Impact of more informative priors on KSC model calibration under SBC

Investigate whether employing more informative prior distributions for the static parameters in the Kim–Shephard–Chib Gaussian off-set mixture stochastic volatility model improves calibration as measured by Simulation Based Calibration, and characterize the conditions under which such prior adjustments enhance calibration.

Background

The authors use the priors specified in the original KSC paper and observe poor calibration for the KSC algorithm under SBC.

They speculate that more informative priors might improve calibration but note that this claim is conjectural and requires empirical validation.