From: Hakan Berument <berument@Bilkent.EDU.TR>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: q:Goodness of fit measure for IV
Date: Tue, 2 Sep 1997 13:53:12 +0300 (EET DST)
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Mailer: Mercury MTS (Bindery) v1.30
I am using two stage least square method to estimate a linear model.
What can I use as a gooness of fit measure for this linear model.
Thanks in advance for your responses.
Hakan Berument Department of Economics
e-mail: berument@bilkent.edu.tr Bilkent University
Phone: + 90-312-241-1224 06533 Bilkent Ankara
Fax : + 90-312-266-4150 Turkey
Homepage: http://www.bilkent.edu.tr/~berument
---------- End of message ----------
From: "Gregory, Richard" <RGregory@VEDP.state.va.us>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: Goodness of fit measure for IV
Date: Tue, 2 Sep 1997 08:02:10 -0400
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.837.3 (via Mercury MTS (Bindery) v1.30)
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
A good place to start might be:
Wooldridge, J.M. (1990) A Note on the Lagrange Multiplier and
F-statistics for Two Stage Least Square Regressions. Economics Letters,
34, 151-5.
Good Luck.
>----------
>From: Hakan Berument[SMTP:berument@Bilkent.EDU.TR]
>Sent: Tuesday, September 02, 1997 6:53 AM
>To: RATS Discussion List
>Subject: q:Goodness of fit measure for IV
>
>I am using two stage least square method to estimate a linear model.
>What can I use as a gooness of fit measure for this linear model.
>
>Thanks in advance for your responses.
>
>
>Hakan Berument Department of
>Economics
>e-mail: berument@bilkent.edu.tr Bilkent
>University
>Phone: + 90-312-241-1224 06533 Bilkent
>Ankara
>Fax : + 90-312-266-4150
>Turkey
>Homepage: http://www.bilkent.edu.tr/~berument
>
>
---------- End of message ----------
From: "John O. Velis, Indiana University" <jvelis@indiana.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: restricted MLE's
Date: Fri, 5 Sep 1997 11:51:04 -0500 (EST)
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Mailer: Mercury MTS (Bindery) v1.30
Greetings:
can any one suggest a good way to run MLE (specifically a GARCH model)
with restrictions (i.e. 0<beta<1 ) on RATS.
Clarification: I am not asking how to do teh GARCH model itself, just how
to code the restrictions.
Thanks
John O. Velis
Indiana University School of Business
Department of Business Economics and Public Policy
1309 East Tenth Street, BU 451
Bloomington, IN 47405-1701
Phone: (812) 855-9219
Fax: (812) 855-3354
Email: jvelis@indiana.edu
---------- End of message ----------
From: "John O. Velis, Indiana University" <jvelis@indiana.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: regime-switching
Date: Fri, 5 Sep 1997 11:52:37 -0500 (EST)
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Mailer: Mercury MTS (Bindery) v1.30
Does anyone have any code (or can he/she refer me to a source) to estimate
a mixture normal and/or normal-jump distribution with time-varying mixing
parameters?
thanks.
-jov
John O. Velis
Indiana University School of Business
Department of Business Economics and Public Policy
1309 East Tenth Street, BU 451
Bloomington, IN 47405-1701
Phone: (812) 855-9219
Fax: (812) 855-3354
Email: jvelis@indiana.edu
---------- End of message ----------
From: "Dipl.-Kfm. Carl-Heinrich Kehr" <ckehr@wiwi.uni-frankfurt.de>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: No Subject
Date: Mon, 8 Sep 1997 21:23:51 +0100
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-mailer: Pegasus Mail for Windows (v2.01) (via Mercury MTS (Bindery) v1.30)
unsubscribe
Dr. Carl-Heinrich Kehr
Mozartweg 2B
D-63225 Langen/Hessen
Federal Republic of Germany
Tel. : **49/6103/78556
e-Mail: ckehr@wiwi.uni-frankfurt.de
---------- End of message ----------
From: Dadanee Vuthipadadorn <dvuthipa@vayu.mof.go.th>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: unsubscribe
Date: Tue, 09 Sep 1997 10:04:52 +0700
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Windows Eudora Light Version 3.0.1 (32) (via Mercury MTS (Bindery) v1.30)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
unsubscribe
---------- End of message ----------
From: Renato Carandang <RCARAN1@allstate.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: unsubscribe
Date: Tue, 09 Sep 1997 07:41:48 -0600
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Novell GroupWise 4.1 (via Mercury MTS (Bindery) v1.30)
unsubscribe
Renato Carandang
---------- End of message ----------
From: XHAJA Agim <xhaja5@hei.unige.ch>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Time-Varying procedure question.
