Re: st: xtabond - testing coefficients

[Giovanni Bruno <giovanni.bruno@uni-bocconi.it>]

At 10.46 17/06/02 +1000, you wrote:
> Hi
>
> I'm trying to better understand the 'xtabond' command. In particular, I'm 
> wondering if it is possible to carry out tests on individual variables 
> that are part of an xtabond model. Eg, I'm running the model:
>
> xtabond indmfp indgap , lags(2) pre(indpdfl1 indpdfl2 , lagstruct (0, 2)) 
> robust
>
> and I want to know if indpdfl1 and indpdfl2 are jointly significantly 
> different from zero? My attempts to do this through the 'test' command are 
> producing r(111) ('variable not found', etc) error messages. Is it 
> possible to conduct this type of test after 'xtabond'? (As I said, I'm 
> still trying to get head around how this model works.)

Tim

Perhaps you are failing to include the variables following -test- in first 
differences (and with the appropriate lags).

-xtabond- transforms your  model in first differences. So, if a variable x 
is an explanatory variable of your model (whether exogenous or 
predetermined), -xtabond- takes it in first differences and the variable 
actually recognized  by -test- is D.x. A log file of mine illustrating this 
follows.

. xtabond  LnNE Lnw, lags(2) pre( LnP90 LnVA)

Arellano-Bond dynamic panel data                Number of obs      =       672
Group variable (i): sector                      Number of groups   =        32

                                                Wald chi2(5)       =   6617.38

Time variable (t): year                         min number of obs  =        21
                                                max number of obs  =        21
                                                mean number of obs =        21

One-step results
------------------------------------------------------------------------------
LnNE         |      Coef.   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
LnNE         |
          LD |   .8781118   .0341416    25.72   0.000     .8111955    .9450282
         L2D |  -.0778004   .0322619    -2.41   0.016    -.1410327   -.0145682
LnP90        |
          D1 |  -.0263407   .0081334    -3.24   0.001    -.0422819   -.0103996
LnVA         |
          D1 |   .1523871   .0131332    11.60   0.000     .1266465    .1781277
Lnw          |
          D1 |  -.1606018   .0131561   -12.21   0.000    -.1863874   -.1348162
_cons        |  -.0001503   .0009943    -0.15   0.880     -.002099    .0017984
------------------------------------------------------------------------------
Sargan test of over-identifying restrictions:
         chi2(754) =   768.71     Prob > chi2 = 0.3469

Arellano-Bond test that average autocovariance in residuals of order 1 is 0:
         H0: no autocorrelation   z = -15.33   Pr > z = 0.0000
Arellano-Bond test that average autocovariance in residuals of order 2 is 0:
         H0: no autocorrelation   z =   1.14   Pr > z = 0.2547

. test D.LnVA + D.LnP90=0

 ( 1)  D.LnP90 + D.LnVA = 0.0

           chi2(  1) =   70.70
         Prob > chi2 =    0.0000


I hope this helps
Giovanni


*
*   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/