Dynamic conditional correlation

[danroler]

Hi,
I am new to econometrics and am using the student version (so no programming or addins). I am trying to calculate conditional correlation between 2 time series (Australian and Argentian stock market index). I know that I can use ARCH estimation but it gives me something as below. I am lost at to how can I plot the conditional correlation between the two time series. The only thing I can plot is the conditional variance and conditional standard deviation Please help!!




*****************************************************************
Dependent Variable: ARGENTINA
Method: ML - ARCH (Marquardt) - Normal distribution
Date: 07/18/15 Time: 02:10
Sample: 1/07/1997 4/28/2015
Included observations: 834
Convergence achieved after 49 iterations
Presample variance: backcast (parameter = 0.7)
GARCH = C(3) + C(4)*RESID(-1)^2 + C(5)*GARCH(-1)

Variable Coefficient Std. Error z-Statistic Prob.

AUSTRALIA 2.616632 0.011854 220.7312 0.0000
C -0.724314 0.034277 -21.13145 0.0000

Variance Equation

C 0.000310 6.80E-05 4.560587 0.0000
RESID(-1)^2 0.901517 0.172282 5.232795 0.0000
GARCH(-1) 0.075978 0.077262 0.983377 0.3254

R-squared 0.816914 Mean dependent var 6.868356
Adjusted R-squared 0.816694 S.D. dependent var 0.344671
S.E. of regression 0.147569 Akaike info criterion -2.228724
Sum squared resid 18.11803 Schwarz criterion -2.200389
Log likelihood 934.3777 Hannan-Quinn criter. -2.217860
Durbin-Watson stat 0.040681

[Tapatobben]

Hi - I guess you have daily data. Quick and dirty solution to your problem: Get rid of the trend in both time series by taking the first difference. Then calculate/estimate the correlation between the series. Evaluate whether the correlation is significant. If you want to plot the correlation, then you need to estimate correlations based on sub-samples.

Sent fra min SM-N9005 via Tapatalk

[trubador]

Hi,
I am new to econometrics and am using the student version (so no programming or addins). I am trying to calculate conditional correlation between 2 time series (Australian and Argentian stock market index). I know that I can use ARCH estimation but it gives me something as below. I am lost at to how can I plot the conditional correlation between the two time series. The only thing I can plot is the conditional variance and conditional standard deviation Please help!!

You need to build a bivariate GARCH model. Use system estimation (Object->New Object->System) and select ARCH as the "Estimation Method". Please refer to a time series econometrics textbook to get some background in Multivariate GARCH models. As for the estimation of these models in EViews you may need to go over the manual and search through similar posts in the forum.

[danroler]

Hi - I guess you have daily data. Quick and dirty solution to your problem: Get rid of the trend in both time series by taking the first difference. Then calculate/estimate the correlation between the series. Evaluate whether the correlation is significant. If you want to plot the correlation, then you need to estimate correlations based on sub-samples.

Sent fra min SM-N9005 via Tapatalk

Thanks, when I calculate the unconditional correlations between the 2 time series I used log difference, since that would give me returns. Are you saying that I need to further take a first difference...bit confused..
thanks again.