Date: Tue, 09 Sep 1997 19:23:15 +0200 (MET DST)
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
MIME-version: 1.0
Content-type: TEXT/PLAIN; charset=US-ASCII
X-Mailer: Mercury MTS (Bindery) v1.30
Dear Readers of the Rats list,
I am a beginner in Rats- sorry if my questions are trivial, but I have no
alternative persons to turn to.
I am trying to estimate a regression where only one regressor
is a time-varying coefficient (i.e. follows a random walk).
Does RATS allow me to do it and if possible how do I proceed?
My second concern is:
When I try to regress the above mentioned equation with all
coefficients varying in time and I follow step by step the
procedure descripted in rats manual, chapter 13.
I don't get any output and any error message!!!( Once, only I have
received a "VAR4" error message .)
I would appreciate if you can help me in solving those problems.
Thank you very much,
A. Xhaja
---------- End of message ----------
From: "Deitch, Jonathan" <B0DEITCH@hq.penfed.org>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: Time-Varying procedure question.
Date: Tue, 9 Sep 1997 15:42:17 -0400
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Internet Mail Service (5.0.1458.49) (via Mercury MTS (Bindery) v1.30)
It would probably help if you posted the code...
> -----Original Message-----
> From: XHAJA Agim [SMTP:xhaja5@hei.unige.ch]
> Sent: Tuesday, September 09, 1997 1:23 PM
> To: RATS Discussion List
> Subject: Time-Varying procedure question.
>
>
>
> Dear Readers of the Rats list,
>
>
> I am a beginner in Rats- sorry if my questions are trivial, but I have
> no
> alternative persons to turn to.
>
> I am trying to estimate a regression where only one regressor
> is a time-varying coefficient (i.e. follows a random walk).
>
>
> Does RATS allow me to do it and if possible how do I proceed?
>
> My second concern is:
>
> When I try to regress the above mentioned equation with all
> coefficients varying in time and I follow step by step the
> procedure descripted in rats manual, chapter 13.
>
> I don't get any output and any error message!!!( Once, only I have
> received a "VAR4" error message .)
>
> I would appreciate if you can help me in solving those problems.
>
> Thank you very much,
>
> A. Xhaja
>
>
>
>
>
---------- End of message ----------
From: XHAJA Agim <xhaja5@hei.unige.ch>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: Time-Varying procedure question.
Date: Tue, 09 Sep 1997 22:02:54 +0200 (MET DST)
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
MIME-version: 1.0
Content-type: MULTIPART/MIXED; BOUNDARY="Boundary_(ID_PWr1mKa+cvesYuLIh+YY+A)"
X-Mailer: Mercury MTS (Bindery) v1.30
This message is in MIME format. The first part should be readable text,
while the remaining parts are likely unreadable without MIME-aware tools.
Send mail to mime@docserver.cac.washington.edu for more info.
--Boundary_(ID_PWr1mKa+cvesYuLIh+YY+A)
Content-type: TEXT/PLAIN; charset=US-ASCII
I am using the Rats 4.2 version and this is the prog i wrote.
thanx Agim
On Tue, 9 Sep 1997, Deitch, Jonathan wrote:
> It would probably help if you posted the code...
>
> > -----Original Message-----
> > From: XHAJA Agim [SMTP:xhaja5@hei.unige.ch]
> > Sent: Tuesday, September 09, 1997 1:23 PM
> > To: RATS Discussion List
> > Subject: Time-Varying procedure question.
> >
> >
> >
> > Dear Readers of the Rats list,
> >
> >
> > I am a beginner in Rats- sorry if my questions are trivial, but I have
> > no
> > alternative persons to turn to.
> >
> > I am trying to estimate a regression where only one regressor
> > is a time-varying coefficient (i.e. follows a random walk).
> >
> >
> > Does RATS allow me to do it and if possible how do I proceed?
