single equation GMM, too much different results

[maxchen]

Ver: EViews7Patch_062211.exe

Code: Select all

%ex7 = "\Example Files\EV7 Manual Data\" %ch20 = "Chapter 20 - Instrumental Variables and GMM" %wf = @evpath + %ex7 +%ch20 +"\klein.wf1" wfopen %wf equation eq02 c = 0 eq02.gmm(method=cue,instwgt=hac,instlag=1,instkern=thann,instbw=andrews,nodf,deriv=aa) i c y y(-1) k(-1) @ c p(-1) k(-1) x(-1) tm wg g t freeze(tb020) eq02.output c = 1 eq02.gmm(method=cue,instwgt=hac,instlag=1,instkern=thann,instbw=andrews,nodf,deriv=aa) i c y y(-1) k(-1) @ c p(-1) k(-1) x(-1) tm wg g t freeze(tb021) eq02.output

the only different is the value of built-in C. However, compare the results, they use different bandwidth, and what is more, most of the betas in tb020 are not significant, but all are significant in tb021, and with signs change.

[EViews Gareth]

Simply a starting value issue, at a guess. CUE is incredibly sensitive to starting values, plus you only have 21 observations which is going to make any optimisation problematic. You'll note that if you don't use CUE, the equations appear to perform much more "nicely".

[maxchen]

but even with showopts option, it does not show the starting value

Code: Select all

eq02.gmm(method=cue,instwgt=hac,instlag=1,instkern=thann,instbw=andrews,nodf,deriv=aa,showopts) i c y y(-1) k(-1) @ c p(-1) k(-1) x(-1) tm wg g t