Lando Cox-process default model
Links stochastic default intensity to observable risk factors and supports rating-state transitions.
How it works
In a Cox model, the default intensity is stochastic and can depend on observable or latent credit factors. Conditional on the intensity path, survival to time T is exp(−∫₀ᵀ λ(s) ds). Averaging across paths gives the survival curve under the chosen measure. A random clock provides a simulation implementation.
Worked example
With an illustrative intensity of 1% in year one and 3% in year two, integrated intensity is 0.04. Survival at two years is exp(−0.04), or about 96.08%. If the two intensities themselves are random, compute the exponential path by path before averaging. Exponentiating the average cumulative intensity generally gives a different answer.
Validation and limitations
Calibrate under a pricing measure for derivative marks and under a historical measure for default forecasts. Check positivity, the behaviour under stressed credit factors and how recovery is coupled to default. A curve fitted to one issuer does not specify dependence across a portfolio.