Return-Path: 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 OAA196358 for ; Mon, 2 Nov 1998 14:17:47 -0500 From: maiser@efs.mq.edu.au Received: (from root@localhost) by corot.bc.edu (8.8.7/8.8.7) with X.500 id OAA43426 for baum@mail1.bc.edu; Mon, 2 Nov 1998 14:17:34 -0500 Received: from sunb.ocs.mq.edu.au (sunb.ocs.mq.edu.au [137.111.1.11]) by corot.bc.edu (8.8.7/8.8.7) with ESMTP id OAA53278 for ; Mon, 2 Nov 1998 14:17:19 -0500 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 GAA21257 for ; Tue, 3 Nov 1998 06:17:07 +1100 (EST) Received: from EFS1/SpoolDir by efs1.efs.mq.edu.au (Mercury 1.40); 3 Nov 98 06:18:11 GMT+1000 Received: from SpoolDir by EFS1 (Mercury 1.40); 3 Nov 98 06:18:11 GMT+1000 To: baum@bc.edu Date: Tue, 3 Nov 98 6:18:10 GMT+1000 Subject: Re: Message-ID: <1AD44B31BE4@efs1.efs.mq.edu.au> From: wobauer@studi.unizh.ch To: "RATS Discussion List" Subject: Forecasting statistics Date: Wed, 30 Sep 1998 16:43:32 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 would like to calculate MAE, RMSE and Theils U on my forecasts that are stored in one series. The actual values are stored in another series. So no estimation has to be done. How can this extremely simple task be done with the THEIL instruction? Thank you for your hints Wolfgang Bauer ---------- End of message ---------- From: Simon.Van-Norden@hec.ca To: "RATS Discussion List" Subject: Re: contrainted max. Likelihood Date: Wed, 30 Sep 1998 11:18:58 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 think your problem sounds hard if you want to do it properly by constrained maxlik. Would you consider using a cheap & solution? If so, why not regress a transformation of Bt, f(Bt) on Xt. Then you just choose a transformation that maps between (-inf,inf) and (-inf,1). SvN WOO KAI YIN wrote: > 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 -- 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" To: "RATS Discussion List" Subject: (Fwd) Mail Delivery Failure. Date: Wed, 30 Sep 1998 13:10:19 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 calculate MAE, RMSE and Theils U on my forecasts > that are stored in one series. The actual values are stored in another > series. So no estimation has to be done. How can this extremely > simple task be done with the THEIL instruction? > I haven't actually tried this recently, but you should just be able to read in your data, define the appropriate equation and assign the estimated coefficients (using EQUATION and ASSOCIATE respectively), and then use THEIL to generate the statistics. Note that you'll need to have all the data for the model available as well (i.e. the explanatory variables in the equation). If you only have the actual and forecasted series, and not the estimated coefficients or explanatory data, then I don't see any way to use THEIL for this--you would have to code up the statistical calculations manually, probably using COMPUTE's and DO loops. 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: James Peery Cover To: "RATS Discussion List" Subject: Re: contrainted max. Likelihood Date: Wed, 30 Sep 1998 14:37:40 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: University of Alabama X-Mailer: Mozilla 4.05 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable why not make the coefficiet something like [abs(alpha)/(1+abs(alpha))]? abs represents the absolute value. (I did not look up the rats command for that. Simon.Van-Norden@hec.ca wrote: >=20 > I think your problem sounds hard if you want to do it properly by > constrained maxlik. >=20 > Would you consider using a cheap & solution? > If so, why not regress a transformation of Bt, f(Bt) on Xt. Then you j= ust > choose a transformation that maps between (-inf,inf) and (-inf,1). >=20 > SvN >=20 > WOO KAI YIN wrote: >=20 > > Dear RATS users, > > > > I want to regree Bt onto Xt. But I want to contraint the fitted valu= e > > 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 y= ou > > very much. > > > > REgards, > > > > Kevin > > Senior Lecturer, > > HONg Kong Shue Yan College >=20 > -- > Simon van Norden Professeur invit=E9 > 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, =C9cole des H.E.C. > 3000 Cote-Sainte-Catherine, Montreal QC, CANADA H3T 2A7 --=20 James Peery Cover Department of Economics and Finance Culverhouse College of Commerce University of Alabama P. O. Box 870224 Tuscaloosa, AL 35487-0224 ---------- End of message ---------- From: Jae-Kwang Hwang To: "RATS Discussion List" Subject: 95%-quantiles of the asymptotic tables Date: Thu, 01 Oct 1998 23:19:25 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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: When I run CATS for cointegration test, it gives 90%-quantiles of the asymptotic tables for the rank tests. Is there any way to get 95%-quantiles tables for the rank tests. Thank you so much. ---------- End of message ---------- From: Pierre.SLADDEN@DG2.cec.be To: "RATS