EViews.com

Dear eviews programming experts,

i have the following problem. I'm a beginner in eviews and have to program for my master thesis in econometrics. So far, i have modelled a vectorautoregression with three variables. Now, i want to calculate one-step-ahead and six-step-ahead forecasts for this model in order to get the forecast errors. For this purpose, the forecasts should be based on a recursive forecasting scheme.
I have tried to write a corresponding program. I'm relatively sure, that this program contains errors. First, i have seen, that only the last one-step-ahead-forecast was saved, for example in the stahl_bereinigt_0 series. How can I manage, that the forecasts over the entire loop are saved in table or a series? How can i store the forecasts? And second, is the conversion from the system to the model right, because i simply defined a system, estimated the system, and then built the model. Third, are there other problems with my program? I thank you very much in advance. Below the code:

wfcreate Masterarbeit m "1991-01" "2012-02"
' creates workfile

!first=@dateval("1991-01")
!last=@dateval("2012-02")

!varnr = 3

matrix(254,!varnr) datamat
' generates matrix for data

datamat.read(a2,s=sheet2) "E:\Empirische Wirtschaftsforschung\Masterarbeit\Datensatz Masterarbeit.xls"
' read the data from the excel dataset into a data matrix

series stahl_bereinigt
series pkw_bereinigt
series logmaschinenbau
' generates series for the different variables

group variables stahl_bereinigt pkw_bereinigt logmaschinenbau
' generates a group including the individual series

mtos(datamat,variables)
' convert the data matrix to the group defined above

genr DS2 = @seas(2)-1/12
genr DS3 = @seas(3)-1/12
genr DS4 = @seas(4)-1/12
genr DS5 = @seas(5)-1/12
genr DS6 = @seas(6)-1/12
genr DS7 = @seas(7)-1/12
genr DS8 = @seas(8)-1/12
genr DS9 = @seas(9)-1/12
genr DS10 = @seas(10)-1/12
genr DS11 = @seas(11)-1/12
genr DS12 = @seas(12)-1/12
' generates centered seasonal dummies

genr DLS_masch_1 = @recode(@date>@dateval("1992-11"),1,0)
genr DLS_pkw = @recode(@date>@dateval("1992-12"),1,0)
genr DLS_stahl = @recode(@date>@dateval("2008-10"),1,0)
genr DLS_masch_2 = @recode(@date>@dateval("2008-12"),1,0)
genr DLS_masch_3 = @recode(@date>@dateval("2009-03"),1,0)
' generates dummies for the level shifts

system VAR_system
VAR_system.append stahl_bereinigt = C(1)+C(2)*@trend(1991M01)+C(3)*stahl_bereinigt(-1)+C(4)*stahl_bereinigt(-2)+C(5)*stahl_bereinigt(-3)+C(6)*pkw_bereinigt(-3)+C(7)*pkw_bereinigt(-5)+C(8)*pkw_bereinigt(-6)+C(9)*pkw_bereinigt(-7)+C(10)*logmaschinenbau(-2)+C(11)*logmaschinenbau(-3)+C(12)*logmaschinenbau(-5)+C(13)*logmaschinenbau(-6)+C(14)*DS2+C(15)*DS7+C(16)*DS8+C(17)*DS9+C(18)*DS11+C(19)*DS12+C(20)*DLS_stahl
VAR_system.append pkw_bereinigt = C(21)+C(22)*@trend(1991M01)+C(23)*stahl_bereinigt(-1)+C(24)*stahl_bereinigt(-2)+C(25)*stahl_bereinigt(-6)+C(26)*stahl_bereinigt(-7)+C(27)*pkw_bereinigt(-2)+C(28)*pkw_bereinigt(-3)+C(29)*pkw_bereinigt(-5)+C(30)*pkw_bereinigt(-7)+C(31)*logmaschinenbau(-4)+C(32)*logmaschinenbau(-5)+C(33)*logmaschinenbau(-6)+C(34)*logmaschinenbau(-7)+C(35)*DS2+C(36)*DS3+C(37)*DS4+C(38)*DS8+C(39)*DS9+C(40)*DS10+C(41)*DS11+C(42)*DS12+C(43)*DLS_pkw
VAR_system.append logmaschinenbau = C(44)+C(45)*@trend(1991M01)+C(46)*stahl_bereinigt(-1)+C(47)*stahl_bereinigt(-5)+C(48)*stahl_bereinigt(-7)+C(49)*pkw_bereinigt(-1)+C(50)*pkw_bereinigt(-5)+C(51)*pkw_bereinigt(-6)+C(52)*pkw_bereinigt(-7)+C(53)*logmaschinenbau(-1)+C(54)*logmaschinenbau(-2)+C(55)*logmaschinenbau(-3)+C(56)*DS2+C(57)*DS3+C(58)*DS4+C(59)*DS5+C(60)*DS6+C(61)*DS7+C(62)*DS8+C(63)*DS9+C(64)*DS10+C(65)*DS11+C(66)*DS12+C(67)*DLS_masch_1+C(68)*DLS_masch_2+C(69)*DLS_masch_3
' defines the systems of equations of the VAR model

