From: "R.Perman" <r.perman@strath.ac.uk>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: KALMAN filter
Date: Mon, 05 Jul 1999 15:15:06 +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: QUALCOMM Windows Eudora Light Version 3.0.3 (32) (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

Dear RATS listers

I would appreciate some advice on whether the following model can be
estimated in RATS using the Kalman filter, and if it can, how. (The model
is taken from a recent paper by Pindyck on the long run evolution of energy
prices):


P(t) = c1 + c2P(t-1) + b1(t) + b2(t)*TREND + e(t)

b1(t) = c3*b1(t-1) + v1(t)

b2(t) = c4*b2(t-1) + v2(t)


where e, v1 and v2 are error processes that are independent, and P is some
energy price. 

I find the section in the manual on the Kalman filter very difficult to
follow (and to generalise), hence my request.

Thanks in advance for any help that might be given me,

Best wishes

Roger Perman
Department of Economics
University of Strathclyde





---------- End of message ----------

From: Luca Cazzulani <luca.cazzulani@uni-bocconi.it>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: bidimensional kernel
Date: Tue, 06 Jul 1999 16:05:42 +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: QUALCOMM Windows Eudora Light Version 3.0.6 (32) (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

Is there a publicy available rats routine that performs bivariate kernel
estimation?

Luca Cazzulani


---------- End of message ----------

From: Marcello Pericoli <pericoli@mail.nexus.it>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: (no subject)
Date: Sat, 08 May 1999 21:26:46 -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: Mozilla 2.02 (Win16; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Alfonso.ARPAIA@EUROSTAT.cec.be wrote:
> 
> please help me to unsubscribe the following recipient from the  rats-l
>  mailing list   Alfonso.ARPAIA@EUROSTAT.cec.be
> despite of my efforts and following the right commands i can not get the
> 
> unsubscribing for this recipient which does not exist anymore
> thanks for your help
> mli mailfail management / vpCaro Alfonso,
anche tu come me hai da pena' per togliere tutta sta spazzatura!

Saluti Marcello

---------- End of message ----------

From: Binelli Maurizio <mbinelli@IFAM.IT>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: 
Date: Fri, 9 Jul 1999 17:25:12 +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>
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.0.1460.8) (via Mercury MTS (Bindery) v1.40)
Content-Type: text/plain

INDEX

---------- End of message ----------

From: Binelli Maurizio <mbinelli@IFAM.IT>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: 
Date: Fri, 9 Jul 1999 19:16:18 +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>
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.0.1460.8) (via Mercury MTS (Bindery) v1.40)
Content-Type: text/plain

LOOKUP <BVAR>

---------- End of message ----------

From: Binelli Maurizio <mbinelli@IFAM.IT>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: LOOKUP <BVAR>
Date: Mon, 12 Jul 1999 09:26:15 +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>
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.0.1460.8) (via Mercury MTS (Bindery) v1.40)
Content-Type: text/plain

LOOKUP <BVAR>

---------- End of message ----------

From: Binelli Maurizio <mbinelli@IFAM.IT>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: SEND RATS9602.ARC
Date: Mon, 12 Jul 1999 11:14:33 +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>
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.0.1460.8) (via Mercury MTS (Bindery) v1.40)
Content-Type: text/plain

SEND RATS9602.ARC

---------- End of message ----------

From: Rob Trevor <robt@efs.mq.edu.au>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: LIST COMMANDS
Date: Tue, 13 Jul 1999 16:20:24 +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" ; format="flowed"
X-Mailer: Mercury MTS (Bindery) v1.40

At 5:25 PM +0200 9/7/99, Binelli Maurizio wrote:
>INDEX

List commands (SUBSCRIBE, UNSUBSCRIBE, INDEX, etc) must be sent to 
the listSERVER address <Maiser@efs.mq.edu.au> rather than to the list 
itself.

Please do not send any more commands to this list.

