9 p.m.
Re: Gretl texbook
Tom asks:
In my question about documentation I guess I am asking if there is a
textbook that explains times series analysis as it is implemented in Gretl,
i.e., is there a times series textbook where the author planned on the
students using Gretl (like ITSM is used in Brockwell and Davis).
One might consider
Using gretl for Principles of Econometrics, 3rd edition This book is a manual for using
gretl suitable for use in an introductory econometrics class. It has been written
specifically for Hill, Griffiths, and Lim's Principles of Econometrics, 3rd edition,
published by John Wiley and Sons, 2008. In the book, I show you how to use gretl to work
all of the examples in POE. The data for the examples is packaged for installation into
gretl and can be found by following the links below.
(Description from the web site: http://www.learneconometrics.com/gretl.html
-----Original Message-----
From: gretl-users-bounces(a)lists.wfu.edu on behalf of gretl-users-request(a)lists.wfu.edu
Sent: Mon 2/4/2008 3:40 AM
To: gretl-users(a)lists.wfu.edu
Subject: Gretl-users Digest, Vol 13, Issue 3
Send Gretl-users mailing list submissions to
gretl-users(a)lists.wfu.edu
To subscribe or unsubscribe via the World Wide Web, visit
http://lists.wfu.edu/mailman/listinfo/gretl-users
or, via email, send a message with subject or body 'help' to
gretl-users-request(a)lists.wfu.edu
You can reach the person managing the list at
gretl-users-owner(a)lists.wfu.edu
When replying, please edit your Subject line so it is more specific
than "Re: Contents of Gretl-users digest..."
Today's Topics:
1. Tests for Residuals (Tom La Bone)
2. Re: Tests for Residuals (Sven Schreiber)
3. Re: Re: Unit Root Test (Allin Cottrell)
4. RE: Tests for Residuals (Tom La Bone)
5. RE: Tests for Residuals (Allin Cottrell)
6. new string handling (was Re: Unit Root Test) (Allin Cottrell)
7. RE: Tests for Residuals (Allin Cottrell)
8. Re: Tests for Residuals (Sven Schreiber)
----------------------------------------------------------------------
Message: 1
Date: Sun, 3 Feb 2008 10:30:22 -0500
From: "Tom La Bone" <labone(a)gforcecable.com>
Subject: [Gretl-users] Tests for Residuals
To: <gretl-users(a)lists.wfu.edu>
Message-ID: <000301c86679$b2445dc0$6401a8c0@Boozoo>
Content-Type: text/plain; charset="us-ascii"
Good morning,
1) Does Gretl have tests for randomness of residuals accessible as options
from the GUI?
2) From the command line?
3) What references are recommended for folks who interested in the
algorithms used in Gretl? In other words, what references discuss in detail
how Gretl does what it does?
Tom
------------------------------
Message: 2
Date: Sun, 03 Feb 2008 18:31:32 +0100
From: Sven Schreiber <svetosch(a)gmx.net>
Subject: Re: [Gretl-users] Tests for Residuals
To: labone(a)gforcecable.com, Gretl list <gretl-users(a)lists.wfu.edu>
Message-ID: <47A5FA74.6080106(a)gmx.net>
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Am 03.02.2008 16:30, Tom La Bone schrieb:
Good morning,
1) Does Gretl have tests for randomness of residuals accessible as options
from the GUI?
2) From the command line?
Short answer: yes. Long answer: In contrast to some other open-source
projects, gretl has excellent documentation; I'm almost certain the user
guide has the detailed answers to your question.
3) What references are recommended for folks who interested in the
algorithms used in Gretl? In other words, what references discuss in detail
how Gretl does what it does?
Again, in some cases the documentation is fairly explicit, but
ultimately Allin and Jack would probably say: use the source, Luke.
