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

st: xtdpd and dgmmiv: restricting the LDV instruments

From   "Mogues, Tewodaj (IFPRI)" <T.MOGUES@CGIAR.ORG>
To   <>
Subject   st: xtdpd and dgmmiv: restricting the LDV instruments
Date   Tue, 24 Jun 2008 19:25:26 -0400

Dear Statalist:

I sought to estimate a dynamic model using xtdpd in Stata 10, and wanted to restrict the number of lags of the dependent variable that would be used as instruments. Let Y be my dependent variable. Without restricting, the command would include -dgmmiv(Y)- . To restrict the number of lags, I changed this to -dgmmiv(Y, lagrange(. 5))- , but this gives me an error command. In fact, even when I include -dgmmiv(Y, lagrange(. .))-  I get an error. Although the Stata Options explain that the lag range can be specified in this way. I don't have a problem when I use -lagrange- to specify the lag range of lags of predetermined or endogenous variables. E.g. let Z be a predetermined variable. Then, I can include -dgmmiv(Z, lagrange(1 .))- in the command to introduce lags of Z as instruments for the differenced equation, without getting an error sign. Why does this not work for restricting the instrument set emerging from lags of the dependent var?
Interestingly, I can do exactly this using xtdpdsys, by specifying for example:

xtdpdsys Y X, maxldep(5)

where X is an exogenous variable. To reproduce this using xtdpd, I would have had to use:

xtdpd L(0/1).Y X, div(X) dgmmiv(Y, lagrange(. 5)) lgmmiv(Y)

but this, as I mentioned, gives an error message, because the command somehow refuses -lagrange()- when it is embedded in the dgmmiv() command for Y, while it doesn't refuse it in a dgmmiv() command for other variables. Why is this? Insights very welcome!


*   For searches and help try:

© Copyright 1996–2017 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index