X12 additive with .spc file in EV (error)

[fmramos]

Hi!

I've been used to seasonal adjustment with x12 in EViews. Usually I have my own .spc file to implement multiplicative (with log transform) adjustments with some moving holidays (eg. Carnival in Brazil).

The problem is the additive adjustment. I think (I really think) there's some problem to additive adjustment when we input a spc file (eg. of command: ser01.x12(sfile="c:\exemple.spc",save="d11 d12")).

EV returns me the message that 'multiplicative only with non-zero/negative data'. But the .spc file contains an additive spec. Moreover, when I do it by buttons, I can implement with no problems additive adjustment (to series with negative values, for instance).

Tks in advance.

Fabio

[EViews Gareth]

Could you post your .spc file?

[fmramos]

The X12 .spc generated by EViews 'buttons', followed by my .spc one (and used by command line 'series.x12'). The differences between the two ones are small, but u can try only the first one in a command mode and u'll get the same error message. If u prefer, I can send u the data in a xls format. Tks in advance, Fabio

PS. I tried it in two different EViews. The version 5.1 with my co-worker's PC and at mine's one (EV 6). The command usually applied is: 'ser01.x12(sfile="C:\teste.spc",save="d11 d12")' or something like that. In fact, everything works ok for only positive data in 'mult' mode.

---

series{
title = "SER01"
start = 1985.1
name = "SER01"
file = "C:\EV_TEMP\EVX12TMP.DAT"
}
automdl{
file="c:\x12a.mdl"
method=best
identify=all
}
x11{
mode = add
sigmalim = (1.5,2.5)
print = ( +ftestd8 +residualseasf +x11diag +qstat +specsa +specirr)
save = ( D10 D11 D12 D13)
savelog = (q,q2,fb1,fd8,msf)
}

------------------------------------------

transform {function=none}

automdl {
file="c:\x12a.mdl"
method=best
identify=all
}

outliers {types=all}

x11 {
mode=add
save = ( D10 D11 D12 D13)
}

[EViews Gareth]

Could you try it after installing the latest patch?

http://eviews.com/download/download.html

[fmramos]

Tks, but it doesn't work. I tried the same routine (additive) to the series in the form 'series ser02=(ser01^2)^(1/2)' and it works. The problem is the negative numbers in a programmed way.

Tks. Fabio

[EViews Gareth]

try adding mode=a to the EViews command:

ser01.x12(mode=a, sfile="C:\teste.spc",save="d11 d12")

[fmramos]

Great, Gareth. Now it's working. So, the 'mode=a' is only necessary for additive adjustments cause de multiplicative one (or log-additive) is the default?

Tks for the really fast fix.

Fabio