If you think that the equation in levels correctly expresses the functional relationship between the variables, and are differencing because of stationarity concerns, then differencing one side of the equation calls for differencing the other side as well. If you don't do that you will be changing the meaning of the coefficients you estimate. For consistency, both sides of the original equation should be integrated to the same order. If that is not the case it may suggest that the original equation is not specified properly. Researchers often ignore these precepts and difference opportunistically in order to achieve stationarity, without regard to the meanings of the equations they are estimating. David Greenberg, Sociology Department, New York University
----- Original Message -----
From: Joel Miller <[email protected]>
Date: Friday, June 26, 2009 6:40 pm
Subject: st: Time series differencing: Should I do both left AND righ hand side ?
To: [email protected]
> Dear colleagues,
>
> I'm contemplating differencing my dependent variable in a (pooled)
> time series analysis to deal with some issues of
> stationarity/autocorrelation.
>
> However, if I do this, do I need to difference my independent
> variables too? I remember from high school maths, that what you do on
> the left hand side you need to do on the right hand side. Does this
> apply to differenced time series data?
>
> Grateful for any explanation on this.
>
> Regards
>
> Joel Miller
> *
> * 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/
*
* 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/
