Risk Management

An approach to computing the Comprehensive Risk Measure

Presentation by Dherminder Kainth, Jan Kwiatowski & Douglas Muirden

This presentation outlines the approach that the bank has taken to computing the regulatory Comprehensive Risk Measure (CRM).

Aggregating stress test scenarios using conditional probabilities.

Paper by Jan Kwiatowski & Riccardo Rebonato

This paper describes a method of aggregating stress test scenarios using conditional probabilities.

Technical note on allocation of risk capital in credit portfolios

Note by Jan Kwiatkowski

This paper describes a method of risk capital allocation based on a paper by Jan Kwiatkowski and James Burridge in the Journal of Credit Risk, Vol 4, number 1 (2008). It is shown how to achieve considerable computational savings by a simple adaptation of the model.

Modelling Default Risk in the Trading Book: Accurate Allocation of Incremental Default Risk Charge

Presentation by Jan Kwiatkowski, Risk Minds, Geneva (2007)

This introduces a method of allocation of Incremental Default Risk Charge in the Trading Book, as computed using the Andersen Sidenius and Basu (ASB) algorithm. Using Bayes’ theorem it is shown how accurate allocation can be achieved by inversion of the ASB algorithm.

A Flexible Analytical Method to Calculate the Specific Risk Surcharge

Paper by Riccardo Rebonato, Jan Kwiatkowski and Lorenzo Liesch (2007)

Is there a simple, efficient way to calculate the specific risk surcharge for defaultable trading-book positions? Riccardo Rebonato, Jan Kwiatkowski and Lorenzo Liesch offer an easy-to-implement, analytic method to estimate specific risk and an “implied” method to calibrate it.Their method is conceptually similar to the Basel II approach and also meets the accord’s comparability requirement for banking-book treatment of specific risk and allows for concentration effects.

Theory and Practice of Model Risk Management

Paper by Riccardo Rebonato (2002)

Model risk is a topic of great, and growing, interest in the risk management arena. Financial institutions are obviously concerned about the possibility of direct losses arising from mis-marked complex instruments. They are becoming even more concerned, however, about the implications that evidence of model risk mismanagement can have on their reputation, and their perceived ability to control their business. Model risk inhabits, by definition, the opaque area where the value of instruments cannot be fully and directly ascertained in the market. The valuation of these products must be arrived at by means of marking-to-model, and therefore always contain a subjective component. In these days of heightened sensitivity to aggressive or opaque accounting, the ability to rely on sound practices to control model risk can have share-price implications well beyond the monetary value of the mis-marked positions. It is also for this reason that financial institutions place increasing importance on the effective management of model risk. Unfortunately, there is a widespread lack of clarity as to what model risk management should achieve, and about which tools should be used for the purpose.

Reconstructing a valid correlation matrix from invalid data

Paper by Peter Jäckel and Riccardo Rebonato (1999)

Correlation matrices calculated from incomplete data or taken from news services sometimes don't comply with the requirement of symmetry and positive semi-definiteness. Whilst it is easy to amend the symmetry requirement by manual intervention, it is not always straightforward to see how to adjust a given correlation matrix to become usable for factor analysis or simulation purposes. In this document, two methods are described that can be used to best-match an invalid correlation matrix given the constraint of positive semi-definiteness.