Estimated reading time: 2 minutes
When making decisions under uncertainty we typically want to know how well our model of the world has performed in the past. And preferably in different environments. This would help us to know if the model can help us to make better decisions, also in the future. We therefore want to review how our models have performed in reality and to what extend they can really help us in the decision making process. In other words, we want to perform a model validation.
Interest rate risk models
We start with a some interest rate risk models from the financial sector. As their primary focus is the determination of capital needs for possible losses caused by extreme changes, they are of particular interest. Consider for example the following models:
- Basel focusses on the banking sector in virtually all western countries
- Solvency II focusses on the insurance sector in Europe
- Wtp focusses on the pension fund industry in the Netherlands
These models are often used as a reference model in risk reports. This raises the question: How well do these models perform? and Is it possible to do better?
In this series of articles we will perform a model validation or backtest on a few of these interest rate risk models. We want to compare them with our Base and Regime models:
- Solvency II models used in the insurance sector in Europe (SII)
- Wtp models used in the pension fund industry in the Netherlands (Wtp)
- Purely statistical models based on historical data (BASE)
- Regime based models (REGIME-BASED)
An evaluation and comparison of the performance of the different methods can be found in theoptimal model selection article.
