Yefin Dain writes:
Could you please advise a tool for modeling VAR (vector autoregression) multilevel models? All multilevel tools I could find information about were designed for cross-sectional data, or I missed something? I am interested in modeling multilevel regression of time series, which assumes endogenous time series(dependent variables) are regressed on its own lags and exogenous time series(independent variables). It is often named as dynamic regression.
I don’t know. Maybe Mike West has some software for this? He did the Bayesian dynamic regression stuff 20 years ago or so.
Sorry, I've been off celebrating the 100th anniversary of the t-distribution in Dublin, so I didn't see this. But:
a) BUGS can do VAR models (naturally!)
b) I was looking at VARMA this some months ago, and apparently they are not identifiable (I haven't got to the bottom of this, so I don't know if none are identifiable, or if only a few aren't). I was told that econometricians have given up on VARMA models and moved onto something else. Hence there may not be too much in the way of software out there.
how can i learn VAR. is there is any software for vector auto regression
the simplest intro to VAR is Brandt and Williams little green sage book (2007). I am a sociologist, but economists have been using VAR since 1980 and political scientists since the late 1980s. Sociologists, to my knowledge, have not done much with it. I was talking to an economist today and he said the field has moved pretty rapidly since then…but basic unrestricted VARs might be useful in disciplines where they have not previously been applied–always good to learn the basics, either way