!obstotal = @obssmpl
' counting the number of observations in the sample

!window = 30
' sets the forecast window

!hori = 1
' sets the forecast horizon

!foreper = !window-!hori+1
' determines the length of the forecast period and hence the series of forecasts

matrix(!foreper,!varnr) foremat
' generates matrix for the model forecasts

for !j=1 to !foreper step 1
' starts a loop for the calculation of the model forecasts

smpl "1991-01" "2012-02"-!window+!j-1

VAR_system.sur
' estimates the VAR model

VAR_system.makemodel(VAR_model)
'transforms the estimated VAR system into a model

smpl "2012-02"-!window+!j "2012-02"-!window+!j+!hori-1
' sets the forecast period

VAR_model.solve
' calculates forecasts by solving the specified VAR model

next

Yohan,

I encountered this problem in a single equation setting. What needs to be done is to store the forecast in a temporary series first and then retrieve the forecast in the period of interest. Parts of the code that I essentially followed is in this thread.

Now, I'm also encountering the same problem in a multiple equation case. It's a matter of finding a way to store the model solution in a temporary series so that the same process can be done. So, after the VAR_model.solve line, an additional line that copies the 6-step ahead solutions to a number of holding series. Unfortunately, I haven't figured out how to do this! I'd appreciate it if you could share the code if you figure it out. Thanks!

errorterm

Hello errorterm,

thanks for your reply, which was the starting point for my idea. Perhaps, i haven't found the easiest way to store the forecasts, but my program is suitable for one-step-ahead as well as more-step-ahead-forecasts. I would thank you, if you can further check whether this program is absolutely correct and test it with your own data. I have calculated several forecasts from the eviews button mode and the resulting forecasts are the same as those of my program.

The code, but remember, that i haven't commented out my changes yet. I will do this later.

wfcreate Masterarbeit m 1991m1 2012m2
' creates workfile

!varnr = 3
' counts the number of variables in the model

matrix(254,!varnr) datamat
' generates matrix for data

datamat.read(a2,s=sheet2) "E:\Empirische Wirtschaftsforschung\Masterarbeit\Datensatz Masterarbeit.xls"
' read the data from the excel dataset into a data matrix

series stahl_bereinigt
series pkw_bereinigt
series logmaschinenbau
' generates series for the different variables

group variables stahl_bereinigt pkw_bereinigt logmaschinenbau
' generates a group including the individual series

mtos(datamat,variables)
' converts the data matrix to the group defined above

genr DS_2 = @seas(2)-1/12
genr DS_3 = @seas(3)-1/12
genr DS_4 = @seas(4)-1/12
genr DS_5 = @seas(5)-1/12
genr DS_6 = @seas(6)-1/12
genr DS_7 = @seas(7)-1/12
genr DS_8 = @seas(8)-1/12
genr DS_9 = @seas(9)-1/12
genr DS_10 = @seas(10)-1/12
genr DS_11 = @seas(11)-1/12
genr DS_12 = @seas(12)-1/12
' generates centered seasonal dummies

genr DLS_MAB_1 = @recode(@date>=@dateval("1992m12"),1,0)
genr DLS_PKW = @recode(@date>=@dateval("1993m1"),1,0)
genr DLS_STA = @recode(@date>=@dateval("2008m11"),1,0)
genr DLS_MAB_2 = @recode(@date>=@dateval("2009m1"),1,0)
genr DLS_MAB_3 = @recode(@date>=@dateval("2009m4"),1,0)
' generates dummies for the level shifts

