Model design, build and management
30 March 2016
Model inventory as an enterprise enabler. ‘The aim is to help organisations lower cost while identifying and eliminating model risk throughout the lifecycle, resulting in an industrialised yet adaptable modelling framework. When new models are required, these can be designed, built, managed and run with total confidence.’
ACHIEVING MODEL CONFIDENCE
Until recently models were seen as strategic and secretive assets, enabling banks to price and hedge exposures more effectively than competitors. Models were closely guarded and there was very little scrutiny as to what they cost.
Increasing regulatory pressure to demonstrate control along with the convergence of models at portfolio level – driven by CVA, collateral and funding needs – mean that these assets are quickly becoming a liability.
The costs of developing and managing models has spiralled while the competitive advantages they offer have fallen.
A legacy of models, built up over decades, can become a significant inhibitor to business transformation and cost reduction.
Riskcare assists customers in mastering their models one step at time – both in isolation and within the context of infrastructure, other models, data flows, controls and governance.
The aim is to help organisations lower cost while identifying and eliminating model risk throughout the lifecycle, resulting in an industrialised yet adaptable modelling framework. When new models are required, these can be designed, built, managed and run with total confidence.
MODEL MANAGEMENT DRIVERS: AN INTRODUCTION
Regulators are demanding far higher standards of consistency, control and audit around models. Significant steps have been taken at many institutions to record inventory, ownership and status, and to impose a level of centralised management. However, the governance process is usually disconnected from system-level controls and there can be a lack of transparency around code, documentation and model usage.
In response to regulatory demands, there is now a need to demonstrate much more evidence of control and this is increasingly forcing proprietary models to be opened for scrutiny.
This presents a significant opportunity for organisations that can achieve an integrated end-to-end environment for building, documenting, controlling and managing their models. By automating large parts of this process, costs can be lowered significantly while increasing confidence that models are operating as intended at all times.
Evolution in the sell-side broking model is driving an expansion in low-touch flow and automated quotation for more complex products. This is bringing a number of modelling challenges with the related impact on credit, liquidity, funding and collateral models.
With each change comes the need to further streamline and productionise run-time models and to extend the depth of centralised and automated oversight.
The first step towards achieving this is to fully manage each part of the model lifecycle, introducing robust industrial design and software development practices from the earliest stages of model specification. This can continue into ensuring highly flexible and performant modelling environments, fully automated run time controls, and delivering absolute confidence in pricing and management of exposure.
Many organisations still pay lip service to these requirements with models that are numerically correct on paper but which are not professionally managed at run time, or where advanced methodology is deeply embedded within opaque systems and offers no flexibility.
Step one towards removing these barriers is to expose and industrialise models and to create appropriate service layers between model and trading workflow.
Market pressures – including the capital controls already mentioned – increased transparency and digitisation in the markets and falling liquidity have all resulted in falling margins on formerly profitable business lines. At the same time, banks face the requirement to meet increasing standards of regulatory scrutiny. The choice here is either to exit business lines or find a way to strip out cost of ownership.
Models are typically costly and time consuming to build, and very hard to reverse engineer once internal skills move on. This presents a significant opportunity to benefit from investment in industrialisation of the model lifecycle – bringing to bear best practices from engineering, industry and software design to deliver models quickly and reliably at far lower cost on a standardised platform.
To view the full article, please click the link below.