Return-Path: <maiser@efs.mq.edu.au>
Received: from corot.bc.edu (corot.bc.edu [136.167.2.209])
	by monet.bc.edu (8.8.7/8.8.7) with ESMTP id JAA59828
	for <baum@monet.bc.edu>; Fri, 2 Oct 1998 09:18:33 -0400
From: maiser@efs.mq.edu.au
Received: (from root@localhost)
	by corot.bc.edu (8.8.7/8.8.7) with X.500 id JAA191172
	for baum@mail1.bc.edu; Fri, 2 Oct 1998 09:18:32 -0400
Received: from baldrick.ocs.mq.edu.au (baldrick.ocs.mq.edu.au [137.111.1.12])
	by corot.bc.edu (8.8.7/8.8.7) with ESMTP id JAA298588
	for <baum@bc.edu>; Fri, 2 Oct 1998 09:18:22 -0400
Received: from efs1.efs.mq.edu.au (efs1.efs.mq.edu.au [137.111.64.8])
	by baldrick.ocs.mq.edu.au (8.9.0/8.9.0) with ESMTP id XAA22994
	for <baum@bc.edu>; Fri, 2 Oct 1998 23:18:14 +1000 (EST)
Received: from EFS1/SpoolDir by efs1.efs.mq.edu.au (Mercury 1.40);
    2 Oct 98 23:18:12 GMT+1000
Received: from SpoolDir by EFS1 (Mercury 1.40); 2 Oct 98 23:18:09 GMT+1000
To: baum@bc.edu
Date: Fri, 2 Oct 98 23:18:08 GMT+1000
Subject: Re: 
Message-ID: <57976CC567A@efs1.efs.mq.edu.au>

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Logit as a SUR
Date: Mon, 31 Aug 1998 10:04:33 -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)

> I have 2 separate questions:
> 
> (1): Is there any way I can estimate a set of logit
> regressions in RATs and then set it up as a SUR
> just for sake of estimation?
>  
> RATs will not allow me to, for example, use:
> 
> LGT(equation=EQ) LY
> 
> as this only allows LINREG and then one may
> set up a SUR using
> 
> SUR(VCV) 2
> # EQ
> 

There are several issues to consider here. First, the reason you
can't do what you describe above is simply that Logit is a
non-linear estimation, so it can't be expressed as an EQUATION or
estimated with Seemingly Unrelated Regression techniques. 

Moreover, Logit models cannot even be expressed as a least-squares 
type of problem, so they also cannot be estimated using Non-linear 
least squares (or as a system of non-linear least squares problems).

Logit models are instead estimated using maximum likelihood
estimation, as described on p. 14-140, and in Chapter 12. In
addition to the built-in LGT instruction, you can use MAXIMIZE
directly for estimating LGT and related models (see the multinomial
logit example on page 12-4, for example).

We're not aware of any methodology for dealing with a "system" of 
logit models (which I guess would be a case where multiple 
simultaneous choices are made), so I can't suggest how one might go 
about estimating such a beast. That doesn't mean it can't be 
done--just that we don't know how it would be done.

Sincerely,
Tom Maycock
Estima


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

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

From: Sune Karlsson <stsk@hhs.se>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Logit as a SUR
Date: Mon, 31 Aug 1998 18:21:07 +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: Mozilla 4.05 [en] (WinNT; I) (via Mercury MTS (Bindery) v1.40)
Content-type: text/plain; charset=us-ascii
Content-transfer-encoding: 7BIT

A system of logit models or a multivariate logit is pretty much
impossible since there is no "reasonable" extension of the logistic
distribution to the multivariate case. I have seen one extension to the
bivariate case but it only allows for positive correlations.

A multivariate probit on the other hand is fairly straightforward since
it is based on the normal distribution. It gets computationally messy
when you go beyond the bivariate case since you need to evaluate the
multivariate normal cdf. Limdep and probably Stata can estimate
multivariate probits.

/Sune

Estima wrote:
> 
> > I have 2 separate questions:
> >
> > (1): Is there any way I can estimate a set of logit
> > regressions in RATs and then set it up as a SUR
> > just for sake of estimation?
> >
> > RATs will not allow me to, for example, use:
> >
> > LGT(equation=EQ) LY
> >
> > as this only allows LINREG and then one may
> > set up a SUR using
> >
> > SUR(VCV) 2
> > # EQ
> >
> 
> There are several issues to consider here. First, the reason you
> can't do what you describe above is simply that Logit is a
> non-linear estimation, so it can't be expressed as an EQUATION or
> estimated with Seemingly Unrelated Regression techniques.
> 
> Moreover, Logit models cannot even be expressed as a least-squares
> type of problem, so they also cannot be estimated using Non-linear
> least squares (or as a system of non-linear least squares problems).
> 
> Logit models are instead estimated using maximum likelihood
> estimation, as described on p. 14-140, and in Chapter 12. In
> addition to the built-in LGT instruction, you can use MAXIMIZE
> directly for estimating LGT and related models (see the multinomial
> logit example on page 12-4, for example).
> 
> We're not aware of any methodology for dealing with a "system" of
> logit models (which I guess would be a case where multiple
> simultaneous choices are made), so I can't suggest how one might go
> about estimating such a beast. That doesn't mean it can't be
> done--just that we don't know how it would be done.
> 
> Sincerely,
> Tom Maycock
> Estima
> 
> ------------------------------------------------------------
> |   Estima                    |  Sales:   (800) 822-8038   |
> |   P.O. Box 1818             |  Support: (847) 864-1910   |
> |   Evanston, IL 60204-1818   |  Fax:     (847) 864-6221   |
> |   USA                       |  estima@estima.com         |
> |                             |  http://www.estima.com     |
> ------------------------------------------------------------

-- 
Sune Karlsson                      | stsk@hhs.se Fax: + 46 8 34 81 61
Stockholm School of Economics      | Phone: + 46 8 736 92 39
Box 6501, 113 83 Stockholm, Sweden | http://www.hhs.se/personal/SuneK/
http://www.hhs.se/stat/            | http://swopec.hhs.se/SWoPEc/

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

From: gerard.gregoire@imag.fr
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Logit as a SUR
Date: Mon, 31 Aug 1998 22:17:27 +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 3.01 [fr]C-NSCP  (Win95; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Rumi Masih wrote:
> 
> Dear RATs Users:
> 
> I have 2 separate questions:
> 
> (1): Is there any way I can estimate a set of logit
> regressions in RATs and then set it up as a SUR
> just for sake of estimation?
> 
> RATs will not allow me to, for example, use:
> 
> LGT(equation=EQ) LY
> 
> as this only allows LINREG and then one may
> set up a SUR using
> 
> SUR(VCV) 2
> # EQ
> 
> etc.
> 
> (2): Does RATS allow the estimation of unbalanced panels
> (either fixed or random effects)
> 
> Thanks for your assistance.
> 
> Rumi
> 
> ***************************************************************
> Rumi Masih, Faculty of Economics,
> University of Cambridge, England
> 
> Address for Correspondence:
> Rumi Masih, Wolfson College, Cambridge CB3 9BB, U.K.
>  Fax: +44-1223-335-908; E-Mail: rm212@cus.cam.ac.uk
> Web Page: http://www.ozemail.com.au/~masih/rumi/
> ***************************************************************
Using SAS ( proc CATMOD) you can deal with some types  of multivariate
logit models 
Namely:
Log(p_j/p_0)= alpha_j0+alpha_j1*x_1+...+ alpha_jp*x_p,    j=1,...,k.
(k equations with p variables)
Inference about the (p+1)*k  parameters is performed via ML method
(as pointed out by Tom Maycock).

Basic references on multivariate logit models can be found in (among
others) 
Moshe Ben-Akiva and Steven R. Lerman 
Discrete choice models; (MIT Press, 1985).
Some other interesting references can be founded
in biological literature.

Sincerely.
Gerard GREGOIRE


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

From: Carl Bonham <bonham@hawaii.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: problems skipping lines in fortran format
Date: Mon, 31 Aug 1998 11:23:44 -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) v1.40


I am trying to read a large number of  text files that have five lines of
extraneous
info at the very top of each file.  The remaining lines are simply data
and could be read free format if it were not for the first five lines I
need to skip.

Here is a small sample of the data file "OCCRATE.LIN"

  88
 0
OCCRATE
LINEAR SERIES
  4
       68.00000000
       70.23330000
       88.46670000
       70.63330000
       75.70000000
       73.06670000
       90.00000000
       74.96670000
       78.50000000

It looks like the fortran format I used below skips five lines, reads one,
and then skips five more.  The error message indicates it is thinks it is
reading observation 80:02 when it has actually reached the end of the file.

CALENDAR 60 1 4
ALLOCATE 0 1998:3

OPEN DATA C:\GRAPH\SERIES\OCCRATE.LIN
DATA(ORG=OBS, FORMAT= '(/////(F18.8))')  1976:03 1998:02 OCCRATE
## IO10. Unexpected end of file while processing line 94. (series OCCRATE
entry 80:02)


Any help with writing the Fortran format (or other format)  would be
appreciated.