(Nothing I ever do since I'm not literate in C, so I'm not trying to be
arrogant here, it's just the fact of open source that unlike proprietary
packages everything is accessible in principle.)
good luck,
sven
------------------------------
Message: 3
Date: Sun, 3 Feb 2008 13:00:05 -0500 (EST)
From: Allin Cottrell <cottrell(a)wfu.edu>
Subject: Re: [Gretl-users] Re: Unit Root Test
To: Gretl list <gretl-users(a)lists.wfu.edu>
Message-ID: <alpine.LRH.1.00.0802031228550.309(a)ricardo.ecn.wfu.edu>
Content-Type: TEXT/PLAIN; charset=US-ASCII
On Fri, 1 Feb 2008, Marcos Larios wrote:
> could you be more specific, please? Single equation or VAR,
> ARMA or not, etc. etc.
Hi Sven, I'm refering to the Roots in an ARMA model, ...
At present the roots are computed and printed (as you know), but
they're not fetchable via a "$" variable. That probably should
be added.
In the meantime this might be helpful: if you estimate an ar(i)ma
model using the --x-12-arima option in gretl, a plain
text file named varname.rts (where "varname" is the name of the
dependent variable) is written. This is in a simple fixed format
and contains all the roots information. It will be written to
your x12arima working directory. You can find out where that is
using
print "@x12adir"
in gretl.
the thing is that I need to build the graph for the unit root
test in ARMA models, for which I require acces (through code) to
the Roots, Imaginary, Modulus and Frecuency in the Unit Root
Test Results in the ARMA models, I also need to know how to
"draw" a circle in a graph with gretl, it seems I can't get
gnuplot to recognize the circle equation...
We currently have code for graphing the roots when you estimate a
VAR. You can see how we do this in either of these ways.
(a) Estimate a arbitrary VAR in the GUI. In the VAR output window
select Graphs/VAR inverse roots. Click on the graph and select
"Save to session as icon". Open the icon view window (View/Icon
view). Right-click on the graph icon and select "Edit plot
commands". You'll get a window showing the gnuplot commands for
making this sort of graph, which you could save under another name
and use as a template.
(b) Look at the C code that constructs the graph commands for this
case.
http://gretl.cvs.sourceforge.net/*checkout*/gretl/gretl/lib/src/graphing....
and look for "gretl_VAR_roots_plot".
Allin Cottrell
------------------------------
Message: 4
Date: Sun, 3 Feb 2008 13:04:42 -0500
From: "Tom La Bone" <labone(a)gforcecable.com>
Subject: RE: [Gretl-users] Tests for Residuals
To: "'Sven Schreiber'" <svetosch(a)gmx.net>, "'Gretl
list'"
<gretl-users(a)lists.wfu.edu>
Message-ID: <000401c8668f$41aaacc0$6401a8c0@Boozoo>
Content-Type: text/plain; charset="us-ascii"
Sven, thanks for your timely response.
In case I have not blown my cover already, I must confess that I am a stat
grad student taking his first time series course. Our professor has given us
the freedom to use any software we want for the course, and I am taking the
opportunity to learn Gretl. I chose Gretl because of many of the things you
mention, i.e., it is well documented, open source (I can look at the code if
I dare), and in the worst case it is a nice front-end for sending data to R
for further analysis.
The main downside of using Gretl is that nobody in the department is
familiar with the program so I am "on my own" in my efforts to learn how to
use it. I make an honest effort to look through the manual for answers to my
questions before asking for help. I have gotten some useful answers and
guidance that were not immediately obvious from the manual. On the other
hand I have gotten some guidance that goes into the "I have not gotten to
that chapter yet" bin. It would be helpful to be able to search the archives
for answers to my questions that have already been posted, but it appears
that is not possible.
In my question about documentation I guess I am asking if there is a
textbook that explains times series analysis as it is implemented in Gretl,
i.e., is there a times series textbook where the author planned on the
students using Gretl (like ITSM is used in Brockwell and Davis).
Tom
-----Original Message-----
From: Sven Schreiber [mailto:svetosch@gmx.net]
Sent: Sunday, February 03, 2008 12:32 PM
To: labone(a)gforcecable.com; Gretl list
Subject: Re: [Gretl-users] Tests for Residuals
Am 03.02.2008 16:30, Tom La Bone schrieb:
Good morning,
1) Does Gretl have tests for randomness of residuals accessible as options
from the GUI?