' defines the systems of equations of the VAR model

!obstotal = @obssmpl
' counting the number of observations in the sample

!window = 50
' sets the forecast window

!hori = 1
' sets the forecast horizon

!foreper = !window-!hori+1
' determines the length of the forecast period and hence the series of forecasts

matrix(!foreper,!obstotal) dummymat_sta
matrix(!foreper,!obstotal) dummymat_pkw
matrix(!foreper,!obstotal) dummymat_mab
' generates matrix for the model forecasts

matrix(!foreper,!varnr) foremat

scalar DAT_MAB_3
scalar DAT_MAB_2
scalar DAT_STA
scalar endsample
scalar endrange

DAT_MAB_3 = @dtoo("2009m4")
DAT_MAB_2 = @dtoo("2009m1")
DAT_STA = @dtoo("2008m11")
endrange = @dtoo("2012m2")

for !j=1 to !foreper step 1
' starts a loop for the calculation of the model forecasts

smpl 1991m1 2012m2-!window+!j-1
' sets the estimation period for the model
endsample = endrange-!window+!j-1

group variables stahl_bereinigt pkw_bereinigt logmaschinenbau

if endsample>=DAT_MAB_3 then

system VAR_system
VAR_system.append stahl_bereinigt = C(1)+C(2)*@trend(1991M01)+C(3)*stahl_bereinigt(-1)+C(4)*stahl_bereinigt(-2)+C(5)*stahl_bereinigt(-3)+C(6)*pkw_bereinigt(-3)+C(7)*pkw_bereinigt(-5)+C(8)*pkw_bereinigt(-6)+C(9)*pkw_bereinigt(-7)+C(10)*logmaschinenbau(-2)+C(11)*logmaschinenbau(-3)+C(12)*logmaschinenbau(-5)+C(13)*logmaschinenbau(-6)+C(14)*DS_2+C(15)*DS_7+C(16)*DS_8+C(17)*DS_9+C(18)*DS_11+C(19)*DS_12+C(20)*DLS_STA
VAR_system.append pkw_bereinigt = C(21)+C(22)*@trend(1991M01)+C(23)*stahl_bereinigt(-1)+C(24)*stahl_bereinigt(-2)+C(25)*stahl_bereinigt(-6)+C(26)*stahl_bereinigt(-7)+C(27)*pkw_bereinigt(-2)+C(28)*pkw_bereinigt(-3)+C(29)*pkw_bereinigt(-5)+C(30)*pkw_bereinigt(-7)+C(31)*logmaschinenbau(-4)+C(32)*logmaschinenbau(-5)+C(33)*logmaschinenbau(-6)+C(34)*logmaschinenbau(-7)+C(35)*DS_2+C(36)*DS_3+C(37)*DS_4+C(38)*DS_8+C(39)*DS_9+C(40)*DS_10+C(41)*DS_11+C(42)*DS_12+C(43)*DLS_PKW
VAR_system.append logmaschinenbau = C(44)+C(45)*@trend(1991M01)+C(46)*stahl_bereinigt(-1)+C(47)*stahl_bereinigt(-5)+C(48)*stahl_bereinigt(-7)+C(49)*pkw_bereinigt(-1)+C(50)*pkw_bereinigt(-5)+C(51)*pkw_bereinigt(-6)+C(52)*pkw_bereinigt(-7)+C(53)*logmaschinenbau(-1)+C(54)*logmaschinenbau(-2)+C(55)*logmaschinenbau(-3)+C(56)*DS_2+C(57)*DS_3+C(58)*DS_4+C(59)*DS_5+C(60)*DS_6+C(61)*DS_7+C(62)*DS_8+C(63)*DS_9+C(64)*DS_10+C(65)*DS_11+C(66)*DS_12+C(67)*DLS_MAB_1+C(68)*DLS_MAB_2+C(69)*DLS_MAB_3

