Page 1 of 1

Correlation

Posted: Sun Nov 30, 2014 10:10 am
by oakway17
Hello all,

I have been asked by my boss to learn how to use eviews and he has given me some practice questions and before I go and see him I was wondering if anyone could assist me so I go into the meeting with the correct answers.

For one of the questions I have been asked to:
Compute, plot and interpret the Auto-Correlation Function (ACF) and the Partial AutoCorrelation Function (PACF) of inflation. What AR, MA or ARMA process seems most appropriate?

For this I did a correllogram as shown in the attachment. From this I decided that the model was an ARMA(2,14) however I feel like this is incorrect and it could be a MA(2). If anyone can assist me it'd be very much appreciated.

Oscar

Re: Correlation

Posted: Sun Nov 30, 2014 11:01 am
by startz
It can't be an MA(2). In an MA(2) the autocorrelations go to zero after two lags.

Re: Correlation

Posted: Sun Nov 30, 2014 11:27 am
by oakway17
Hi,

Sorry I meant AR(2)? I've been staring at this for a few weeks straight that I'm finally getting all my answers crossed...

This is because the PAC is more than -1/1? This makes me think its AR(2)?
The reason for it being ARMA(2,14) is that the PAC at 2 lags is more than -1 and the AC is more than 1 until 14 lags?

This is a new topic to me sorry I know some of this is probably pretty basic..

What would you make it?