Rob Trevor


---------- End of message ----------

From: "VIJAYA SARATHI N." <SARATHI@FPMLAN.iimahd.ernet.in>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: 
Date: Thu, 15 Jul 1999 02:35:59 GMT+0530
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:  IIM, Ahmedabad
X-mailer: Pegasus Mail v3.31 (via Mercury MTS (Bindery) v1.40)

hello colleagues,

i was running the following code and i got the follwing error 
message. could you help me understand the problem
****************
cal(daily) 96 4 4
all 0 98:8:31
open data f:\thesis2\ind1.wk1
data(format=wks,org=obs) / call fx y2 y6 y10 nifty
compute gstart=96:4:8, gend=98:8:31
nonlin  b11 b12 b13 b14 b15 b16 b17 $
    b21 b22 b23 b24 b25 b26 b27 $
    b31 b32 b33 b34 b35 b36 b37 $
    b41 b42 b43 b44 b45 b46 b47 $
    b51 b52 b53 b54 b55 b56 b57 $
    b61 b62 b63 b64 b65 b66 b67 $

FRML RESID1=call-b11-b12*call{1}-b13*fx{1}-b14*y2{1}-b15*y6{1}-
b16*y10{1}-b17*nifty{1}

## SX22. Expected Type FRML(REAL), Got RECTANGULAR Instead
>>>>=call-b11-b12*call{<<<<
************

regards,

sarathi

---------- End of message ----------

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: 
Date: Thu, 15 Jul 1999 12:51:12 -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) v1.40)

> 
> FRML RESID1=call-b11-b12*call{1}-b13*fx{1}-b14*y2{1}-b15*y6{1}-
> b16*y10{1}-b17*nifty{1}
> 
> ## SX22. Expected Type FRML(REAL), Got RECTANGULAR Instead
> >>>>=call-b11-b12*call{<<<<
> ************

You just need to put at least one space before the "=" sign. For 
example:

frml resid1 = call-b11-...

Without the space, RATS interprets "RESID1=..." as an expression
assigning a value to RESID1, which ends up defining RESID1 as a
variable type other than FRML, which then results in the error.

Sincerely,
Tom Maycock

--
------------------------------------------------------------
|   Estima                    |  Sales:   (800) 822-8038   |
|   P.O. Box 1818             |           (847) 864-8772   |
|   Evanston, IL 60204-1818   |  Support: (847) 864-1910   |
|   USA                       |  Fax:     (847) 864-6221   |
|   http://www.estima.com     |  estima@estima.com         |
------------------------------------------------------------

---------- End of message ----------

From: "VIJAYA SARATHI N." <SARATHI@FPMLAN.iimahd.ernet.in>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: 
Date: Wed, 21 Jul 1999 01:08:58 GMT+0530
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:  IIM, Ahmedabad
X-mailer: Pegasus Mail v3.31 (via Mercury MTS (Bindery) v1.40)

Hello colleagues,

could you tell me what diagnostics will help me detect if the MAXIMIZE 
command is getting struck in a local maximum ?

regards,

sarathi
IIM, India.

---------- End of message ----------

From: Simon van Norden <Simon.van-Norden@hec.ca>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Maxlik and local maxima
Date: Tue, 20 Jul 1999 21:59:31 -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>
Organization: Ecole des Hautes Etudes Commerciales
X-Mailer: Mozilla 4.51 [en] (Win98; U) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

There are almost none, and there are no really good ones.
I'd suggest

1) Try many different (randomly generated?) starting values. If you alway=
s wind
up in the same place regardless of where you start, you probably have a g=
lobal
maxima. However, I've run into problems where you can't seem to find *any=
* local
maxima unless you start near one, and certainly running lots of different
starting values takes much more time.

2) Think about the economics of the problem.  Sometimes, the values of th=
e
parameter estimates are wildly unreasonable because the maximization algo=
rithm
"gets stuck" somewhere in the parameter space that we don't usually think=
 about
but where the likelihood function is badly behaved. If the parameters see=
m out
by several orders of magnitude, worry.

3) Try simulated annealing. For code, see Goffe et al, in J. Econometrics=
 a few
years back. It takes a long time to run, but I've found nothing more
bullet-proof.

Hope this helps.

SvN


"VIJAYA SARATHI N." wrote:
>=20
> Hello colleagues,
>=20
> could you tell me what diagnostics will help me detect if the MAXIMIZE
> command is getting struck in a local maximum ?
>=20
> regards,
>=20
> sarathi
> IIM, India.

