Risk Management in a Post-Covid World: Highlights from “RiskMinds International 2020”

The year 2020 has borne witness to an unprecedented pandemic, which has left its indelible mark on history. This global pandemic has touched every single industry sector and has left the financial services sector especially bruised.  

In second week of December, I attended the RiskMinds International 2020 conference, organized by Informa Connect. 

Despite being moved to a virtual format, the event was an enriching experience that encouraged active participation and learning. This conference provided a great space to discuss emerging trends in risk management, and exchange ideas with seasoned risk professionals.  I have highlighted four salient themes that stood out from the conference below:

  • The Covid-19 Crisis – Impact of on model methodologies, banks resilience, and capital regulations
  • Model Risk – Strategy Management, Automation, and Quantification using Advanced Methodologies
  • Regulatory Landscape in 2021 – IBOR, FRTB, Basel IV, and Climate Change Risk
  • Latest Innovations in Risk Modelling – Application of AI / ML and Quantum Computing techniques in data and model management

The Covid-19 Crisis has dominated the majority of discussions during the conference.

During these discussions, it was highlighted that Quantitative Analysts from various banks have to tackle peculiar modelling issues in terms of handling negative interest rates, the breakdown of relationships between dependent and predictor variables, and large swings in asset prices & related indices. Risk modellers, while facing this conundrum, had to assess if Covid-19 related data points were to be treated as data anomalies or as a true Black Swan event. However, the majority of the risk modellers are of the opinion that the Covid-19 data points were to be treated as a true Black Swan event and hence, need to be considered in the modelling process even though it might result in a level shift within the model equation.

This crisis also left indelible impact on stress testing scenario analysis and tested the resilience of banks. Scenario managers expressed the need to recalibrate stress testing & reserve stress testing models using Covid-19 data. In addition, they discussed different ways changes can be incorporated in prominent scenario variables (such as the unemployment rate and S&P 500 index). Dr. Gernot Stania – Deputy Head of Stress Test Expert Division at European Central Bank, during a session surrounding the vulnerabilities of Eurozone banks in pandemic recession, mentioned that the central bank’s scenario analysis resulted in a CET 1 ratio depletion of -1.9% (to 12.6% in 2022 when data is aggregated for all Eurozone banks). Despite EU banks being well capitalized to tackle short-lived recession, a delayed economic recovery as projected by severe recession would result in CET 1 ratio depletion of -5.7% (to 8.8%in 2022 when data is aggregated for all Eurozone banks).   In general, the participants concluded that the lessons from 2008’s sub-prime crisis helped banks and financial institutions to better prepare for this crisis. In another session on post Covid-19 regulations, Gonzalo Gasos – Senior Director of Prudential Policy and Supervision at European Banking Federation (EBF), mentioned that banks are well capitalized (in the range of 13-14% of capital buffers); however, buffers of capital are required to be reassessed as these are currently very complex and are implemented at various jurisdictions.  

So, in the near future (next 2-3 years of time) these are required to be more flexible and releasable to get maximum benefits of capitals. It was also discussed that certain elements of the prudential framework (e.g. EBIT ratio, Basel 2.5 Market Risk and Loan Provisioning standards) and some accounting rules are found to be very procyclical especially in case of high stress periods. Hence, regulators are adjusting their prudential regulatory framework to counter this issue of procyclicality. In case of post pandemic regulatory focus, central banks are now focusing on regulations with due consideration for loan moratoria, treatment of government guarantees, and impact on non-performing assets once government support is withdrawn.  

The Model Risk Management topic focused on discussing strategy management, automation, and quantification using advanced methodologies.

Slava Obraztsov – Managing Director, Global Head of Model Risk at Nomura, presented his vision to treat Model Risk as a risk class that should be managed as much as other types of risks. It was further discussed that model risk management should use a step wise approach that includes identification, quantification, performance monitoring, periodic model approval, and aggregation of model risk. A session on model risk and changes in design & functionality in post Covid-19 has gathered a lot of interest. In session, the panelists Steve Lindo – Course Designer & Instructor, Financial Risk Management at Columbia University, Dr. Anna Slodka-Turner – Global Leader, Risk and Compliance Practice at Evalueserve and Xiaobo Liu – Managing Director, Corporate Model Risk Management at Wells Fargo introduced innovative ideas such as automating model testing and documentation, along with digital regulatory reporting. Different cutting-edge techniques in automation such as robotic process automation (RPA) for testing automation, and natural language processing (NLP) for documentation automation were discussed. Most of the participants believe that these techniques will help increase the efficiency of managing model risk.  