else
if endsample>=DAT_MAB_2 and endsample<DAT_MAB_3 then

system VAR_system
VAR_system.append stahl_bereinigt = C(1)+C(2)*@trend(1991M01)+C(3)*stahl_bereinigt(-1)+C(4)*stahl_bereinigt(-2)+C(5)*stahl_bereinigt(-3)+C(6)*pkw_bereinigt(-3)+C(7)*pkw_bereinigt(-5)+C(8)*pkw_bereinigt(-6)+C(9)*pkw_bereinigt(-7)+C(10)*logmaschinenbau(-2)+C(11)*logmaschinenbau(-3)+C(12)*logmaschinenbau(-5)+C(13)*logmaschinenbau(-6)+C(14)*DS_2+C(15)*DS_7+C(16)*DS_8+C(17)*DS_9+C(18)*DS_11+C(19)*DS_12+C(20)*DLS_STA
VAR_system.append pkw_bereinigt = C(21)+C(22)*@trend(1991M01)+C(23)*stahl_bereinigt(-1)+C(24)*stahl_bereinigt(-2)+C(25)*stahl_bereinigt(-6)+C(26)*stahl_bereinigt(-7)+C(27)*pkw_bereinigt(-2)+C(28)*pkw_bereinigt(-3)+C(29)*pkw_bereinigt(-5)+C(30)*pkw_bereinigt(-7)+C(31)*logmaschinenbau(-4)+C(32)*logmaschinenbau(-5)+C(33)*logmaschinenbau(-6)+C(34)*logmaschinenbau(-7)+C(35)*DS_2+C(36)*DS_3+C(37)*DS_4+C(38)*DS_8+C(39)*DS_9+C(40)*DS_10+C(41)*DS_11+C(42)*DS_12+C(43)*DLS_PKW
VAR_system.append logmaschinenbau = C(44)+C(45)*@trend(1991M01)+C(46)*stahl_bereinigt(-1)+C(47)*stahl_bereinigt(-5)+C(48)*stahl_bereinigt(-7)+C(49)*pkw_bereinigt(-1)+C(50)*pkw_bereinigt(-5)+C(51)*pkw_bereinigt(-6)+C(52)*pkw_bereinigt(-7)+C(53)*logmaschinenbau(-1)+C(54)*logmaschinenbau(-2)+C(55)*logmaschinenbau(-3)+C(56)*DS_2+C(57)*DS_3+C(58)*DS_4+C(59)*DS_5+C(60)*DS_6+C(61)*DS_7+C(62)*DS_8+C(63)*DS_9+C(64)*DS_10+C(65)*DS_11+C(66)*DS_12+C(67)*DLS_MAB_1+C(68)*DLS_MAB_2

else
if endsample>=DAT_STA and endsample<DAT_MAB_2 then

system VAR_system
VAR_system.append stahl_bereinigt = C(1)+C(2)*@trend(1991M01)+C(3)*stahl_bereinigt(-1)+C(4)*stahl_bereinigt(-2)+C(5)*stahl_bereinigt(-3)+C(6)*pkw_bereinigt(-3)+C(7)*pkw_bereinigt(-5)+C(8)*pkw_bereinigt(-6)+C(9)*pkw_bereinigt(-7)+C(10)*logmaschinenbau(-2)+C(11)*logmaschinenbau(-3)+C(12)*logmaschinenbau(-5)+C(13)*logmaschinenbau(-6)+C(14)*DS_2+C(15)*DS_7+C(16)*DS_8+C(17)*DS_9+C(18)*DS_11+C(19)*DS_12+C(20)*DLS_STA
VAR_system.append pkw_bereinigt = C(21)+C(22)*@trend(1991M01)+C(23)*stahl_bereinigt(-1)+C(24)*stahl_bereinigt(-2)+C(25)*stahl_bereinigt(-6)+C(26)*stahl_bereinigt(-7)+C(27)*pkw_bereinigt(-2)+C(28)*pkw_bereinigt(-3)+C(29)*pkw_bereinigt(-5)+C(30)*pkw_bereinigt(-7)+C(31)*logmaschinenbau(-4)+C(32)*logmaschinenbau(-5)+C(33)*logmaschinenbau(-6)+C(34)*logmaschinenbau(-7)+C(35)*DS_2+C(36)*DS_3+C(37)*DS_4+C(38)*DS_8+C(39)*DS_9+C(40)*DS_10+C(41)*DS_11+C(42)*DS_12+C(43)*DLS_PKW
VAR_system.append logmaschinenbau = C(44)+C(45)*@trend(1991M01)+C(46)*stahl_bereinigt(-1)+C(47)*stahl_bereinigt(-5)+C(48)*stahl_bereinigt(-7)+C(49)*pkw_bereinigt(-1)+C(50)*pkw_bereinigt(-5)+C(51)*pkw_bereinigt(-6)+C(52)*pkw_bereinigt(-7)+C(53)*logmaschinenbau(-1)+C(54)*logmaschinenbau(-2)+C(55)*logmaschinenbau(-3)+C(56)*DS_2+C(57)*DS_3+C(58)*DS_4+C(59)*DS_5+C(60)*DS_6+C(61)*DS_7+C(62)*DS_8+C(63)*DS_9+C(64)*DS_10+C(65)*DS_11+C(66)*DS_12+C(67)*DLS_MAB_1

