In addition to the reestimation of the 4 standard models, the DMBS model included, we treat the following two subjects.
- A very intuitive solution to the problem of fixing prepayments at next unpublished term date.
- Generation of group level poolfactors and estimation of models using these.
1.
In the two prior releases we have been dealing with the issue of fixing prepayments at the next unpublished term date across time and loan groups. This is a subtle issue since bluntly fixing prepayments in an arbitrary way will induce biases into the valuation procedure. In this release we propose a solution to the problem, which in a very elegant way allows for fixing of prepayments. The solution method is introduced and effects of fixing prepayments using this solution are analyzed.
2.
All models estimated in SPV have traditionally used a series level measure of burnout. The main reason for this is that the values needed to calculate a series level poolfactor, namely prepayments and sizes of new issues, are readily available from data. Because of the structure of the Danish data, this is not directly possible using the raw input data. In this release we propose a method for generating group level poolfactors, using the generated group level prepayments which are used in the estimation of the SPV models. We give a thorough introduction to the generation procedure and provide examples of the results.
Two models using these group level poolfactors are estimated; one being an extension of the DMBS model, and the estimation and valuation results are compared.