Dear Statalist
in a panel test, to what extent is it acceptable a model with no lagged dep variable among the indep variables? Is it always necessary to include a lag dep variable? Moreover: what do you think about a model that has all variables (both y and x) at the same year? Please: is there a reference that clarify these issues? Thank you. * * 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/ 
<> The entire [XT] manual is your reference of choice, I would say, particularly since you have it in pdf form on your machine. Look at the examples in help xtreg, where you won`t see lagged values... HTH Martin Ursprüngliche Nachricht Von: [hidden email] [mailto:[hidden email]] Im Auftrag von Fabio Zona Gesendet: Freitag, 24. September 2010 12:04 An: [hidden email] Betreff: st: Panel data and lagged dep variable Dear Statalist in a panel test, to what extent is it acceptable a model with no lagged dep variable among the indep variables? Is it always necessary to include a lag dep variable? Moreover: what do you think about a model that has all variables (both y and x) at the same year? Please: is there a reference that clarify these issues? Thank you. * * 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/ 
In reply to this post by Fabio Zona
Good day,
No it's not always necessary as we have in this case the static panel data model, the inclusion of the lagged dependant variable among the explanatory gives as the dynamic approach. I hope those classe notes may give you some insights. http://www.cemfi.es/~arellano/staticpanelsclassnote.pdf Ama * * 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/ 
In reply to this post by Fabio Zona
Hi, my understanding is static model presents longterm/equilibrium relationship, while the dynamic model presents shortterm/disequilibrium relation.

In reply to this post by Fabio Zona
<>
On Sep 25, 2010, at 2:33 AM, Fabio wrote: > in a panel test, to what extent is it acceptable a model with no lagged dep variable among the indep variables? Is it always necessary to include a lag dep variable? Moreover: what do you think about a model that has all variables (both y and x) at the same year? > Please: is there a reference that clarify these issues? In terms of an economic model, if you write down y_{i,t} = f (X) where the Xs may contain current and past values, but not including lagged y, you are arguing that the data observed at each point in time represent equilibria. If you believe that it takes more than one of your time periods for economic agents to adjust to changes in the Xs (or to adjust to shocks), then the LDV is warranted. As you can always estimate the model (although not with fixed or random effects!) including the LDV, you can test whether it is warranted in your particular setup. If you do that, it would probably be a good idea to include time dummies as well, as the absence of those factors could confound the analysis of the LDV's significance. Kit Kit Baum  Boston College Economics & DIW Berlin  http://ideas.repec.org/e/pba1.html An Introduction to Stata Programming  http://www.statapress.com/books/isp.html An Introduction to Modern Econometrics Using Stata  http://www.statapress.com/books/imeus.html * * 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/ 
Hi Kit, as you indictaed "If you believe that it takes more than one of your time periods for economic agents to adjust to changes in the Xs (or to adjust to shocks), then the LDV is warranted." does that mean current value of Y is closely related with previous value of Y? e.g company's current income is closely related with previous income.

In reply to this post by Fabio Zona
<>
Hi Kit, as you indictaed "If you believe that it takes more than one of your time periods for economic agents to adjust to changes in the Xs (or to adjust to shocks), then the LDV is warranted." does that mean current value of Y is closely related with previous value of Y? e.g company's current income is closely related with previous income. In a predictive sense, if the coeff on the LDV is not statistically distinguishable from zero, then current income is not closely related to previous income, cet.par. But the real question is whether a change in X causes income to change only in the current period or whether it has effects on subsequent periods. If it does, then something like a partial adjustment process is in order, and the LDV should be included. You can't just look at the ACF of current income. KIt Kit Baum  Boston College Economics & DIW Berlin  http://ideas.repec.org/e/pba1.html An Introduction to Stata Programming  http://www.statapress.com/books/isp.html An Introduction to Modern Econometrics Using Stata  http://www.statapress.com/books/imeus.html * * 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/ 
Hi Kit, thank you very much for letting me understand LDV much better and deeper.
As you mentioned, the prupose of involving LDV is to account the subsequent influence of change in Xs. You also indicate we should include time dummies to in favor of LDV. My understanding of time dummies is to absorb timespecific changes (e.g shocks). i understand LDV and time dummies should be consistent, but exactely in what extant, the time dummies and LDV are consitent in the regression? Pan 
In reply to this post by Fabio Zona
<>
Hi Kit, thank you very much for letting me understand LDV much better and deeper. As you mentioned, the prupose of involving LDV is to account the subsequent influence of change in Xs. You also indicate we should include time dummies to in favor of LDV. My understanding of time dummies is to absorb timespecific changes (e.g shocks). i understand LDV and time dummies should be consistent, but exactely in what extant, the time dummies and LDV are consitent in the regression? If there are significant timespecific common factors (e.g., business cycle factors) appearing in the response variable, then by omitting time dummies you are misspecifying the model, and estimates of any coefficients (including the coefficient of the LDV) are likely to be biased and inconsistent. Include the dummies, do an Ftest for their joint significance, and if you don't need them, out they go. But determining whether a static or dynamic model is needed should be performed in the context of a wellspecified model. Kit * * 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/ 
Kit Baum replied to Pan Wang:
> If there are significant timespecific common factors (e.g., business cycle factors) appearing in the response variable, > then by omitting time dummies you are misspecifying the model, and estimates of any coefficients (including > the coefficient of the LDV) are likely to be biased and inconsistent. Include the dummies, do an Ftest for their > joint significance, and if you don't need them, out they go. But determining whether a static or dynamic model > is needed should be performed in the context of a wellspecified model. Absolutely right, of course. But frequently I find myself estimating unbalanced pooledpanel models with singleton unit dummies, which naturally drop out of the model. Consequently, I have to circumvent this issue by demeaning the explanatory variables within units and then reestimating. But, as you'll know, that typically comes at the cost of reduce the reducing the amount of variation in the response variable my model would have had a chance of explaining. Your answer to this might be: "Then collect more data". But, as you'll also know, that's not always possible.  Clive Nicholas [Please DO NOT mail me personally here, but at <[hidden email]>. Please respond to contributions I make in a list thread here. Thanks!] "My colleagues in the social sciences talk a great deal about methodology. I prefer to call it style."  Freeman J. Dyson. * * 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/ 
Powered by Nabble  Edit this page 