To the shareholders the report shows that the Bank has roughly equal current

To the shareholders the report shows that the bank

This preview shows page 48 - 52 out of 52 pages.

To the shareholders the report shows that the Bank has roughly equal current risk for credit risk and market risk.
Image of page 48
49 Model risk The use of quantitative methods in risk estimation generates its own risks (an operational risk). The more models are used the more operational risk! Model Risk is defined as the potential loss a firm may incur, as a consequence of decisions made principally based on the output of models, due to errors in the development, implementation or use of such models. Sources of model risk are: Inaccurate or unrepresentative data Age of model and deteriorating assumptions Human errors in fitting the model Human error in programming the model Incorrect use of model Poor explainability and poor management understanding resulting in lack of challenge With the growth in use of Machine Learning (Artificial Intelligence) organisations are increasingly exposed to model risk.
Image of page 49
50 Example of why data can be a challenge Shown is quarterly US unemployment rates. The impact of COVID-19 is a shock outside of data used to fit models making it difficult to reliably forecast economic and financial data trends. Instead, businesses will be relying on scenario analysis and qualitative estimates. Size of economic stress only predictable through scenario analysis
Image of page 50
51 Scenario analysis is an important method in Black Swan events Quantitative Methods Inputs How qualitative? Where best used VaR, CFaR, EC Historical data Nil These methods are poor at predicting the future when there is significant uncertainty. Work best in business as uasual conditions. Stress testing Historical data with judgemental adjustment to some inputs (eg. Stress correlations, or use worst case calibration inputs). Some judgement. These methods are better at dealing with uncertainty but still rely on historical data to calibrate models. Scenario analysis Mostly judgement in designing the scenarios but may have some quantitative modelling to estimate outcomes from scenarios Most judgement. Could even be completely qualitative. Best at dealing with extreme uncertainty eg. Business forecasting in Black Swan event like COVID-19 or 2018 GFC.
Image of page 51
52 Copyright @ 2020 NOT FOR RESALE. All materials produced for this course of study are reproduced under Part VB of the Copyright Act 1968, or with permission of the copyright owner or under terms of database agreements. These materials are protected by copyright. Monash students are permitted to use these materials for personal study and research only. Use of these materials for any other purposes, including copying or resale, without express permission of the copyright owner, may infringe copyright. The copyright owner may take action against you for infringement.
Image of page 52

You've reached the end of your free preview.

Want to read all 52 pages?

  • Spring '19
  • risk principles

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture