Fixed Income (Credit) fair value models in the literature tend to be variations on cross-sectional regressions. For a recent example in a factor-model setting, see here.
My understanding is that this kind of model is not considered state-of-the-art by many buy-side firms, but it is very hard to find literature on these.
I'm familiar with three (broad) additional classes of models:
- Stochastic pricing models with two factors, one for the call option, as a function of interest rates (calibrated to swaptions, for example), and one for the "default-option", where credit-quality is a proxy for how out-of-the-money the option is (calibrated to something like a transition matrix based on historical-defaults).
- A structural model based on a Merton-type framework.
- A combination of the two: for example parametrizing the default space in terms of distance-to-default.
Which relative value models are considered world-class? Are there any good references in this space? What "works"?