> >
> > My second concern is:
> >
> > When I try to regress the above mentioned equation with all
> > coefficients varying in time and I follow step by step the
> > procedure descripted in rats manual, chapter 13.
> >
> > I don't get any output and any error message!!!( Once, only I have
> > received a "VAR4" error message .)
> >
> > I would appreciate if you can help me in solving those problems.
> >
> > Thank you very much,
> >
> > A. Xhaja
> >
> >
> >
> >
> >
>
--Boundary_(ID_PWr1mKa+cvesYuLIh+YY+A)
Content-id: <Pine.SOL.3.96.970909220254.24429B@hei.unige.ch>
Content-type: TEXT/PLAIN; name=shtator2; charset=US-ASCII
Content-description: this was my source program
Content-disposition: ATTACHMENT; FILENAME=shtator2
Content-transfer-encoding: BASE64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--Boundary_(ID_PWr1mKa+cvesYuLIh+YY+A)--
---------- End of message ----------
From: Robert Skarupke <skarr000@goofy.zdv.Uni-Mainz.de>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RATS with Panel-Data
Date: Wed, 10 Sep 1997 12:39:16 +0200 (MET DST)
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-Mailer: Mercury MTS (Bindery) v1.30
Hello everybody!
Can anyone help me with RATS-Instructions on Panel-Data?
I have two major questions:
1) I want to test for heteroscedasticity in a one-way error component
model with individual effects, so I use the SPREAD-Option in PSTATS. Is
the Test I get from RATS the Bartlett-Test as it is - like the Bartlett
Test - X2 distributed with (N-1) degrees of freedom?
2) In my case this test shows heteroscedasticity. How can I control for it
in the PANEL-Instruction using the SPREAD-Option. Can you give me an
example for such a command?
Thanks a lot for your help!
Robert (Mainz, Germany)
---------- End of message ----------
From: Renato Carandang <RCARAN1@allstate.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: unsubscribe
Date: Wed, 10 Sep 1997 07:53:50 -0600
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Novell GroupWise 4.1 (via Mercury MTS (Bindery) v1.30)
unsubscribe
---------- End of message ----------
From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: Time-Varying procedure question.
Date: Wed, 10 Sep 1997 10:42:54 -0600
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
X-mailer: Pegasus Mail for Win32 (v2.54) (via Mercury MTS (Bindery) v1.30)
Agim:
I don't know how you got a VAR4 error out of this, but there are a
few problems you need to fix.
Looking at your code:
> calendar 1967 1 12
> allocate 1991:2
> open data dataagim.dat
> data(org=obs) / y1 y3 y6 y9 y12
> open data data31.dat
> data(org=var) / p1 p3 p6 p9 p12
> set p31 = p3 - p1
> set y31 = y3 - y1
> equation kalmeq p31
> # constant y31
> compute gseesq=.9*%seesq
>
In this last line, you set GSEESQ equal to .9 times %SEESQ (the
reserved variable RATS uses to store the standard error of estimate
squared from a regression).
However, you haven't done any regressions, so at this point %SEESQ is
NA (i.e. is set to the missing value code). Thus GSEESQ will also be
equal to NA, which will cause major problems (all your KALMAN results
will be NA's as well). You either need to set this to some specific
value, or else do a preliminary regression, such as:
linreg p31
# constant y31
Then do:
compute gseesq=.9*%seesq
The next section looks OK:
> system kalmeq
> kfset(constant,noscale,likelihood=likely) xxx
> # gseesq
> tvarying tvx
> end(system)
> dim xxx(2,2) tvx(2,2)
> compute xxx=%const(0.0)
> compute xxx(1,1)=.04*gseesq
> compute xxx(2,2)=.025
> compute tvx=.00000001*xxx
> do time=67:1,91:2
> if time==67:1
> kalman(startup=time)
> else
> kalman printflag testflag
> end do time
>
Except for your use of "PRINTFLAG" and "TESTFLAG". Not sure what you
were expecting this to do--you haven't defined PRINTFLAG anywhere,
so you're not goint to get any output whatsoever. See the section on
KALMAN in Chapter 14 (particularly the 1st example on page 14-135)
for details on using logical expressions on these parameters to have
KALMAN produce output periodically.
