# Interpretation of y-axis in impulse response function graphs for VAR models

I read different papers which try to analyse the relationship between oil and macroeconomics with the help of a VAR model. The results are explained in graphics which show the impulse response functions. The question is now how to interpret the Y-axis. There is no graphic with an description of the Y-axis. I read in a learning book that it is in some cases only "how strong is the dependence of the variables " but in other cases it is the percentage of change in the variables. As example: 0.5 in period 1 = 0,5% change in variable x in period one / or only positive relationship. • Right now the question is a bit unclear and thus hard to answer. Do you have an example of such a graph and can you show us the model specification? – Andy May 26 '15 at 11:55
• Yes sure. pic-upload.de/view-27157416/HELP.png.html Here are three examples. My opinion is, that number one and two only have got the statement if there is a positiv or negativ relationship. Number three could be interpret as percentage changes in the values. – Mark May 26 '15 at 13:49
• As example in number one: oil supply shock /oil quantity: is there an 1% depreciation in oil quanity? or is there only a niegative relationship? – Mark May 26 '15 at 13:52

## 1 Answer

The vertical axis is expressed in units of the $Y$ variable. The solid line is a point estimate for the amount $Y$ is expected to change following a unit impulse after the number of periods on the horizontal axis.

Look at the top left graph, of Oil Quantity vs Oil Supply Shock. Using the point estimate (the solid line), a one-unit Oil Supply Shock (whatever the units are) is expected to lead to a change in Oil Quantity of -0.5 units after 10 periods

• Maybe I should write that before. In all of these VAR Modells the x-axis is the periods, in the case of picture one in month. And what are the Units? And is it every new Period again the amount of units? – Mark May 26 '15 at 16:53
• And when you look at picture two: it is the respond of stockindex to oil shocks. why there is only a .002 percent or unit shock? the author discribes the reactions as significant positiv or negativ. so i think its only expression of how much positive ( very positive , small positive etc.)not how much percent. – Mark May 26 '15 at 16:56
• @Mark if the horizontal axis is in periods, then it's the effect of a one unit impulse after that many periods. I edited my post – shadowtalker May 26 '15 at 16:58
• I don't know what you're trying to say here. It's possible that "significant" means "significantly different from zero based on the confidence interval" but without seeing the quote from the paper it's hard to say. – shadowtalker May 26 '15 at 17:02
• As for why it's only 0.002%? That's what the model predicts. I don't think you're saying what you mean to be saying – shadowtalker May 26 '15 at 17:02