Discussion List" Subject: graphs printing problem Date: Fri, 2 Oct 1998 15:35:20 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mercury MTS (Bindery) v1.40 Hi, I wish to print multiple graphs with SPGRAPH on the same page using the whole space. With the proc below the page is only half-completed, despited the UNFIX ON, the printer orientation on 'portrait', ... I can change the text and labels size, modify ticks, ... but how to modify the size of graph's lines? *spgraph(vfields=4,hfields=2) *dofor x = 'as' 'be' 'ca' 'dn' 'fi' 'fr' 'ud' 'de' *label 1 *# x+z+'14' *clear 1 *open data c:\temp\orl2.rat *data(format=rats) / 1 *set b = %valid(a) *accumulate b / c *if c(1997:1)<>0 *{ *graph(header='Machinery & Equip./ Gross K (Px 90), Agriculture',subhead=x) 1 *# a *} *else *{ *next *} *end dofor *spgraph(done) Thanks all, Pierre Sladden European Commission Brussels ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: graphs printing problem Date: Fri, 2 Oct 1998 10:03:45 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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) > Hi, > > I wish to print multiple graphs with SPGRAPH on the same page using the whole space. > With the proc below the page is only half-completed, despited the UNFIX ON, the printer orientation on 'portrait', ... > I can change the text and labels size, modify ticks, ... but how to modify the size of graph's lines? > As shown on page 10 of the 4.2 manual supplement, this is the way portrait mode output is supposed to look (i.e. fills up top half of the page). After the manual was printed, we made a late decision to impelement the GRPARM(PORTRAIT/[NOPORTRAIT]) option in the Windows versions, rather than just in the DOS version. The two settings of this GRPARM option plus the two choices of output orientation (Portrait vs. Landscape) that are available in the Print Setup dialog box give a total of 4 different output orientations. Rather than describe them here, I would suggest just trying the 4 combinations to see for yourself. 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" To: "RATS Discussion List" Subject: Re: graphs printing problem Date: Fri, 02 Oct 1998 12:00:20 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 Being too lazy to conduct an empirical experiment, my question: did this change make it into UNIX version 4.30? Thx Kit Baum --On Fri, Oct 2, 1998 10:03 -0600 Estima wrote: >> Hi, >> >> I wish to print multiple graphs with SPGRAPH on the same page using the whole space. >> With the proc below the page is only half-completed, despited the UNFIX ON, the printer orientation on 'portrait', ... >> I can change the text and labels size, modify ticks, ... but how to modify the size of graph's lines? >> > > As shown on page 10 of the 4.2 manual supplement, this is the way > portrait mode output is supposed to look (i.e. fills up top half of > the page). > > After the manual was printed, we made a late decision to impelement > the GRPARM(PORTRAIT/[NOPORTRAIT]) option in the Windows versions, > rather than just in the DOS version. > > The two settings of this GRPARM option plus the two choices of output > orientation (Portrait vs. Landscape) that are available in the Print > Setup dialog box give a total of 4 different output orientations. > Rather than describe them here, I would suggest just trying the 4 > combinations to see for yourself. > > 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: "Stephen A. DeLurgio" To: "RATS Discussion List" Subject: Re: Cutting and pasteing with office 97. Date: Fri, 02 Oct 1998 11:14:21 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: University of Missouri X-Mailer: Mozilla 4.02 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I have noted some changes in the quality of graphs that are imported/shared directly from a copy menu in RATS and then pasted into a word or wordperfect file since converting to office 97. Has anyone else had this problem, and is there a fix for this problem. The graphs do not have the clarity of those which were imported using office 95. The vertical axis and key boxes have strange fonts, and the appearance is rather unappealing. This problem did exist prior to the conversion to office 97. Thanks for any input. Steve DeLurgio ---------- End of message ---------- From: "Wai Lee" To: "RATS Discussion List" Subject: Re: Cutting and pasteing with office 97. Date: Fri, 2 Oct 1998 12:11:02 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-type: text/plain; charset=us-ascii X-Mailer: Mercury MTS (Bindery) v1.40 I have had the same problem, not only with Office 97, but quality of graph deteriorates quite a bit whenever I paste it to other applications. I'm too lazy to find out why though. W. Lee "Stephen A. DeLurgio" on 10/02/98 12:14:21 PM Please respond to "RATS Discussion List" To: "RATS Discussion List" cc: (Wai Lee) Subject: Re: Cutting and pasteing with office 97. I have noted some