First, though, I would recommend that you simply use the PRINT
option on KALMAN--RATS will produce output at each step, which
should help you figure out how your model is working and so on. So,
replace:
do time=67:1,91:2
if time==67:1
kalman(startup=time)
else
kalman printflag testflag
end do time
with:
do time=67:1,91:2
if time==67:1
kalman(startup=time)
else
kalman(print)
end do time
Sincerely,
Tom Maycock
Estima
--
+-----------------------------+-----------------------------------------+
| Estima | |
| P.O. Box 1818 | Voice: (847) 864-8772 |
| Evanston, IL 60204-1818 | Fax: (847) 864-6221 |
| U.S.A | BBS: (847) 864-8816 |
| e-mail: estima@estima.com | CompuServe: 73140,2202 |
|-----------------------------------------------------------------------|
| Web Site: http://www.estima.com |
| RATS Internet Mailing List: New members can join by sending e-mail to |
| MAISER@EFS.MQ.EDU.AU with the message: SUBSCRIBE RATS-L |
+-----------------------------------------------------------------------+
---------- End of message ----------
From: Paulo Freire-de-Oliveira <de.oliveira@mail.telepac.pt>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Unsubscribe
Date: Tue, 09 Sep 1997 15:01:22 +0100
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Mozilla 3.0 (Win95; I) (via Mercury MTS (Bindery) v1.30)
MIME-Version: 1.0
---------- End of message ----------
From: Kenneth Leong <kleong@tartarus.uwa.edu.au>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Periodic AR/ARIMA estimation coeds.
Date: Thu, 11 Sep 1997 11:37:08 +0800
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Windows Eudora Light Version 3.0.1 (32) (via Mercury MTS (Bindery) v1.30)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Dear all,
If I have the following periodic autoregressive model of order 1 (PAR(1)):
y(t) = a(s) + b(s)y(t-1) + e(t),
where t is quarterly observed in season s (s=1,2,3,4),
I would estimate this model using seasonal dummy variables (ie,
D1,D2,D3,D4), as in
y(t) = D1(t) + D2(t) + D3(t) + D4(t)
+ D1(t)*y(t-1) + D2(t)*y(t-1) + D3(t)*y(t-1) + D4(t)*y(t-1) + e(t).
My question is, if I have a periodic ARIMA model such as:
(1-a(s)L)(1-b(s)L^4)y(t) = e(t),
what would be the estimable equivalent and how would I estimate this in RATS?
Any replies will be much appreciated.
Kenneth.
---------- End of message ----------
From: "Craig Ingram" <atype3d@ozemail.com.au>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: unsubscribe
Date: Thu, 11 Sep 1997 15:00:18 +1000
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
MIME-Version: 1.0
Content-Type: text/plain;
Content-Transfer-Encoding: 7bit
X-Mailer: Microsoft Outlook Express 4.71.1008.3 (via Mercury MTS (Bindery) v1.30)
---------- End of message ----------
From: Rob Trevor <robt@efs.mq.edu.au>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: restricted MLE's
Date: Thu, 11 Sep 1997 15:05:08 +1000
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
X-Mailer: Mercury MTS (Bindery) v1.30
At 2:51 AM +1000 6/9/97, John O. Velis, Indiana University wrote:
> ...
> can any one suggest a good way to run MLE (specifically a GARCH model)
> with restrictions (i.e. 0<beta<1 ) on RATS.
>
> Clarification: I am not asking how to do teh GARCH model itself, just how
> to code the restrictions.
> ...
Try something like 'sqrt(beta)*sqrt(beta)' instead of just 'beta' in your
code. Its always worked for me...
Rob
-----------------------------------------------------------------------
Rob Trevor Associate Professor
Centre for Studies in Money, Banking and Finance
Macquarie University, NSW, AUSTRALIA 2109
Internet: robt@efs.mq.edu.au
VOICE: +61 (2) 9850-8447 FAX: +61 (2) 9850-7281 or +61 (2) 9418-1847
-----------------------------------------------------------------------
---------- End of message ----------
From: Economic Group <pgallagh@ozemail.com.au>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Adjusting standard errors in time varying parameters model
Date: Mon, 22 Sep 1997 09:56:43 +1000
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
X-Mailer: Mercury MTS (Bindery) v1.30
Hi,
I've been estimating a model with time varying parameters using the KALMAN
command in RATS 4.3. The modelling work is essentially complete now, but
after running some standard tests, I've found autocorrelation in the
residuals in the model.
