On the Observed-Data Deviance Information Criterion for Volatility Modeling

Joshua Chan and Angelia Grant (2016)
Journal of Financial Econometrics, 14(4), 772-802
[ Journal Version | Working Paper | Code ]
(Previous title: Issues in Comparing Stochastic Volatility Models Using the Deviance Information Criterion)

This code estimates seven stochastic volatility models. For each model, it computes two variants of the DIC: the observed-data DIC and the conditional DIC. We recommend that only the observed-data DIC should be used.

The methodology is demonstrated using daily returns on the S&P500 index. The seven stochastic volatility models are:

1. SV: stochastic volatility model where the log volatility follows a stationary AR(1)
2. SV-2: same as SV but the log volatility follows a stationary AR(2)
3. SV-J: same as SV but the prices equation has a "jump" component
4. SV-M: same as SV but the log volatility enters the prices equation as a covariate
5. SV-MA:same as SV but the observation error follows an MA(1)
6. SV-L: SV with a leverage effect
7. SV-t: same as SV but the observation error follows a Student's t distribution