Re: [Gretl-users] problem with system instrumental variables

Hello listers, greetings from Paris ! Still on my system estimation commands, I have a few difficulties with the use of instruments. I would like to estimate a system of two equations with one different variable and one instrumented variable (income). So I use the syntax for tsls to specifiy regressors and instruments on the same line : =================================================== script exerpt ============================================================== list socdemo = CHILD LOCATAIRE AUTOMOBILE system name=AIDS equation WLOG const lnPLOGCOMP lnPCARBCOMP lnDEPTOTR socdemo MRLOG ; const lnPLOGCOMP lnPCARBCOMP lnREVTOTUCR socdemo MRLOG equation WCAR const lnPLOGCOMP lnPCARBCOMP lnDEPTOTR socdemo MRCAR ; const lnPLOGCOMP lnPCARBCOMP lnREVTOTUCR socdemo MRCAR end system restrict AIDS b[1,3]-b[2,2]=0 end restrict estimate AIDS method=3sls =================================================== end script exerpt ============================================================== but the command fails as gretl says : "variable 0 duplicated in commands list" GRETL seems not to accept my instruments, in this AIDS model total expense (lnDEPTOTR) is instrumented by income (lnREVTOTUCR) ; all other variables being the same except the last ones (MRLOG and MRCAR) which are specific to each equation (and must remain different). If i use both MLOG and MCAR in each equation, system estimation runs ok but does not makes sense on a statistical perspective. Any advice from the list ?