Smart About Risk  
Model Risk

Model Risk

Nowadays many companies (e.g. banks, energy, telco, automotive etc.) use different mathematical, statistical as well as expert models to support their business. The models may not only increase efficiency through quicker decisions, cost reduction and simplification of complex problems, but also increase objectivity of decisions.

On the other hand, application of models may potentially incur loss as a consequence of decisions based on incorrect or misused outputs of these models. The potential loss connected with models is usually denoted as model risk.

Due to the increasing influence of models on many various processes, model risk, its identification, measurement and management are topics of growing importance. The requirements for implementation of a model risk management system are often formulated by owners or management of a company, or in banking industry by regulators.

ARM would like to offer you consultancy services aimed at model risk management and related areas.

The offered services are structured as follows:

Development / independent review of the methodology describing model risk management

  • definition of what is understood as a “model”
  • classification of models (specification of model type, e.g. statistical/expert models, internal/external models)
  • definition of model risk management procedures throughout the model lifecycle for individual types of models
  • setting of rules and criteria for assessing:
    • relative importance of every used model (in terms of size of portfolio, type and frequency of supported activity, etc.)
    • quantitative impact of model errors (e.g. quantification of probability of error, average loss impact of an error and expected number of applications)
    • size of model risk for individual types of models
  • definition of limit values for individual criteria (e.g. in terms of “traffic light system” for managerial decisions) and list of possible corrective measures
  • model governance framework, especially the definition of roles and responsibilities
  • setting of reporting rules


Specialized workshops and ad-hoc consultations during the preparation of a methodology describing model risk management

  • introducing possible approaches to model risk identification, measurement and management
  • discussion of specific tasks in terms of client’s models
  • clarifying the regulatory requirements (if any)
  • providing comments and expert opinion during the methodology preparation


Technical support during the software preparation in case of purchase and implementation of existing software solution

  • help with the preparation of check-list that will describe the minimum needs for the software operation and required functionalities
  • preparation of user functional specification for the functionalities which are not currently included in the solution
  • supporting function to project management


Technical support during the software preparation in case of tailor-made development

  • preparation of detailed user functional specification for the software in order to:
    • assure the smooth development of the software in terms of stipulated time schedule
    • full involvement of all functionalities required by the client (e.g. workflow related to model development, validation, acceptance, implementation, testing and using; storage of related model documentation)
    • enable comprehensible management of the tool development (i.e. the programmers fully understand the technical requirements and they do not need to understand the inner logic of risk management; client’s experts have the core basis for the monitoring of developed functionalities)
  • preparation of testing scenarios (user tests)
  • performance of user-testing procedures
  • support for project management