Basically, I want to know if there is some way I can adjust the standard
errors in the model for the autocorrelation (a la the ROBUSTERRORS command
in LINREG). I've applied the standard Newey-West procedure to the model,
but am unsure as to whether it is still valid in the presence of time
varying parameters (more bluntly, I'm almost certain it is not valid).
Any information/feedback on:
a) Whether the standard errors need adjusting
b) How to do it
c) Whether the Newey-West procedure (ie. the method which the ROBUSTERRORS
option uses to adjust the standard errors) is appropriate
would be appreciated.
Thanks
James Vickery
Reserve Bank of Australia
Economic Analysis Department Email enquiries:
Economic Group PH: +612 9551 8817
Reserve Bank of Australia FAX: +612 9551 8833
---------- End of message ----------
From: Frieder Knuepling <knueplin@ruf.uni-freiburg.de>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Computational Accuracy.
Date: Tue, 23 Sep 1997 15:02:50 +0200
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Mozilla 3.03Gold (Win95; I) (via Mercury MTS (Bindery) v1.30)
MIME-Version: 1.0
Content-Type: multipart/mixed; boundary="------------6D2D2093452"
This is a multi-part message in MIME format.
--------------6D2D2093452
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
I have writte an estimation procedure calculating among others a number
of probabilities that should by construction sum up to unity, but for
RATS in some cases they don't.
I have attached a sample program and the output produced by RATS386
showing the computational inaccuracy.
Does any one have an explanation for the second line of the output?
Do you get different results when running the program?
Have you come across this problem before?
Yours,
Frieder
--
Frieder Knuepling
Albert-Ludwigs-Universitaet Freiburg
Institut fuer Allgemeine Wirtschaftsforschung
Abteilung Statistik und Oekonometrie
Belfortstr. 24
D-79098 Freiburg
Tel +49 761 / 203 - 2341
Fax +49 761 / 203 - 2298
--------------6D2D2093452
Content-Type: text/plain; charset=us-ascii; name="example.prg"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline; filename="example.prg"
declare real f1 f2 f3 f4 ff r1 r2 r3 r4 rr
com f1 = 1.0/7.0
com f2 = 2.0/7.0
com f3 = 3.0/7.0
com f4 = 4.0/7.0
com ff = f1+f2+f3+f4
com r1 = f1/ff
com r2 = f2/ff
com r3 = f3/ff
com r4 = f4/ff
dis 1.0-(r1+r2+r3+r4)
dis 1.0-r1-r2-r3-r4
--------------6D2D2093452
Content-Type: text/plain; charset=us-ascii; name="Output.txt"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline; filename="Output.txt"
0.00000
-1.11022e-016
--------------6D2D2093452--
---------- End of message ----------
From: Laurent =?iso-8859-1?Q?S=E9bastien?= <u114120@vm1.ulg.ac.be>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject:
Date: Tue, 23 Sep 1997 15:02:23 +0200
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Windows Eudora Light Version 3.0.1 (32) (via Mercury MTS (Bindery) v1.30)
Mime-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
HOW IS IT POSSIBLE TO ESTIMATE A WEIGHTED PROBIT MODEL ?
THANKS.
********************************************************
Laurent S=E9bastien * ******
CREPP-ULG * *
Tel.:04/366.31.07 * ******
Secr=E9tariat :04/366.31.08 * *=09
Fax :04/366.31.06 ****** ****** =09
********************************************************
---------- End of message ----------
From: Laurent =?iso-8859-1?Q?S=E9bastien?= <u114120@vm1.ulg.ac.be>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject:
Date: Tue, 23 Sep 1997 15:20:36 +0200
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Windows Eudora Light Version 3.0.1 (32) (via Mercury MTS (Bindery) v1.30)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
HOW IS IT POSSIBLE TO ESTIMATE A WEIGHTED PROBIT MODEL ?
THANKS.
