Impulse response chart

[J2Summers]

Hi,

I'm using v7.2

When estimating an impulse reponse from a VAR the resulting chart legend indicates that the series shown is the series which should have received the innovation and not the responding series. The data values appear to be correct relative to 'shocked series,' so I'm thinking the only thing wrong is the legend series title selection. Anyone know if this is a known bug?

Thanks,

John

[J2Summers]

To clarify ... The data values appear to be correct such that the series displayed is the response series. The legend indicates that it is instead the shocked series. Tx.

[EViews Gareth]

Could you provide an example?

[J2Summers]

Sure,

I've attached a PowerPoint which lays out the model I'm referring to.

Thanks,

John

[EViews Gareth]

Not quite sure I follow. Your first slides (nicely done, by the way), show that LEADING_SERIES cause changes in SERIES_OF_INTEREST, and not the other way round.

Then you show a graph of an impulse response that shows that LEADING_SERIES causes a response in SERIES_OF_INTEREST. Which is perfectly in line with the first slides.

Where is the contradiction?

[J2Summers]

Thanks.

Shouldn't the blue series in the impulse response chart be SERIES_OF_INTEREST, i.e. the series whose response we are interested in, and not LEADING_SERIES, i.e. the series we shocked?

A One S.D. shock to LEADING_SERIES should be ~4.12 (slide 2), given the coefficient on LEADING_SERIES(-2) in the VAR, the value 2.07 shown in the impulse response should be attributable to SERIES_OF_INTEREST, yes?

[EViews Gareth]

Ah, true. But EViews doesn't put that legend box on the impulse response graph. It just puts the title you have. I assume you went in and turned the legend back on. Yes, the text in the legend will be wrong if you do that.

[J2Summers]

Gareth,

Okay, cool. Yes I did turn the legend on. I can change the text no problem, just wanted to confirm that the legend is incorrect by default.

Thanks again for your help!