Carl Bonham

`````````````````````````````````````````````````````````````````````
Carl Bonham                                    bonham@Hawaii.Edu
Department of Economics                phone: 808-956-7605
University of Hawaii at Manoa         fax: 808-956-4347
2424 Maile Way, Room 542            http://www2.hawaii.edu/~bonham/
Honolulu, HI  96822	         http://www2.hawaii.edu/~uhero
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~



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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: problems skipping lines in fortran format
Date: Mon, 31 Aug 1998 17:04:30 -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)

> I am trying to read a large number of  text files that have five lines of
> extraneous
> info at the very top of each file.  The remaining lines are simply data
> and could be read free format if it were not for the first five lines I
> need to skip.
> 
> It looks like the fortran format I used below skips five lines, reads one,
> and then skips five more.  

That's exactly what happens--the program skips 5, reads 1, skips 5, 
reads 1, etc. I don't know of any FORTRAN format that would just skip 
5 lines and then read all subsequent lines.

Fortunately, there are several ways around that. The easiest is 
probably to use a READ instruction to "use up" those first 5 lines, 
then apply the DATA command. For example:

OPEN DATA carl.dat
dec label junk
do i=1,5
 read(unit=data) junk
end do
DATA(ver,format=free,org=obs)  1976:03 1998:02 OCCRATE

This should do the trick.

Sincerely,
Tom Maycock
Estima
 

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

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

From: ALEJANDRO FRANCISCO JARA <ajara@ucla.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Hodrick-Prescott Filter
Date: Mon, 31 Aug 1998 16:16:46 -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

Hi,

There is anybody who knows how to compute a Hodrick-Prescott filter
in RATS?.

Thanks, 
Alejandro Jara
Department of Economics at UCLA




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

From: Simon.Van-Norden@hec.ca
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Hodrick-Prescott Filter
Date: Mon, 31 Aug 1998 19:53:24 -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 4.05 [en] (Win95; U) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

Here's a procedure (due, I think , to Estima.)

SvN

ALEJANDRO FRANCISCO JARA wrote:

> Hi,
>
> There is anybody who knows how to compute a Hodrick-Prescott filter
> in RATS?.
>
> Thanks,
> Alejandro Jara
> Department of Economics at UCLA



--
Simon van Norden        Professeur invité
simon.van-norden@hec.ca or svn@alum.mit.edu or http://www.hec.ca/~p280
(514)340-6781 or fax (514)340-5632 or cell (514)993-0466
Service de l'enseignement de la finance, Ecole des H.E.C.
3000 Cote-Sainte-Catherine, Montreal QC, CANADA   H3T 2A7



*
*  HPFILTER.SRC
*  Executes a Hodrick-Prescott Filter (Hodrick, R. & Prescott, E.,
*  "Post-War U.S. Business Cycles: An Empirical Investigation",
*   Carnegie-Mellon working paper, 1980.
*
*  Syntax:
*
*    @HPFILTER  series  start  end  growth component
*
*  Options:
*    LAMBDA=weight on squared 2nd difference of growth component [1600]
*
*  HPFILTER computes the estimated growth component using the DP
algorithm
*  to solve the problem:
*
*  Minimize sum over t of
*
*  (*) (y(t)-g(t))**2 + lambda * (g(t)-2g(t-1)+g(t-2))**2
*
*  where g is the growth component of y.
*
*      X(t)=(g(t),g(t-1),1)'
*      u(t)=g(t)-2g(t-1)+g(t-2)
*
*      X(t) = A X(t-1) + B u(t), where
*
*             2 -1  0        1
*        A =  1  0  0    B = 0
*             0  0  1        0
*
*  Using these definitions, the minimand (*) can be written
*
*     (c(t)X(t))**2 + lambda * u(t)**2
*       c(t)=(1,0,-y(t))
*
*  With cost-to-go functional  X(t)'R(t)X(t), the solution is of the
form
*  u(t)=TEMP1(t)*X(t-1), and X(t)=(A+B*TEMP1(t))*X(t-1)
*
PROCEDURE  HPFILTER  SERIES  START  END  OUTSER
TYPE SERIES  SERIES
TYPE INTEGER START  END
TYPE SERIES *OUTSER
*
OPTION  REAL  LAMBDA 1600
*
LOCAL INTEGER STARTL ENDL
*
LOCAL RECT A B C
LOCAL SYMM R RR
LOCAL RECT D
LOCAL RECT DMATS
LOCAL RECT TEMP1
LOCAL VECT RB X
LOCAL INTEGER TOTAL I J
*
IF .NOT.(%DEFINED(SERIES).AND.%DEFINED(OUTSER)) {
  DISPLAY 'Syntax: @HPFILTER  series  start  end   growth component'
  RETURN
}

INQUIRE(SERIES=SERIES) STARTL>>START ENDL>>END
COMPUTE TOTAL=ENDL-STARTL

DIM A(3,3) B(3,1) C(1,3) R(3,3) X(3)
DIM DMATS(9,TOTAL)

COMPUTE A=||2,-1,0|1,0,0|0,0,1||
COMPUTE B=||1|0|0||
COMPUTE C=||1,0,0||
COMPUTE R=%CONST(0.0)
*
*  DP transition matrices are set up.  Loop over periods until 2 above
*  beginning of data.
*
DO I=ENDL,STARTL+2,-1
   COMPUTE C(1,3)=-SERIES(I)
   OVERLAY DMATS(1,I-STARTL) WITH D(3,3)
   COMPUTE R=R+TR(C)*C
   COMPUTE SCALAR=LAMBDA+%QFORM(R,B)
   COMPUTE TEMP1=(-1.0/SCALAR)*TR(B)*R*A
   COMPUTE D=A+B*TEMP1
   COMPUTE R=%MQFORM(R,D)+LAMBDA*TR(TEMP1)*TEMP1
END DO I
*
*  Compute minimizer for the initial two periods.  Since s(1) and s(0)
*  are free initial parameters, u(1) and u(2) can be set equal to zero.
*  Thus, we only need to deal with the (y(t)-s(t))**2 terms.
*
COMPUTE C(1,3)=-SERIES(STARTL+1)
COMPUTE R=R+TR(C)*C
COMPUTE C=||0.0,1.0,-SERIES(STARTL)||
COMPUTE R=R+TR(C)*C
*
OVERLAY R(1,1) WITH RR(2,2)
OVERLAY R(3,1) WITH RB(2)
*
SET OUTSER STARTL ENDL = 0.0
COMPUTE RB=INV(RR)*RB
COMPUTE X(1)=OUTSER(STARTL+1)=-RB(1)
COMPUTE X(2)=OUTSER(STARTL)=-RB(2)
COMPUTE X(3)=1.0
*
DO I=STARTL+2,ENDL
  OVERLAY DMATS(1,I-STARTL) WITH D(3,3)
  COMPUTE X=D*X
  COMPUTE OUTSER(I)=X(1)
END DO I
END



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

From: Carlos Quintanilla <carlosq@umich.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Hodrick-Prescott Filter
Date: Mon, 31 Aug 1998 19:54:26 -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: QUALCOMM Windows Eudora Pro Version 4.0.1  (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: multipart/mixed;

--=====================_735282==_
Content-Type: text/plain; charset="us-ascii"

Alejandro:
The code for the HP Filter comes with RATS. If you don't 
have it, I've attached a copy of it.

	--CArlos


--=====================_735282==_
Content-Type: text/plain; charset="us-ascii"
Content-Disposition: attachment; filename="Hpfilter.src"

*
*  HPFILTER.SRC
*  Executes a Hodrick-Prescott Filter (Hodrick, R. & Prescott, E.,
*  "Post-War U.S. Business Cycles: An Empirical Investigation",
*   Carnegie-Mellon working paper, 1980.
*
*  Syntax:
*
*    @HPFILTER  series  start  end  growth component
*
*  Options:
*    LAMBDA=weight on squared 2nd difference of growth component [1600]
*
*  HPFILTER computes the estimated growth component using the DP algorithm
*  to solve the problem:
*
*  Minimize sum over t of
*
*  (*) (y(t)-g(t))**2 + lambda * (g(t)-2g(t-1)+g(t-2))**2
*
*  where g is the growth component of y.
*
*      X(t)=(g(t),g(t-1),1)'
*      u(t)=g(t)-2g(t-1)+g(t-2)
*
*      X(t) = A X(t-1) + B u(t), where
*
*             2 -1  0        1
*        A =  1  0  0    B = 0
*             0  0  1        0
*
*  Using these definitions, the minimand (*) can be written
*
*     (c(t)X(t))**2 + lambda * u(t)**2
*       c(t)=(1,0,-y(t))
*
*  With cost-to-go functional  X(t)'R(t)X(t), the solution is of the form
*  u(t)=TEMP1(t)*X(t-1), and X(t)=(A+B*TEMP1(t))*X(t-1)
*
PROCEDURE  HPFILTER  SERIES  START  END  OUTSER
TYPE SERIES  SERIES
TYPE INTEGER START  END
TYPE SERIES *OUTSER
*
OPTION  REAL  LAMBDA 1600
*
LOCAL INTEGER STARTL ENDL
*
LOCAL RECT A B C
LOCAL SYMM R RR
LOCAL RECT D
LOCAL RECT DMATS
LOCAL RECT TEMP1
LOCAL VECT RB X
LOCAL INTEGER TOTAL I J
*
IF .NOT.(%DEFINED(SERIES).AND.%DEFINED(OUTSER)) {
  DISPLAY 'Syntax: @HPFILTER  series  start  end   growth component'
  RETURN
}

INQUIRE(SERIES=SERIES) STARTL>>START ENDL>>END
COMPUTE TOTAL=ENDL-STARTL

DIM A(3,3) B(3,1) C(1,3) R(3,3) X(3)
DIM DMATS(9,TOTAL)

COMPUTE A=||2,-1,0|1,0,0|0,0,1||
COMPUTE B=||1|0|0||
COMPUTE C=||1,0,0||
COMPUTE R=%CONST(0.0)
*
*  DP transition matrices are set up.  Loop over periods until 2 above
*  beginning of data.
*
DO I=ENDL,STARTL+2,-1
   COMPUTE C(1,3)=-SERIES(I)
   OVERLAY DMATS(1,I-STARTL) WITH D(3,3)
   COMPUTE R=R+TR(C)*C
   COMPUTE SCALAR=LAMBDA+%QFORM(R,B)
   COMPUTE TEMP1=(-1.0/SCALAR)*TR(B)*R*A
   COMPUTE D=A+B*TEMP1
   COMPUTE R=%MQFORM(R,D)+LAMBDA*TR(TEMP1)*TEMP1
END DO I
*
*  Compute minimizer for the initial two periods.  Since s(1) and s(0)
*  are free initial parameters, u(1) and u(2) can be set equal to zero.
*  Thus, we only need to deal with the (y(t)-s(t))**2 terms.
*
COMPUTE C(1,3)=-SERIES(STARTL+1)
COMPUTE R=R+TR(C)*C
COMPUTE C=||0.0,1.0,-SERIES(STARTL)||
COMPUTE R=R+TR(C)*C
*
OVERLAY R(1,1) WITH RR(2,2)
OVERLAY R(3,1) WITH RB(2)
*
SET OUTSER STARTL ENDL = 0.0
COMPUTE RB=INV(RR)*RB
COMPUTE X(1)=OUTSER(STARTL+1)=-RB(1)
COMPUTE X(2)=OUTSER(STARTL)=-RB(2)
COMPUTE X(3)=1.0
*
DO I=STARTL+2,ENDL
  OVERLAY DMATS(1,I-STARTL) WITH D(3,3)
  COMPUTE X=D*X
  COMPUTE OUTSER(I)=X(1)
END DO I
END

--=====================_735282==_
Content-Type: text/plain; charset="us-ascii"


-------------------------------------------------------------------------
Carlos Quintanilla
University of Michigan
phone : (313) 747 7956
e-mail: carlosq@umich.edu
s-mail: P.O.Box 7643
        Ann Arbor, MI 48107
-------------------------------------------------------------------------

--=====================_735282==_--


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

From: arturo jose galindo <agalindo@students.uiuc.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Hodrick-Prescott Filter
Date: Mon, 31 Aug 1998 19:07:32 -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

There is a procedure called hpfilter.src or something similar in the RATS
program.

On Mon, 31 Aug 1998, ALEJANDRO FRANCISCO JARA wrote:

> Hi,
> 
> There is anybody who knows how to compute a Hodrick-Prescott filter
> in RATS?.
> 
> Thanks, 
> Alejandro Jara
> Department of Economics at UCLA
> 
> 
> 


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

From: Marco Hell <marco.hell@ki.comcity.de>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: ARFIMA
Date: Wed, 02 Sep 1998 16:05:09 +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: Comcity
X-Mailer: Mozilla 4.04 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Dear all,

is it possible to estimate an ARFIMA-modell with RATS?

Thank you.

--
        \\\|||||///
        \\  - -  //
         (  @ @  )
+------oOOo-(_)-oOOo---- +----------------------------------------+
|                        | Marco Hell                             |
| Have a nice day !      | Tel: 0049 (0)431/80.67.66              |
|                        | e-Mail: marco.hell@ki.comcity.de       |
+---------------Oooo-----+----------------------------------------+
        oooO   (   )
       (   )    ) /
        \ (    (_/
         \_)





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

From: "VIJAYA SARATHI N." <SARATHI@FPMLAN.iimahd.ernet.in>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: graphing impulse responses
Date: Wed, 2 Sep 1998 23:34:47 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)

Hi all:

I'm trying to graph the standardized impulse responses from a 11 
variable VAR model. The example no 8-3 in the RATS handbook(page 8-
15) gives a program, but it allows for a maximum of 9 variables.

could you advise me regarding graphing standardised responses of a 11-
variable model.

Thanks in advance for your help.

regards,

sarathi
doctoral candidate, 
IIM-A, India.

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

From: Ricardo Chaves Lima <rcl@netpe.com.br>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Multinomial Logit/Probit
Date: Wed, 02 Sep 1998 18:51:12 -0300
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 4.01 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Does anyone know a procedure to do multinomial
logit/probit on RATS?

thanks

Ricardo C. Lima
rcl@netpe.com.br


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

From: laurent ferrara CML RRC <Laurent.Ferrara@ratp.fr>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: ARFIMA
Date: Thu, 03 Sep 1998 18:29:00 -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 3.01 [fr] (Win16; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

There are many ways to estimate an ARFIMA model with RATS.
The procedure HURST.SRC available on the web's site of Estima allows to
estimate the d parameter. The David Montgomery's home page
(http://www2.hawaii.edu/~dmontgom/) contains the procedures GPH.SRC
which estimates the d and the procedure ARFIMA.SRC which estimates (MLE)
all the parameters simultaneously.
I would like to know if somebody wrote a procedure which makes it
possible to use the method of Whittle to estimate a FARIMA(p d q).

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

From: David S Montgomery <dmontgom@hawaii.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: ARFIMA
Date: Thu, 3 Sep 1998 11:02:03 -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) v1.40

My home page at www2.hawaii.edu/~dmontgom contains a rats procedure that
implements Sowell's method of MLE of an arfima model.  Note that this
procedure will only work for a select order of arma models.  

Good Luck,

David Montgomery


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

From: szuniga@entelchile.net (SERGIO ZUNIGA  JARA)
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: GMM Estimation of Cox-Ingersoll-Ross
Date: Thu, 3 Sep 1998 17:46:11 -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>
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)

Wolfgang:  Rats have an "almost ready to run" instructions for bivariate
GMM. Take a look to the Rats user's manual for details.
Any way, following CKLS here you have an example. Note that "term_1" is a
30-days interst rate time serie (for example).

CAL(WEEKLY) 1993 5 17
ALLOCATE 261
OPEN DATA c:\A_tesis\ratsdata\cap1c\data_ytm.txt
DATA(FORMAT=PRN, ORG=OBS)  /

TABLE
Series             Obs       Mean       Std Error      Minimum       Maximum
TERM_1              237   7.745000000   0.784509590   6.090000000
11.720000000

NONLIN ALFA BETA SIGMA2 GAMA
FRML H1 = term_1 - term_1{1} - ALFA - BETA*term_1{1}
FRML H2 = (DEPASO=H1(T)), DEPASO**2 - SIGMA2*term_1{1}**(2*GAMA)

COMPUTE ALFA=.05, BETA=-.06 ,SIGMA2=1. ,GAMA=1.

INSTRUMENTS CONSTANT  term_1{1}
* (CKLS used the constant and only one lag)
NLSYSTEM(INST,zudep) / H1 H2
CDF CHISQR %UZWZU 12*1+1

Because in our case GMM seems have more problems than ML, see Brenner et al.
(JFQA v.31 no1 march 1996. p 94).

Comments are welcome!!

Regards,

          \\  - -  //
         (  @ @  )
+--oOOo-(_)-oOOo-- +------------------------------------------+
|   Sergio Zuniga       |  U. Catolica del Norte - Chile |
| Tel: 56-51-327248  |  szuniga@entelchile.net |
+------------------------------+----------------------------------------+

-----Original Message-----
From: wobauer@studi.unizh.ch <wobauer@studi.unizh.ch>
To: szuniga@entelchile.net <szuniga@entelchile.net>
Date: Jueves 3 de Septiembre de 1998 03:35 AM
Subject: GMM Estimation of Cox-Ingersoll-Ross



>Hi,

>have you implemented the GMM approach for CIR estimation? Thats

>what I tried to do for Swiss data according to Chan et al...
...
>Wolfgang Bauer




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

From: Rumi Masih <rm212@cus.cam.ac.uk>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Fixed-Effect Logit?
Date: Thu, 03 Sep 1998 23:29:50 +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 3.0.5 (32) (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"


Tom: thanks a lot for your help.

I just had a realted question whilst I am
on the topic: it is possible to estimate
a logit on cross sectional pooled
data. That I know of. But have you heard
of a fixed effects estimation using logit?

Using LSDV is fair play, but when one
applies logit, is this valid? I have
pooled country data where my dependent
variable is dichotomous, with 4 RHS binary
and 2 continuous variables. If I then supplement
this with extra country dummies I get distorted
probabilities due to the 'fixed-effect'.

I am sure this estimation procedure
is not perfectly valid but cannot think
of the exact reason. RATs will do this
mechanically, but I am more interested
in the statistical validity. 

Would be glad if you could lend me
your assistance,

Best wishes, Rumi Masih

At 10:04 AM 8/31/98 -0600, you wrote:
>> I have 2 separate questions:
>> 
>> (1): Is there any way I can estimate a set of logit
>> regressions in RATs and then set it up as a SUR
>> just for sake of estimation?
>>  
>> RATs will not allow me to, for example, use:
>> 
>> LGT(equation=EQ) LY
>> 
>> as this only allows LINREG and then one may
>> set up a SUR using
>> 
>> SUR(VCV) 2
>> # EQ
>> 
>
>There are several issues to consider here. First, the reason you
>can't do what you describe above is simply that Logit is a
>non-linear estimation, so it can't be expressed as an EQUATION or
>estimated with Seemingly Unrelated Regression techniques. 
>
>Moreover, Logit models cannot even be expressed as a least-squares 
>type of problem, so they also cannot be estimated using Non-linear 
>least squares (or as a system of non-linear least squares problems).
>
>Logit models are instead estimated using maximum likelihood
>estimation, as described on p. 14-140, and in Chapter 12. In
>addition to the built-in LGT instruction, you can use MAXIMIZE
>directly for estimating LGT and related models (see the multinomial
>logit example on page 12-4, for example).
>
>We're not aware of any methodology for dealing with a "system" of 
>logit models (which I guess would be a case where multiple 
>simultaneous choices are made), so I can't suggest how one might go 
>about estimating such a beast. That doesn't mean it can't be 
>done--just that we don't know how it would be done.
>
>Sincerely,
>Tom Maycock
>Estima
>
>
>------------------------------------------------------------
>|   Estima                    |  Sales:   (800) 822-8038   |
>|   P.O. Box 1818             |  Support: (847) 864-1910   |
>|   Evanston, IL 60204-1818   |  Fax:     (847) 864-6221   |
>|   USA                       |  estima@estima.com         |
>|                             |  http://www.estima.com     |
>------------------------------------------------------------
>
>
***************************************************************
Rumi Masih, Faculty of Economics,
University of Cambridge, England

Address for Correspondence:
Rumi Masih, Wolfson College, Cambridge CB3 9BB, U.K.
 Fax: +44-1223-335-908; E-Mail: rm212@cus.cam.ac.uk
Web Page: http://www.ozemail.com.au/~masih/rumi/
***************************************************************

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

From: collinss@RBNZ.GOVT.NZ
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Maximum likelihood estimation of Kalman filter
Date: Fri, 4 Sep 1998 10:17:50 +1200
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=ISO-8859-1