--=20
Simon van Norden, Prof. agr=E9g=E9, www.hec.ca/pages/simon.van-norden
Service de l'enseignement de la finance, =C9cole des H.E.C.
3000 Cote-Sainte-Catherine, Montreal QC, CANADA   H3T 2A7
simon.van-norden@hec.ca or (514)340-6781 or fax:(514)340-5632

---------- End of message ----------

From: Van & Bethanie Newby <newby@micron.net>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: 
Date: Tue, 20 Jul 1999 22:53:09 -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: QUALCOMM Windows Eudora Light Version 3.0.3 (32) (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

I think one possibility is to run MAXIMIZE for awhile with the SIMPLEX
option.  This should get you going on the right track for a global maximum.
 Since this doesn't fully answer your question (e.g., how many iterations
you should try before you know for sure that you're heading towards a
global maximum), let's see what others say.

Van Newby
Marriott School of Business
Brigham Young University

At 01:08 AM 7/21/1999 GMT+0530, you wrote:
>Hello colleagues,
>
>could you tell me what diagnostics will help me detect if the MAXIMIZE 
>command is getting struck in a local maximum ?
>
>regards,
>
>sarathi
>IIM, India.
>
>


---------- End of message ----------

From: "Konstantinos Tsatsaronis" <Konstantinos.Tsatsaronis@bis.org>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: GMM test specification
Date: Wed, 21 Jul 1999 17:37:55 +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: Novell GroupWise 5.5 (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable

Dear all,

RATS is a useful tool, but its manual is rather cryptic at times. My =
problem is quite simple but I am not sure I read the manual right.

I am estimating a simple linear regression of the type=20
y =3D a +b*X + u
and  I want to test the orthogonality conditions between a set of =
instruments Z (the X's are not among the Z's) and the error term u using =
GMM.

I am following the sample program shown in the manual (p. 5-22 in the =
section Method of Moments). My question is whether to include the X's in =
the set of instruments listed in the INSTRUMENT command or not. My guess =
is that the answer is yes, but I wonder whether thre is a soul out there =
that has looked into this issue more carefully.



Kostas Tsatsaronis
Monetary and Economic Dept.
Bank for International Settlements
CH-4051,  Basel, Switzerland

tel: +41 61 280 8082
fax: +41 61 280 9100

DISCLAIMER: Any e-mail messages from the Bank for International Settlements are sent in good faith, but shall not be binding nor construed as constituting any obligation on the part of the Bank.

CONFIDENTIALITY NOTICE: This e-mail contains confidential information, which is intended only for the use of the recipient(s) named above. If you have received this communication in error, please notify the sender immediately via e-mail and return the entire message. Thank you for your assistance.

 

---------- End of message ----------

From: "Jian Yang" <jian-yang@tamu.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Phillips and Hansen (1990) estimator
Date: Thu, 22 Jul 1999 17:08:22 -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>
X-Mailer: Novell GroupWise 5.5 (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: quoted-printable

Dear RATS users,

I would like to do some extensive tests on cointegration.  Can someone out =
there tell me where I can find help about programming of Phillips and =
Hansen (1990) fully modified maximum likelihood estimator.  Your help is =
highly appreciated.  Thank you in advance.


Jian Yang

Ph.D. candidate

Texas A&M University




---------- End of message ----------

From: richard.priestley@bi.no
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Andrews structural stability tests.
Date: Fri, 23 Jul 1999 11:05:05 +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>
Mime-Version: 1.0
Content-type: text/plain; charset=us-ascii
X-Mailer: Mercury MTS (Bindery) v1.40



I have a garch-m model where the mean is a function of a set of
instruments. I want to test for structural stability of the coefficients
wrt the instruments using Andrews, 1993, Tests for parameter instability
and structural change with unknown chnage point, Econometrica 61, 821-856,
type tests. This type of test has been used by Ghysels, 1997, On stable
factor structures in the pricing of risk...., Journal of Finance, in a GMM
framework.

Does anybody have RATS code for this type of test that may help me?

