Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: TSCS estimator selection for unequally spaced data


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   RE: st: TSCS estimator selection for unequally spaced data
Date   Thu, 14 Aug 2008 17:14:56 +0100

It's not the main question, but I would never 

replace year = year/5 

Rather I would 

gen time = year/5 

The original variable -year- should remain as is, say for plotting data. 

On the main question, you can still use time as a predictor even if you don't have regularly spaced values. It is just that some kind of regression rather than some kind of time series model is likely to be the main vehicle of analysis. Also, it is easy enough to keep an eye on serial (auto) correlation of residuals etc., perhaps through a variogram rather than an autocorrelation function approach. 

However, that may be off the mark given whatever your data and scientific problem actually are. 

Nick 
n.j.cox@durham.ac.uk 

Augusto Cadenas

yes, of course the first thing I would do is tssetting the dataset
after having changed the time variable in such a way that the minimal
interval is 1, as Nick suggests. (in this case: replace year =
year/5).

But then again, I still don't know how to estimate a dynamic panel in
this case, where some observations are 1 time unit apart, some are 2
units apart, some are 4 units apart.

Would, alternatively, -xtregar- be a decent subsitute? If I can't
model the exact dynamic evolution of the dependent variable, could it
be meaningful to subsume the dynamics under an autocorrelated error
term?

On Mon, Aug 4, 2008 at 1:45 PM, Nick Cox <n.j.cox@durham.ac.uk> wrote:

> You may be slightly better off using something like year/5 for your time variable. Otherwise no magic fix is apparent to me.

Augusto Cadenas

> what if you have a very normal panel data set, with the DV being
> measured in all panels at the same time (as opposed to Ken's dataset),
> but the time periods are unevenly spaced? In my specific case, 20
> years apart at the beginning of the dataset, 10 years after that, and
> 5 years towards the end. I am very reluctant to drop observations in
> order to achieve evenly spaced intervals. What if one now wants to
> estimate a dynamic panel in this case?
>
> The problem I am thinking of is that if the autocorrelation factor is
> equal to, say, "a" for two periods that are 5 years apart, it should
> be "a^2" for two periods that are ten years apart, and "a^4" in the
> earliest part of the dataset.
>
> Does anyone of you know whether this problem has been dealt with in
> the literature in the past, (and if there are any implementations for
> that in stata)?

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index