Need Help on Dummy variable rep.

[Ahmad]

Hi everybody.

I am Ahmad, just join this group. I need help on how to specify year dummies in Eview.
I am running a panel data ( want to run panel VAR) I have series on 29 banks for 8 years period. I want to analyze year effect as well as individual firms. I have a problem in representing years as dommies in my data. I tried several times but when I want to run the analysis eview will tell me " Near singularity" how can I solve this problem.

Please kindly guide me.

Thanks.
Ams.

[startz]

Hi everybody.

I am Ahmad, just join this group. I need help on how to specify year dummies in Eview.
I am running a panel data ( want to run panel VAR) I have series on 29 banks for 8 years period. I want to analyze year effect as well as individual firms. I have a problem in representing years as dommies in my data. I tried several times but when I want to run the analysis eview will tell me " Near singularity" how can I solve this problem.

Please kindly guide me.

Thanks.
Ams.

Leave out one year dummy and one firm dummy and include a constant

[Jack]

Hey!

I'm experiencing the same problem and I am pretty certain that my equation is correctly specified. It seems Eviews is having problem handling some of my variables when applying dummies. When I remove said variables, the regressions turn out the way I want them (filtering out year-specific effects) with the rest of my data but once I include them, Eviews again complains "Near singular matrix".

Any hints?

[startz]

Hey!

I'm experiencing the same problem and I am pretty certain that my equation is correctly specified. It seems Eviews is having problem handling some of my variables when applying dummies. When I remove said variables, the regressions turn out the way I want them (filtering out year-specific effects) with the rest of my data but once I include them, Eviews again complains "Near singular matrix".

Any hints?

Remember that a regression cannot have more than one complete set of dummies, nor a complete set of dummies with a constant.

[Jack]

Remember that a regression cannot have more than one complete set of dummies, nor a complete set of dummies with a constant.

Thank you for responding.

However, mine doesn't. It has a constant, 26 dummies for 27 years and 7 regressors. That's it. As I said (or tried to), my dummies don't act weird all the time. It's just when I include some variables that they seem to interact with them in this way.

[startz]

Remember that a regression cannot have more than one complete set of dummies, nor a complete set of dummies with a constant.

Thank you for responding.

However, mine doesn't. It has a constant, 26 dummies for 27 years and 7 regressors. That's it. As I said (or tried to), my dummies don't act weird all the time. It's just when I include some variables that they seem to interact with them in this way.

You might post your workfile here, including the equation that doesn't work.

[Jack]

Remember that a regression cannot have more than one complete set of dummies, nor a complete set of dummies with a constant.

Thank you for responding.

However, mine doesn't. It has a constant, 26 dummies for 27 years and 7 regressors. That's it. As I said (or tried to), my dummies don't act weird all the time. It's just when I include some variables that they seem to interact with them in this way.

You might post your workfile here, including the equation that doesn't work.

I'll attach it. It is not a matter of life or death, but it would be very kind of you to have a look at it, since I am very new to this (d'uh).

The regression I want to run, but cannot, is simply:

Code: Select all

ls log(odac) c log(oda(-1)/gdp(-1)) log(gdpc(-1)) log(odapgdp) extdebt openness inflatio cim fh d1980 d1981 d1982 d1983 d1984 d1985 d1986 d1987 d1988 d1989 d1990 d1991 d1992 d1993 d1994 d1995 d1996 d1997 d1998 d1999 d2000 d2001 d2002 d2003 d2004 d2005

An example of code that Eviews can execute is:

Code: Select all

ls log(odac) c log(gdp(-1)/oda(-1)) cim openness log(odapgdp) inflatio extdebt d1980 d1981 d1982 d1983 d1984 d1985 d1986 d1987 d1988 d1989 d1990 d1991 d1992 d1993 d1994 d1995 d1996 d1997 d1998 d1999 d2000 d2001 d2002 d2003 d2004 d2005

I'm fairly certain by now that the problem is regarding the variable "fh".

Cheers
J

[startz]

Thank you for responding.

However, mine doesn't. It has a constant, 26 dummies for 27 years and 7 regressors. That's it. As I said (or tried to), my dummies don't act weird all the time. It's just when I include some variables that they seem to interact with them in this way.

You might post your workfile here, including the equation that doesn't work.

I'll attach it. It is not a matter of life or death, but it would be very kind of you to have a look at it, since I am very new to this (d'uh).

The regression I want to run, but cannot, is simply:

Code: Select all

ls log(odac) c log(oda(-1)/gdp(-1)) log(gdpc(-1)) log(odapgdp) extdebt openness inflatio cim fh d1980 d1981 d1982 d1983 d1984 d1985 d1986 d1987 d1988 d1989 d1990 d1991 d1992 d1993 d1994 d1995 d1996 d1997 d1998 d1999 d2000 d2001 d2002 d2003 d2004 d2005

An example of code that Eviews can execute is:

Code: Select all

ls log(odac) c log(gdp(-1)/oda(-1)) cim openness log(odapgdp) inflatio extdebt d1980 d1981 d1982 d1983 d1984 d1985 d1986 d1987 d1988 d1989 d1990 d1991 d1992 d1993 d1994 d1995 d1996 d1997 d1998 d1999 d2000 d2001 d2002 d2003 d2004 d2005

I'm fairly certain by now that the problem is regarding the variable "fh".

Cheers
J[/quote]

I'm happy to give it a quick try. Others may be as well.
The workfile smpl is set to 1 3537. I get a message "log of non-positive number" Is that the sample you're using?

[Jack]

The workfile smpl is set to 1 3537. I get a message "log of non-positive number" Is that the sample you're using?

It is. I do not get that error message though (using Eviews 5). At what point does it appear? Come to think about it, I did experience something similar when generating a variable couple of days ago... What happens if you use log(oda/populati) instead of log(odac), they should be almost the same things.

[startz]

The workfile smpl is set to 1 3537. I get a message "log of non-positive number" Is that the sample you're using?

It is. I do not get that error message though (using Eviews 5). At what point does it appear? Come to think about it, I did experience something similar when generating a variable couple of days ago... What happens if you use log(oda/populati) instead of log(odac), they should be almost the same things.

The problem is that your variable fh is perfectly explained by the year dummies. (The question about the logs turned out to be irrelevant.) To see the problem, try

Code: Select all

ls fh d1980 d1981 d1982 d1983 d1984 d1985 d1986 d1987 d1988 d1989 d1990 d1991 d1992 d1993 d1994 d1995 d1996 d1997 d1998 d1999 d2000 d2001 d2002 d2003 d2004 d2005

[Jack]

The problem is that your variable fh is perfectly explained by the year dummies.

Okay. I guess I'm just going to have to without those dummies then...

Anyway, thanks a bundle for clarifying. It was really helpful!