Thanks in advance


Richard Priestley
Associate Professor of Financial Economics
Department of Financial Economics
Norwegian School of Management
Elias Smiths vei 15
Sandvika
N1301
Norway
TEL: + 47 67557104
FAX: + 47 67557675
Email: richard.priestley@bi.no



---------- End of message ----------

From: "VIJAYA SARATHI N." <SARATHI@FPMLAN.iimahd.ernet.in>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: 
Date: Sat, 24 Jul 1999 20:23:05 GMT+0530
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:  IIM, Ahmedabad
X-mailer: Pegasus Mail v3.31 (via Mercury MTS (Bindery) v1.40)

hello colleagues,

could you advice regarding the incorporation of positivity 
constraints on regression coefficients in an ARCH model. how do i do 
it.

regards,

sarathi 

---------- End of message ----------

From: Rob Trevor <robt@efs.mq.edu.au>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re:
Date: Mon, 26 Jul 1999 16:42:29 +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" ; format="flowed"
X-Mailer: Mercury MTS (Bindery) v1.40

At 8:23 PM +1000 24/7/99, VIJAYA SARATHI N. wrote:
>could you advice regarding the incorporation of positivity
>constraints on regression coefficients in an ARCH model. how do i do
>it.

You can use something like abs(b1) or sqrt(b1)*sqrt(b1) (which is 
differentiable), or you can add penalty functions to the 
log-likelihood, etc.

Rob Trevor


---------- End of message ----------

From: Simon.Van-Norden@hec.ca
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: LINREG(CMOM) and EXCLUDE
Date: Mon, 26 Jul 1999 10:36:41 -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>
Organization: Ecole des Hautes Etudes Commerciales
X-Mailer: Mozilla 4.51 [en] (Win98; U) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: quoted-printable

I'm using LINREG(CMOM) in a loop to speed some calculations by modifying =
the
%cmom matrix myself every time through the loop. To make sure that I'm do=
ing the
calculations properly, I check the results against another program that j=
ust
does the usual LINREG.

Here's the problem. My LINREG(CMOM) gives the same %beta as the ordinary =
LINREG
every time through the loop.
However, when I then test a restriction on %beta using EXCLUDE, I get dif=
ferent
results from the two.

Is EXCLUDE using something other than %CMOM that I need to set for it to =
work
properly?=20


--=20
Simon van Norden, Prof. agr=E9g=E9, www.hec.ca/pages/simon.van-norden
Service de l'enseignement de la finance, =C9cole des H.E.C.
3000 Cote-Sainte-Catherine, Montreal QC, CANADA   H3T 2A7
simon.van-norden@hec.ca or (514)340-6781 or fax:(514)340-5632

---------- End of message ----------

From: szuniga@entelchile.net (SERGIO ZUNIGA  JARA)
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: How can I read that in RATS?
Date: Tue, 27 Jul 1999 00:35:13 -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;
Content-Transfer-Encoding: 7bit
X-Mailer: Microsoft Outlook Express 4.72.3110.5 (via Mercury MTS (Bindery) v1.40)

Dear Rats users:

I have a long matrix data (irregular in time), that contains only two
series: yield and frecuency.
Frecuency is times that yield must be repited for each observation.
How can I read that in RATS?
Thanks in advance,

***************************************************
Sergio Zuniga       szuniga@entelchile.net
Universidad Catolica del Norte
Larrondo 1281
Coquimbo - Chile
Tel.: 56-51-327248
***************************************************


---------- End of message ----------

From: Ashley Lyman <alyman@uidaho.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: 
Date: Wed, 28 Jul 1999 10:52:40 -0700 (PDT)
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) v1.40

Tom Maycock,

	The Linreg command uses some algorithm that, when sufficient
collinearity is present in the independent variables, will still elect to
print estimation results --but with some variables displayed with zeros
for their associated statistics in the display table.  My quuestions are:
[1] what algorithm is being used? and, [2] how is the exclusion of
variables (and hence zeros for results) arrived at? ( I haven't found a
reference in the manual.)

Sincerely,

Ashley Lyman
Department of Economics
University of Idaho
Moscow, Idaho 83844-3240






---------- End of message ----------

From: anjun zhou <a-zhou@students.uiuc.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: MAXIMIZE with BHHH 
Date: Wed, 28 Jul 1999 15:38:25 -0500 (CDT)
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) v1.40


HI, I tried to compare the performance of a model with its simplified 
version, using MAXIMIZE with BHHH method. According to the likelihood
ratio test, the two models are not significantly different from each
other. However, all the coefficients that appear only in the more complex 
model are significant. Can this be possible?

