EViews.com

Hi Dear Users
I'm not sure if my question is related to this forum or not, but I would be grateful if someone reply me!
I am looking for an Econometric method to estimate the time varying co-movement between the series. What I need is not DCC-GARCH class of models, because they deal with conditional correlations between volatilities not the the returns. For instance if I want to estimate dynamic or time varying co-movement between two series of returns or other kind of data like inventories, etc...is time varying copula a proper approach? any other suggestion?

Thank you in advance

Definition of co-movement depends on the context, so it affects the choice of appropriate method. As far as I understand, you are working with the return series of financial assets. Then the time-varying copula approach would be fine. As for other types of data, cointegration, common trend or time-varying parameter analyses might be useful depending on the research question.

Definition of co-movement depends on the context, so it affects the choice of appropriate method. As far as I understand, you are working with the return series of financial assets. Then the time-varying copula approach would be fine. As for other types of data, cointegration, common trend or time-varying parameter analyses might be useful depending on the research question.

Thanks for your reply!
the series that I'm going to estimate their time varying co-movement are 1. inventories in commodity markets and 2. liquidity in commodity markets. However, as I know with cointegration approaches only we can estimate if the series are moving together in the long run or not, it is not possible to obtain a time varying series of their co-movement, isn't it?
Another question is that is it possible to perform time-varying copula approach with Eviews? I was not able to find anything about it on the forum.
Thanks in advance

Yes, cointegration postulates that there is a long-term relationship among the endogenous variables. I think it fits to this setting, but it is your call. You can also consider dynamic factor approach, which would be the most appropriate method for your case, I believe.

As for the time varying copula approach, it is not an easy task to do in EViews, but it is doable (esp. for the bivariate case).

Yes, cointegration postulates that there is a long-term relationship among the endogenous variables. I think it fits to this setting, but it is your call. You can also consider dynamic factor approach, which would be the most appropriate method for your case, I believe.

As for the time varying copula approach, it is not an easy task to do in EViews, but it is doable (esp. for the bivariate case).

Thanks for your reply, after practicing sspace environment to perform a dynamic factor model I faced with one point: if I have a daily data set, something like 1600 observations, and I need to obtain the time series of dynamic co-movement between the series within this time period, say 1600 values of coefficients, what I recognized is that within a sspace environment, Eviews doesn't generate the time varying coefficients as a time series, however I need to obtain it one by one via "Specifications-Coefficient values", am I right? then obtaining 1600 coefficients and save them to make a time series is a hard job, very hardly feasible!
Would I ask you is there another way to obtain the coefficients as a time series within a sspace framework?

Dynamic factor is the common component and it changes over time. That's why it is specified as a (latent) state variable. What you refer as "coefficients" are the values of state variable, I think. If you are deliberately referring to "time varying coefficients" estimation within the state space framework, then you can search for the examples or discussions in the forum. Here, you can find the most recent one: http://forums.eviews.com/viewtopic.php?f=4&t=11859