Statalist


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

RE: st: xtdpd and dgmmiv: restricting the LDV instruments


From   "David M. Drukker, StataCorp" <[email protected]>
To   [email protected]
Subject   RE: st: xtdpd and dgmmiv: restricting the LDV instruments
Date   Wed, 25 Jun 2008 08:37:00 -0500

Tewodaj Mogues <[email protected]> is having trouble limiting the lags that
-xtdpd- uses to create moment conditions.

The short answer is that Tewodaj should specify numbers for the first and
last lags used, e.g. dgmmiv(y, lagrange(2 3)) will create moment conditions
using lags 2 and 3 and the differenced errors.

What follows is more a detailed answer.

Tewodaj was on the right path.  Reproducing -xtabond- and -xtdpdsys- results
with is a good place to start answering questions about the -xtdpd- syntax.
When using this strategy, remember that all three dynamic panel-data
commands have a footer that says which instruments were used.

Let's begin with an -xtabond- example that only uses the first 5 available
lags of the dependent variable to create instruments.

. webuse abdata

. xtabond n w k , maxldep(5)

Arellano-Bond dynamic panel-data estimation  Number of obs         =       751
Group variable: id                           Number of groups      =       140
Time variable: year
                                             Obs per group:    min =         5
                                                               avg =  5.364286
                                                               max =         7

Number of instruments =     28               Wald chi2(3)          =   1169.50
                                             Prob > chi2           =    0.0000
One-step results
------------------------------------------------------------------------------
           n |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
           n |
         L1. |   .4656755   .0601311     7.74   0.000     .3478208    .5835302
           w |  -.6198203   .0509389   -12.17   0.000    -.7196587   -.5199819
           k |   .3469762   .0307059    11.30   0.000     .2867938    .4071586
       _cons |   2.635807   .1895574    13.91   0.000     2.264281    3.007333
------------------------------------------------------------------------------
Instruments for differenced equation
        GMM-type: L(2/6).n
        Standard: D.w D.k
Instruments for level equation
        Standard: _cons

The output indicates that -xtabond- is using lags 2 through 6 to create
GMM-type instruments for the difference equation.  This implies that we
should specify the option -lagrange(2 6)- in the dgmmiv() option in -xtdpd-.
Here is an example

. xtdpd L(0/1).n w k , dgmmiv(n, lagrange(2 6)) div(w k)

Dynamic panel-data estimation                Number of obs         =       891
Group variable: id                           Number of groups      =       140
Time variable: year
                                             Obs per group:    min =         6
                                                               avg =  6.364286
                                                               max =         8

Number of instruments =     28               Wald chi2(3)          =   1169.50
                                             Prob > chi2           =    0.0000
One-step results
------------------------------------------------------------------------------
           n |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
           n |
         L1. |   .4656755   .0601311     7.74   0.000     .3478208    .5835302
           w |  -.6198203   .0509389   -12.17   0.000    -.7196587   -.5199819
           k |   .3469762   .0307059    11.30   0.000     .2867938    .4071586
       _cons |   2.635807   .1895574    13.91   0.000     2.264281    3.007333
------------------------------------------------------------------------------
Instruments for differenced equation
        GMM-type: L(2/6).n
        Standard: D.w D.k
Instruments for level equation
        Standard: _cons


-xtdpdsys- works analogously.  Specifying -maxldep(5)- causes -xtdpdsys- to
use the first 5 available lags, which will be lags 2 through 6.  Here is an
example.

. xtdpdsys n w k , maxldep(5)

System dynamic panel-data estimation         Number of obs         =       891
Group variable: id                           Number of groups      =       140
Time variable: year
                                             Obs per group:    min =         6
                                                               avg =  6.364286
                                                               max =         8

Number of instruments =     35               Wald chi2(3)          =   4644.61
                                             Prob > chi2           =    0.0000
One-step results
------------------------------------------------------------------------------
           n |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
           n |
         L1. |   .5292982   .0419181    12.63   0.000     .4471402    .6114562
           w |  -.6249517   .0509272   -12.27   0.000    -.7247673   -.5251362
           k |   .3358337   .0236543    14.20   0.000     .2894721    .3821954
       _cons |   2.577968   .1774118    14.53   0.000     2.230247    2.925688
------------------------------------------------------------------------------
Instruments for differenced equation
        GMM-type: L(2/6).n
        Standard: D.w D.k
Instruments for level equation
        GMM-type: LD.n
        Standard: _cons

The output says that -xtdpdsys- used lags 2 through 6 of -n- to create
instruments for the differenced errors. Here is an example in which
-xtdpd uses the same set of instruments.

. xtdpd L(0/1).n w k , dgmmiv(n, lagrange(2 6)) div(w k) lgmmiv(n) 

Dynamic panel-data estimation                Number of obs         =       891
Group variable: id                           Number of groups      =       140
Time variable: year
                                             Obs per group:    min =         6
                                                               avg =  6.364286
                                                               max =         8

Number of instruments =     35               Wald chi2(3)          =   4644.61
                                             Prob > chi2           =    0.0000
One-step results
------------------------------------------------------------------------------
           n |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
           n |
         L1. |   .5292982   .0419181    12.63   0.000     .4471402    .6114562
           w |  -.6249517   .0509272   -12.27   0.000    -.7247673   -.5251362
           k |   .3358337   .0236543    14.20   0.000     .2894721    .3821954
       _cons |   2.577968   .1774118    14.53   0.000     2.230247    2.925688
------------------------------------------------------------------------------
Instruments for differenced equation
        GMM-type: L(2/6).n
        Standard: D.w D.k
Instruments for level equation
        GMM-type: LD.n
        Standard: _cons

I hope that these examples are helpful.

     --David
     <[email protected]>
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index