2) From the command line?
Short answer: yes. Long answer: In contrast to some other open-source
projects, gretl has excellent documentation; I'm almost certain the user
guide has the detailed answers to your question.
3) What references are recommended for folks who interested in the
algorithms used in Gretl? In other words, what references discuss in
detail
how Gretl does what it does?
Again, in some cases the documentation is fairly explicit, but
ultimately Allin and Jack would probably say: use the source, Luke.
(Nothing I ever do since I'm not literate in C, so I'm not trying to be
arrogant here, it's just the fact of open source that unlike proprietary
packages everything is accessible in principle.)
good luck,
sven
------------------------------
Message: 5
Date: Sun, 3 Feb 2008 16:47:34 -0500 (EST)
From: Allin Cottrell <cottrell(a)wfu.edu>
Subject: RE: [Gretl-users] Tests for Residuals
To: labone(a)gforcecable.com, Gretl list <gretl-users(a)lists.wfu.edu>
Message-ID: <alpine.LRH.1.00.0802031640270.309(a)ricardo.ecn.wfu.edu>
Content-Type: TEXT/PLAIN; charset=US-ASCII
On Sun, 3 Feb 2008, Tom La Bone wrote:
In case I have not blown my cover already, I must confess that I
am a stat grad student taking his first time series course. Our
professor has given us the freedom to use any software we want
for the course, and I am taking the opportunity to learn Gretl.
I chose Gretl because of many of the things you mention...
OK, then we'll try to be extra-helpful ;-)
1) Does Gretl have tests for randomness of residuals accessible as
options from the GUI?
Yes, various aspects of the "randomness" (or otherwise) of the
residuals are explored by almost all the items under the "Tests"
menu in the model output window. You could also save the
residuals and do a runs test if you wanted.
2) From the command line?
help lmtest
help testuhat
help runs
3) What references are recommended for folks who interested in the
algorithms used in Gretl? In other words, what references discuss
in detail how Gretl does what it does?
The Gretl User's Guide, latest version at
http://ricardo.ecn.wfu.edu/pub/gretl/manual/
plus of course the C source, as Sven said.
Allin Cottrell
------------------------------
Message: 6
Date: Sun, 3 Feb 2008 21:15:41 -0500 (EST)
From: Allin Cottrell <cottrell(a)wfu.edu>
Subject: [Gretl-users] new string handling (was Re: Unit Root Test)
To: Gretl list <gretl-users(a)lists.wfu.edu>
Message-ID: <alpine.LRH.1.00.0802032105470.3269(a)ricardo.ecn.wfu.edu>
Content-Type: TEXT/PLAIN; charset=US-ASCII
On Sun, 3 Feb 2008, I wrote:
In the meantime this might be helpful: if you estimate an
ar(i)ma model using the --x-12-arima option in gretl, a plain
text file named varname.rts (where "varname" is the name of the
dependent variable) is written.
Around Christmas I added some functionality for handling strings.
I think the timing was bad and it wasn't much noticed, and as a
result was not much debugged. However, reading roots information
from a X-12-ARIMA .rts file is a handy case in point. I've had a
go at this, and in the process have fixed a few bugs and added
some more functionality. Find below a function that reads an .rts
file and returns a roots matrix. The new string-handling idioms
are explained in (a) "help sscanf" and (b) section 11.2 of the
User's Guide (CVS and Windows snapshot).
<script>
function read_x12a_roots (series y, int type[1:2:1])
# compose the name of the X12A roots file...
sprintf yfile, "%s/%s.rts", @x12adir, argname(y)
# and grab its content into a string
string rts = readfile(@yfile)
matrix r
matrix R
if (type = 1)
string targ = "AR"
else
string targ = "MA"
endif
string line = strstr(@rts, @targ)
loop while isstring(line) --quiet
# offset on line to find the numbers
string line = @line + 18
# get 4 values: real, imaginary, modulus, frequency
sscanf @line, "%m", r
R = R|r
string line = strstr(@line, @targ)
endloop
return matrix R
end function
open data9-7
arma 2 1 ; QNC --x-12-arima
AR_roots = read_x12a_roots(QNC, 1)
MA_roots = read_x12a_roots(QNC, 2)
print AR_roots MA_roots
</script>
Allin.