---------- End of message ----------
From: Laurent =?iso-8859-1?Q?S=E9bastien?= <u114120@vm1.ulg.ac.be>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject:
Date: Tue, 23 Sep 1997 15:25:33 +0200
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Windows Eudora Light Version 3.0.1 (32) (via Mercury MTS (Bindery) v1.30)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
HOW IS IT POSSIBLE TO ESTIMATE A WEIGHTED PROBIT MODEL ?
THANKS.
---------- End of message ----------
From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Computational Accuracy.
Date: Tue, 23 Sep 1997 15:52:46 -0600
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
X-mailer: Pegasus Mail for Win32 (v2.54) (via Mercury MTS (Bindery) v1.30)
> I have writte an estimation procedure calculating among others a number
> of probabilities that should by construction sum up to unity, but for
> RATS in some cases they don't.
> I have attached a sample program and the output produced by RATS386
> showing the computational inaccuracy.
> Does any one have an explanation for the second line of the output?
> Do you get different results when running the program?
> Have you come across this problem before?
>
Dear Frieder:
This is simply an artifact of the fact that computers handle all
numbers using binary representations, rather than base-10
representations. For example, the decimal value 0.1 is represented
in binary as:
..11001100 x 2^-3
while 0.3 is represented as:
..100110011 x 2^-1
These are actually repeating decimals--I've truncated the values
here.
The important thing to note is the difference in the exponent.
Basically, this means that these two numbers are represented to
different precisions. Operations involving numbers that are
represented to different precisions can return slightly different
results depending on the order of the operation.
By way of demonstration, consider an example with much lower
precision than a PC, where numbers less than 1/2 are represented by
five decimals, and numbers greater than 1/2 represented by only four
decimals.
First, add 1/3 plus 1/3 plus 1/3, and note that you get exactly 1.0:
.33333
+ .33333
--------
.6667
+ .33333
--------
1.0000 (because the result is truncated to 4 decimals)
Now subtract off 1/3, 1/3, and 1/3, and note that you don't get
exactly 0.0:
1.0000
- .33333
--------
.6667 (because it is truncated to 4 decimals)
- .33333
--------
.33337
- .33333
--------
.00004
Finally, take your original example, but add 1.0 to each of the "R"
terms. This forces each component to have the same precision. Note
that the results are now exact:
com ff = f1+f2+f3+f4
com r1 = 1.0 + f1/ff
com r2 = 1.0 + f2/ff
com r3 = 1.0 + f3/ff
com r4 = 1.0 + f4/ff
dis 4.0-(r1+r2+r3+r4)
-1.00000
dis 4.0-r1-r2-r3-r4
-1.00000
Finally, remember that for an application using double-precision
reals for floating point operations (as RATS does), a result on the
order of 1e-16 is indistinguishible from zero if you are doing
operations on numbers of the order 1.0.
Hope this makes sense.
Sincerely,
Tom Maycock
--
+-----------------------------+-----------------------------------------+
| Estima | |
| P.O. Box 1818 | Voice: (847) 864-8772 |
| Evanston, IL 60204-1818 | Fax: (847) 864-6221 |
| U.S.A | BBS: (847) 864-8816 |
| e-mail: estima@estima.com | CompuServe: 73140,2202 |
|-----------------------------------------------------------------------|
| Web Site: http://www.estima.com |
| RATS Internet Mailing List: New members can join by sending e-mail to |
| MAISER@EFS.MQ.EDU.AU with the message: SUBSCRIBE RATS-L |
+-----------------------------------------------------------------------+
---------- End of message ----------
From: "Philippe PROTIN" <PROTIN@esa.upmf-grenoble.fr>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Hadamard matix product
Date: Wed, 24 Sep 1997 09:23:24 +0200
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
Organization: Ecole Superieure des Affaires
X-mailer: Pegasus Mail/Windows (v1.22) (via Mercury MTS (Bindery) v1.30)
Hi,
I just wonder if there exist a rats instruction to perform Hadamard
matrix product (element by element) ?
Thanks for help
Philippe PROTIN
ESA-CERAG
BP 47
38040 GRENOBLE CEDEX 9
--------------------------------
E-mail : protin@esa.upmf-grenoble.fr
--------------------------------
---------- End of message ----------
From: anas@mailccip.ccip.fr (M. ANAS - Centre d'Observation Economique)
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Computational Accuracy.