Content-transfer-encoding: quoted-printable
X-Mailer: TFS Gateway /310000000/300006000/300001000/ (via Mercury MTS (Bindery) v1.40)

*** TFS Evaluation Copy of module : Novell GroupWise ***
Does anybody know of routines (or of a way) to estimate variance
covariance matrix for Kalman filter in RATS? =20

I presume it could in theory be done using the RATS routines for maximizing
functions, but it seems like it would be a pain=2E

Thanks,
Sean Collins
Reserve Bank of New Zealand
email: collinss@rbnz=2Egovt=2Enz
*****************************************************
The information contained in this message may be
CONFIDENTIAL and is intended for the addressee only=2E
Any unauthorized use, dissemination of the information,
or copying of this message is prohibited=2E If you
are not the addressee, please notify collinss@RBNZ=2EGOVT=2ENZ
immediately by return e-mail and delete this message=2E
Thank you=2E
*****************************************************

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

From: Daniel Porath <porath@jura1.jura.uni-mainz.de>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: ARFIMA
Date: Fri, 04 Sep 1998 12:45:10 +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 4.05 [en] (WinNT; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit



laurent ferrara CML RRC wrote:

> There are many ways to estimate an ARFIMA model with RATS.
> The procedure HURST.SRC available on the web's site of Estima allows to
> estimate the d parameter. The David Montgomery's home page
> (http://www2.hawaii.edu/~dmontgom/) contains the procedures GPH.SRC
> which estimates the d and the procedure ARFIMA.SRC which estimates (MLE)
> all the parameters simultaneously.
> I would like to know if somebody wrote a procedure which makes it
> possible to use the method of Whittle to estimate a FARIMA(p d q).


Dear Laurent,

I didn't write a procedure for Whittle-Estimation of ARIMA-models, but a
rather automatized program that perhaps could be of some use. It implements
full Whittle-Estimation as proposed by Smith, Sowell, Zin (1997 Empirical
Economics) or Hauser (1997 Empirical Economics) using as starting values the
GPH-estimator. It does not use the approximation by Fox/ Taqqu. Extensions
to this kind of estimator or higher lag orders are straightforward though.

If you have any comments, I would be grateful to hear of you.
Daniel

**** Whittle-Estimation of ARFIMA-Models

*set reihe = select a series
diff reihe / dreihe
statistics(noprint) dreihe
compute time = %nobs

***
source(noecho) c:\winrats\gph.src
@gph ddreihe
compute anfd = -%beta(2)
display anfd
***
frequency 6 %freqsize(time)
compute [integer] mint = %freqsize(time)/2 +1
compute [real] m = mint
rtoc 2 time+1 1
# dreihe
# 1
*
fft 1
cmultiply(scale=1./time) 1 1 / 2
*
ctor 2 mint 1     ***frequency 0 is suppressed
# 2
# period
***
smpl 1 mint-1
set lambda = 0.5*t/(m-1)

*
nonlin beta1 beta2 beta3 beta4 alpha1 alpha2 alpha3 alpha4  seq d

frml ma0 = 1.0
frml ma1 = (1.0 - beta1*cos(2*%pi*lambda))**2 +
(beta1**2)*(sin(2*%pi*lambda))**2
frml ma2 = (1.0 - beta1*cos(2*%pi*lambda)- beta2*cos(4*%pi*lambda))**2$
+ (beta1*(sin(2*%pi*lambda))+beta2*(sin(4*%pi*lambda)))**2

frml ma3 = (1.0 - beta1*cos(2*%pi*lambda)- beta2*cos(4*%pi*lambda)-
beta3*cos(6*%pi*lambda))**2 $
+(beta1*(sin(2*%pi*lambda))+beta2*(sin(4*%pi*lambda))+beta3*(sin(6*%pi*lambda)))**2

frml ma4 = (1.0 - beta1*cos(2*%pi*lambda)- beta2*cos(4*%pi*lambda)-
beta3*cos(6*%pi*lambda)-$
beta4*cos(8*%pi*lambda))**2 +
(beta1*(sin(2*%pi*lambda))+beta2*(sin(4*%pi*lambda))+$
beta3*(sin(6*%pi*lambda))+beta4*(sin(8*%pi*lambda)))**2


frml ar0 = 1.0
frml ar1 = (1.0 - alpha1*cos(2*%pi*lambda))**2 +
(alpha1**2)*(sin(2*%pi*lambda))**2
frml ar2 = (1.0 - alpha1*cos(2*%pi*lambda)- alpha2*cos(4*%pi*lambda))**2$
+ (alpha1*(sin(2*%pi*lambda))+alpha2*(sin(4*%pi*lambda)))**2

frml ar3 = (1.0 - alpha1*cos(2*%pi*lambda)- alpha2*cos(4*%pi*lambda)-
alpha3*cos(6*%pi*lambda))**2 $
+(alpha1*(sin(2*%pi*lambda))+alpha2*(sin(4*%pi*lambda))+alpha3*(sin(6*%pi*lambda)))**2

frml ar4 = (1.0 - alpha1*cos(2*%pi*lambda)- alpha2*cos(4*%pi*lambda)-
alpha3*cos(6*%pi*lambda)-$
alpha4*cos(8*%pi*lambda))**2 +
(alpha1*(sin(2*%pi*lambda))+alpha2*(sin(4*%pi*lambda))+$
alpha3*(sin(6*%pi*lambda))+alpha4*(sin(8*%pi*lambda)))**2


dofor [frml] autoreg = ar0 ar1 ar2 ar3 ar4
   dofor [frml] movav = ma0 ma1 ma2 ma3 ma4
    frml arma = seq*(movav/autoreg)
    frml arfima = arma*((2.*sin(%pi*lambda))**(-2.*d))
    frml bruch = period/arfima

    frml like = -log(arfima) - bruch
    compute beta1=0., alpha1=0., alpha2=0., alpha3=0., alpha4=0., beta2=0.,
$
    beta3=0., beta4=0., seq=200.0, d=anfd

    nlpar(subiterations=50,cvcrit=0.0001)
    maximize(iterations=100) like
    compute zahl = 0
    do index = 1,%nreg,1
    if %xx(index,index) == 0.0
       compute zahl = zahl + 1
    else
       compute zahl = zahl + 0
    end do index

    compute anz = %nreg - zahl - 1
compute aic = (-2.*%funcval + 2*anz)/%nobs
compute sc = (-%funcval + 0.5*anz*log(%nobs))/%nobs
display "aic - min = " aic
display "sc - min = " sc
display "number of regressors" anz
end do index

   end dofor j
end dofor i
***********






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

From: Wolfgang Bauer <wobauer@studi.unizh.ch>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Newey West covariance estimator
Date: Fri, 4 Sep 1998 14:26:02 +0200 (CEST)
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

Dear RATS users,

I d like to use Newey/West s (1987) estimator for covariances for a 
autocorrelated model. As I understand from a first glance at the 
handbook (14-157), another estimator (Whites?) is implemented in 
the programme. Has anyone code for the Newey/West estimator?

Thanks a lot
Wolfgang Bauer




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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Newey West covariance estimator
Date: Fri, 4 Sep 1998 10:08:53 -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)

> 
> I d like to use Newey/West s (1987) estimator for covariances for a 
> autocorrelated model. As I understand from a first glance at the 
> handbook (14-157), another estimator (Whites?) is implemented in 
> the programme. Has anyone code for the Newey/West estimator?
> 


To get the Newey-West estimator, you need to use the option DAMP=1 in 
addition to the ROBUSTERRORS and LAGS options. See the MCOV section 
of Chapter 14 for details.

Sincerely,
Tom Maycock
Estima


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

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

From: Wolfgang Bauer <wobauer@studi.unizh.ch>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Sums in FRML
Date: Fri, 4 Sep 1998 19:53:51 +0200 (CEST)
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

Dear RATS users,
I try to do a GMM estimation. The moment restrictions contain sum (like
from i=1 to n). How can I put this into a FRML without writing out every
single term?

Thanks in advance for advice

Wolfgang Bauer




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

From: "VIJAYA SARATHI N." <SARATHI@FPMLAN.iimahd.ernet.in>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: SRC procedures
Date: Fri, 4 Sep 1998 23:50:55 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)

Hi all:

I'm trying to incorporate some SRC procedures from the estima web 
site. I'm using a Linux web browser. I get only ascii characters when 
i download the procedure from the ftp site.

could you help find a way of downloading the procedures using a unix 
browser.

Thanks for your help.

Regards,

Sarathi
Doctoral Candidate
IIM-A, India.

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

From: wobauer@studi.unizh.ch
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Fortran Format
Date: Tue, 8 Sep 1998 22:21:31 +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
Content-transfer-encoding: 7BIT
X-mailer: Pegasus Mail for Win32 (v3.01b) (via Mercury MTS (Bindery) v1.40)

Dear Rats users,
I'd like to write a matrix to a file with a specified number of digits 
ritght to the decimal point. The decimal format (fd.w) works, but how 
can I get a matrix form, i.e. a (MxN) table?

Thanks 

Wolfgang Bauer

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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Fixed-Effect Logit?
Date: Tue, 8 Sep 1998 16:37:34 -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)

> 
> I just had a realted question whilst I am
> on the topic: it is possible to estimate
> a logit on cross sectional pooled
> data. That I know of. But have you heard
> of a fixed effects estimation using logit?
> 
> Using LSDV is fair play, but when one
> applies logit, is this valid? I have
> pooled country data where my dependent
> variable is dichotomous, with 4 RHS binary
> and 2 continuous variables. If I then supplement
> this with extra country dummies I get distorted
> probabilities due to the 'fixed-effect'.
> 
 
Rumi:

If I understand you correctly, I don't think there should be any
problems with what you are proposing. For logit models, you can't use 
implement fixed effects estimation by using the PANEL instruction to 
transform the variables. However, using dummy variables to implement 
fixed effects should be perfectly valid.

Please let me know if you have any questions.

Sincerely,
Tom Maycock
Estima

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

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

From: David Edgerton <david.edgerton@nek.lu.se>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: Fixed-Effect Logit?
Date: Wed, 9 Sep 1998 10:15:32 +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 (5.5.1960.3) (via Mercury MTS (Bindery) v1.40)
Content-Type: multipart/mixed;

This message is in MIME format. Since your mail reader does not understand
this format, some or all of this message may not be legible.

------ =_NextPart_000_01BDDBCA.03D93ED0
Content-Type: text/plain

Hi!
The fixed effects logit suggested by Tom is not valid  due to the
nusiance parameter problem. Quite simply, the dummy variable
coefficients are not consistent for fixed T and increasing N. The trick
of differencing as in LSDV will not work here since the nuisance
parameters no longer dissapear.
The correct procedure to use is Chamberlin's conditional logit, which
maximises the likelihood conditioned on a minimal sufficient statstic,
namely Siyit.This is described in e.g. Greene's textbook or Baltagi's
book on panel data. The procedure is implemented in LIMDEP and STATA.
Sincerely
David
-----------------------------------------------------------------
David Edgerton
Dept of Economics 
P.O. Box 7082
S-220 07 Lund 
Sweden

tel: (+46) 46 222 8678
fax: (+46) 46 222 4613
e-mail: David.Edgerton@nek.lu.se


	-----Original Message-----
	From:	Estima [SMTP:estima@estima.com]
	Sent:	09 September 1998 00:38
	To:	RATS Discussion List
	Subject:	Re: Fixed-Effect Logit?

	> 
	> I just had a realted question whilst I am
	> on the topic: it is possible to estimate
	> a logit on cross sectional pooled
	> data. That I know of. But have you heard
	> of a fixed effects estimation using logit?
	> 
	> Using LSDV is fair play, but when one
	> applies logit, is this valid? I have
	> pooled country data where my dependent
	> variable is dichotomous, with 4 RHS binary
	> and 2 continuous variables. If I then supplement
	> this with extra country dummies I get distorted
	> probabilities due to the 'fixed-effect'.
	> 
	 
	Rumi:

	If I understand you correctly, I don't think there should be any
	problems with what you are proposing. For logit models, you
can't use 
	implement fixed effects estimation by using the PANEL
instruction to 
	transform the variables. However, using dummy variables to
implement 
	fixed effects should be perfectly valid.

	Please let me know if you have any questions.

	Sincerely,
	Tom Maycock
	Estima

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

------ =_NextPart_000_01BDDBCA.03D93ED0
Content-Type: application/ms-tnef
Content-Transfer-Encoding: base64
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------ =_NextPart_000_01BDDBCA.03D93ED0--

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

From: wobauer@studi.unizh.ch
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Fortran Format
Date: Tue, 8 Sep 1998 22:21:31 +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
Content-transfer-encoding: 7BIT
X-mailer: Pegasus Mail for Win32 (v3.01b) (via Mercury MTS (Bindery) v1.40) (via Mercury MTS (Bindery) v1.40)

Dear Rats users,
I'd like to write a matrix to a file with a specified number of digits 
ritght to the decimal point. The decimal format (fd.w) works, but how 
can I get a matrix form, i.e. a (MxN) table?

Thanks 

Wolfgang Bauer

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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Fixed-Effect Logit?
Date: Tue, 8 Sep 1998 16:37:34 -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) (via Mercury MTS (Bindery) v1.40)

> 
> I just had a realted question whilst I am
> on the topic: it is possible to estimate
> a logit on cross sectional pooled
> data. That I know of. But have you heard
> of a fixed effects estimation using logit?
> 
> Using LSDV is fair play, but when one
> applies logit, is this valid? I have
> pooled country data where my dependent
> variable is dichotomous, with 4 RHS binary
> and 2 continuous variables. If I then supplement
> this with extra country dummies I get distorted
> probabilities due to the 'fixed-effect'.
> 
 
Rumi:

If I understand you correctly, I don't think there should be any
problems with what you are proposing. For logit models, you can't use 
implement fixed effects estimation by using the PANEL instruction to 
transform the variables. However, using dummy variables to implement 
fixed effects should be perfectly valid.

Please let me know if you have any questions.

Sincerely,
Tom Maycock
Estima

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

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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Fortran Format
Date: Wed, 9 Sep 1998 10:19:30 -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)


> Dear Rats users,
> I'd like to write a matrix to a file with a specified number of digits 
> ritght to the decimal point. The decimal format (fd.w) works, but how 
> can I get a matrix form, i.e. a (MxN) table?
> 

Just use the same type of FORMAT in a WRITE instruction. For example:


dec rect r(5,10)
ewise r(i,j) = %ran(200)
open copy mat.txt
write(unit=copy,format='(10f15.5)') r

Sincerely,
Tom Maycock
Estima

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

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

From: Jean-Francois Guilly <guillyj@CIRANO.UMontreal.CA>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: How to edit the content of a formula ?
Date: Wed, 9 Sep 1998 14:11:35 -0400 (EDT)
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=ISO-8859-1
Content-Transfer-Encoding: QUOTED-PRINTABLE
X-Mailer: Mercury MTS (Bindery) v1.40


Dear Rats users,

I have a little question for you. I want to understand the content of a
Rats program I have been given.

In the program, there is a part like the one below. It is quite easy to
understand. But I would like to be sure of the content of the formula
created below.

Before the part below, a few steps were done. After a stepwise regression
where DIFF_INFL((USD) is the explained variable, REGLIST has been created
which contains the list of regressors from the stepwise. S_COEFFS is a
vector of the coefficients of the stepwise regression.
=20

IF PAYS=3D=3DUSD) {
  EQUATION(IDENTITY) EQ_INFL_USD DIFF_INFL(USD)
  # REGLIST
  ASSOCIATE EQ_INFL_USD S_COEFFS
  FRML(IDENT,EQUATION=3DEQ_INFL_USD) F_INFL_USD
 =FD=20

How must I understand the FRML formula ?=20
Can it be edited ? If so, which command must I use ?
Is it only a vector of regression coefficients ?

Or can it be understood as:

EQUATION would give something like

Y =3D a + b1*X1 + b2*X2 + b3*X3 + epsilon

and FORMULA would give something like:

Y =3D ahat + b1hat*X1 + b2hat*X2 + b3hat*X3 + epsilon


Thanks in advance.


Jean-Francois Guilly
Canada


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

From: Jean-Francois Guilly <guillyj@CIRANO.UMontreal.CA>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Initial approximation for GROUP command
Date: Wed, 9 Sep 1998 17:13:51 -0400 (EDT)
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

Dear Rats users,

Does anyone know the initial approximation used in the Gauss-Siedel
algorithm used by RATS (when using the GROUP command).

Thanks in advance

Jean-Francois Guilly
Canada


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

From: Klaus Fischer <Klaus.Fischer@fas.ulaval.ca>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: How to edit the content of a formula ?
Date: Wed, 9 Sep 1998 19:43:48 -0400 (EDT)
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 Pro Version 2.2 (16) (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: 8bit

Hello J-P:

Tomorrow morning I'll answer all your messages....See you

KF

At 14:11 98-09-09 -0400, you wrote:
>
>Dear Rats users,
>
>I have a little question for you. I want to understand the content of a
>Rats program I have been given.
>
>In the program, there is a part like the one below. It is quite easy to
>understand. But I would like to be sure of the content of the formula
>created below.
>
>Before the part below, a few steps were done. After a stepwise regression
>where DIFF_INFL((USD) is the explained variable, REGLIST has been created
>which contains the list of regressors from the stepwise. S_COEFFS is a
>vector of the coefficients of the stepwise regression.
> 
>
>IF PAYS==USD) {
>  EQUATION(IDENTITY) EQ_INFL_USD DIFF_INFL(USD)
>  # REGLIST
>  ASSOCIATE EQ_INFL_USD S_COEFFS
>  FRML(IDENT,EQUATION=EQ_INFL_USD) F_INFL_USD
> ý 
>
>How must I understand the FRML formula ? 
>Can it be edited ? If so, which command must I use ?
>Is it only a vector of regression coefficients ?
>
>Or can it be understood as:
>
>EQUATION would give something like
>
>Y = a + b1*X1 + b2*X2 + b3*X3 + epsilon
>
>and FORMULA would give something like:
>
>Y = ahat + b1hat*X1 + b2hat*X2 + b3hat*X3 + epsilon
>
>
>Thanks in advance.
>
>
>Jean-Francois Guilly
>Canada
>
>
>
Klaus P. Fischer, Ph.D.
Faculté des sciences de l'adminsitration et
Centre de recherche en économie et finance appliquées (CREFA)
Université Laval
Quebec, G1K 7P4, CANADA
Ph. 418-656-2131 X3679
Fax 418-656-7746
Internet: Klaus.Fischer@fas.ulaval.ca


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

From: Klaus Fischer <Klaus.Fischer@fas.ulaval.ca>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: How to edit the content of a formula ?
Date: Wed, 9 Sep 1998 19:48:38 -0400 (EDT)
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 Pro Version 2.2 (16) (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: 8bit

Dear Colleagues

Sorry, I sent a personal message to the Rats Discussion list as a non-intended