------------------------------
Message: 7
Date: Sun, 3 Feb 2008 21:51:24 -0500 (EST)
From: Allin Cottrell <cottrell(a)wfu.edu>
Subject: RE: [Gretl-users] Tests for Residuals
To: labone(a)gforcecable.com, Gretl list <gretl-users(a)lists.wfu.edu>
Message-ID: <alpine.LRH.1.00.0802032147370.3269(a)ricardo.ecn.wfu.edu>
Content-Type: TEXT/PLAIN; charset=US-ASCII
On Sun, 3 Feb 2008, Tom La Bone wrote:
In my question about documentation I guess I am asking if there is a
textbook that explains times series analysis as it is implemented in Gretl,
i.e., is there a times series textbook where the author planned on the
students using Gretl (like ITSM is used in Brockwell and Davis).
Try Lee Adkins's site:
http://www.learneconometrics.com/gretl.html
Allin Cottrell
------------------------------
Message: 8
Date: Mon, 04 Feb 2008 09:39:48 +0100
From: Sven Schreiber <svetosch(a)gmx.net>
Subject: Re: [Gretl-users] Tests for Residuals
To: Gretl list <gretl-users(a)lists.wfu.edu>
Message-ID: <47A6CF54.5080202(a)gmx.net>
Content-Type: text/plain; charset=ISO-8859-15; format=flowed
Am 04.02.2008 03:51, Allin Cottrell schrieb:
On Sun, 3 Feb 2008, Tom La Bone wrote:
> In my question about documentation I guess I am asking if there is a
> textbook that explains times series analysis as it is implemented in Gretl,
> i.e., is there a times series textbook where the author planned on the
> students using Gretl (like ITSM is used in Brockwell and Davis).
Try Lee Adkins's site:
http://www.learneconometrics.com/gretl.html
Very interesting link; maybe worth adding it (or even an entire section
with related stuff) to the website, if it's not there already?
-sven
------------------------------
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
End of Gretl-users Digest, Vol 13, Issue 3
******************************************
11:27 a.m.
New subject: problem with system instrumental variables
Hello listers, greetings from Paris !
Still on my system estimation commands, I have a few difficulties with
the use of instruments.
I would like to estimate a system of two equations with one different
variable and one instrumented variable (income). So I use the syntax for
tsls to specifiy regressors and instruments on the same line :
=================================================== script exerpt
==============================================================
list socdemo = CHILD LOCATAIRE AUTOMOBILE
system name=AIDS
equation WLOG const lnPLOGCOMP lnPCARBCOMP lnDEPTOTR socdemo MRLOG ;
const lnPLOGCOMP lnPCARBCOMP lnREVTOTUCR socdemo MRLOG
equation WCAR const lnPLOGCOMP lnPCARBCOMP lnDEPTOTR socdemo MRCAR ;
const lnPLOGCOMP lnPCARBCOMP lnREVTOTUCR socdemo MRCAR
end system
restrict AIDS
b[1,3]-b[2,2]=0
end restrict
estimate AIDS method=3sls
=================================================== end script exerpt
==============================================================
but the command fails as gretl says : "variable 0 duplicated in commands
list"
GRETL seems not to accept my instruments, in this AIDS model total
expense (lnDEPTOTR) is instrumented by income (lnREVTOTUCR) ; all other
variables being the same except the last ones (MRLOG and MRCAR) which
are specific to each equation (and must remain different).
If i use both MLOG and MCAR in each equation, system estimation runs ok
but does not makes sense on a statistical perspective.
Any advice from the list ?
sincerely
Franck
11:52 a.m.
New subject: problem with system instrumental variables
Am 07.02.2008 12:27, nadaud(a)centre-cired.fr schrieb:
Hello listers, greetings from Paris !
