SR3 Error. Tried to Use Series Number 125, Only 23 Are Available

[GraemeC]

Hi Tom , sorry to bother you but I am having trouble with the above error popping up and just can't figure out where the error is. Its a error correction equation from one of the Reserve Bank of Australia's forecasting / policy testing models(Martin) which I am trying to replicate. A cut down version of the RBA's data file is attached relating to the variables in the household consumption equation to be estimated. I must be making some fundamental error in my specification I think. I know the equation and data works as I have managed to run it in Gretl and Eviews test version with similar coefficient estimates to the RBA's. The coefficient names are the ones the RBA used in the equation specification in Eviews. I would greatly appreciate any help / guidedance you could provide. My preference is to create and estimate the full model in RATS.
Rats Code:

Code: Select all

OPEN DATA "C:\Users\graeme\Desktop\Martin\testcons_martin_public.xlsx"
CALENDAR(Q) 1958:1 2029:04
allocate 2029:04
DATA(FORMAT=XLSX,ORG=COLUMNS) 1958:01 2029:04 rc dum_rc hdy hnw $
rcr hcoe hoy tdlla tdllpop tdllhpp lur
set lrc = log(rc)
set lhcoe = log(hcoe)
set  lhdy = log(hdy)
set lhnw = log(hnw)
set lhoy = log(hoy)
diff lur / ldur
diff lrc / ldrc
diff lhcoe / ldhcoe
diff lhnw / ldhnw
diff lhoy / ldhoy
diff lur / dlur
set  tdcomp = tdlla + tdllpop + tdllhpp
nonlin c1 c11 c2 c31 c32 c60 c50
frml olsc  ldrc = c1 +c50*dum_rc-c11*(lrc(1)-c2*lhdy(1)-(1-c2)*lhnw(2)+ 0.5/100*rcr(1))+ 0.15*ldhcoe + c31*ldhoy(2) + c32*ldhnw(1) + (1-0.15-c31-c32)*(tdlla(1) + tdllpop(1) + tdllhpp(1)) + c60*dlur(2)/100
compute c1 = 0.30
compute c2 = 0.83
compute c50  = 0.008
compute c11 = 0.077
compute c32 = 0.059
compute c31 = 0.012
compute c60 = -0.417
nlls(frml=olsc) 1989:1 2018:2

[TomDoan]

There are a couple of issues. First, the syntax in RATS for lags of series is series{lag} not series(lag). So your FRML should read:

frml olsc ldrc = c1 +c50*dum_rc-c11*(lrc{1}-c2*lhdy{1}-(1-c2)*lhnw{2}+ 0.5/100*rcr{1})+ 0.15*ldhcoe + $
c31*ldhoy{2} + c32*ldhnw{1} + (1-0.15-c31-c32)*(tdlla{1} + tdllpop{1} + tdllhpp{1}) + c60*dlur{2}/100

The other is that the syntax of NLLS has

NLLS depvar start end

not

NLLS start end

While you included the depvar in the frml definition, you still need to allow at least for a slot for it on the NLLS so either

nlls(frml=olsc) * 1989:1 2018:2

or

nlls(frml=olsc) ldrc 1989:1 2018:2

[GraemeC]

Thank you so much Tom, those changes all worked and the equation estimated correctly with coefficient estimates similar to the RBA paper. Really appreciate your help. Regards Graeme