Maybe it has to do with the way the standard errors are estimated in the
procedure MAXIMIZE with BHHH method. So my question is: are they
derived from the inverse of the second-order derivatives or from the
product of the first-order derivatives? If the latter is used, then how
accurate are the standard error estimates? If there is a way to look at 
the content of MAXIMIZE function, it'd be great too.

Thank you very much for your help.

anjun zhou
University of Illinois


---------- End of message ----------

From: "Estima" <tomm@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Posting guidlines
Date: Wed, 28 Jul 1999 16:05:14 -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>
Organization: Estima
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
X-mailer: Pegasus Mail for Win32 (v3.11) (via Mercury MTS (Bindery) v1.40)

A few recent posts suggest that it is time to re-iterate our (that is, 
Estima's) general policy regarding the RATS-L mailing list. I offer 
this in the interest of mainting this list as a valuable resource for 
RATS users and for ensuring that folks who need technical support 
from us know how to go about getting it.  

1) Remember, we don't run the list, Rob Trevor does. Thus Rob 
always has final say on what is and is not appropriate use of the 
list.

2) If you want to send a question or comment to the RATS mailing 
list community, then send it to the mailing list address:
  RATS-L@efs.mq.edu.au

3) If, however, you want to ask Estima a tecnical support question, 
please send it directly to us at:
 support@estima.com (or estima@estima.com)

We do respond to many, but certainly not all, of the questions or 
comments posed on the mailing list.

4) Please don't do both (3) and (4) simultaneously. Doing so 
threatens to waste the time of a lot of folks through duplication of 
effort. Pick your target.

5) If you tried the list first and no one there can help you after a day 
or two, feel free to repeat the question directly to us.

6) Please, please, please don't send automated list commands 
(like unsubscribe, etc.) to this list. See our web site for instructions 
on subscribing, unsubscribing, etc.

Thanks for your time everyone.

Sincerely,
Tom Maycock
Estima

--
------------------------------------------------------------
|   Estima                    |  Sales:   (800) 822-8038   |
|   P.O. Box 1818             |           (847) 864-8772   |
|   Evanston, IL 60204-1818   |  Support: (847) 864-1910   |
|   USA                       |  Fax:     (847) 864-6221   |
|   http://www.estima.com     |  estima@estima.com         |
------------------------------------------------------------

---------- End of message ----------

From: "Estima" <tomm@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: multicollinearity
Date: Wed, 28 Jul 1999 16:05:37 -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>
Organization: Estima
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
X-mailer: Pegasus Mail for Win32 (v3.11) (via Mercury MTS (Bindery) v1.40)

> 	The Linreg command uses some algorithm that, when sufficient
> collinearity is present in the independent variables, will still elect to
> print estimation results --but with some variables displayed with zeros
> for their associated statistics in the display table.  My quuestions are:
> [1] what algorithm is being used? and, [2] how is the exclusion of
> variables (and hence zeros for results) arrived at? ( I haven't found a
> reference in the manual.)

If the tr(X)*X matrix is non-invertible (the basic sign that something's 
wrong), RATS applies what is effectively a generalized inverse 
computation that involves computing a Choleski factorization. 

If a diagonal element of the result effectively has no precision 
relative to the values that went into the computation (or is negative), 
RATS will "flag" the corresponding regressor, and report zeros for 
that regressor in the output. This is effectively like regressing one 
regressor on the preceiding regressors (based on the order listed 
on the supplementary card).

You can usually be sure that the regressor(s) flagged as zeros are 
directly involved in the cointegration. For example, if 

C = A + B

and you regress Y on A, B, C, and some other variables, A and B 
will show normal results while C will be flagged with zeros because 
that's when the problem will first show itself.

Sincerely,
Tom Maycock

--
------------------------------------------------------------
|   Estima                    |  Sales:   (800) 822-8038   |
|   P.O. Box 1818             |           (847) 864-8772   |
|   Evanston, IL 60204-1818   |  Support: (847) 864-1910   |
|   USA                       |  Fax:     (847) 864-6221   |
|   http://www.estima.com     |  estima@estima.com         |
------------------------------------------------------------

---------- End of message ----------