response to Jean-Francois Guilly.

K. Fischer
U. Laval

At 14:11 98-09-09 -0400, you wrote:
>
>Dear Rats users,
>
>I have a little question for you. I want to understand the content of a
>Rats program I have been given.
>
>In the program, there is a part like the one below. It is quite easy to
>understand. But I would like to be sure of the content of the formula
>created below.
>
>Before the part below, a few steps were done. After a stepwise regression
>where DIFF_INFL((USD) is the explained variable, REGLIST has been created
>which contains the list of regressors from the stepwise. S_COEFFS is a
>vector of the coefficients of the stepwise regression.
> 
>
>IF PAYS==USD) {
>  EQUATION(IDENTITY) EQ_INFL_USD DIFF_INFL(USD)
>  # REGLIST
>  ASSOCIATE EQ_INFL_USD S_COEFFS
>  FRML(IDENT,EQUATION=EQ_INFL_USD) F_INFL_USD
> ý 
>
>How must I understand the FRML formula ? 
>Can it be edited ? If so, which command must I use ?
>Is it only a vector of regression coefficients ?
>
>Or can it be understood as:
>
>EQUATION would give something like
>
>Y = a + b1*X1 + b2*X2 + b3*X3 + epsilon
>
>and FORMULA would give something like:
>
>Y = ahat + b1hat*X1 + b2hat*X2 + b3hat*X3 + epsilon
>
>
>Thanks in advance.
>
>
>Jean-Francois Guilly
>Canada
>
>
>
Klaus P. Fischer, Ph.D.
Faculté des sciences de l'adminsitration et
Centre de recherche en économie et finance appliquées (CREFA)
Université Laval
Quebec, G1K 7P4, CANADA
Ph. 418-656-2131 X3679
Fax 418-656-7746
Internet: Klaus.Fischer@fas.ulaval.ca


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

From: wobauer@studi.unizh.ch
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Sums in FRML 2
Date: Thu, 10 Sep 1998 10:44:18 +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
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
X-mailer: Pegasus Mail for Win32 (v3.01b) (via Mercury MTS (Bindery) v1.40)

Dear RATS users,
unfortunately I haven't received any hints to my question below. As 
an important part of my current work depends on the answer I pose 
it again hoping for more success.
I try to do a GMM estimation. The moment restrictions contain a sum (like
from i=1 to n). How can I put this into a FRML without writing out every
single term? Or is there another way to put up more complicated moment conditions

Thanks in advance for advice

Wolfgang Bauer





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

From: Marco Hell <marco.hell@ki.comcity.de>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Error message in CATS
Date: Fri, 11 Sep 1998 19:34:06 +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: Comcity
X-Mailer: Mozilla 4.04 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Dear all,

I've got a question about CATS. After starting the CATS-procedure there
comes the error message YOU HAVE MISSING OBSERVATIONS IN THE ENDOGENOUS
SERIES
THIS IS NOT ALLOWED IN THIS PROCEDURE. But if I check my endogenous
variables then they are complete. If anybody has an idea, then let me
know.

Greets,
Marco Hell.

--
        \\\|||||///
        \\  - -  //
         (  @ @  )
+------oOOo-(_)-oOOo---- +----------------------------------------+
|                        | Marco Hell                             |
| Have a nice day !      | Tel: 0049 (0)431/641168                |
|                        | e-Mail: marco.hell@ki.comcity.de       |
+---------------Oooo-----+----------------------------------------+
        oooO   (   )
       (   )    ) /
        \ (    (_/
         \_)



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

From: "geronimi" <geronimi@infobiogen.fr>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: KALMAN FILTER PROCEDURE
Date: Sun, 13 Sep 1998 20:06:08 +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: Microsoft Internet Mail 4.70.1155 (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Dear all,=20

Trying to modelize a time varying system using Kalman filter's procedure,=
 I
am wondering if it is possible to introduce a non null variance Mt in the
state process equation
Bt =3D Bt-1 + vt for the study of time varying system (chapter 13.2 of RA=
TS
USER MANUAL).
=20
Thanks

Vincent Geronimi
Ma=EEtre de Conf=E9rences
Universit=E9 Versailles-Saint-Quentin
France
geronimi@infobiogen.fr

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

From: David Gelernter <gelernter.1@osu.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: computing a numerical gradient 
Date: Mon, 14 Sep 1998 18:07:04 -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: QUALCOMM Windows Eudora Pro Version 4.0.1  (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

Dear RATS users:

To compute a Wald test of a nonlinear function of the coefficients of  a
VAR, I'm having a problem with the gradient of the function with respect to
the VAR coefficients.  Specifically, I have some function f(b), where the
b=(b1, b2, ..., bn) is a vector of coefficients.  

I need the vector df(b)/db', where the ith element of this vector is the
partial derivative of f with respect to the ith element of b.

To compute the numerical partial derivative of f with respect to the ith
element of b, I'm using:

df/db(i) = [ f ( b1, b2, . . ., bi+e, . . . , bn) - f (b1, b2, . . ., bi ,
. . .bn) ] / e  , where e is a small number.

I think this is the standard way of doing this (for example in Hamilton's
textbook).  My problem is in the choice of the number e.  

I get very different results when I choose, say e=0.0000001 versus
e=0.000000000000000001.  Theoretically, the smaller number should be
better, but in practice, the smaller number can be a problem if RATS is
rounding during its calculations.  I think the fact that some of the
elements of b are quite small (like 0.000005) and others are large (like
200), suggests that I should choose the value of e depending on which
element of b I'm varying.

The User's manual says (on page 4-5) that "RATS does the computations at
full precision".  Does this mean that I should not have a problem with the
smaller number?  

I would appreciate any thoughts.

Thanks.





S. David Gelernter
Department of Economics
410 Arps Hall
1945 North High Street
Columbus, OH 43210
Phone: (614) 292-0275
E-mail: gelernter.1@osu.edu



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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: computing a numerical gradient 
Date: Tue, 15 Sep 1998 09:59:34 +0000
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 Windows (v2.54) (via Mercury MTS (Bindery) v1.40)

> Dear RATS users:
> 
> To compute a Wald test of a nonlinear function of the coefficients of  a
> VAR, I'm having a problem with the gradient of the function with respect to
> the VAR coefficients.  Specifically, I have some function f(b), where the
> b=(b1, b2, ..., bn) is a vector of coefficients.  
> 
> I need the vector df(b)/db', where the ith element of this vector is the
> partial derivative of f with respect to the ith element of b.
> 
> To compute the numerical partial derivative of f with respect to the ith
> element of b, I'm using:
> 
> df/db(i) = [ f ( b1, b2, . . ., bi+e, . . . , bn) - f (b1, b2, . . ., bi ,
> . . .bn) ] / e  , where e is a small number.
> 
> I think this is the standard way of doing this (for example in Hamilton's
> textbook).  My problem is in the choice of the number e.  
> 
> I get very different results when I choose, say e=0.0000001 versus
> e=0.000000000000000001.  Theoretically, the smaller number should be
> better, but in practice, the smaller number can be a problem if RATS is
> rounding during its calculations.  I think the fact that some of the
> elements of b are quite small (like 0.000005) and others are large (like
> 200), suggests that I should choose the value of e depending on which
> element of b I'm varying.

e=10e-18 will probably produce 0 derivatives in most contexts. In 
fact, 1+10e-18=1 in machine arithmetic.

For models where coefficients are expected to be a few orders of
magnitude one way or the other from 1.0 (Box-Jenkins models, e.g.),
I use e=.0000001. MAXIMIZE is a bit trickier because it is quite
possible to have coefficients with a natural scale which is very
small or very large. What RATS does there, beginning with iteration
2, is to use .00001*sqrt(stderr(i)) where i is the current estimate
of the standard error for the coefficient, with some extra logic to
make sure it doesn't take too large an e if the standard error is
really large, as in an underidentified model. For your situation, 
that might work pretty well.


> The User's manual says (on page 4-5) that "RATS does the computations at
> full precision".  Does this mean that I should not have a problem with the
> smaller number?  

The citation on 4-5 refers to the fact that the summary statistics 
(such as %RSS), coefficients and standard errors are not printed at 
the full 14-15 digit precision, so that you can get very different 
results from putting printed output values through a calculator 
versus doing the same calculation symbolically using RATS 
instructions.

RATS stores numbers in double precision. Depending upon the machine 
and compiler, some intermediate calculations might be done in the 
slightly higher precision used by the floating point processor, but 
effectively it does double precision.

> I would appreciate any thoughts.
> 
> Thanks.
> 
> S. David Gelernter
> Department of Economics
> 410 Arps Hall
> 1945 North High Street
> Columbus, OH 43210
> Phone: (614) 292-0275
> E-mail: gelernter.1@osu.edu
--
+-----------------------------+-----------------------------------------+
| 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: Simon.Van-Norden@hec.ca
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: computing a numerical gradient
Date: Tue, 15 Sep 1998 11:16:48 -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>
Organization: Ecole des Hautes Etudes Commerciales
X-Mailer: Mozilla 4.05 [en] (Win95; U) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

I'd just like to add to Estima's comments regarding the effects of the scale of
the coefficients.

Old, Wise men have often advised me that nonlinear estimation works better when
all the parameters you seek to estimate are roughly the same scale (the largest
is, say, no more that 100 time the smallest.)  The reason for this is precisely
the kind of problem Estima is mentioning.  It is therefore often amazingly useful
to just rescale some of your parameters or data series in order to help get
accurate estimation of the gradients.

--
Simon van Norden   Professeur invité
E-mail:  simon.van-norden@hec.ca   or    svn@alum.mit.edu
Home Page:       http://www.hec.ca/pages/simon.van-norden
(514)340-6781 or fax (514)340-5632 or  cell (514)993-0466
Service de l'enseignement de la finance, École des H.E.C.
3000 Cote-Sainte-Catherine, Montreal QC, CANADA   H3T 2A7



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

From: "Coleman, Mark" <MColeman@dlj.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Interest rate process
Date: Tue, 15 Sep 1998 11:22:56 -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>
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.1960.3) (via Mercury MTS (Bindery) v1.40)
Content-Type: text/plain

Greetings,

I apologize in advance if this is not quite an appropriate question for
this list. I am searching for source code implementing a single factor
stochastic interest rate model, e.g., Vasicek, Black-Derman-Toy, etc.
Source code were in RATS it would be fabulous. I will happily translate
source code from some other language into RATS, however, and then place
the RATS code in the public domain if there is interest.

We are working on a bond pricing model and have used RATS to estimate
key valuation parameters and functions. I would like to develop the rest
of the system in RATS and avoid the costs of switching to C, C++, Java,
or some other statistical development package. I realize that RATS may
not optimal for this type of project, but RATS has performed admirably
for us and I'm a die-hard fan of the software!

Any help would be much appreciated!

Best Regards,

Mark
mcoleman@dlj.com

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

From: <kohns@IIW.UNI-BONN.DE>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Automatizing the ENVIRONMENT command
Date: Tue, 15 Sep 1998 18:07:16 MET-1MEST
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: IIW - University of Bonn
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
X-mailer: Pegasus Mail for Windows (v2.42a) (via Mercury MTS (Bindery) v1.40)

Dear RATS users,

 I have a relatively large number of data files, each of which contains 
data series (employment, wages etc.) for a specific industrial sector. 
The labels of the series are the same in all files. Since the 
same type of estimation is to be applied to all the files, I would 
like to automatize the reading of the data. 
To this end I tried the following:
1. I defined a vector FILELIST of labels that conains the filenames.
2. I created a DOFOR loop with a variable FILENAME that loops 
over the entries of FILELIST.
3. Finally, I used ENV ratsdata = FILENAME do read the data.
Unfortunately, RATS does not interprete FILENAME as a variable 
but as a filename (and consequently asks wether do create a new file 
FILENAME).
 Does anyone have an idea how to make RATS understand 
FILENAME as a variable?
 Any help would be greatly appreciated.

 Stephan

***************************************************
* Dipl.-Volkswirt Stephan Kohns                   *
* Institute for International Economics           *
* University of Bonn                              *
* Germany                                         *
* Tel.: 0049-228-739234                           *
* E-Mail: kohns@iiw.uni-bonn.de                   *
***************************************************

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

From: "Christopher F Baum" <baum@bc.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Automatizing the ENVIRONMENT command
Date: Tue, 15 Sep 1998 14:10:36 -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: Mulberry (MacOS) [1.3.4, s/n P020-300786-009] (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Stephan,

This is how it may be done. This code fragment reads a list of 26 dates from
filename.dat. I have 26 files with names like s19980101.dat; in each file
there is a line indicating how many observations to read (of a single
variable). This code fragment does not go farther than that, but this should
do what you need.

all 2 20
dec vector[string] date(2)
dec int nobs1 nobs2 last

open data filename.dat
do i=1,26
input(unit=data) date(i)
end do i

*this loops over dates 
do I=1,26
*
*define filenames for read
*
compute [string] in1='s'+&date(i)+'.dat'

open data &in1
input(unit=data) nobs1
dis &date(i) nobs1
close data
end do l
END


Kit Baum
Boston College
SSC Archive maintainer
http://ideas.uqam.ca/ideas/data/bocbocode.html


--On Tue, Sep 15, 1998 18:07 +0000 kohns@IIW.UNI-BONN.DE wrote: 

> Dear RATS users,
> 
>  I have a relatively large number of data files, each of which contains 
> data series (employment, wages etc.) for a specific industrial sector. 
> The labels of the series are the same in all files. Since the 
> same type of estimation is to be applied to all the files, I would 
> like to automatize the reading of the data. 
> To this end I tried the following:
> 1. I defined a vector FILELIST of labels that conains the filenames.
> 2. I created a DOFOR loop with a variable FILENAME that loops 
> over the entries of FILELIST.
> 3. Finally, I used ENV ratsdata = FILENAME do read the data.
> Unfortunately, RATS does not interprete FILENAME as a variable 
> but as a filename (and consequently asks wether do create a new file 
> FILENAME).
>  Does anyone have an idea how to make RATS understand 
> FILENAME as a variable?
>  Any help would be greatly appreciated.
> 
>  Stephan
> 
> ***************************************************
> * Dipl.-Volkswirt Stephan Kohns                   *
> * Institute for International Economics           *
> * University of Bonn                              *
> * Germany                                         *
> * Tel.: 0049-228-739234                           *
> * E-Mail: kohns@iiw.uni-bonn.de                   *
> ***************************************************



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

From: "VIJAYA SARATHI N." <SARATHI@FPMLAN.iimahd.ernet.in>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: VAR with I(0) and I(1) variables
Date: Wed, 16 Sep 1998 03:21:45 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)

Dear colleagues:

I'm estimating a VAR with some I(1) and some I(0) variables. what is  
the effect of this difference in the order of integration amongst 
the variables on the inferences drawn from impulse response analysis 
and variance decomposition?

thank you in advance for your thoughts.

Regards,

sarathi
Doctoral Candidate
IIM-A, India.

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

From: Ying-Man Chung <y_mchung@yahoo.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: 
Date: Tue, 15 Sep 1998 21:36:23 -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


I conducted a single-equation partial vector error correction model
and found that ARCH(4) is significant. Then I think the results for
coefficients may not be reliable, so that I want to use Monte Carlo
simulation method to confirm that my results empirically reliable.
However, I want to ask for the RATS code for ARCH simulation.

The model for ARCH is:

et = ao + a1 et-1 + a2 et-2 +...

May I ask how to use RATS command to simulate et, please??

Thank you.

REgards,

Sarah
HKSYC 



_________________________________________________________
DO YOU YAHOO!?
Get your free @yahoo.com address at http://mail.yahoo.com


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

From: Ying-Man Chung <y_mchung@yahoo.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: ARCH simulation
Date: Tue, 15 Sep 1998 21:35:13 -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


I conducted a single-equation partial vector error correction model
and found that ARCH(4) is significant. Then I think the results for
coefficients may not be reliable, so that I want to use Monte Carlo
simulation method to confirm that my results empirically reliable.
However, I want to ask for the RATS code for ARCH simulation.

The model for ARCH is:

et = ao + a1 et-1 + a2 et-2 +...

May I ask how to use RATS command to simulate et, please??

Thank you.

REgards,

Sarah
HKSYC 



_________________________________________________________
DO YOU YAHOO!?
Get your free @yahoo.com address at http://mail.yahoo.com


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

From: "Massimo Fabiano" <massimofabiano@hotmail.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: univariate equation and standard error
Date: Wed, 16 Sep 1998 08:19:48 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>
Content-Type: multipart/mixed; boundary="====================987654321_0==_"