changes in the quality of graphs that are imported/shared directly from a copy menu in RATS and then pasted into a word or wordperfect file since converting to office 97. Has anyone else had this problem, and is there a fix for this problem. The graphs do not have the clarity of those which were imported using office 95. The vertical axis and key boxes have strange fonts, and the appearance is rather unappealing. This problem did exist prior to the conversion to office 97. Thanks for any input. Steve DeLurgio ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: graphs printing problem Date: Fri, 2 Oct 1998 12:08:28 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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) > Being too lazy to conduct an empirical experiment, my question: > did this change make it into UNIX version 4.30? > Yes, the GRPARM(PORTRAIT/[NOPORTRAIT]) option is functional in the UNIX platforms. I believe this was true with all UNIX releases (4.0 through 4.3). The difference there is that the Windows version offers Portrait and Landscape mode PostScript exports (as well as the Portrait vs. Landscape choice offered by Windows-based printer drivers). To achieve the same effect in UNIX requires compiling two different versions of the RGF2PST program: one for Portrait mode one for Landscape mode (which I belive you guys have already 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: "Estima" To: "RATS Discussion List" Subject: Re: Cutting and pasteing with office 97. Date: Fri, 2 Oct 1998 12:26:18 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 noted some changes in the quality of graphs that are imported/shared directly from a copy menu in RATS and then > pasted into a word or wordperfect file since converting to office 97. Has anyone else had this problem, and is there a fix > for this problem. > > The graphs do not have the clarity of those which were imported using office 95. The vertical axis and key boxes have > strange fonts, and the appearance is rather unappealing. This problem did exist prior to the conversion to office 97. > Steve et al: There's a simple solution (using "Edit-Paste Special" rather than "Edit-Paste"), which requires a bit of explanation: When you copy a graph to the clipboard under Windows, most programs, including WinRATS, actually store the graph in several different versions. We do Picture (aka "Windows Metafile") and Bitmap formats. When you do an Edit-Paste operation, the target application picks one of these formats to import. With Word 95 and most other older applications, doing Edit-Paste will give you the Picture format, which usually provides the best appearance. However, Word 97 and many other newer applications now also support an "Enhanced Metafile" format, which WinRATS does not yet provide. Unfortunately, at least in Word 97, if you do Edit-Paste, the program selects the "enhanced metafile" version. In effect, you getting a standard metafile translated through an "enhanced metafile" filter, producing a lower-quality result (the fonts seem to be affected the most). Fortunately, most applications including Word 97 also offer an "Edit-Paste Special" operation, allowing you to select from the available formats. If you select "Picture (Enhanced Metafile)", you'll see the same funky results you get from doing Edit-Paste. But, if you select the format simply labelled "Picture", you'll get the standard metafile format which produces the (better) results you're used to seeing. Note that Paste Special can be useful in almost any Copy-and-paste situation--whether you are pasting text, graphics, spreadsheets, etc. 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: "Wai Lee" To: "RATS Discussion List" Subject: Re: Cutting and pasteing with office 97. Date: Fri, 2 Oct 1998 13:49:15 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-type: text/plain; charset=us-ascii X-Mailer: Mercury MTS (Bindery) v1.40 Tom's suggestions do not work for me. To be exact, I have Word version 7.0a for Windows 95 but running on NT. When I do Edit-Paste Special, I have 3 choices: "Picture," "Bitmap," and "Device Independent Bitmap." Of course, I don't have choice when I do just Edit-Paste. I have just tested all combinations, and all of them give funcky graphs in Word. I usually do OPEN COPY and then plot the graphs I need by other means. This could be an exceptional case unique to the set up of my machine. Nevermind. W. Lee "Estima" on 10/02/98 02:26:18 PM Please respond to "RATS Discussion List" To: "RATS Discussion List" cc: (Wai Lee) Subject: Re: Cutting and pasteing with office 97. > I have noted some changes in the quality of graphs that are imported/shared directly from a copy menu in RATS and then > pasted into a word or wordperfect file since converting to office 97. Has anyone else had this problem, and is there a fix > for this problem. > > The graphs do not have the clarity of those which