Model to calculate the expected losses for a mortgage portfolio
The task at hand was to develop a credit risk model, compliant to the IFRS9 regulation. The model should calculate the expected losses for the Achmea mortgage portfolio. In particular, the outcome of the model should be coherent with the current state of the macro economy. Additionally, in contrast to the newly developed AIRB models, the expected loss should refer to the full lifetime of the client.
As the RiskQuest team was heavily involved in the previous AIRB development, we were already acquainted with the data and business processes at Achmea. We made sure the base structure of the AIRB model was convenient for an IFRS9 model as well. This gave us a head start and allowed for an efficient and short project timelines.
The IFRS9 project started with an investigation how macro-economic information such as GDP and house prices affects both ‘probability of default’ and ‘loss given default’ historically. In the present, the covid-19 crisis started and was stirring up some macro-economic drivers. Not always in an intuitive way. This lead to discussions about how relations found in the past, are applicable for the future.
RiskQuest was in the lead for the IFRS9 modelling, working alongside modellers from the bank. While developing the model, business experts are involved to discuss model choices are validate results. In the end, the model was completed and accompanied by an explanatory document on the modelling process, choices, results and final model. As a result, the provisions for the mortgage portfolio are now better in the line with the current economy.
Based on RiskQuest experience in credit risk models we were able to set out a detailed modelling roadmap in preparation of the project. This allowed us to work efficiently and minimize chances for unwanted surprises.