
Furthermore, customers must be able to understand what business leaders are finding. Big data and data science are closely associated. Information volume and variety have reached unprecedented levels for institutions today.
When analyzing these different sources of information, data scientists bring a wider and deeper perspective than traditional analysts. A data scientist analyzes data from many angles, determines its meaning, and recommends ways to use it.
As part of conducting a comprehensive, extensive information analysis, it is often challenging to obtain access to multiple data sets locked away in organizational silos – whether they belong to different lines of business, different companies within an industry, or different institutions in society as a whole.
Data science is capable of solving complex problems by connecting, synchronizing, and analyzing previously disjointed data sets. Building a predictive risk management system requires data, but it is only one piece of the puzzle.
In order to assess the validity of high-impact, hard-to-predict black swan events, firms need to work across data sets and silos. Experts often discover that even when the necessary information was present, we failed to anticipate catastrophic black swan events like airline crashes, financial crises, or terrorist attacks. The data was spread across different organizations and never properly brought together.
With access to oceans of data, firms are turning to data science to find valuable insights within that data. All kinds of critical issues will be addressed this way.
Furthermore, while big data may produce more information, the information must be useful and add value to the decision-maker, as well as the business model of the firm. We need Big Data that generates signals (and does not want noise or Big Data that remains silent on Black Swans) to help mitigate the impact of future Black Swans. Big Four companies, SAP, IBM, SAS, GE, and the Big Four, are very adept at data analytics & consulting.