X-Mailer: Mercury MTS (Bindery) v1.40

--====================987654321_0==_
Content-Type: text/plain; charset="us-ascii"

Dear RATS users,
I need help to solve a problem you will find in the attached file.

Regards,

Massimo Fabiano
Bocconi University

______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com


--====================987654321_0==_
Content-Type: application/octet-stream; name="DOC.DOC"
Content-Transfer-Encoding: x-uuencode
Content-Disposition: attachment; filename="DOC.DOC"

begin 644 DOC.DOC
MT,\1X*&Q&N$`````````````````````/@`#`/[_"0`&```````````````!
M````+0``````````$```,`````$```#^____`````"X```#_____________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M_______________________LI<$`60`0!```"!*_````````$```````!```
M[P@```X`8FIB:O-7\U<````````````````````````0!!8`'A(``)$]`0"1
M/0$`[P0```````````````````````````````````````#__P\`````````
M``#__P\```````````#__P\``````````````````````%T``````*8`````
M````I@```*8`````````I@````````"F`````````*8`````````I@```!0`
M`````````````+H`````````N@````````"Z`````````+H`````````N@``
M``P```#&````%````+H`````````00@``+8```#F````%@```/P`````````
M_`````````#\`````````/P`````````_`````````#\`````````/P`````
M````!@@```(````("`````````@(````````"`@````````("`````````@(
M````````"`@``"0```#W"```]`$``.L*``">````+`@``!4`````````````
M````````````I@````````#\``````````````````````````````#\````
M`````/P`````````_`````````#\`````````"P(``````````<```````"F
M`````````*8`````````_````````````````````/P`````````Y@``````
M````!P`````````'``````````<```````#\````!`8``*8`````````_```
M``````"F`````````/P`````````!@@`````````````````````````````
MN@````````"Z`````````*8`````````I@````````"F`````````*8`````
M````_``````````&"``````````'```&`0````<```````````````````8(
M````````I@````````"F````````````````````````````````````````
M````````````````````````````````````````````!@@```````#\````
M`````-H````,````8%LR3W?AO0&Z`````````+H```````````<````````&
M"```````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````1&5A<B!20513('5S97)S+`U)(&YE960@:&5L<"!T;R!S;VQV92!A
M('!R;V)L96TN#4D@97-T:6UA=&5D(&$@=6YI=F%R:6%T92!E<75A=&EO;B!O
M9B!T:&ES('1Y<&4@.@T3($5-0D5$($5Q=6%T:6]N+C,@(!0!%0UA;F0@22!C
M;VUP=71E9"!T:&4@(&1Y;F%M:6,@;75L=&EP;&EE<B!O9B!Y="!C875S960@
M8GD@82!P97)M86YE;G0@=F%R:6%T:6]N(&]F($UT(&]V97(@,S8@<&5R:6]D
M.@U9,3T@*#`[#5DR/2@P*R@Q*R@Q*"@P*0U9,STH,2@H,"LH,2LH,2@H,"DI
M*R@R*"`H,"DK("@P*R@Q*R@R#2X-+B`@#0U)('=O=6QD(&QI:V4@=&\@:VYO
M=R!I9B`@86YY;VYE(&ES(&%W87)E(&]F(&$@9V%U<W,@;W(@4F%T<R!R;W5T
M:6YE('1O(&-O;7!U=&4@=&AE('-T86YD87)D(&5R<F]R(&]F('1H:7,@:6UP
M=6QS92!R97-P;VYS92!F=6YC=&EO;BX@#0T-#0T-=&4@82!.975R86P@3F5T
M=V]R:S\@("`@('-O;65O;F4@02!M;V1E;"`H;V8@<VEG;6$@8V]N=F5R9V5N
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M;V1E;"DL(&)O=&@@=6YD97(@=&AE($AO.G(],`UA;F0@2&\Z<CTQ/S\_(`T-
M4F5G87)D<RP-````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M``````````````````0``&($``!C!```=@0``'<$``!X!```>00``*4$``"F
M!```RP0``,P$``#>!```WP0``.$$``#B!```XP0``.8$``#G!```Z`0``.D$
M``#J!```ZP0``.P$``#M!```[@0``.\$``#P!```\00``/($``#S!```]@0`
M`/<$``#X!```^00``/H$``#[!```_`0``/T$``#^!```_P0````%```!!0``
M`@4```,%```$!0``!04```8%```)!0``"@4```L%```-!0``#@4```\%```2
M!0``$P4``!0%```5!0``%@4``!<%```8!0``&04``!L%``#O"````/H`\.GZ
M`.<`YP#G`./G`.<`X^<`X^<`W^<`X^<`YP#?YP#CYP#CYP#?YP#CYP#?YP#C
MYP#CYP#CYP#CYP``````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M```````````&"6H!`&?P``8):@$`;/```T@J`@T#:@````!%2/#_50@!$P-J
M@#VJ.0H(`54(`58(`6U(``0)`VH`````50@!`#X`!```$00``#$$``!B!```
M>@0``-T$``#E!```]00``!L%```=!0``(04``"(%``"H!0``J04``*H%``"K
M!0``K`4``*T%```7!@``7@8``)\&``"N!@``O`8``,H&``#=!@``[P8``!H'
M``!C!P``K`<``.$'``#]````````````````_0```````````````/T`````
M``````````#]````````````````_0```````````````/T`````````````
M``#]````````````````_0```````````````/T```````````````#]````
M````````````_0```````````````/T```````````````#]````````````
M````_0```````````````/T```````````````#]````````````````_0``
M`````````````/T```````````````#]````````````````_0``````````
M`````/T```````````````#]````````````````_0```````````````/T`
M``````````````#]````````````````_0```````````````/T`````````
M``````#]````````````````_0`````````````````````````````!````
M'0`$```1!```,00``&($``!Z!```W00``.4$``#U!```&P4``!T%```A!0``
M(@4``*@%``"I!0``J@4``*L%``"L!0``K04``!<&``!>!@``GP8``*X&``"\
M!@``R@8``-T&``#O!@``&@<``&,'``"L!P``X0<``/,'``#T!P``!@@``$L(
M``"."```U@@``.4(``#F"```[P@`````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M```````````````````````FX0<``/,'``#T!P``!@@``$L(``"."```U@@`
M`.4(``#F"```[P@``/T```````````````#]````````````````_0``````
M`````````/T```````````````#]````````````````_0``````````````
M`/T```````````````#]````````````````_0``````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M`````````````````````````````````````````0````D<`!^P@BX@L,9!
M(;!N!"*P;@0CD(D%))!N!"6P````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M`````````),"``!$`&0``````````@````````````````"4#,P!Z`/H`P``
M````````````````````````````````````````#P`$\$(```"R!`KP"```
M``$$````"@``4P`+\!X````$00$````_`0``!@"!`1$``!"_`0P`'P#_`0``
M"````!#P!````````(`R``?P_0$```,$O[7D7I=_RSI'C[;2_P!S@?\`V0$`
M``$```!$``````"&`&`A&_#1`0``O[7D7I=_RSI'C[;2_P!S@?X#````````
M`````"`4``#@`@``&#(?`'AT!`"?`0```/YXG(V3STL"013'WSQUU55LU22+
M#HM$ES!2ZA`1^`_8(3T'5DL)KALHA3<]>>W>O]!_8`>O_0==BB6J6^"I2]@V
M/];!W=AJEV'F,^_[YKTW/PC$`4(.`4#(`?LBM,61C<*T(2%\AJ#C.'QFD^3Y
M3`F)JT[@S"^)-NJ+&3I:5Q9@!1SF!!KE,1V-:+.S`#=4GW`U2:@VNN?UWH4!
MT.?*3QQ/>2)PS0(@_;5:SSRV6B*C(IV-T;ZL[N)>A$E>)3\E!8MU=.?W=0CO
M)W*]`>Y$F71?\D-JGBOXEF%<E'RI"A;Q^E]_Y2WB$=??)J,$DVY(SH>]_$Q$
M_-S/>B!,UZ\W3:.C'QA7^J%E-MJ0#JAL`B\*.Z3R]JRR6UYI2=J'*:_]=('9
MMV2E[ZK0^S,A`9DL!]0\P$>%^30E]U/S7$$KZ^4AW_.NY'O-:U]2O?:/F->N
M$&'/^<XH*/.@/;3)6IKY5"5;<<8]R:LAQB>2CT#8_;>#_/-V3/D9M"07>&5G
MH+IO1^7OBU\&-X8&44YW_$42Q$*MU^D:)K3]-86X[AM];XJ2````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M```````````!`````@````,````$````!0````8````'````"`````D```#^
M____"P````P````-````#@````\````0````$0```/[____]____%````!<`
M``#^____&````"`````9````&@```!L````<````'0```!X````?````_O__
M__[_________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M_________________________________U(`;P!O`'0`(`!%`&X`=`!R`'D`
M``````P````,````!```````````````"````````````0`````````````6
M``4!__________\*````!@D"``````#`````````1@````#@=-9-=^&]`>`4
M74]WX;T!%@`````2```"````1`!A`'0`80````````````,````#````!`8`
M````````````"````````````0`````````````+````"P````H``@'_____
M__________\\`$``Q-I!`*0%``"D!0``5R_7$```B`48`,$`!0`8`0```$`*
M`````!````````!7`&\`<@!D`$0`;P!C`'4`;0!E`&X`=````&(`_____R@`
M````````I`4```,```"+`````@```*(````8````&@`"`0D```#_________
M_Z0%```"````=`````(```"+````&````"#,8@#_____*``````````>$@``
M`0```$\`8@!J`&4`8P!T`%``;P!O`&P```#__R@`````````I`4```@````^
M`````@```%4````8````(,QB`/____\6``$!#````/____\$````````````
M``````````````````#@-^].=^&]`>`474]WX;T!!P`````````"````7P`Y
M`#8`-P`T`#4`.``Q`#<`-@``````````````````````````````````````
M`````````````````````!8``0'__________P8````"S@(``````,``````
M``!&`````*!?^$YWX;T!@"@)3W?AO0$````````````````!`$\`;`!E````
M````````````````````````````````````````````````````````````
M````````````"@`"`?_______________P``````````````````````````
M```````````````````````````4``````````$`0P!O`&T`<`!/`&(`:@``
M````````````````````````````````````````````````````````````
M```2``(!!0````<```#_____````````````````````````````````````
M`````````````0```&8``````````P!/`&(`:@!)`&X`9@!O````````````
M`````````````````````````````````````````````````````!(``@'_
M____"````/____\`````````````````````````````````````````````
M```#````!`````````#^____`@```/[____^____!0````8````'````_O__
M_PD````*````"P````P````-````#@```/[___\0````$0```!(````3````
M%````!4````6````_O___Q@````9````&@```!L````<````'0```!X````?
M````(````"$````B````(P```"0````E````)@```"<````H````*0```"H`
M```K````+````"T````N````+P```#`````Q````,@```#,````T````-0``
M`#8````W````.````#D````Z````.P```#P````]````/@```#\```!`````
M00```$(```!#````1````$4```#^____1P```/[_____________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M_________________________________________P$```(`````````````
M````````````````````````````````````````````````````````````
M```````!`/[_`PH``/____\"S@(``````,````````!&%P```$UI8W)O<V]F
M="!%<75A=&EO;B`S+C``#````$13($5Q=6%T:6]N``L```!%<75A=&EO;BXS
M`/0YLG$`````````````````````````````````````````````````````
M`P``````````````````````````````````````````````````````````
M```````````````````````<`````@#XU<P`````````L&9)`(AA20``````
M`P$!`P`*`1*#>0`##P$`"P$2@W0``!$`"@*&/0`2@V,``H8K``,>`0`!`H2S
M`P,/`0`+`1*#:0``$0`*$H-Y``,/`0`+`1*#=``"AA(B`H@Q```1``H"ABL`
M`QX!``$"A+L#`P\!``L!$H-I```1``H2@TT``P\!``L!$H-T``*&$B(2@VD`
M`!$```$2@VD``H8]``*(,````0*(,@`"B#0```T"AA$B```+`1*#:0`"ACT`
M`H@Q```!`H@R``*(-```#0*&$2(````%````````````````````````````
M````10!Q`'4`80!T`&D`;P!N`"``3@!A`'0`:0!V`&4`````````````````
M`````````````````````````````"```@#_______________\`````````
M```````````````````````````````````````$````Z``````````Q`%0`
M80!B`&P`90``````````````````````````````````````````````````
M````````````````````#@`"``$````#````_____P``````````````````
M`````````````````````````````!<```")"P````````4`4P!U`&T`;0!A
M`'(`>0!)`&X`9@!O`'(`;0!A`'0`:0!O`&X`````````````````````````
M```````````H``(!`@````L```#_____````````````````````````````
M````````````````````"````+`!````````!0!$`&\`8P!U`&T`90!N`'0`
M4P!U`&T`;0!A`'(`>0!)`&X`9@!O`'(`;0!A`'0`:0!O`&X`````````````
M`#@``@'_______________\`````````````````````````````````````
M```````````/````[`$```````#^_P``!``"```````````````````````!
M````X(6?\OE/:!"KD0@`*R>SV3````"``0``$`````$```"(`````@```)``
M```#````K`````0```"X````!0```-@````'````Y`````@```#T````"0``
M`!0!```2````(`$```H````\`0``#````$@!```-````5`$```X```!@`0``
M#P```&@!```0````<`$``!,```!X`0```@```.0$```>````$0```$1E87(@
M4D%44R!U<V5R<RP``&0`'@````$`````96%R'@```!8```!5;FEV97)S:73@
M($PN($)O8V-O;FD`.``>`````0````!N:78>````!P```$YO<FUA;`!I'@``
M`!8```!5;FEV97)S:73@($PN($)O8V-O;FD`.``>`````@```#$`:78>````
M$P```$UI8W)O<V]F="!7;W)D(#@N,`!N0`````!\"=X%````0``````RB7IE
MX;T!0`````#L:#AWX;T!`P````$````#````M@````,````1!````P``````
M````````````````````````_O\```0``@```````````````````````@``
M``+5S=6<+AL0DY<(`"LL^:Y$````!=7-U9PN&Q"3EP@`*RSYKE0!```0`0``
M#`````$```!H````#P```'`````%````E`````8```"<````$0```*0````7
M````K`````L```"T````$````+P````3````Q````!8```#,````#0```-0`
M```,````\0````(```#D!```'@```!D```!5;FEV97)S:73@($QU:6=I($)O
M8V-O;FD``$1E`P````@````#`````@````,```#^!````P```&H0"``+````
M``````L`````````"P`````````+`````````!X0```!````$0```$1E87(@
M4D%44R!U<V5R<RP`#!````(````>````!@```%1I=&QE``,````!`````)@`
M```#`````````"`````!````-@````(````^`````0````(````*````7U!)
M1%]'54E$``(```#D!```00```$X```![`#$`1@`R`#4`00!!`#@`-P`M`#0`
M1``V`$$`+0`Q`#$`1``R`"T`00`U`#@`.0`M`#``,``P`#$`0P`X`#0`,P`U
M`#@`,@`S`'T````````````````````````````````````2``\`"@`!`%L`
M#P`"`````````"8``$#Q_P(`)@````<`3@!O`'(`;0!A`&P`90````(````$
M`&U($`0```````````````````````!.`$%`\O^A`$X````?`$,`80!R`&$`
M=`!T`&4`<@!E`"``<`!R`&4`9`!E`&8`:0!N`&D`=`!O`"``<`!A`'(`80!G
M`'(`80!F`&\``````````````````````.\$```&```2`````/____\`!```
M[P@```4`````!```X0<``.\(```&````"``````$``#O"```!P```&(```!V
M````>````.\$```3.I3_E8``````!`````H````/````$P```!<````8````
M'````!T````?````(````"4````H````+P```#,````\````/P```$D```!*
M````4@```%8```!:````6P```%\```!B````>0```'H```!]````?@```'\`
M``"`````B````(D```",````C@```)4```"6````H````*0```"F````IP``
M`*T```"S````O````+T```#&````R@```,P```#5````VP```"0!```I`0``
M*@$``"X!```O`0``,0$``#<!```Y`0``.P$``$$!``!"`0``1`$``$4!``!*
M`0``60$``%T!``!F`0``:`$``&D!``!P`0``<0$``'0!``!^`0``@P$``(<!
M``"+`0``C`$``),!``"4`0``G`$``)T!``"E`0``L@$``+@!``#&`0``S0$`
M`-`!``#5`0``X`$``.L!``#M`0``[P$``/`!``#V`0````(```D"```*`@``
M#0(```X"```5`@``&P(``"("```C`@``)P(``"P"```P`@``,0(``#H"```[
M`@``/P(``$X"``!1`@``5`(``%D"``!C`@``90(``&8"``!K`@``;`(``'`"
M``!Q`@``>`(``'P"``""`@``B`(``(T"``".`@``D`(``)$"``"4`@``E0(`
M`)X"``"?`@``I@(``-T"``#@`@``[P(``/("```'`P``#P,``!H#```=`P``
M'@,``"$#```B`P``*0,``"H#```P`P``,0,``#,#```T`P``-P,``#@#```]
M`P``00,``$,#``!%`P``20,``$H#``!,`P``3@,``%`#``!2`P``50,``%P#
M``!?`P``90,``&D#``!M`P``=`,``'H#``!]`P``@`,``(,#``"&`P``D@,`
M`)0#``"7`P``F@,``)\#``"I`P``JP,``+8#``"Y`P``N@,``+T#``"^`P``
MQP,``,@#``#*`P``RP,``-$#``#4`P``V0,``-H#``#<`P``W0,``-\#``#A
M`P``Y`,``.4#``#H`P``Z0,``.T#``#V`P``^@,``/T#```$!```!@0```T$
M```.!```$P0``!0$```8!```&00``!L$```<!```(`0``"@$```J!```*P0`
M`#0$```U!```/@0``#\$``!!!```2P0``%4$``!6!```8`0``&<$``!Q!```
M>`0``'L$``!\!```A00``(D$``"-!```C@0``)4$``"9!```G`0``)T$``"I
M!```J@0``+<$``"X!```O00``,`$``#$!```RP0``,X$``#6!```V00``.8$
M``#M!```\00``!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'
M`!P`!P`<``<`'``'``,`!P`3``<``P`'`!P`!P`<``<`'``'`!P`!P`<``<`
M'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<
M``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`
M!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'
M`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`
M'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<
M``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`
M!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'
M`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`
M'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<
M``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`__\$````%0!5`&X`:0!V`&4`
M<@!S`&D`=`#@`"``3``N`"``0@!O`&,`8P!O`&X`:0`R`$,`.@!<`%0`10!-
M`%``7`!3`&$`;`!V`&$`=`!A`&<`9P!I`&\`(`!A`'4`=`!O`&T`80!T`&D`
M8P!O`"``9`!I`"``(``@`$0`;P!C`'4`;0!E`&X`=`!O`#$`+@!A`',`9``5
M`%4`;@!I`'8`90!R`',`:0!T`.``(`!,`"X`(`!"`&\`8P!C`&\`;@!I`"``
M0P`Z`%P`1`!O`&,`=0!M`&4`;@!T`&D`7`!$`&4`80!R`"``4@!!`%0`4P`@
M`'4`<P!E`'(`<P`N`&0`;P!C`/]``(`!````````````;"Z$``$``0``````
M`````````````````A``````````[P0``&````@`0````P```$<6D`$```("
M!@,%!`4"`P0#`````````````````````0````````!4`&D`;0!E`',`(`!.
M`&4`=P`@`%(`;P!M`&$`;@```#46D`$"``4%`0(!!P8"!0<`````````$```
M````````````@`````!3`'D`;0!B`&\`;````#,FD`$```(+!@0"`@("`@0#
M`````````````````````0````````!!`'(`:0!A`&P````B``0`,`B(&```
MQ`(``!L!`````%^#*6;F@REF``````$`*@```+8````1!````0`"````!``#
M$`@```````````````$``0````$`````````(0,`````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````I0;`![0`M`"``!(P````
M``````````````#^!```````````````````````````````````````````
M``````````````````````````````````````````````(``````/__$@``
M```````0`$0`90!A`'(`(`!2`$$`5`!3`"``=0!S`&4`<@!S`"P`````````
M%0!5`&X`:0!V`&4`<@!S`&D`=`#@`"``3``N`"``0@!O`&,`8P!O`&X`:0`5
M`%4`;@!I`'8`90!R`',`:0!T`.``(`!,`"X`(`!"`&\`8P!C`&\`;@!I````
M````````````````````````````````````````````````````````````
M``````````````````````````````````````$`_O\#"@``_____P8)`@``
M````P````````$89````1&]C=6UE;G1O($UI8W)O<V]F="!7;W)D``H```!-
M4U=O<F1$;V,`$````%=O<F0N1&]C=6UE;G0N.`#T.;)Q````````````````
M`````````````````````````````0!#`&\`;0!P`$\`8@!J````````````
M`````````````````````````````````````````````````````!(``@#_
M______________\`````````````````````````````````````````````
M``!&````:P``````````````````````````````````````````````````
M`````````````````````````````````````````````````/__________
M_____P``````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````________________````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M``````````````````````````````#_______________\`````````````
M``````````````````````````````````````````````````#^_P``!``"
M```````````````````````!````X(6?\OE/:!"KD0@`*R>SV3````"``0``
M$`````$```"(`````@```)`````#````K`````0```"X````!0```-@````'
M````Y`````@```#T````"0```!0!```2````(`$```H````\`0``#````$@!
M```-````5`$```X```!@`0``#P```&@!```0````<`$``!,```!X`0```@``
M`.0$```>````$0```$1E87(@4D%44R!U<V5R<RP``&0`'@````$`````96%R
M'@```!8```!5;FEV97)S:73@($PN($)O8V-O;FD`.``>`````0````!N:78>
M````!P```$YO<FUA;`!I'@```!8```!5;FEV97)S:73@($PN($)O8V-O;FD`
M.``>`````@```#(`:78>````$P```$UI8W)O<V]F="!7;W)D(#@N,`!N0```
M``!.4TD&````0``````RB7IEX;T!0`````!R9L6$X;T!`P````$````#````
M00````,```!V`0```P``````````:0`5`%4`;@!I`'8`90!R`',`:0!T`.``
M(`!,`"X`(`!"`&\`8P!C`&\`;@!I````````````````````````````0SI<
M1&]C=6UE;G1I7/[_```$``(```````````````````````(````"U<W5G"X;
M$).7"``K+/FN1`````75S=6<+AL0DY<(`"LL^:Y4`0``$`$```P````!````
M:`````\```!P````!0```)0````&````G````!$```"D````%P```*P````+
M````M````!````"\````$P```,0````6````S`````T```#4````#````/$`
M```"````Y`0``!X````9````56YI=F5R<VETX"!,=6EG:2!";V-C;VYI``!$
M90,````#`````P````$````#````RP$```,```!J$`@`"P`````````+````
M``````L`````````"P`````````>$````0```!$```!$96%R(%)!5%,@=7-E
M<G,L``P0```"````'@````8```!4:71L90`#`````0````"8`````P``````
M```@`````0```#8````"````/@````$````"````"@```%]0241?1U5)1``"
M````Y`0``$$```!.````>P`Q`$8`,@`U`$$`00`X`#<`+0`T`$0`-@!!`"T`
M,0`Q`$0`,@`M`$$`-0`X`#D`+0`P`#``,``Q`$,`.``T`#,`-0`X`#(`,P!]
M``````````````````!#.EQ$;V-U;65N=&E<``0``&($``!C!```=@0``'<$
M``!X!```>00``*4$``"F!```RP0``,P$``#>!```WP0``.$$``#B!```XP0`
M`.8$``#G!```Z`0``.D$``#J!```ZP0``.P$``#M!```[@0``.\$``#P!```
M\00``/($``#S!```]@0``/<$``#X!```^00``/H$``#[!```_`0``/T$``#^
M!```_P0````%```!!0```@4```,%```$!0``!04```8%```)!0``"@4```L%
M```-!0``#@4```\%```2!0``$P4``!0%```5!0``%@4``!<%```8!0``&04`
M`!L%``#O"```1!(```#Z`/#I^@#G`.<`YP#CYP#G`./G`./G`-_G`./G`.<`
MW^<`X^<`X^<`W^<`X^<`W^<`X^<`X^<`X^<`X^<`````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M```````````````````````````````&"6H!`&?P``8):@$`;/```T@J`@T#
M:@````!%2/#_50@!$P-J@#VJ.0H(`54(`58(`6U(``0)`VH`````50@!`#\<
M`!^P@BX@L,9!(;!N!"*P;@0CD(D%))!N!"6P```-`"``#0!-`&$`<P!S`&D`
M;0!O`"``(`!&`&$`8@!I`&$`;@!O````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M`````````````````````.$'``#S!P``]`<```8(``!+"```C@@``-8(``#E
M"```Y@@``.\(```@$@``)!(``/T```````````````#]````````````````
M_0```````````````/T```````````````#]````````````````_0``````
M`````````/T```````````````#]````````````````_0``````````````
M`/T```````````````#]````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M``````````````````````````````````````$````+$@`/``H``0!;``\`
M`@`````````F``!`\?\"`"8````'`$X`;P!R`&T`80!L`&4````"````!`!M
M2!`$````````````````````````3@!!0/+_H0!.````'P!#`&$`<@!A`'0`
M=`!E`'(`90`@`'``<@!E`&0`90!F`&D`;@!I`'0`;P`@`'``80!R`&$`9P!R
M`&$`9@!O``````````````````````#&`0``!```$@````#_____`````*D!
M``"J`0``LP$``+4!``#(`0``G````````````````(````"`F```````````
M`````(````"`F````````````````(````"`F````````````````(````"`
MF````````````````(````"```0``$02```%``````0``.$'```D$@``!@``
M``@`````!```[P@```<```!B````=@```'@```#&`0``$SJ4_Y6```````0`
M```*````#P```!,````7````&````!P````=````'P```"`````E````*```
M`"\````S````/````#\```!)````2@```%(```!6````6@```%L```!?````
M>@```'T```"`````B````(D```",````C@```)4```"6````H````*0```"F
M````IP```*T```"S````O````+T```#&````R@```,P```#5````VP```"0!
M```I`0``*@$``"X!```O`0``,0$``#<!```Y`0``.P$``$$!``!"`0``1`$`
M`$4!``!*`0``60$``%T!``!F`0``:`$``&D!``!P`0``<0$``'0!``!^`0``
M@P$``(<!``"+`0``C`$``),!``"4`0``G`$``)T!``"E`0``J@$``+$!``"^
M`0``Q0$``,@!```<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`
M!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'
M`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`
M'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`!``'`/__"```
M`!4`50!N`&D`=@!E`'(`<P!I`'0`X``@`$P`+@`@`$(`;P!C`&,`;P!N`&D`