Date: Wed, 24 Sep 1997 13:56:41 +0200
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Windows Eudora Version 1.4.4 (via Mercury MTS (Bindery) v1.30)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
>What is strange is that you get 0 if you change the order (see example
below). I still do not understand what it could be !!
>Yours,
> Jacques
>
>--
>
>dis 1.0-(r2+r4+r1+r3)
>>
> 0.00000
>
>
Jacques Anas
CCIP-COE
27, avenue de Friedland
75008 PARIS
Tel.(1) 42-89-70-89
---------- End of message ----------
From: Christopher F Baum <baum@bc.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Hadamard matix product
Date: Wed, 24 Sep 1997 08:34:48 -0400
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Simeon for MacPPC Version 4.1.1 Build (16) (via Mercury MTS (Bindery) v1.30)
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; CHARSET=US-ASCII
ewise, perhaps?
On Wed, 24 Sep 1997 09:23:24 +0200 Philippe PROTIN
<PROTIN@esa.upmf-grenoble.fr> wrote:
> Hi,
> I just wonder if there exist a rats instruction to perform Hadamard
> matrix product (element by element) ?
>
> Thanks for help
> Philippe PROTIN
> ESA-CERAG
> BP 47
> 38040 GRENOBLE CEDEX 9
> --------------------------------
> E-mail : protin@esa.upmf-grenoble.fr
> --------------------------------
>
---------------------------------------------------------------------
Christopher F Baum Boston College Econ, Chestnut Hill MA 02167 USA
baum@bc.edu fax 617 552 2308 http://fmwww.bc.edu/EC-V/baum.fac.html
---------- End of message ----------
From: Frieder Knuepling <knueplin@ruf.uni-freiburg.de>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Speed.
Date: Tue, 30 Sep 1997 15:04:23 +0100
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
X-Mailer: Mozilla 3.03Gold (Win95; I) (via Mercury MTS (Bindery) v1.30)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Dear RATS users,
I am wondering why good old RATS386 runs about twice as fast (in a DOS
window under Windows 95) than my recently purchased WinRATS-32 (4.3),
using buffered or unbuffered text output in both versions. This applies
to processing data and an iterative estimation procedure using a number
of nested loops I have written.
Estima announced in their WWW Product Information that "WinRATS-32 runs
at about the same speed as RATS386".
Frieder Knuepling
--
Frieder Knuepling
Albert-Ludwigs-Universitaet Freiburg
Institut fuer Allgemeine Wirtschaftsforschung
Abteilung Statistik und Oekonometrie
Belfortstr. 24
D-79098 Freiburg
Tel +49 761 / 203 - 2341
Fax +49 761 / 203 - 2298
---------- End of message ----------
From: "Steve Cunningham" <Cunning@uconnvm.uconn.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Speed.
Date: Tue, 30 Sep 1997 09:36:57 -0500
Errors-to: <rats-l-owner@efs.mq.edu.au>
Reply-to: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Sender: Maiser@efs1.efs.mq.edu.au
X-listname: <RATS-L@efs1.efs.mq.edu.au>
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
X-mailer: Pegasus Mail for Win32 (v2.54) (via Mercury MTS (Bindery) v1.30)
On 30 Sep 97 at 15:04, Frieder Knuepling wrote:
> I am wondering why good old RATS386 runs about twice as fast (in a
> DOS window under Windows 95) than my recently purchased WinRATS-32
> (4.3), using buffered or unbuffered text output in both versions.
> This applies to processing data and an iterative estimation
> procedure using a number of nested loops I have written. Estima
> announced in their WWW Product Information that "WinRATS-32 runs at
> about the same speed as RATS386".
One suggestion that I can make. Windows has a setting which allocates
time (processor clock cycles= "computing power") to the various tasks
that may be ongoing at any given time. The default setting for a
windows program is 50%. Curious coincidence? You can change this so
that your Windows program gets 100% and won't be slowed by background
tasks. This will probably increase the performance of WinRATS-32
enormously.
_________________________
Steven R. Cunningham
Cunning@UConnVM.UConn.Edu
Department of Economics, University of Connecticut
and Connecticut Center for Economic Analysis
341 Mansfield Road, U-63
Storrs, Connecticut 06269-1063 USA
Voice: (860) 486-3022 FAX: (860) 486-4463
---------- End of message ----------