Still on my system estimation commands, I have a few difficulties with
the use of instruments.
I would like to estimate a system of two equations with one different
variable and one instrumented variable (income). So I use the syntax for
tsls to specifiy regressors and instruments on the same line :
=================================================== script exerpt
==============================================================
list socdemo = CHILD LOCATAIRE AUTOMOBILE
system name=AIDS
equation WLOG const lnPLOGCOMP lnPCARBCOMP lnDEPTOTR socdemo MRLOG ;
const lnPLOGCOMP lnPCARBCOMP lnREVTOTUCR socdemo MRLOG equation WCAR
const lnPLOGCOMP lnPCARBCOMP lnDEPTOTR socdemo MRCAR ; const lnPLOGCOMP
lnPCARBCOMP lnREVTOTUCR socdemo MRCAR
end system
restrict AIDS
b[1,3]-b[2,2]=0
end restrict
estimate AIDS method=3sls
=================================================== end script exerpt
==============================================================
but the command fails as gretl says : "variable 0 duplicated in commands
list"
GRETL seems not to accept my instruments, in this AIDS model total
expense (lnDEPTOTR) is instrumented by income (lnREVTOTUCR) ; all other
variables being the same except the last ones (MRLOG and MRCAR) which
are specific to each equation (and must remain different).
If i use both MLOG and MCAR in each equation, system estimation runs ok
but does not makes sense on a statistical perspective.
Any advice from the list ?
I believe that variable 0 refers to the constant. And for TSLS there was
recently some change in whether the constant is automatically added or
not. Don't know whether this applies to 3SLS as well.
So until someone knowledgeable answers this, I can think of two
workarounds that may or may not work:
1) try leaving out the const and check afterwards whether gretl has
added it automatically
2) update to a newer version; if you already use 1.7.1 maybe use the
latest windows snapshot (or cvs, respectively).
No guarantees, but good luck!
-sven
12:24 p.m.
New subject: problem with system instrumental variables
Hello again, and thank you Sven for your advice
I believe that variable 0 refers to the constant. And for TSLS there
was recently some change in whether the constant is automatically
added or not. Don't know whether this applies to 3SLS as well.
Yes that is
right, I add the constant because in my AIDS system,
demographics translate the constant.
So until someone knowledgeable answers this, I can think of two
workarounds that may or may not work:
1) try leaving out the const and check afterwards whether gretl has
added it automatically
2) update to a newer version; if you already use 1.7.1 maybe use the
latest windows snapshot (or cvs, respectively).
No guarantees, but good luck!
And after doing both, the problem persist !
I wonder if it is possible to estimate systems with just one different
variables ? Looking at the example scripts it seems so. Also with the
tsls syntax i can normally specify different instruments for both
equations nope ?
Well let me try something else...
sincerely
Franck
12:44 p.m.
New subject: problem with system instrumental variables
Am 07.02.2008 13:24, nadaud(a)centre-cired.fr schrieb:
And after doing both, the problem persist !
Sorry, I'm out of ideas. Maybe it's simply a bug; but I have never done
such systems with gretl so far.
I wonder if it is possible to estimate systems with just one different
variables ? Looking at the example scripts it seems so. Also with the
tsls syntax i can normally specify different instruments for both
equations nope ?
Could you rephrase that, I don't understand what you mean, sorry.
-sven
1:59 p.m.
New subject: problem with system instrumental variables
Hi Sven
Well, so for the system ! It work quite well and is a neat feature. As
you may micro data analysis often done with system estimators
>
> I wonder if it is possible to estimate systems with just one
> different variables ? Looking at the example scripts it seems so.
> Also with the tsls syntax i can normally specify different
> instruments for both equations nope ?
>
Could you rephrase that, I don't understand what you mean, sorry.
Yep ! Well, up to now I was working on the AIDS demand system on
microdata (I sent a few days ago a sample of my data and the
accompanying script).
In this context all was easy because in the standard formulation of AIDS
the same variables appear in each equation.
