? ============================ cov1ssys =========================
proc cov1ssys neq ivs imask an1 ;

? This procedure makes a covariance matrix for the system GMM
? estimator. The structure of the covariance matrix is the same
? as one in p9 in "Panel Data estimation using DPD for Ox",
? edited by Doornik, Arellano, and Bond (June 20, 2001), except
? for the scaling.
? The system GMM estimator is introduced in Blundell and Bond
? (Journal of Econometrics, 1998) in detail.
?
? That is, for example, when T= 5,
?
?      |  2 -1  0  0  0 |
?      | -1  2 -1  0  0 |
? HD = |  0 -1  2 -1  0 |
?      |  0  0 -1  2 -1 |
?      |  0  0  0 -1  2 |
?
?      |  1  0  0  0  0 |
?      |  0  1  0  0  0 |
? I =  |  0  0  1  0  0 |
?      |  0  0  0  1  0 |
?      |  0  0  0  0  1 |
?
?      |  0  0  0  0  0 |
?      |  0  0  0  0  0 |
? O =  |  0  0  0  0  0 |
?      |  0  0  0  0  0 |
?      |  0  0  0  0  0 |
?
? H =  |  HD  O |
?      |  O   I |
?
? Remark: Write the ivs list in the following order:
?         [Instruments List for Difference Equation]
?         [Instruments List for Level Equation]
?         and
?         in gmm command, write the equation list
?         in the following order:
?         gmm(...) difference_eq_list level_eq_list ;
?
? This procedure is made by Yoshitsgu Kitazawa
? (probably August/03), refering to the cov1step.tsp
? procedure designed by Clint Cummins (12/17/97).

?  neq   - number of equations (time periods (T-2)*2),
?         where first T-2 is number of 1st dif eq while
?         last T-2 is number of level eq.
?  ivs   - a list of instrumental variables.
?  imask  - a ninst by neq matrix of zeros and ones that specifies 
?          which instruments are used for which equation. 
?  an1   - the covariance matrix estimate (returned). 

   local h i i1 diagm ;
   local neqh neqhp1 ;

?  Make the theoretical covariance matrix of the equation residuals 
?  (MA(1)).

?  In TSP 4.4 (rev. 8/97), this can be done with the BAND option to
?  MFORM:
?mmake i 2 -1;  ? vector of band values, starting with main diagonal
?mform(band,nrow=neq) h = i;

  set neqh = neq / 2 ;
  set neqhp1 = neqh + 1 ;

  mat h = 2*ident(neq) ;
  if neq .gt. 2 ; then ; do ;
    do i = 2 neqh ;
      set i1 = i-1 ;
      set h(i,i1) = -1 ;
    enddo ;
    do i = neqhp1 neq ;
      set h(i,i) = h(i,i) / 2 ;
    enddo ;
  enddo ;
  
  moment(noprint) ivs ;  ? @mom = (z'z)/@nob
  mat an1 = sym(h) # @mom ;
  ?
  ?  Apply the mask to an1, to work around a bug in GMM
  ?  that was fixed on 12/17/97
  ?
  mat diagm = diag(vec(imask));
  mat an1 = diagm'an1*diagm;

endproc cov1ssys ;
? ==================================================================