In the case of model risk quantification, Alexy Masyutin – Managing Director, Validation Department at Sberbank, discussed that the multiclass model’s share has significantly increased, compared to model regression or binary classification of models. Moreover, multiclass machine learning (ML) models are now being used more often when solving complex business problems such as IT-support queries categorization, product offers, and new monitoring to help with trading decisions. These ML models generally use algorithms such as NER, Word2vec +, XGBoost, and Random Forest that showed significant lift over conventional models. In a follow-up session from Dr. Peter Quell Head of Portfolio Modelling for Market & Credit Risk at DZ BANK, it was discussed that the advanced machine learning models are able to help track the right signals from a noisy cloud of data, which can be further used for tracking performance of VaR like models.

Regulatory Landscape in 2021 was focused on topics such as IBOR transitioning & FRTB, status of implementation, and emerging regulations such as Basel IV.

In the case of IBOR transitioning – priorities for the banks were discussed. GBP, JPY, EUR, and CHF based LIBORs are expected to be discontinued in 2021. FCA expects banks to move away from LIBOR as far as the issuance of new products are concerned. However, switching from LIBOR based products to RFR based products is not straight forward due to the fundamental difference between curves. In addition, banks need to build liquidity around these new products and get the general market to adapt. Efforts to gain liquidity around these products is progressing well since CCPs are moving to new RFR based products and hence, are helping to create required liquidity (e.g. long end of SONIA curve). However, getting marking adaption will be relatively slow as banks need to convince corporates to adapt to new RFR-based products. IBOR transitions will also have an impact on downstream risk functions since it will lead to changes in proxies, and additional basis risks and selection of fall backs will have to be considered. Adolfo Montoro – Director of Global Market Risk Analytics at Bank of America, also discussed the impact of IBOR transitioning on FRTB regulations. The choice of RFRs will have significant impact on modellable risk factors. This is due to the non-availability of timeseries from historical periods. Hence, regulators (like ECB) are prescribing banks to use proxies for enriching data.

It was also discussed that in order to remain compliant with FRTB regulations, banks are developing tactical solutions for analysis and reporting of FRTB numbers across the business lines. There is greater need to streamline processes (between FO to Risk Management to Finance/ Balance Sheet management teams) and automate analysis and reporting. Basel III reforms (also known as Basel IV) was discussed in terms of its readiness and implementation timelines. To be compliant with Basel IV regulations by 2024, the banks need to make substantial efforts in terms of implementing reforms for both global regulations (like Basel) and local jurisdictions.  All these changes will have a significant impact on data, modelling, and reporting.  The climate change topic largely revolved around finding ways to incorporate climate change risk in scenario analysis and credit risk management. The consensus was that banks need to manage their portfolio in such a way that Brown companies at the time of acquisition and portfolio management should be managed to avoid risk in terms of extreme events and transition risks.

Latest Innovations in Risk Modelling were presented and discussed. AI / ML methods are finding greater applicability in data and model management.

Gilles Artaud – Head of Model Risk Audit at Crédit Agricole, presented a session on ML model’s explainability and interpretability. ML models can be applied to many use cases in various areas of risk such as a credit risk alert system and retail & corporate credit risk models.  Artaud also highlighted a real-life example of ML models being used by bringing-up Google’s neighborhood search algorithm. Despite the explainability remaining a big question, ML was able to find complex and non-linear patterns in large volumes of data.  

It was also observed that AI / ML models hosted on traditional computing systems took a huge amount of time finding out convergence. Hence, advanced technology like quantum computing has helped to enhance the computing power required for these complex calculations. This technique is also useful when solving optimization algorithms (like asset allocation and mean / variance portfolio settlement) and Monte Carlo simulations. Quantum computing shows promising flexibility in terms of using a cloud hosted quantum computer and simulating it on a traditional computer, which reduces the cost of infrastructure surrounding quantum computing.  

This year has brought extraordinary challenges for financial services, especially around the subject of risk management. RiskMinds International 2020 was a great opportunity to (virtually) gather with risk professionals around the world and reflect on the challenges we encountered and discuss changes for the future. To learn more about model risk management, contact MRMsolutions@evalueserve.com or speak with an expert.


Amit Inamdar
Head of AI ML Innovations Labs Posts

Latest Posts