Model risk
Bot’s job? Quants question AI’s model validation powers
But supervisors cautiously welcome next-gen model risk management

EBA: more climate risk supervisory reporting is coming
Official anticipates effort to identify climate impact on internal models, concentration risk

As interest rates surge, bankers fret over last year’s models
IRRBB modellers trying to predict client behaviour have little relevant data to fall back on
Model risk in mortality-linked contingent claims pricing
The authors investigate the influence of model risk on pricing life products and demonstrate that classical Lee-Carter-type models can be less accurate than the proposed model.
Quantification of model risk with an application to probability of default estimation and stress testing for a large corporate portfolio
This paper discusses the building of obligor-level rather than segment-level hazard rate corporate probability of default models for stress testing.
HSBC hits up rivals to strengthen US risk team
Bank hires from BNPP for US CRO and BofA for head of model risk management
ECL model forecasts are off-target, researchers find
Anticipated slowdown will be first major test for new generation of expected credit loss models
Model risk quantification based on relative entropy
This paper proposes a minimum relative entropy technique for challenging derivatives pricing models that can also assess the model risk of a target portfolio.
EU Parliament creates multiverse of SA-CCR outcomes
Some MEPs want to ease rules further than EC draft; others want return to undiluted Basel text
Covid chaos spurs on search for model risk aggregation
Many models failed in pandemic, but analysing them in clusters easier than whole-bank view
Can we take the “stress” out of stress testing? Applications of generalized structural equation modeling to consumer finance
This paper provides a practical introduction to the GSEM statistical framework in risk management, and it illustrates the game-changing potential of this methodology with two empirical applications.
Correlations in operational risk stress testing: use and abuse
The paper presents an analysis of correlation effects of economic factors on the operational risk losses of a medium-large UK retail bank, and it recommends that causal factors that effect operational risk should be identified.
Top US banks record 14 VAR breaches
JPM, Morgan Stanley, BofA, Citi, Goldman and State Street wrong-footed in volatile end to 2021
Apra’s overlay pushes CBA’s market RWAs up 30%
Market risk is at the highest level since Q4 2020
Hints on quantification approaches
Tiziano Bellini, head of risk integration competence line, international markets at Prometeia, examines the key components of successful model risk management, focusing on the importance of integration, processes, governance and IT solutions to…
Don’t follow the models: they’re lost, too – risk managers
Risk USA: managers cite Covid, repo crisis and geopolitical risks as examples of model failures
Scalability could trump complexity in machine learning debate
Risk USA: banks “on the precipice” of adopting more complex models, says Goldman exec
FRTB starts ‘tug of war’ between front and back offices
Risk USA: dealers face trade-off between accuracy of pricing models and level of capital charges
Model risk management: building trust and governance
As organisations increasingly rely on models that cover a wide range of business functions, there is an increasing need to create and maintain a comprehensive model inventory for enhanced collaboration and regulatory compliance across multiple regions…
BNY Mellon incurred a VAR breach in Q2
The highest losses-to-VAR ratio was 145.75%, in the first backtesting exception reported by the bank since 2018
Stronger together: CLS’s chief risk officer on risk culture
Deborah Hrvatin discusses integrated risk management, mega-hacks and model risk
Show your workings: lenders push to demystify AI models
Machine learning could help with loan decisions – but only if banks can explain how it works. And that’s not easy
New BoE climate risk scenarios spark race for extra data
Banks need information from clients or third parties to populate 1,760 stress test variables
Podcast: Claudio Albanese on how bad models survive
Darwin’s theory of natural selection could help quants detect flawed models and strategies