To sum up, because of theoritical constraints, the system covariance
matrix is singular. Estimation is done by dropping one equation and
homogeneity in prices is imposed costlessly by the use of price
relatives of deleted good.
My concern now is that I have a nasty problem with my data : some are
censored with nil consumption. Thats means serious biais proportional to
the degree of censoring. Unfortunately, one of my goods suffers from
substantial censoring (>40 %). To overcome this I use first a probit
model and compute the fitted Mills ratios [ f(x) /F(x) = dnorm(yhat) /
cnorm(yhat) ]that are used as extra instruments in the AIDS system.
Finally, my problem is that to do so, I have a system where each
equation have all the same variables in common, except one => the Mills
ratio. GRETL does not estimate the system when i specify using the tsls
syntax which can in principle allow equation specific instruments lists.
here is where i am.
cheers & thanx
Franck
8:04 p.m.
New subject: problem with system instrumental variables
On Thu, 7 Feb 2008, nadaud(a)centre-cired.fr wrote:
Finally, my problem is that to do so, I have a system where each
equation have all the same variables in common, except one =>
the Mills ratio. GRETL does not estimate the system when i
specify using the tsls syntax which can in principle allow
equation specific instruments lists. here is where i am.
I see your point. The use of tsls-style syntax to specify
differing instruments per equation should now (CVS and Windows
snapshot) work OK.
Thanks for keeping us honest on system estimation!
Allin.
10:45 a.m.
New subject: problem with system instrumental variables ==> runs OK
Hello gretl listers ! Greetings from Paris !
First, thank you Allin for the quick fix, you save my day (and maybe my
week !)
Just installed winsnapshot and tried the 500 observations sample script
with the tsls syntax command.
All runs smoothly and perfectly ! Once again thank you for the good job.
I will try later my that day the estimation on the whole sample .
I observed that the TSLS type syntax is available for the following
sytem estimators ('method=xxx' switch) : TSLS ; 3SLS but not for SUR and
OLS (I did not try the others).
That's perfect as I can do my special Heiein-Wessels 2 step estimation.
By the way, i am currently working intensively with system estimation on
microdata, and this is the reason of my modest contribution. I am also
fond of GRETL and try to do most of my work the jewel, encountering
these little problems.
Again, thanx, and continue the good job !
sincerely
Franck
2:48 p.m.
New subject: problem with system instrumental variables
On Thu, 7 Feb 2008, nadaud(a)centre-cired.fr wrote:
I would like to estimate a system of two equations with one
different variable and one instrumented variable (income). So I
use the syntax for tsls to specifiy regressors and instruments
on the same line...
Yes, that method is supposed to work, but it seems to have got
broken at some point (assuming it ever really worked!).
I'm investigating this now and will report back before long.
Allin.
7:09 p.m.
New subject: problem with system instrumental variables
On Thu, 7 Feb 2008, Allin Cottrell wrote:
On Thu, 7 Feb 2008, nadaud(a)centre-cired.fr wrote:
> I would like to estimate a system of two equations with one
> different variable and one instrumented variable (income). So I
> use the syntax for tsls to specifiy regressors and instruments
> on the same line...
Yes, that method is supposed to work, but it seems to have got
broken at some point (assuming it ever really worked!).
I'm investigating this now and will report back before long.
Please try the new snapshot
http://ricardo.ecn.wfu.edu/pub/gretl/gretl_install.exe
The more obvious problems are fixed, I think. I will do some
checking against R.
Allin.
8 p.m.
New subject: problem with system instrumental variables
On Thu, 7 Feb 2008, Allin Cottrell wrote:
Please try the new snapshot
http://ricardo.ecn.wfu.edu/pub/gretl/gretl_install.exe
The more obvious problems are fixed, I think. I will do some
checking against R.
We get the same results as R for the problem Franck outlined.
Allin.
5921
days inactive
5925
days old
10 comments
4 participants
participants (4)
-
Allin Cottrell -
Mixon, Wilson -
nadaud＠centre-cired.fr -
Sven Schreiber