Hybrid term structure models

In this paper, we study the properties of so-called hybrid models in which the various components of a multi-dimensional vector describing the state of the market, are both affine and non-affine models. Such models allows to combine the strengths of both affine and non-affine models. Using example of a hybrid model quadratic – Duffy – Kan the different properties were found for the main functions of the term structure – the forward rate and the yield curve. We found conditions on the parameters of the model, when curves are increases (decreases). Width of the bands characterizing the flexibility of the model fitting to the real data was studied. As a result, it is shown that expanding affine models to hybrid models by adding non-affine factors does not add additional complexity to analysis, but increases flexibility of the model.