Re: st: Invalid Lags message - gmm, a system of two simultaneous equations

["Brian P. Poi" <bpoi@stata.com>]

On Mon, 15 Feb 2010, Ari Dothan wrote:

> Dear Statalist participants,
>
> The two equations are:
>
> (1) C_t = B1*P_t-1 + B2*C_t-1 + B3*A_t-1 +  B4*CM_t-1 + alpha  +  u_it
> (2) P_t = B5*C_t-1 + B6*P_t-1 + B7*PM_t-1 + alpha1 +  v_it
>
> The first stage equation alone in Stata code:
> gmm (D.C_t - {rho}*LD.C_t -{xb: LD.P_t LD.A_t  LD.CM_t}), xtinstruments(C_t,
> lags (2/4)) instruments (LD.P_t LD.A_t  LD.CM_t , noconstant)
> deriv(/rho=-1*LD.w10_or)  deriv (/xb=-1) winitial (xt D) onestep
>
> Both stages are combined in Stata code:
> gmm (eq:D.C_t - {rho1}*LD.C_t -{xb: LD.P_t LD.A_t  LD.CM_t}) (eq2:D.P_t -
> {rho1}* LD.P_t - {xc: LD.C_t  LD.PM_t-1}),
> xtinstruments(eq1:C_t,lags(2/4)) (eq2:P_t, lags(2/4)) instruments
> (eq1:LD.(P_t - A_t) LD.CM_t) instruments(eq2:LD.C_t LD.CM_t) deriv(eq1:/rho
> = - 1*LD.C_t) (eq2:/rho1 = - 1*LD.P_t) deriv(eq1:/xb = -1) (eq2:/xc = -1)
> winitial(xt D) wmatrix(robust) onestep
>
> I am aware of the fact that I am trying to run 2 level equations and two
> difference equations, and hope that I understand the procedure as explained
> in the Stata manual on gmm.
>
> The problem is that I get an "invalid lags" message.


With dynamic panel models, -gmm- allows you to specify at most two 
equations.  If you specify two equations, then one equation should be in 
differenced form and the other in levels form.  In that case the 
winitial() option should either be winitial(xt DL) or winitial(xt LD) 
depending on whether you specify the differenced or level equation first.

  -- Brian Poi
  -- bpoi@stata.com
*
*   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/