DCC GARCH Model with Metropolis-Hastings

[TomDoan]

The following uses Independence Chain Metropolis to analyze the conditional correlations generated by a DCC model. The conditional correlations are a complicated function of the data and parameters so getting an estimate of their precision (and not just point estimates) isn't simple. This uses M-H to sample from the posterior of the model, saves the histories of the conditional correlations and generates graphs of upper and lower bounds. Over half the program is for dealing with the rather complicated structure required for saving and analyzing many draws of several complete (and long) time series.

garchmvdccgibbs.rpf

Program file

(3.45 KiB) Downloaded 159 times

g10xrate.xls

Data file (included in RATS v8 distribution)

(1.2 MiB) Downloaded 144 times