else
if endsample<DAT_STA then

system VAR_system
VAR_system.append stahl_bereinigt = C(1)+C(2)*@trend(1991M01)+C(3)*stahl_bereinigt(-1)+C(4)*stahl_bereinigt(-2)+C(5)*stahl_bereinigt(-3)+C(6)*pkw_bereinigt(-3)+C(7)*pkw_bereinigt(-5)+C(8)*pkw_bereinigt(-6)+C(9)*pkw_bereinigt(-7)+C(10)*logmaschinenbau(-2)+C(11)*logmaschinenbau(-3)+C(12)*logmaschinenbau(-5)+C(13)*logmaschinenbau(-6)+C(14)*DS_2+C(15)*DS_7+C(16)*DS_8+C(17)*DS_9+C(18)*DS_11+C(19)*DS_12
VAR_system.append pkw_bereinigt = C(21)+C(22)*@trend(1991M01)+C(23)*stahl_bereinigt(-1)+C(24)*stahl_bereinigt(-2)+C(25)*stahl_bereinigt(-6)+C(26)*stahl_bereinigt(-7)+C(27)*pkw_bereinigt(-2)+C(28)*pkw_bereinigt(-3)+C(29)*pkw_bereinigt(-5)+C(30)*pkw_bereinigt(-7)+C(31)*logmaschinenbau(-4)+C(32)*logmaschinenbau(-5)+C(33)*logmaschinenbau(-6)+C(34)*logmaschinenbau(-7)+C(35)*DS_2+C(36)*DS_3+C(37)*DS_4+C(38)*DS_8+C(39)*DS_9+C(40)*DS_10+C(41)*DS_11+C(42)*DS_12+C(43)*DLS_PKW
VAR_system.append logmaschinenbau = C(44)+C(45)*@trend(1991M01)+C(46)*stahl_bereinigt(-1)+C(47)*stahl_bereinigt(-5)+C(48)*stahl_bereinigt(-7)+C(49)*pkw_bereinigt(-1)+C(50)*pkw_bereinigt(-5)+C(51)*pkw_bereinigt(-6)+C(52)*pkw_bereinigt(-7)+C(53)*logmaschinenbau(-1)+C(54)*logmaschinenbau(-2)+C(55)*logmaschinenbau(-3)+C(56)*DS_2+C(57)*DS_3+C(58)*DS_4+C(59)*DS_5+C(60)*DS_6+C(61)*DS_7+C(62)*DS_8+C(63)*DS_9+C(64)*DS_10+C(65)*DS_11+C(66)*DS_12+C(67)*DLS_MAB_1

endif
endif
endif
endif

' defines the system of model equations

VAR_system.sur(deriv=aa)
' estimates the VAR system with analytical derivatives

VAR_system.makemodel(VAR_model)
' transforms the estimated VAR system into a model

smpl 2012m2-!window+!j 2012m2-!foreper+!j
' sets the forecast period

VAR_model.solve
' calculates forecasts by solving the specified VAR model

smpl 1991m1 2012m2-!window+!j+!hori-1
series dummy_sta = stahl_bereinigt_0
series dummy_pkw = pkw_bereinigt_0
series dummy_mab = logmaschinenbau_0

for !i=1 to !obstotal step 1

dummymat_sta(!j,!i)=dummy_sta(!i)
dummymat_pkw(!j,!i)=dummy_pkw(!i)
dummymat_mab(!j,!i)=dummy_mab(!i)

next

foremat(!j,1) = dummymat_sta(!j,!obstotal-!foreper+!j)
foremat(!j,2) = dummymat_pkw(!j,!obstotal-!foreper+!j)
foremat(!j,3) = dummymat_mab(!j,!obstotal-!foreper+!j)

next

smpl 2012m2-!foreper+1 2012m2

series foresta
series forepkw
series foremab

group foreser foresta forepkw foremab

mtos(foremat,foreser)