This is an example of estimation of a VECM-GARCH model (specifically an asymmetric BEKK model) based loosely on Pardo and Torro(2007), "Trading with asymmetric volatility spillovers", JBFA, vol. 34(9-10), 1548-1568, using a different data set.
Note that the data were taken from Yahoo Finance and are in reversed chronological order. RATS v9.1 automatically detects that and rearranges the data. If you want to run this with an earlier version of RATS, you'll need to use a different data set.
VECM-GARCH Model
Re: VECM-GARCH Model
Dear Mr. Doan,
In this code there are comments about the relationship between futures and spot prices. It is mentioned that
the theoretical relationship would make unnecessary the estimation of the cointegration vector. I would assume that the relationship is then a vector (-1,1). How could that be included in the model form to be usable in GARCH:
The original is:
set dx1 = logdjia-logdjia{1}
set dx2 = logrut-logrut{1}
*
set ect = %eqnprj(ecteq,t)
*
system(model=garchvecm)
var dx1 dx2
Lags 1 to nlags-1
det constant ect{1}
end(system)
*
So it would be ect equal to a series of (-1,1) ?
set ect = ?
Thanks,
In this code there are comments about the relationship between futures and spot prices. It is mentioned that
the theoretical relationship would make unnecessary the estimation of the cointegration vector. I would assume that the relationship is then a vector (-1,1). How could that be included in the model form to be usable in GARCH:
The original is:
set dx1 = logdjia-logdjia{1}
set dx2 = logrut-logrut{1}
*
set ect = %eqnprj(ecteq,t)
*
system(model=garchvecm)
var dx1 dx2
Lags 1 to nlags-1
det constant ect{1}
end(system)
*
So it would be ect equal to a series of (-1,1) ?
set ect = ?
Thanks,
Re: VECM-GARCH Model
set ect = logdjia-logrut