M,@!#`#H`7`!4`$4`30!0`%P`4P!A`&P`=@!A`'0`80!G`&<`:0!O`"``80!U
M`'0`;P!M`&$`=`!I`&,`;P`@`&0`:0`@`"``(`!$`&\`8P!U`&T`90!N`'0`
M;P`Q`"X`80!S`&0`%0!5`&X`:0!V`&4`<@!S`&D`=`#@`"``3``N`"``0@!O
M`&,`8P!O`&X`:0`@`$,`.@!<`$0`;P!C`'4`;0!E`&X`=`!I`%P`1`!E`&$`
M<@`@`%(`00!4`%,`(`!U`',`90!R`',`+@!D`&\`8P`5`%4`;@!I`'8`90!R
M`',`:0!T`.``(`!,`"X`(`!"`&\`8P!C`&\`;@!I`#<`0P`Z`%P`5`!%`$T`
M4`!<`%,`80!L`'8`80!T`&$`9P!G`&D`;P`@`&$`=0!T`&\`;0!A`'0`:0!C
M`&\`(`!D`&D`(``@`"``1`!E`&$`<@`@`%(`00!4`%,`(`!U`',`90!R`',`
M+@!A`',`9``5`%4`;@!I`'8`90!R`',`:0!T`.``(`!,`"X`(`!"`&\`8P!C
M`&\`;@!I`"``0P`Z`%P`1`!O`&,`=0!M`&4`;@!T`&D`7`!$`&4`80!R`"``
M4@!!`%0`4P`@`'4`<P!E`'(`<P`N`&0`;P!C`/]``X`!`,4!``#%`0``^`JU
M`&(`8@#%`0```````,4!`````````E@`````````J0$``*H!``"R`0``LP$`
M`,$!``#%`0``Q@$``$````@`0```0`!8"P!```!!`,P1`$```$``'A(`````
M0``@$@````!!`#P2`````$``W!$`0````P```$<6D`$```("!@,%!`4"`P0#
M`````````````````````0````````!4`&D`;0!E`',`(`!.`&4`=P`@`%(`
M;P!M`&$`;@```#46D`$"``4%`0(!!P8"!0<`````````$```````````````
M@`````!3`'D`;0!B`&\`;````#,FD`$```(+!@0"`@("`@0#````````````
M`````````0````````!!`'(`:0!A`&P````B``0`<0B(&```Q`(``!L!````
M`%^#*69/A"EF``````(`+0```$$```!V`0```0`!````!``#$`,`````````
M``````$``0````$`````````(0,`````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````I0;`![0`M`"``!(P````````````````
M``#+`0``````````````````````````````````````````````````````
M``````````````````````````````````(``````/__$@`````````0`$0`
M90!A`'(`(`!2`$$`5`!3`"``=0!S`&4`<@!S`"P`````````%0!5`&X`:0!V
M`&4`<@!S`&D`=`#@`"``3``N`"``0@!O`&,`8P!O`&X`:0`5`%4`;@!I`'8`
M90!R`',`:0!T`.``(`!,`"X`(`!"`&\`8P!C`&\`;@!I````````````````
M`````````````&X`80`@`%@`(`!D`&D`(`!9`````@`1``0`;4@0!"0`_F_Q
M_R(!)`````8`00!U`'0`;P!R`&4````"`!(`!`!M2!`$.@#^;_'_,@$Z````
M$0!$`&$`=`!A`"``9`!I`"``8P!R`&4`80!Z`&D`;P!N`&4````"`!,`!`!M
M2!`$/`#^;_'_0@$\````$@!$`&$`=`!A`"``=0!L`'0`:0!M`&$`(`!S`'0`
M80!M`'``80````(`%``$`&U($`1*`/YO\?]2`>REP0!9`!`$```<$+\`````
M```0```````$``#O"```#@!B:F)J\U?S5P```````````````````````!`$
M%@``%```D3T!`)$]`0#&`0``````````````````````````````````````
M`/__#P```````````/__#P```````````/__#P``````````````````````
M70````````````````"F````I@````````"F`````````*8`````````I@``
M``````"F````%```````````````N@`````````L`0```````"P!````````
M+`$````````L`0``#````#@!```4````N@````````#Q!0``M@```%@!```6
M````;@$```````!N`0```````&X!````````;@$```````!N`0```````&X!
M````````;@$```````!N!0```@```'`%````````<`4```````!P!0``````
M`'`%````````<`4```````!P!0``)````*<&``#T`0``FP@``)X```"4!0``
M70````````````````````````"F`````````&X!````````````````````
M`````````&X!````````;@$```````!N`0```````&X!````````E`4`````
M``!>`P```````*8`````````I@````````!N`0``````````````````;@$`
M``````!8`0```````%X#````````7@,```````!>`P```````&X!``#P`0``
MI@````````!N`0```````*8`````````;@$```````!N!0``````````````
M``````````````"Z`````````+H`````````I@````````"F`````````*8`
M````````I@````````!N`0```````&X%````````7@,``!`"``!>`P``````
M````````````;@4```````"F`````````*8`````````````````````````
M``````````````````````````````````````````````````````````!N
M!0```````&X!````````3`$```P```#`(<_BA.&]`;H```!R````+`$`````
M``!>`P```````&X%````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M``````````````````````!2`&\`;P!T`"``10!N`'0`<@!Y```````,````
M#`````0```````````````@```````````$`````````````%@`%`?______
M____"@````8)`@``````P````````$8`````X'3637?AO0'`(<_BA.&]`18`
M``!`&P```@```$0`80!T`&$````````````#`````P````0&````````````
M``@```````````$`````````````"P````L````*``(!_____PT```#_____
M/`!``,3:00"D!0``I`4``%<OUQ```(@%&`#!``4`&`$```!`"@`````0````
M````5P!O`'(`9`!$`&\`8P!U`&T`90!N`'0```!B`/____\H`````````*0%
M```#````BP````(```"B````&````!H``@$)````__________^D!0```@``
M`'0````"````BP```!@````@S&(`_____R@````K`````!0```$```!/`&(`
M:@!E`&,`=`!0`&\`;P!L````__\H`````````*0%```(````/@````(```!5
M````&````"#,8@#_____%@`!`0P```#_____!```````````````````````
M````````X#?O3G?AO0'@%%U/=^&]`0<``````````@```/__________`P``
M``0````C````_____P<````E````__________\+````#`````T````.````
M#P```!`````1````_O________________________\V````+P```/______
M____&P```!P````=````'@```!\````Q````________________!@```/[_
M__\D````__________________________\L`````@```#@```#]_____O__
M__[___\R````,P```#0````U````_O___S<````:````%P```/__________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M____________________________________________________________
M________`0!#`&\`;0!P`$\`8@!J````````````````````````````````
M`````````````````````````````````!(``@#_______________\`````
M``````````````````````````````````````````!&````:P`````````P
M`%0`80!B`&P`90``````````````````````````````````````````````
M````````````````````````#@`"`/_______________P``````````````
M`````````````````````````````````$@````Y"0``````````````````
M````````````````````````````````````````````````````````````
M````````````````````________________````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M``````````#_______________\`````````````````````````````````
M``````````````````````````````#^____`@```/[____^____!0````8`
M```'````_O___PD````*````"P````P````-````#@```/[___\0````$0``
M`!(````3````%````!4````6````_O___Q@````9````&@```!L````<````
M'0```!X````?````(````"$````B````(P```"0````E````)@```"<````H
M````*0```"H````K````+````"T````N````+P```#`````Q````,@```#,`
M```T````-0```#8````W````.````#D````Z````.P```#P````]````/@``
M`#\```!`````00```$(```!#````1````$4```#^____1P```/[___])````
M2@```$L```!,````30```$X```!/````4````%$```!2````4P```%0```!5
M````5@```%<```!8````60```%H```!;````7````%T```!>````7P```&``
M``!A````8@```&,```!D````90```&8```!G````:````&D```!J````:P``
M`&P```#^____________________________________________________
M_____________________________________________________Q(`#P`*
M``$`6P`/``(`````````)@``0/'_`@`F````!P!.`&\`<@!M`&$`;`!E````
M`@````0`;4@0!````````````````````````$X`04#R_Z$`3@```!\`0P!A
M`'(`80!T`'0`90!R`&4`(`!P`'(`90!D`&4`9@!I`&X`:0!T`&\`(`!P`&$`
M<@!A`&<`<@!A`&8`;P``````````````````````Q@$```0``!(`````____
M_P````"I`0``J@$``+,!``"U`0``R`$``)P```````````````"`````@)@`
M``````````````"`````@)@```````````````"`````@)@`````````````
M``"`````@)@```````````````"`````@``$``!$$@``!0`````$``#A!P``
M)!(```8````(``````0``.\(```'````8@```'8```!X````Q@$``!,ZE/^5
M@``````$````"@````\````3````%P```!@````<````'0```!\````@````
M)0```"@````O````,P```#P````_````20```$H```!2````5@```%H```!;
M````7P```'H```!]````@````(@```")````C````(X```"5````E@```*``
M``"D````I@```*<```"M````LP```+P```"]````Q@```,H```#,````U0``
M`-L````D`0``*0$``"H!```N`0``+P$``#$!```W`0``.0$``#L!``!!`0``
M0@$``$0!``!%`0``2@$``%D!``!=`0``9@$``&@!``!I`0``<`$``'$!``!T
M`0``?@$``(,!``"'`0``BP$``(P!``"3`0``E`$``)P!``"=`0``I0$``*H!
M``"Q`0``O@$``,4!``#(`0``'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'
M`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`
M'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<
M``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'`!P`!P`<``<`'``'``0`
M!P#__P@````5`%4`;@!I`'8`90!R`',`:0!T`.``(`!,`"X`(`!"`&\`8P!C
M`&\`;@!I`#(`0P`Z`%P`5`!%`$T`4`!<`%,`80!L`'8`80!T`&$`9P!G`&D`
M;P`@`&$`=0!T`&\`;0!A`'0`:0!C`&\`(`!D`&D`(``@`"``1`!O`&,`=0!M
M`&4`;@!T`&\`,0`N`&$`<P!D`!4`50!N`&D`=@!E`'(`<P!I`'0`X``@`$P`
M+@`@`$(`;P!C`&,`;P!N`&D`(`!#`#H`7`!$`&\`8P!U`&T`90!N`'0`:0!<
M`$0`90!A`'(`(`!2`$$`5`!3`"``=0!S`&4`<@!S`"X`9`!O`&,`%0!5`&X`
M:0!V`&4`<@!S`&D`=`#@`"``3``N`"``0@!O`&,`8P!O`&X`:0`W`$,`.@!<
M`%0`10!-`%``7`!3`&$`;`!V`&$`=`!A`&<`9P!I`&\`(`!A`'4`=`!O`&T`
M80!T`&D`8P!O`"``9`!I`"``(``@`$0`90!A`'(`(`!2`$$`5`!3`"``=0!S
M`&4`<@!S`"X`80!S`&0`%0!5`&X`:0!V`&4`<@!S`&D`=`#@`"``3``N`"``
M0@!O`&,`8P!O`&X`:0`@`$,`.@!<`$0`;P!C`'4`;0!E`&X`=`!I`%P`1`!E
M`&$`<@`@`%(`00!4`%,`(`!U`',`90!R`',`+@!D`&\`8P#_0`.``0#%`0``
MQ0$``/@*M0!B`&(`Q0$```````#%`0````````)8`````````*D!``"J`0``
ML@$``+,!``#!`0``Q0$``,8!``!````(`$```$``6`L`0```00#,$0!```!`
M`!X2`````$``(!(`````00`\$@````!``-P1`$````,```!'%I`!```"`@8#
M!00%`@,$`P````````````````````$`````````5`!I`&T`90!S`"``3@!E
M`'<`(`!2`&\`;0!A`&X````U%I`!`@`%!0$"`0<&`@4'`````````!``````
M`````````(``````4P!Y`&T`8@!O`&P````S)I`!```""P8$`@("`@($`P``
M``````````````````$`````````00!R`&D`80!L````(@`$`'$(B!@``,0"
M```;`0````!?@REF3X0I9@`````"`"T```!!````=@$```$``0````0``Q`#
M```````````````!``$````!`````````"$#````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M`````````````````````````````````````*4&P`>T`+0`@``2,```````
M````````````RP$`````````````````````````````````````````````
M```````````````````````````````````````````"``````#__Q(`````
M````$`!$`&4`80!R`"``4@!!`%0`4P`@`'4`<P!E`'(`<P`L`````````!4`
M50!N`&D`=@!E`'(`<P!I`'0`X``@`$P`+@`@`$(`;P!C`&,`;P!N`&D`%0!5
M`&X`:0!V`&4`<@!S`&D`=`#@`"``3``N`"``0@!O`&,`8P!O`&X`:0``````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M````````````````````````````````````````````````````````````
M``````````````````````````````````````````````#^_P``!``"````
M```````````````````!````X(6?\OE/:!"KD0@`*R>SV3````"``0``$```
M``$```"(`````@```)`````#````K`````0```"X````!0```-@````'````
MY`````@```#T````"0```!0!```2````(`$```H````\`0``#````$@!```-
M````5`$```X```!@`0``#P```&@!```0````<`$``!,```!X`0```@```.0$
M```>````$0```$1E87(@4D%44R!U<V5R<RP``&0`'@````$`````96%R'@``
M`!8```!5;FEV97)S:73@($PN($)O8V-O;FD`.``>`````0````!N:78>````
M!P```$YO<FUA;`!I'@```!8```!5;FEV97)S:73@($PN($)O8V-O;FD`.``>
M`````@```#(`:78>````$P```$UI8W)O<V]F="!7;W)D(#@N,`!N0`````!.
M4TD&````0``````RB7IEX;T!0`````!R9L6$X;T!`P````$````#````00``
M``,```!V`0```P``````````````````````````````_O\```0``@``````
M`````````````````@````+5S=6<+AL0DY<(`"LL^:Y$````!=7-U9PN&Q"3
MEP@`*RSYKE0!```0`0``#`````$```!H````#P```'`````%````E`````8`
M``"<````$0```*0````7````K`````L```"T````$````+P````3````Q```
M`!8```#,````#0```-0````,````\0````(```#D!```'@```!D```!5;FEV
M97)S:73@($QU:6=I($)O8V-O;FD``$1E`P````,````#`````0````,```#+
M`0```P```&H0"``+``````````L`````````"P`````````+`````````!X0
M```!````$0```$1E87(@4D%44R!U<V5R<RP`#!````(````>````!@```%1I
M=&QE``,````!`````)@````#`````````"`````!````-@````(````^````
M`0````(````*````7U!)1%]'54E$``(```#D!```00```$X```![`#$`1@`R
M`#4`00!!`#@`-P`M`#0`1``V`$$`+0`Q`#$`1``R`"T`00`U`#@`.0`M`#``
M,``P`#$`0P`X`#0`,P`U`#@`,@`S`'T`````````````````````````````
M```````2``\`"@`!`%L`#P`"`````````"8``$#Q_P(`)@````<`3@!O`'(`
M;0!A`&P`90````(````$`&U($`0`````7P`Y`#8`-P`T`#4`.``Q`#<`-@``
M`````````````````````````````````````````````````````````!8`
M`0'__________P8````"S@(``````,````````!&`````*!?^$YWX;T!@"@)
M3W?AO0$````````````````!`$\`;`!E````````````````````````````
M````````````````````````````````````````````````"@`"`?______
M_________P``````````````````````````````````````````````````
M```4``````````$`0P!O`&T`<`!/`&(`:@``````````````````````````
M```````````````````````````````````````2``(!!0````<```#_____
M`````````````````````````````````````````````````0```&8`````
M`````P!/`&(`:@!)`&X`9@!O````````````````````````````````````
M`````````````````````````````!(``@'_____"````/____\`````````
I```````````````````````````````````````#````!```````````
`
end

--====================987654321_0==_
Content-Type: text/plain; charset="us-ascii"

--====================987654321_0==_--

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

From: Norman Morin <m1njm00@Frb.GOV>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: FAME  
Date: Thu, 17 Sep 1998 09:02:12 -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: exmh version 2.0.2.4 7/8/98 (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii

Hello,

Is UNIX RATS the only version which can read data from and write 
data to a FAME database?  

Thanks,

Norm
-- 

                           Norman J. Morin 
                  nmorin@frb.gov or m1njm00@frb.gov         
          --------------------------------------------------
          Division of Research and Statistics * Mail Stop 82
           Board of Governors of the Federal Reserve System 
               Washington, D.C. 20551 * (202) 452-2476
          --------------------------------------------------



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

From: Simon.Van-Norden@hec.ca
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: FAME
Date: Thu, 17 Sep 1998 10:44:28 -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 4.05 [en] (Win95; U) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

Yes.

I think that's because the FRS and the Bank of Canada are all Estima and
FAME customers using unix platforms.  (I think the BoC might even have
supplied the code.)

I asked a while about reading FAME files from RATS and was told there
was no way to do it with the Windows/DOS version.  If you find out
otherwise, could you please post it to the list?

Thanks.

SvN

Norman Morin wrote:

> Hello,
>
> Is UNIX RATS the only version which can read data from and write
> data to a FAME database?
>
> Thanks,
>
> Norm
> --
>
>                            Norman J. Morin
>                   nmorin@frb.gov or m1njm00@frb.gov
>           --------------------------------------------------
>           Division of Research and Statistics * Mail Stop 82
>            Board of Governors of the Federal Reserve System
>                Washington, D.C. 20551 * (202) 452-2476
>           --------------------------------------------------



--
Simon van Norden   Professeur invité
E-mail:  simon.van-norden@hec.ca   or    svn@alum.mit.edu
Home Page:       http://www.hec.ca/pages/simon.van-norden
(514)340-6781 or fax (514)340-5632 or  cell (514)993-0466
Service de l'enseignement de la finance, École des H.E.C.
3000 Cote-Sainte-Catherine, Montreal QC, CANADA   H3T 2A7



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

From: Jae-Kwang Hwang <jhwang1@bcc.cba.ua.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: ERROR CORRECTION TERM
Date: Thu, 17 Sep 1998 10:59:22 -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: Mozilla 4.04 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Dear Rats Users,

I estimated 11 variables with Cats and found 5 or 6 cointegration
vectors. After running cointegration in Cats, I need to have error
correction term for forecasting. Cats ask that if you want to save ECM,
enter 1and it saves ECM at catsdata.dat file. My question is that these
ECMs are error correction term for each equations? Or  are residuals
from estimation eccor correction term?
Thank you. Have a nice day.


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

From: "Massimo Fabiano" <massimofabiano@hotmail.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: univariate equation and standard error
Date: Thu, 17 Sep 1998 10:12:25 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>
Content-Type: text/plain
X-Mailer: Mercury MTS (Bindery) v1.40

Dear RATS users,
I need help to solve a problem.
I estimated a univariate equation and I computed the  dynamic response 
of dependent veriable caused by a permanent variation of the 
contemporaneous independent one over 36 period.
I would like to know if  anyone is aware of a gauss or Rats routine to 
compute the standard error of this impulse response function. 


Regards,
 
Massimo  Fabiano
Bocconi University

______________________________________________________
Get Your Private, Free Email at http://www.hotmail.com

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

From: Wolfgang Bauer <wobauer@studi.unizh.ch>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Factor Analysis
Date: Fri, 18 Sep 1998 15:42:20 +0200 (CEST)
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

Dear Rats users,

I'd like to do a factor analysis according to Joreskog 1967. Has anybody
done this before and could assist me with hints on this?

Thank you very much, have a good weekend

Wolfgang Bauer


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

From: WOO KAI YIN <kywoo@ctimail.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE:restriction test on short-run dynamics of the ECM
Date: Sat, 19 Sep 1998 01:17:24 +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: Mozilla 3.0 (Win95; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Dear RATS users,

May I ask whether there are CATS code / commands to conduct restriction
test on short-run dynamics of the ECM, that is, restrict the
coefficients of the lagged difference terms of the ECM, please?

I would be grateful if you could help me!

Regards.,

KY Woo

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

From: Zafer Mustafaoglu <zmustafa@dpt.gov.tr>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: ERROR CORRECTION TERM
Date: Sat, 19 Sep 1998 15:46:29 +0300
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 4.5b2 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

The ouput of cats gives both the residuals from the cointegrating
relatinships and the res}duals from the single equations of the endogenous
variables in the sytem. If you need the ECMs you press save only ECMs
button (1) . and obtain the long run vectors. However I have just noticed
that these residuals are already lagged one period.So if you want to use
them in a dynamic short run modeli don.'t take the one period lagged
residuals since it will correspond to ECM(-2)

Pelin Kale

Jae-Kwang Hwang wrote:

> Dear Rats Users,
>
> I estimated 11 variables with Cats and found 5 or 6 cointegration
> vectors. After running cointegration in Cats, I need to have error
> correction term for forecasting. Cats ask that if you want to save ECM,
> enter 1and it saves ECM at catsdata.dat file. My question is that these
> ECMs are error correction term for each equations? Or  are residuals
> from estimation eccor correction term?
> Thank you. Have a nice day.


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

From: "Frieder Knüpling" <knueplin@ruf.uni-freiburg.de>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Markov-Switching Models.
Date: Mon, 21 Sep 1998 13:35:44 +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 4.06 [en] (Win98; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Dear RATS users,

I want to estimate Markov-switching models (cf. Hamilton [1989...]) via
numerical methods like those used by MAXIMIZE. The problem is that the
likelihood function in this case is not a sum of the likelihood elements
for t=1,...,T as required by MAXIMIZE. There are rekursive algorithms to
evaluate the likelihood function like the one provided by COSSLETT and
LEE [1985], but I cannot see how they might help in this context. Does
anyone have a suggestion how this problem might be implemented by making
efficient use of  RATS procedures like MAXIMIZE?

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 - 2340



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

From: phsu@mail.stit.edu.tw
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: EGARCH with dummy
Date: Tue, 22 Sep 1998 11:02:17 +0000
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 Windows (v2.41) (via Mercury MTS (Bindery) v1.40)

Dear RATS users,
  I try to add dummy variables to EGARCH model (with
GED), but I always receive the following erro message 
while I run MAXIMIZE:
 ## SR10. Missing Values And/Or SMPL Options Leave 
      No Usable Data Points

Could you please tell me what's wrong with it and what
should I do. Thanks. 
Philip Hsu

