try the follwoing code, data from http://www.eviews.com/Learning/AllTutorials.zip
Code: Select all
%pt = "Z:\2013\DL\EV8\Forecasting\"
%wf = %pt + "Results.wf1"
wfopen %wf
'pptx p48-49, Example 10
smpl sample_est
equation eq10.ls d(payroll) c ism tbill3m @expand(@month, @droplast)
smpl sample_for
freeze(gf10df) eq10.forecast(g,d) eq10_diff_dyn2 eq10_diff_dyn_sd2
freeze(gf10sf) eq10.fit(g,d) eq10_diff_stat2 eq10_diff_stat_sd2
graph gf_diff2.line eq10_diff_dyn2 eq10_diff_stat2
'remark: by forecasting the entire expression rather than the normalized variable
'1 eq10_diff_dyn2 and eq10_diff_stat2 should be equal, and both have the same sample length, not that one longer than the other
'2 for the sd, eq10_diff_dyn_sd2 and eq10_diff_stat_sd2 are equal and the equal sample length
'now we turn to forcast diff by (1) forecast level, (2) compute the diff
freeze(gf10df0) eq10.forecast(g) eq10_dyn eq10_dyn_sd
freeze(gf10sf0) eq10.fit(g) eq10_stat eq10_stat_sd
genr eq10_diff_dyn3 = d(eq10_dyn)
genr eq10_diff_stat3 = d(eq10_stat)
graph gf_diff3.line eq10_diff_dyn3 eq10_diff_stat3
show gf_diff2 gf_diff3
'remark: here dynamic is diff from the static forecast. Compare gf10df0 and gf10sf0, the sd of dynamic is increasing