QRR Rating

With the publication of the newly revised version of the capital regulation for financial institutions by the Committee on Banking Supervision in Basel (Basel II Capital Accord, June 2004), the internal rating systems have acquired a specific weight in the regulatory framework, specially for institutions that adopt the most advanced approach (Internal Rating Based). In this case the internal rating models can be used as an effective tool for calculating the probability of default (pd) of a given counterparty, one of the fundamental parameters involved in the calculation of the regulatory capital.

QRR-Rating is a rating tool based on the estimation of the probability of default of a given counterparty, with a methodological framework based on models both recognized and widely used in the industry.

It allows the creation and management of rating models, spanning different methodologies for the different risk segments and the availability of historical data of default:

SMEs (Small and Medium Enterprises): Use of statistical models for the rating of small and medium enterprises, with a statistically significant probability of default and for which there exists ample historical default information.
Corporations & Financial Institutions: Use of Merton structural model combined with credit derivatives (CDS) information for the rating of Large Corporations and Financial Institutions, where the information on historical defaults is relatively insignificant.
Sovereigns: Use of credit derivatives (CDS) information for the rating of Sovereign Countries, where the probability of default is also very low.

Web application developed entirely in Java using a client-server 3-tier architecture, consistent with the J2EE (Java Enterprise Edition) standard.

Main Features

Workspaces divided into configurable branches. Information relative to counterparties, models, users, etc..., are stored independently for each branch.
Users can interactively input data by uploading CSV files, easily exported from Microsoft Excel�.
By a guided process the user can create and validate credit rating models, based on the available counterparty information.
Possibility of using advanced non lineal regression techniques (Alternate Conditional Expectation) in statistical models.
Monte Carlo model simulation module as an auxiliary tool in the detection and selection of explanatory variables and the identification of models with a high Gini index.
Validation tools for statistical models constructed for the high probability of default segment:
- Bootstrap techniques for the analysis of the sensitivity of the model to the calibration sample checking the Gini index.
- Correlation analysis of the ratings returned by various models, as a model comparison tool.
- What-if analysis, that allows the evaluation of models under particular scenarios as well as variations in the explanatory variables.
Possibility of including qualitative modifications to the obtained ratings based on the judgement of expert analysts or the particular characteristics of a counterparty.
Contrast modules that allow the comparison of the ratings obtained by the application (created models) and the ratings given by external agencies (Fitch, S&P, etc...).
Management of consolidated historic ratings: the information of all ratings consolidated by any user are stored and maintained in the application database.
Use of configurable rating scales to convert between probabilities of default and ratings.
Different user roles with different levels of permissions: Administrator, Analyst, Modeler and Supervisor.
Localized user interface: the application is currently developed in English and Spanish but it is easily adapted to any other language.
PDF report generation.
User operation log. The supervisor may view the operation log to see all operations performed by all users on application data (creation, deletion, update...).