This site will work and look better in a browser that supports web standards, but it is accessible to any browser or Internet device.
The Basel Framework (Basel II) places heavy emphasis on the usage of default risk models that have been developed using historical data on the default/survival of loan transactions. Unfortunately, the necessary data can often be unavailable.
New types of lending involving new products or new customers, by definition, do not come with the historical data necessary to statistically estimate a model of default probabilities.
Even when a bank has been engaged in a particular line of lending business for a long time, there is no guarantee that the available data will include all of the risk factors deemed to be necessary inclusions in a model of default. This problem also adversely impacts on banks that decide to update an existing default risk model to include new risk factors that have not previously been captured.
When sufficient quantities of historical data are unavailable what can banks do? Except in the very few cases where relevant and reliable data can be purchased from third parties, banks are forced to adopt a more subjective approach to default risk assessment, using expert models that are calibrated using input from staff and advisors.
These subjective approaches to default risk modelling come in a variety of forms. Datafactory supports one such approach. Specifically, it enables users to develop an artificial dataset that can be used to calibrate a default risk model. The modelling approach supported by Datafactory has the following benefits:
Given these benefits, banks can include the modelling approach supported by Datafactory as part of a transition programme to a default risk estimation capability that meets the criteria set out in the Basel Framework for internal rating models.
Datafactory is a website that works with you to create an artificial set of loan transaction data that can be used to build a mathematical model of default risk. Clearly, the dataset created is only ever as useful as the information that is provided by Datafactory users. For this reason, usage of Datafactory should be carefully overseen by people who understand the data creation process and who understand the mathematical methods that will be applied to calibrate a mathematical model to the data.