cal(daily) 1996 1 9
all 0 1998:3:23
open data tse.rat
data(format=rats) / return
smpl 5 570
set y = return(t)
set x = return(t-1)
declare series u
declare series v
set u = 0.0
set v = 0.0
set dummy = t>=96:1:9.AND.t<=97:1:9
dec real b0 b1 bd1 p1 q1 l1 c nu qd1 pd1
nonlin b0 b1 bd1 p1 q1 l1 c nu qd1 pd1
frml regresid = y-b0-bd1*dummy-b1*x
FRML G = ABS(u(t)/SQRT(v(t))) - SQRT(2.0/%PI) -$
                  l1*u(t)/SQRT(v(t)) 
FRML garchvar = $
          EXP( c + (pd1*dummy + p1)*LOG(v{1})+ (qd1*dummy +$
          q1)*G{1} )
FRML LAMBDA = EXP((-1./nu)*LOG(2) + $
                              0.5*%LNGAMMA(1./nu) $
                             - 0.5*%LNGAMMA(3./nu))
FRML logl = v(t) = garchvar(t), u(t) = regresid(t), $
                  LOG(nu) - 0.5*ABS(u(t)/(LAMBDA* $
                  SQRT(v(t))))**nu - (LOG(LAMBDA) $
                  + (1+1./nu)*LOG(2) + %LNGAMMA(1./nu)) - $
                  .5*LOG(v(t))
linreg y / u
# constant dummy x
NLPAR(subiter=250,cvcrit=0.0000001)
compute b0=%beta(1), bd1=%beta(2), b1=%beta(3)
compute c=%seesq,p1=.01,q1=.02,nu=5,pd1=.01,qd1=.01,l1=0.01 
set v = %seesq 
maximize(method=bhhh,recursive,iter=100) logl 5 570 
## SR10. Missing Values And/Or SMPL Options Leave
    No Usable Data Points
*********************************************

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

From: "Gregory, Richard" <RGregory@VEDP.state.va.us>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: EGARCH with dummy
Date: Tue, 22 Sep 1998 07:35:14 -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>
MIME-Version: 1.0
X-Mailer: Internet Mail Service (5.5.2232.9) (via Mercury MTS (Bindery) v1.40)
Content-Type: text/plain

I believe it is your sample command conflicting with your set command with
the dummy variable. You are setting the dummy for 96:1:9 to 97:1:9, but your
SMPL command begins at 96:1:14.

> -----Original Message-----
> From:	phsu@mail.stit.edu.tw [SMTP:phsu@mail.stit.edu.tw]
> Sent:	Tuesday, September 22, 1998 7:02 AM
> To:	RATS Discussion List
> Subject:	EGARCH with dummy
> 
> Dear RATS users,
>   I try to add dummy variables to EGARCH model (with
> GED), but I always receive the following erro message 
> while I run MAXIMIZE:
>  ## SR10. Missing Values And/Or SMPL Options Leave 
>       No Usable Data Points
> 
> Could you please tell me what's wrong with it and what
> should I do. Thanks. 
> Philip Hsu
> 
> cal(daily) 1996 1 9
> all 0 1998:3:23
> open data tse.rat
> data(format=rats) / return
> smpl 5 570
> set y = return(t)
> set x = return(t-1)
> declare series u
> declare series v
> set u = 0.0
> set v = 0.0
> set dummy = t>=96:1:9.AND.t<=97:1:9
> dec real b0 b1 bd1 p1 q1 l1 c nu qd1 pd1
> nonlin b0 b1 bd1 p1 q1 l1 c nu qd1 pd1
> frml regresid = y-b0-bd1*dummy-b1*x
> FRML G = ABS(u(t)/SQRT(v(t))) - SQRT(2.0/%PI) -$
>                   l1*u(t)/SQRT(v(t)) 
> FRML garchvar = $
>           EXP( c + (pd1*dummy + p1)*LOG(v{1})+ (qd1*dummy +$
>           q1)*G{1} )
> FRML LAMBDA = EXP((-1./nu)*LOG(2) + $
>                               0.5*%LNGAMMA(1./nu) $
>                              - 0.5*%LNGAMMA(3./nu))
> FRML logl = v(t) = garchvar(t), u(t) = regresid(t), $
>                   LOG(nu) - 0.5*ABS(u(t)/(LAMBDA* $
>                   SQRT(v(t))))**nu - (LOG(LAMBDA) $
>                   + (1+1./nu)*LOG(2) + %LNGAMMA(1./nu)) - $
>                   .5*LOG(v(t))
> linreg y / u
> # constant dummy x
> NLPAR(subiter=250,cvcrit=0.0000001)
> compute b0=%beta(1), bd1=%beta(2), b1=%beta(3)
> compute c=%seesq,p1=.01,q1=.02,nu=5,pd1=.01,qd1=.01,l1=0.01 
> set v = %seesq 
> maximize(method=bhhh,recursive,iter=100) logl 5 570 
> ## SR10. Missing Values And/Or SMPL Options Leave
>     No Usable Data Points
> *********************************************

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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: EGARCH with dummy
Date: Tue, 22 Sep 1998 09:33:42 -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)


> I believe it is your sample command conflicting with your set command with
> the dummy variable. You are setting the dummy for 96:1:9 to 97:1:9, but your
> SMPL command begins at 96:1:14.
> 

Yes, I believe the problem is the "SMPL 5 570" command. I don't think 
the problem is really the dummy variable itself, because your 
model only needs data for that over the estimation range (entries 5 
through 570). 

The real problem is that, because of the SMPL, the V and U series
are are only initialized from 5 through 570, and are thus NA
(missing values) for entries 1 through 4. However, you are trying to 
start the estimation at entry 5, which would require having lagged 
values of U and V at entry 4. Because U and V are NA at entry 4, all 
subsequent entries will also be NA. 

So, I would suggest simply removing the SMPL command.

Sincerely,
Tom Maycock
Estima


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

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

From: "clecourt" <clecourt@mailsc.univ-lille2.fr>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Memory Problem for estimating an ARFIMA model
Date: Tue, 22 Sep 1998 17:05:11 +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 Pro Version 3.0.1 (32) [F] (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

Dear all,
can someone tell me how to solve the following memory problem:
## M4. A memory request for an additional 156999220 bytes cannot be satisfied
The Error Occurred At Location 0535 of MLELOOP

This message occurs when I'm estimating an ARFIMA model (by MLE) with
ARFIMA.SRC

Best Regards.
        \\\|||||///
        \\  - -  //
         (  @ @  )
+------oOOo-(_)-oOOo---- +----------------------------------------+
|   Christelle Lecourt   | Centres de Recherche:    MEDE ET CADRE |      
|       Economiste       | Tel: 0032 (0)85/25.04.06               |
| UNIVERSITE DE LILLE II | e-Mail: clecourt@mailsc.univ-lille2.fr |
+---------------Oooo-----+----------------------------------------+
        oooO   (   )
       (   )    ) /
        \ (    (_/
         \_)

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

From: "E. Roca" <e.roca@qut.edu.au>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: EGARCH with dummy
Date: Wed, 23 Sep 1998 09:10:41 +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
X-Mailer: Windows Eudora Pro Version 3.0 (32) (via Mercury MTS (Bindery) v1.40)
Content-type: text/plain; charset="us-ascii"

Philip:

I haven't met this problem in another situation.  Try adjusting the
starting values of your data with GSTART, say GSTART=3, before you begin
the calculations.

Good luck,

Eduardo
At 07:35 22/09/98 -0400, you wrote:
>I believe it is your sample command conflicting with your set command with
>the dummy variable. You are setting the dummy for 96:1:9 to 97:1:9, but your
>SMPL command begins at 96:1:14.
>
>> -----Original Message-----
>> From:	phsu@mail.stit.edu.tw [SMTP:phsu@mail.stit.edu.tw]
>> Sent:	Tuesday, September 22, 1998 7:02 AM
>> To:	RATS Discussion List
>> Subject:	EGARCH with dummy
>> 
>> Dear RATS users,
>>   I try to add dummy variables to EGARCH model (with
>> GED), but I always receive the following erro message 
>> while I run MAXIMIZE:
>>  ## SR10. Missing Values And/Or SMPL Options Leave 
>>       No Usable Data Points
>> 
>> Could you please tell me what's wrong with it and what
>> should I do. Thanks. 
>> Philip Hsu
>> 
>> cal(daily) 1996 1 9
>> all 0 1998:3:23
>> open data tse.rat
>> data(format=rats) / return
>> smpl 5 570
>> set y = return(t)
>> set x = return(t-1)
>> declare series u
>> declare series v
>> set u = 0.0
>> set v = 0.0
>> set dummy = t>=96:1:9.AND.t<=97:1:9
>> dec real b0 b1 bd1 p1 q1 l1 c nu qd1 pd1
>> nonlin b0 b1 bd1 p1 q1 l1 c nu qd1 pd1
>> frml regresid = y-b0-bd1*dummy-b1*x
>> FRML G = ABS(u(t)/SQRT(v(t))) - SQRT(2.0/%PI) -$
>>                   l1*u(t)/SQRT(v(t)) 
>> FRML garchvar = $
>>           EXP( c + (pd1*dummy + p1)*LOG(v{1})+ (qd1*dummy +$
>>           q1)*G{1} )
>> FRML LAMBDA = EXP((-1./nu)*LOG(2) + $
>>                               0.5*%LNGAMMA(1./nu) $
>>                              - 0.5*%LNGAMMA(3./nu))
>> FRML logl = v(t) = garchvar(t), u(t) = regresid(t), $
>>                   LOG(nu) - 0.5*ABS(u(t)/(LAMBDA* $
>>                   SQRT(v(t))))**nu - (LOG(LAMBDA) $
>>                   + (1+1./nu)*LOG(2) + %LNGAMMA(1./nu)) - $
>>                   .5*LOG(v(t))
>> linreg y / u
>> # constant dummy x
>> NLPAR(subiter=250,cvcrit=0.0000001)
>> compute b0=%beta(1), bd1=%beta(2), b1=%beta(3)
>> compute c=%seesq,p1=.01,q1=.02,nu=5,pd1=.01,qd1=.01,l1=0.01 
>> set v = %seesq 
>> maximize(method=bhhh,recursive,iter=100) logl 5 570 
>> ## SR10. Missing Values And/Or SMPL Options Leave
>>     No Usable Data Points
>> *********************************************
>
>
Eduardo D. Roca
School of Economics and Finance
Faculty of Business
Queensland University of Technology
2 George Street GPO Box 2434
Brisbane, Queensland, Australia 4001
Phone: (07)3864 1233 Fax (07) 3864 1500
International (+61 7)
e-mail: e.roca@qut.edu.au
www: http://www.qut.edu.au

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

From: "E. Roca" <e.roca@qut.edu.au>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: EGARCH with dummy
Date: Wed, 23 Sep 1998 09:13:34 +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
X-Mailer: Windows Eudora Pro Version 3.0 (32) (via Mercury MTS (Bindery) v1.40)
Content-type: text/plain; charset="us-ascii"

Philip: 

I should have said use GSTART=8 or GSTART=9 rather than  GSTART=3, a value
higher than 5 which is what you are using as your starting value for your
data.

Cheers,

Eduardo


Philip:

I haven't met this problem in another situation.  Try adjusting the
starting values of your data with GSTART, say GSTART=3, before you begin
the calculations.

Good luck,

Eduardo
At 07:35 22/09/98 -0400, you wrote:
>I believe it is your sample command conflicting with your set command with
>the dummy variable. You are setting the dummy for 96:1:9 to 97:1:9, but your
>SMPL command begins at 96:1:14.
>
>> -----Original Message-----
>> From:	phsu@mail.stit.edu.tw [SMTP:phsu@mail.stit.edu.tw]
>> Sent:	Tuesday, September 22, 1998 7:02 AM
>> To:	RATS Discussion List
>> Subject:	EGARCH with dummy
>> 
>> Dear RATS users,
>>   I try to add dummy variables to EGARCH model (with
>> GED), but I always receive the following erro message 
>> while I run MAXIMIZE:
>>  ## SR10. Missing Values And/Or SMPL Options Leave 
>>       No Usable Data Points
>> 
>> Could you please tell me what's wrong with it and what
>> should I do. Thanks. 
>> Philip Hsu
>> 
>> cal(daily) 1996 1 9
>> all 0 1998:3:23
>> open data tse.rat
>> data(format=rats) / return
>> smpl 5 570
>> set y = return(t)
>> set x = return(t-1)
>> declare series u
>> declare series v
>> set u = 0.0
>> set v = 0.0
>> set dummy = t>=96:1:9.AND.t<=97:1:9
>> dec real b0 b1 bd1 p1 q1 l1 c nu qd1 pd1
>> nonlin b0 b1 bd1 p1 q1 l1 c nu qd1 pd1
>> frml regresid = y-b0-bd1*dummy-b1*x
>> FRML G = ABS(u(t)/SQRT(v(t))) - SQRT(2.0/%PI) -$
>>                   l1*u(t)/SQRT(v(t)) 
>> FRML garchvar = $
>>           EXP( c + (pd1*dummy + p1)*LOG(v{1})+ (qd1*dummy +$
>>           q1)*G{1} )
>> FRML LAMBDA = EXP((-1./nu)*LOG(2) + $
>>                               0.5*%LNGAMMA(1./nu) $
>>                              - 0.5*%LNGAMMA(3./nu))
>> FRML logl = v(t) = garchvar(t), u(t) = regresid(t), $
>>                   LOG(nu) - 0.5*ABS(u(t)/(LAMBDA* $
>>                   SQRT(v(t))))**nu - (LOG(LAMBDA) $
>>                   + (1+1./nu)*LOG(2) + %LNGAMMA(1./nu)) - $
>>                   .5*LOG(v(t))
>> linreg y / u
>> # constant dummy x
>> NLPAR(subiter=250,cvcrit=0.0000001)
>> compute b0=%beta(1), bd1=%beta(2), b1=%beta(3)
>> compute c=%seesq,p1=.01,q1=.02,nu=5,pd1=.01,qd1=.01,l1=0.01 
>> set v = %seesq 
>> maximize(method=bhhh,recursive,iter=100) logl 5 570 
>> ## SR10. Missing Values And/Or SMPL Options Leave
>>     No Usable Data Points
>> *********************************************
>
>
Eduardo D. Roca
School of Economics and Finance
Faculty of Business
Queensland University of Technology
2 George Street GPO Box 2434
Brisbane, Queensland, Australia 4001
Phone: (07)3864 1233 Fax (07) 3864 1500
International (+61 7)
e-mail: e.roca@qut.edu.au
www: http://www.qut.edu.au

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

From: phsu@mail.stit.edu.tw
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: SRC procedures
Date: Wed, 23 Sep 1998 10:16:20 +0000
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 Windows (v2.41) (via Mercury MTS (Bindery) v1.40)

Dear Sarathi,
  The SRC files are all ASCII files. If you want to get binary files 
and have known where they are, you can use FTP command in 
UNIX (without Browser) to transfer the binary files, for example:
ftp ftp.estima.com
login: ftp
passwd: your email address
binary
get <FileName>

Philip Hsu

> From:          SARATHI@FPMLAN.iimahd.ernet.in
> To:            "RATS Discussion List" <RATS-L@efs.mq.edu.au>
> Subject:       SRC procedures
> Date:          Fri, 4 Sep 1998 23:50:55 GMT+0530
> Reply-to:      "RATS Discussion List" <RATS-L@efs.mq.edu.au>
> Organization:  IIM, Ahmedabad

> Hi all:
> 
> I'm trying to incorporate some SRC procedures from the estima web 
> site. I'm using a Linux web browser. I get only ascii characters when 
> i download the procedure from the ftp site.
> 
> could you help find a way of downloading the procedures using a unix 
> browser.
> 
> Thanks for your help.
> 
> Regards,
> 
> Sarathi
> Doctoral Candidate
> IIM-A, India.
> 
> 

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

From: Simon.Van-Norden@hec.ca
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Markov-Switching Models.
Date: Tue, 22 Sep 1998 22:59:23 -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 4.05 [en] (Win95; U) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

Frieder:

I don't know how to use MAXIMIZE to estimate these kinds of models.  The
problem is essentially one of using FRML commands to construct the filtered
regime-classification probabilities, which might be possible using temporary
variables to store intermediate values (see the RATS manual for details.)

However, you could also use the EM algorithm to completely avoid the use of
the MAXIMIZE command.  To do so, you would have to write a RATS procedure to
estimate the smoothed (i.e. full-sample) regime-classification
probabilities.  The EM algorithm then just consists of iteratively (1)
constructing the smoothed probability weights for given parameter values,
(2) using these weights in WLS to update the parameter estimates.  Details
are given in the Hamilton text.

Of course, I also have code in GAUSS for the estimation of such models, and
James Hamilton makes most of his programs available on his web page.

SvN

Frieder Knüpling wrote:

> Dear RATS users,
>
> I want to estimate Markov-switching models (cf. Hamilton [1989...]) via
> numerical methods like those used by MAXIMIZE. The problem is that the
> likelihood function in this case is not a sum of the likelihood elements
> for t=1,...,T as required by MAXIMIZE. There are rekursive algorithms to
> evaluate the likelihood function like the one provided by COSSLETT and
> LEE [1985], but I cannot see how they might help in this context. Does
> anyone have a suggestion how this problem might be implemented by making
> efficient use of  RATS procedures like MAXIMIZE?
>
> 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 - 2340



--
Simon van Norden   Professeur invité
E-mail:  simon.van-norden@hec.ca   or    svn@alum.mit.edu
Home Page:       http://www.hec.ca/pages/simon.van-norden
(514)340-6781 or fax (514)340-5632 or  cell (514)993-0466
Service de l'enseignement de la finance, École des H.E.C.
3000 Cote-Sainte-Catherine, Montreal QC, CANADA   H3T 2A7



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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Markov-Switching Models.
Date: Wed, 23 Sep 1998 11:25:16 -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)

> 
> I don't know how to use MAXIMIZE to estimate these kinds of models.  The
> problem is essentially one of using FRML commands to construct the filtered
> regime-classification probabilities, which might be possible using temporary
> variables to store intermediate values (see the RATS manual for details.)
> 
 
As there's been a lot of interest in these models lately, we've 
decided to try and put together some Markov switching model examples. 
I hope to be able to post something in the next day or two that may 
be of help.

Also, we're planning on putting out an issue of our newsletter in the 
next few weeks, which will include further information on Markov 
models. As part of our work on Version 5, we're also looking into 
making some improvements that will make handling these models easier 
and extend the range of models that can be handled.

Sincerely,
Tom Maycock
Estima


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

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

From: Price SG <S.G.Price@city.ac.uk>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: panel standard error adjustment
Date: Wed, 23 Sep 1998 19:42:15 +0100 (BST)
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

There's a robusterrors type adjustment called PCSE panel corrected
standard errors by Beck and Katz that is implemented in shazam, stata
and limdep.  I have some old RATS code that does it but are there any
recent friendly procedures out there to do this?


Many thanks.



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

From: "Christopher F Baum" <baum@bc.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: panel standard error adjustment
Date: Wed, 23 Sep 1998 14:52:53 -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: Mulberry (MacOS) [1.3.4, s/n P020-300786-009] (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Probably not what you're hoping to hear, but the xtgls command (which is
id'd as relevant by 'lookup PCSE' in Stata 5.0) is implemented as an
ado-file, xtgls.ado. That means that anyone with a copy of Stata who can
read Stata code and write RATS code can recode that procedure in RATS.
Admittedly a nontrivial enterprise, but beats starting from the econometrics
textbook...

Kit Baum
Boston College

--On Wed, Sep 23, 1998 19:42 +0100 "Price SG" <S.G.Price@city.ac.uk> wrote: 

> There's a robusterrors type adjustment called PCSE panel corrected
> standard errors by Beck and Katz that is implemented in shazam, stata
> and limdep.  I have some old RATS code that does it but are there any
> recent friendly procedures out there to do this?
> 
> 
> Many thanks.
> 
> 



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

From: jtebeka@oddo.fr
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: BVAR models
Date: Thu, 24 Sep 1998 17:11:56 +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 am trying to implement a Bayesian VAR. My system is composed of different
blocs (prices, International,...) that interract. How can i do that with
