Credit related decisions in banking can be divided into two different areas, underwriting of new loans and monitoring of existing portfolio. While banks are putting a lot of emphasis and efforts to assess quality of credit when taking them to their balance sheet, but later when macro factors (e.g. GDP growth or property market) change it is more challenging to review the quality of the portfolio. For the same reason stress-testing of the portfolio requires large resources at banks as it is done on a non-automatic basis. Quantifying underlying macro assumptions and connecting the risk profile of the portfolio to those assumptions solves both the revaluation and the stress-testing problems. RISKAWARE provides help to risk managers to this extent (RISKAWARE/Benefits).
Largest part of banks' balance sheet consists of loans which are not traded on the capital markets thus valuation of those assets requires professional expertise and a developed mark-to-model approach. Models produce an objective and automatic framework when estimating risk profile of banking assets, and this can be valuable as human decisions always include some subjective elements and carry cost and time constraints. In our view the combination of expert analysis and an effective model produces the best result in risk management.
Mainstream statistical (Scorecard) credit rating models use historical default rates to determine riskiness of credits. This gives assistance to risk managers when evaluating risk profile of loans but requires frequent recalibration to reflect changing macroeconomic environment and applies significant simplification to the calculation (e.g. ignores idiosyncratic factors). We believe that introducing Scorecard credit rating models was a big step in credit risk modelling, but a more fundamental and forward-looking model could provide further support to risk management. The Methodology of RISKAWARE is similar to the DCF method in its granularity and forward-looking approach, while Scorecard models are more comparable to the valuation techniques using multiples.