Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: RE: One v. two-step ECMs


From   Neesha Harnam <neesha.harnam@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: RE: One v. two-step ECMs
Date   Fri, 3 Feb 2012 08:43:00 +0700

Dear Eric,

Thank you for your response - that answers my question and more! I
will look up the references you suggested.

Best
Neesha



On Thu, Feb 2, 2012 at 6:26 PM, DE SOUZA Eric
<eric.de_souza@coleurope.eu> wrote:
> The equivalence between the one-step and two-step methods is an asymptotic result. Banerjee, Dolado, Galbraith and Hendry in their 1993 book (I don't have the title at hand) on cointegration have shown that for even large finite samples the bias in the static equation estimated in the first step can be quite large.
>
> You can also Google for a paper by Hendry and Juselius, Explaining Cointegration, Part 1
>
>
> Eric de Souza
> Professor, European Economic Studies
> Director, Library
> College of Europe
> Dyver 11
> BE-8000 Brugge (Bruges)
> Belgium
> Tel.: (32.(0)50) 47 72 23
> Fax:: (32 (0)50) 47 71 10
> http://www.coleurope.eu
>
>
> -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-statalist@hsphsun2.harvard.edu] On Behalf Of Neesha Harnam
> Sent: 02 February 2012 11:18
> To: statalist@hsphsun2.harvard.edu
> Subject: st: One v. two-step ECMs
>
> Dear statalist users,
>
> I have an error-correction model for analyzing panel data (n=70, t=30) and am having difficulty figuring out if and where I went wrong with my coding. The results for the one- and two-step ECMs are very different (especially in terms of coefficients and significance) depending on the model used, and my understanding was that they estimated the same thing. The differences also appear in the indicator variables once I add them to my model. GDP per capita is I(1) but stationary in differences, while H1, a health outcome, is I(0). IND1-4 are dummy variables that show whether a country experienced a particular event in a given year.
>
> Two-step process:
> regress H1 log_gdppc, vce(cluster countrycode) predict e, resid regress DH1 Dlog_gdppc L.e, vce(cluster countrycode)
>
> Two-step with indicators:
> regress DH1 Dlog_gdppc L.e IND1 IND2 IND3 IND4, vce(cluster countrycode)
>
> One-step process:
> regress DH1 Dlog_gdppc L.H1 L.log_gdppc, vce(cluster countrycode)
>
> One-step with indicators:
> regress DH1 Dlog_gdppc L.H1 L.log_gdppc IND1 IND2 IND3 IND4, vce(cluster countrycode)
>
> Any help would be greatly appreciated.
>
> Thank you,
> Neesha
>
>
> *
> *   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/

*
*   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   |   Site index