Rats as it seems to handle only BVAR with SYSTEM....END(SYSTEM)
instruction.
Thanks for your help.
Jacques.



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

From: Durwood Marshall <dmarshal@emerald.tufts.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: BVAR models
Date: Thu, 24 Sep 1998 11:34:00 -0400 (EDT)
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: Mercury MTS (Bindery) v1.40

On Thu, 24 Sep 1998 jtebeka@oddo.fr wrote:
> I am trying to implement a Bayesian VAR. My system is composed of different
> blocs (prices, International,...) that interract. How can i do that with

As an additional tool to supplement Rats, others may be interested in
BUGS for Bayesian Inference.  Check:
http://www.mrc-bsu.cam.ac.uk/bugs/

All the best,
Durwood Marshall - Tufts University
Statistical and Research Computing Consulting
Tufts Computer and Communications Services
617.628.5000 x72180 : Fax 617.627.3667 : E-Mail: dmarshal@tufts.edu
------------------------------------------------------------------


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

From: Michael Hanson <mhanson@econ.lsa.umich.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: How to save SPGRAPHs without printing to the screen?
Date: Thu, 24 Sep 1998 15:30:03 -0400 (EDT)
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 would like to generate some graphs using the SPGRAPH command and
save them to a file but NOT display them to the terminal (console).  A
sample program bit is below.  This works fine, except that it displays the
graph upon completion of the SPGRAPH(DONE) command.  I'm trying to do this
on a Unix (SunOS) machine, although I'd be interested in a cross-platform 
solution, if available.  (Don't ask my motives -- it's unnecessarily
complicated....)  Thanks for any suggestions.

					-- Mike

======================================================================
Michael S. Hanson                                    mhanson@umich.edu
Department of Economics         http://www.econ.lsa.umich.edu/~mhanson
University of Michigan


<--- BEGIN SAMPLE CODE --->

* PLOT IMPULSE RESPONSES

open plot var2a_plot1.rgf

spgraph(vfields=neqn,hfields=neqn)
    do i = 1,neqn
        do j = 1,neqn
            disp(store=header) 'Response of' response(i) $
                                'to a' shock(j) 'Shock:'
            graph(header=header,number=0) 1 i j
            # vma1(i,j) 1 h
        end do j
    end do i
spgraph(done)

<--- END SAMPLE CODE --->


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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: How to save SPGRAPHs without printing to the screen?
Date: Thu, 24 Sep 1998 14:42:36 -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)

> 	I would like to generate some graphs using the SPGRAPH command and
> save them to a file but NOT display them to the terminal (console).  A
> sample program bit is below.  This works fine, except that it displays the
> graph upon completion of the SPGRAPH(DONE) command.  I'm trying to do this
> on a Unix (SunOS) machine, although I'd be interested in a cross-platform 
> solution, if available.  

Just include the command:

ENVIRONMENT NOSHOWGRAPHS

somewhere in your program before the SPGRAPH section. Although I 
haven't tried it recently, I'm pretty sure this works on the UNIX 
version. 

If you want to turn the displaying of graphs back on later in the 
prograj, just do:

ENVIRONMENT SHOWGRAPHS


Sincerely,
Tom Maycock
Estima


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

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

From: "Christopher F Baum" <baum@bc.edu>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: How to save SPGRAPHs without printing to the screen?
Date: Thu, 24 Sep 1998 15:44:11 -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: Mulberry (MacOS) [1.3.4, s/n P020-300786-009] (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

you are using 

env noshowgraphs 

at the top of your program, of course...

Kit

--On Thu, Sep 24, 1998 15:30 -0400 "Michael Hanson"
<mhanson@econ.lsa.umich.edu> wrote: 

> 	I would like to generate some graphs using the SPGRAPH command and
> save them to a file but NOT display them to the terminal (console).  A
> sample program bit is below.  This works fine, except that it displays the
> graph upon completion of the SPGRAPH(DONE) command.  I'm trying to do this
> on a Unix (SunOS) machine, although I'd be interested in a cross-platform 
> solution, if available.  (Don't ask my motives -- it's unnecessarily
> complicated....)  Thanks for any suggestions.
> 
> 					-- Mike
> 
> ======================================================================
> Michael S. Hanson                                    mhanson@umich.edu
> Department of Economics         http://www.econ.lsa.umich.edu/~mhanson
> University of Michigan
> 
> 
> <--- BEGIN SAMPLE CODE --->
> 
> * PLOT IMPULSE RESPONSES
> 
> open plot var2a_plot1.rgf
> 
> spgraph(vfields=neqn,hfields=neqn)
>     do i = 1,neqn
>         do j = 1,neqn
>             disp(store=header) 'Response of' response(i) $
>                                 'to a' shock(j) 'Shock:'
>             graph(header=header,number=0) 1 i j
>             # vma1(i,j) 1 h
>         end do j
>     end do i
> spgraph(done)
> 
> <--- END SAMPLE CODE --->
> 



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

From: Norman Morin <m1njm00@Frb.GOV>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: How to save SPGRAPHs without printing to the screen? 
Date: Thu, 24 Sep 1998 15:50:33 -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: exmh version 2.0.2.4 7/8/98 (via Mercury MTS (Bindery) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii

> > 	I would like to generate some graphs using the SPGRAPH command and
> > save them to a file but NOT display them to the terminal (console).
> 
> Just include the command:
> 
> ENVIRONMENT NOSHOWGRAPHS
> 
> somewhere in your program before the SPGRAPH section. Although I 
> haven't tried it recently, I'm pretty sure this works on the UNIX 
> version. 

It does indeed work as advertised. 

Norm
-- 

                           Norman J. Morin 
                  nmorin@frb.gov or m1njm00@frb.gov         
          --------------------------------------------------
          Division of Research and Statistics * Mail Stop 82
           Board of Governors of the Federal Reserve System 
               Washington, D.C. 20551 * (202) 452-2476
          --------------------------------------------------



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

From: KHL2288@UTARLG.UTA.EDU
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Multivariate GARCH
Date: Thu, 24 Sep 1998 17:56:07 -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

Dear RATS users,
I am trying to estimate multivariate GARCH model with Ding and Engle(1994; UCSD
working paper) specification (especially model 15) which is lately used by
Santis and Gerard(1997;JF). I'd like to know if somebody wrote a procedure for
the Ding and Engles' specification which uses the Hadamard product.

Sincerely,
Kyong Lee
Dept. of Economics
U. of Texas at Arlington

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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: BVAR models
Date: Fri, 25 Sep 1998 09:35:48 -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>
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)


> On Thu, 24 Sep 1998 jtebeka@oddo.fr wrote:
> > I am trying to implement a Bayesian VAR. My system is composed of different
> > blocs (prices, International,...) that interract. How can i do that with
> 
> As an additional tool to supplement Rats, others may be interested in
> BUGS for Bayesian Inference.  Check:
> http://www.mrc-bsu.cam.ac.uk/bugs/
> 

Thanks for the info--we've actually been working on putting together 
Gibbs Sampling examples for the aforementioned newsletter and 
Version 5--it looks like BUGS will give us something to check our 
results against.

Sincerely,
Tom Maycock
Estima
 

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

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

From: KHL2288@UTARLG.UTA.EDU
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Computing square root of a series to test autocorr. of residuals
Date: Fri, 25 Sep 1998 22:56:02 -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

Dear RATS users,
I am trying to test the autocorrelation of residuals from GARCH specification.
I am using Rob's sample program. I could test for (U**2)/H, but not for U/sqrt(H). I got an error message: ##SX22. Expected type REAL, Got SERIES Instead. So, it seems that problem arises from the command(?), sqrt.
Would you help me solve this problem?
Many thanks in advance.
Kyong Lee
Economics Dept.
U. of Texas at Arlington

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

From: KHL2288@UTARLG.UTA.EDU
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Testing the autocorrelation
Date: Sun, 27 Sep 1998 00:04:45 -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: MESSAGE/RFC822
X-Mailer: Mercury MTS (Bindery) v1.40

Return-path: <rats-l-owner@efs.mq.edu.au>
Received: from sunb.ocs.mq.edu.au by UTARLG.UTA.EDU (PMDF V5.1-9 #18538)
 with ESMTP id <01J28OG9SEOA00419K@UTARLG.UTA.EDU> for KHL2288@UTARLG.UTA.EDU;
 Fri, 25 Sep 1998 23:08:19 CDT
Received: from efs1.efs.mq.edu.au (efs1.efs.mq.edu.au [137.111.64.8])
 by sunb.ocs.mq.edu.au (8.9.0/8.9.0) with ESMTP id OAA23537; Sat,
 26 Sep 1998 14:06:20 +1000 (EST)
Received: from EFS1/SpoolDir by efs1.efs.mq.edu.au (Mercury 1.40); Sat,
 26 Sep 1998 14:06:40 +0000
Received: from SpoolDir by EFS1 (Mercury 1.40); Sat, 26 Sep 1998 14:06:14 +0000
Date: Fri, 25 Sep 1998 22:56:02 -0500 (CDT)
From: KHL2288@UTARLG.UTA.EDU
Subject: Computing square root of a series to test autocorr. of residuals
Sender: Maiser@efs1.efs.mq.edu.au
To: RATS Discussion List <RATS-L@efs.mq.edu.au>
Errors-to: rats-l-owner@efs.mq.edu.au
Reply-to: RATS Discussion List <RATS-L@efs.mq.edu.au>
Message-id: <4E03F8450C4@efs1.efs.mq.edu.au>
MIME-version: 1.0
X-Mailer: Mercury MTS (Bindery) v1.40
Content-type: TEXT/PLAIN; CHARSET=US-ASCII
X-Listname: <RATS-L@efs1.efs.mq.edu.au>

Dear RATS users,
I am trying to test the autocorrelation of residuals from GARCH specification.
I am using Rob's sample program. I could test for (U**2)/H, but not for U/sqrt(H). I got an error message: ##SX22. Expected type REAL, Got SERIES Instead. So, it seems that problem arises from the command(?), sqrt.
Would you help me solve this problem?
Many thanks in advance.
Kyong Lee
Economics Dept.
U. of Texas at Arlington

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

From: KHL2288@UTARLG.UTA.EDU
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Ding and Engles'(1994) GARCH specification
Date: Sun, 27 Sep 1998 00:23:13 -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: MESSAGE/RFC822
X-Mailer: Mercury MTS (Bindery) v1.40

Return-path: <KHL2288@UTARLG.UTA.EDU>
Received: from UTARLG.UTA.EDU by UTARLG.UTA.EDU (PMDF V5.1-9 #18538)
 id <01J26Z7HE3JK003MQT@UTARLG.UTA.EDU> for khl2288@UTARLG.UTA.EDU; Thu,
 24 Sep 1998 18:02:49 CDT
Date: Thu, 24 Sep 1998 17:56:07 -0500 (CDT)
From: KHL2288@UTARLG.UTA.EDU
Subject: Multivariate GARCH
To: RATS-L@efs.mq.edu.au
Cc: khl2288@UTARLG.UTA.EDU
Message-id: <01J26Z7T1J3Q003MQT@UTARLG.UTA.EDU>
MIME-version: 1.0
Content-type: TEXT/PLAIN; CHARSET=US-ASCII

Dear RATS users,
I am trying to estimate multivariate GARCH model with Ding and Engle(1994; UCSD
working paper) specification (especially model 15) which is lately used by
Santis and Gerard(1997;JF). I'd like to know if somebody wrote a procedure for
the Ding and Engles' specification which uses the Hadamard product.

Sincerely,
Kyong Lee
Dept. of Economics
U. of Texas at Arlington

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

From: phsu@mail.stit.edu.tw
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Graphic with different symbols
Date: Mon, 28 Sep 1998 14:51:09 +0000
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 Windows (v2.41) (via Mercury MTS (Bindery) v1.40)

Dear RATS user,
  I'd like to have a graphic with three series.
graph(style=symbol) 3
# a1
# a2
# a3
Each of them should be printed in different 
symbols (not different colors).
What should I do? Thanks in advance.
Philip Hsu


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

From: "Eric Thode" <eric.thode@mail.uni-wuerzburg.de>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Graphic with different symbols
Date: Mon, 28 Sep 1998 10:30:25 +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
Content-transfer-encoding: 7BIT
X-mailer: Pegasus Mail for Windows (v2.53/R1) (via Mercury MTS (Bindery) v1.40)


> Dear RATS user,
>   I'd like to have a graphic with three series.
> graph(style=symbol) 3
> # a1
> # a2
> # a3
> Each of them should be printed in different 
> symbols (not different colors).
> What should I do? Thanks in advance.
> Philip Hsu

Use the option PATTERNS in the GRAPH statement.

Cheers,
Eric
Dipl.-Vw. Eric Thode
Dept. of Economics (VWL 4)
University of Wuerzburg
Sanderring 2
97070 Wuerzburg

Tel.: +49-931-312928
Fax.: +49-931-312774

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

From: Richard Priestley <richard.priestley@bi.no>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: 
Date: Tue, 29 Sep 1998 17:36:46 +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) v1.40)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

Rats Users:

Does anybody know why I get the following result when I use the Maximise
command?

Estimation by BFGS
NO CONVERGENCE IN 1 ITERATIONS
LAST CRITERION WAS  0.0000000
SUBITERATIONS LIMIT EXCEEDED. USE NLPAR TO SET HIGHER LIMIT
Weekly Data From 91:07:12 To 98:06:26
Usable Observations    364      Degrees of Freedom   351
Function Value                    4511.58024410

The NLPAR is set to 100 iters, and the number of iterations in both the
SIMPLEX and maximize commands is 100. The estimated paramters are sensible
and the function value is very close to a calculated value for a very
similar converged model. 


Any comments as to why this occurs, what it means and how to overcome the
problem would be greatly appreciated.

Many thanks in advance 

Richard Priestley




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

From: Simon.Van-Norden@hec.ca
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Maximize problem
Date: Tue, 29 Sep 1998 12:31:49 -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 4.05 [en] (Win95; U) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit

Richard:

My guess is that you're starting *too close* to the converged estimates (or in
a very "tough" area of the likelihood function.)

As a result, although the gradient is not quite zero, your maximization
algorithm can't find a steplength that doesn't reduce the value of the
objective function. (That would be why the subiterations limit is exceeded.)
Since it is unable to figure out how to improve on the starting estimates but
has a non-zero gradient, the algorithm quits with an error message (just as it
should.)

If that is the case, it this is by no means a problem specific to RATS. The
solution is to use a *worse* starting point, or play with the parameters
governing steplength selection.

Simon

Richard Priestley wrote:

> Rats Users:
>
> Does anybody know why I get the following result when I use the Maximise
> command?
>
> Estimation by BFGS
> NO CONVERGENCE IN 1 ITERATIONS
> LAST CRITERION WAS  0.0000000
> SUBITERATIONS LIMIT EXCEEDED. USE NLPAR TO SET HIGHER LIMIT
> Weekly Data From 91:07:12 To 98:06:26
> Usable Observations    364      Degrees of Freedom   351
> Function Value                    4511.58024410
>
> The NLPAR is set to 100 iters, and the number of iterations in both the
> SIMPLEX and maximize commands is 100. The estimated paramters are sensible
> and the function value is very close to a calculated value for a very
> similar converged model.
>
> Any comments as to why this occurs, what it means and how to overcome the
> problem would be greatly appreciated.
>
> Many thanks in advance
>
> Richard Priestley



--
Simon van Norden   Professeur invité
E-mail:  simon.van-norden@hec.ca   or    svn@alum.mit.edu
Home Page:       http://www.hec.ca/pages/simon.van-norden
(514)340-6781 or fax (514)340-5632 or  cell (514)993-0466
Service de l'enseignement de la finance, École des H.E.C.
3000 Cote-Sainte-Catherine, Montreal QC, CANADA   H3T 2A7



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

From: "Estima" <estima@estima.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: Re: Maximize problem
Date: Tue, 29 Sep 1998 13:46:29 -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)

> 
> My guess is that you're starting *too close* to the converged estimates (or in
> a very "tough" area of the likelihood function.)
> 
> As a result, although the gradient is not quite zero, your maximization
> algorithm can't find a steplength that doesn't reduce the value of the
> objective function. (That would be why the subiterations limit is exceeded.)
> Since it is unable to figure out how to improve on the starting estimates but
> has a non-zero gradient, the algorithm quits with an error message (just as it
> should.)
> 
> If that is the case, it this is by no means a problem specific to RATS. The
> solution is to use a *worse* starting point, or play with the parameters
> governing steplength selection.
> 

This is exactly correct, and is most likely the cause of the problem. 

I did want to add a couple of comments:

First, even if you don't encounter the "subiters" limit problem, 
starting the estimation with values that are "too good" is generally 
a bad idea. In particular, it may cause the model to converge too 
quickly, leaving you with standard errors that may be unreliable.

Second, if you aren't already doing so, I would strongly recommend 
the use of the TRACE option on MAXIMIZE. This will display the 
coefficient and function values at each iteration. Before concluding 
that a model has converged, it is a good idea to examine the trace 
output to assure yourself that the coefficients have indeed settled 
down in a reasonable fashion over the last few iterations.

Sincerely,
Tom Maycock
Estima


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

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

From: WOO KAI YIN <kywoo@ctimail.com>
To: "RATS Discussion List" <RATS-L@efs.mq.edu.au>
Subject: RE: contrainted max. Likelihood
Date: Wed, 30 Sep 1998 10:09:25 +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: Mozilla 3.0 (Win95; I) (via Mercury MTS (Bindery) v1.40)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit

Dear RATS users,

I want to regree  Bt onto Xt. But I want to contraint the fitted value
of Bt to be less than unity. I think constrainted ML method can do it.
May I ask the RATS codes to deal with the estimation, please? Thank you
very much.


REgards,

Kevin
Senior Lecturer,
HONg Kong Shue Yan College

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