Near Singular Matrix Error

For technical questions regarding estimation of single equations, systems, VARs, Factor analysis and State Space Models in EViews. General econometric questions and advice should go in the Econometric Discussions forum.

Moderators: EViews Gareth, EViews Moderator

Alina03
Posts: 16
Joined: Sat Jul 14, 2018 12:49 am

Re: Near Singular Matrix Error

Postby Alina03 » Sun Jul 15, 2018 6:29 am

The Normality Test shows me if the residual VAR has a normal distribution (at 0.05). The prob. is 0.0000. What can I do in this case?
Attachments
Capture1.JPG
Capture1.JPG (49.33 KiB) Viewed 328 times

startz
Non-normality and collinearity are NOT problems!
Posts: 3370
Joined: Wed Sep 17, 2008 2:25 pm

Re: Near Singular Matrix Error

Postby startz » Sun Jul 15, 2018 8:09 am

My question is why are you concerned about non-normality?

Alina03
Posts: 16
Joined: Sat Jul 14, 2018 12:49 am

Re: Near Singular Matrix Error

Postby Alina03 » Sun Jul 15, 2018 12:27 pm

Because as I know this is one from the other tests that show me that VAR is ok. Doesn't it show us a VAR characteristic? What I want to do is to represent the impulse response function of 6 variable and I want to assure that VAR is correct represented.

startz
Non-normality and collinearity are NOT problems!
Posts: 3370
Joined: Wed Sep 17, 2008 2:25 pm

Re: Near Singular Matrix Error

Postby startz » Sun Jul 15, 2018 12:31 pm

Normality is of almost no importance for estimating an impulse response function. Sometimes people transform variables, for example taking logs, to get more normal residuals. But then of course you have to translate the transformed impulse response function back to the variable you actually care about.

Alina03
Posts: 16
Joined: Sat Jul 14, 2018 12:49 am

Re: Near Singular Matrix Error

Postby Alina03 » Mon Jul 16, 2018 8:46 am

Is there any way to solve this problem?

startz
Non-normality and collinearity are NOT problems!
Posts: 3370
Joined: Wed Sep 17, 2008 2:25 pm

Re: Near Singular Matrix Error

Postby startz » Mon Jul 16, 2018 8:48 am

(1) You haven't said what the problem is. Non-normality does not invalidate a VAR.
(2) Sometimes you can take logs or make other transformations of the data to make residuals more normal.

Alina03
Posts: 16
Joined: Sat Jul 14, 2018 12:49 am

Re: Near Singular Matrix Error

Postby Alina03 » Mon Jul 16, 2018 8:53 am

I log data and after that I applied the first difference, but this didn't help. What other transformation should I do to test the normal distribution? This is for a project so I have to show it.

startz
Non-normality and collinearity are NOT problems!
Posts: 3370
Joined: Wed Sep 17, 2008 2:25 pm

Re: Near Singular Matrix Error

Postby startz » Mon Jul 16, 2018 9:10 am

Perhaps you should ask your instructor why normality matters in this application.

After all, lots of things just aren't normal.

PaulR
Posts: 2
Joined: Mon Apr 09, 2018 11:22 am

Near singular matrix - without dummies

Postby PaulR » Thu Jul 26, 2018 1:15 pm

Hi, I have the "near singular matrix" problem in estimating panel PMG ARDL model in EViews. The strange thing is that I have no dummy variables - my dataset is annual data in a panel with 20 countries and 150 years, the variables are logs of GDP, Prices, Consumption and Money supply.

Different methods (panel VAR, panel VECM, ...) did not report this problem. Estimation was fine and in line with expectations. I cannot find where the problem is. Unfortunately I need the ARDL too.

I attach the file here.

Thank you very much if a prompt reply is possible.

Paul
Attachments
EViews - Sample.xlsx
(174.86 KiB) Downloaded 15 times

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 11846
Joined: Tue Sep 16, 2008 5:38 pm

Re: Near Singular Matrix Error

Postby EViews Gareth » Thu Jul 26, 2018 1:23 pm

You don't have enough data. Brazil, for example, only had 4 data points, which are non-consecutive. As soon as you take lags, you have no data.
Follow us on Twitter @IHSEViews

spradhan
Posts: 8
Joined: Thu Sep 24, 2015 12:55 am

Re: Near Singular Matrix Error

Postby spradhan » Fri Sep 28, 2018 3:15 am

Hi,

I'm trying to run LS estimation using attached data total_earnings is dependent variable and others are independent. I get estimation if I exclude IS_SPORTS, so can you please let me know what's wrong with IS_SPORTS.
All SC_* variables are scaled variables and running similar equation for SC_* variables I get Near Singular matrix error.
So kindly let me know if any other method would work for scaled variables.

Thanks,
Santosh
Attachments
sales_data_test.wf1
(102.87 KiB) Downloaded 5 times

EViews Gareth
Fe ddaethom, fe welon, fe amcangyfrifon
Posts: 11846
Joined: Tue Sep 16, 2008 5:38 pm

Re: Near Singular Matrix Error

Postby EViews Gareth » Fri Sep 28, 2018 7:16 am

action+roleplaying+sports=1
Follow us on Twitter @IHSEViews

spradhan
Posts: 8
Joined: Thu Sep 24, 2015 12:55 am

Re: Near Singular Matrix Error

Postby spradhan » Fri Sep 28, 2018 8:25 am

EViews Gareth wrote:action+roleplaying+sports=1

Thanks Gareth


Return to “Estimation”

Who is online

Users browsing this forum: No registered users and 14 guests