were imported using office 95. The vertical axis and key boxes have > strange fonts, and the appearance is rather unappealing. This problem did exist prior to the conversion to office 97. > Steve et al: There's a simple solution (using "Edit-Paste Special" rather than "Edit-Paste"), which requires a bit of explanation: When you copy a graph to the clipboard under Windows, most programs, including WinRATS, actually store the graph in several different versions. We do Picture (aka "Windows Metafile") and Bitmap formats. When you do an Edit-Paste operation, the target application picks one of these formats to import. With Word 95 and most other older applications, doing Edit-Paste will give you the Picture format, which usually provides the best appearance. However, Word 97 and many other newer applications now also support an "Enhanced Metafile" format, which WinRATS does not yet provide. Unfortunately, at least in Word 97, if you do Edit-Paste, the program selects the "enhanced metafile" version. In effect, you getting a standard metafile translated through an "enhanced metafile" filter, producing a lower-quality result (the fonts seem to be affected the most). Fortunately, most applications including Word 97 also offer an "Edit-Paste Special" operation, allowing you to select from the available formats. If you select "Picture (Enhanced Metafile)", you'll see the same funky results you get from doing Edit-Paste. But, if you select the format simply labelled "Picture", you'll get the standard metafile format which produces the (better) results you're used to seeing. Note that Paste Special can be useful in almost any Copy-and-paste situation--whether you are pasting text, graphics, spreadsheets, etc. 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: "Stephen A. DeLurgio" To: "RATS Discussion List" Subject: Re: Cutting and pasteing with office 97. Date: Fri, 02 Oct 1998 16:28:54 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: University of Missouri X-Mailer: Mozilla 4.02 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: Tom Maycock and Wai Lee Thanks for your help in this matter. I will try it out very soon. Estima wrote: > > I have noted some changes in the quality of graphs that are imported/shared directly from a copy menu in RATS and then > > pasted into a word or wordperfect file since converting to office 97. Has anyone else had this problem, and is there a fix > > for this problem. > > > > The graphs do not have the clarity of those which were imported using office 95. The vertical axis and key boxes have > > strange fonts, and the appearance is rather unappealing. This problem did exist prior to the conversion to office 97. > > > > Steve et al: > > There's a simple solution (using "Edit-Paste Special" rather than > "Edit-Paste"), which requires a bit of explanation: > > When you copy a graph to the clipboard under Windows, most programs, > including WinRATS, actually store the graph in several different > versions. We do Picture (aka "Windows Metafile") and Bitmap formats. > When you do an Edit-Paste operation, the target application picks one > of these formats to import. > > With Word 95 and most other older applications, doing Edit-Paste > will give you the Picture format, which usually provides the best > appearance. > > However, Word 97 and many other newer applications now also support > an "Enhanced Metafile" format, which WinRATS does not yet provide. > > Unfortunately, at least in Word 97, if you do Edit-Paste, the program > selects the "enhanced metafile" version. In effect, you getting a > standard metafile translated through an "enhanced metafile" filter, > producing a lower-quality result (the fonts seem to be affected the > most). > > Fortunately, most applications including Word 97 also offer an > "Edit-Paste Special" operation, allowing you to select from the > available formats. If you select "Picture (Enhanced Metafile)", > you'll see the same funky results you get from doing Edit-Paste. > > But, if you select the format simply labelled "Picture", you'll get > the standard metafile format which produces the (better) results > you're used to seeing. > > Note that Paste Special can be useful in almost any Copy-and-paste > situation--whether you are pasting text, graphics, spreadsheets, etc. > > 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: "Krassimir Petrov" To: "RATS Discussion List" Subject: Re: Cutting and pasteing with office 97. Date: Fri, 02 Oct 1998 17:55:35 PDT Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Content-Type: text/plain X-Mailer: Mercury MTS (Bindery) v1.40 >From: "Wai Lee" >To: "RATS Discussion List" >Subject: Re: Cutting and pasteing with office 97. >Date: Fri, 2 Oct 1998 13:49:15 -0400 >Reply-to: "RATS Discussion List" > >Tom's suggestions do not work for me. > >To be exact, I have Word version 7.0a for Windows 95 but running on NT. >When I do Edit-Paste Special, I have 3 choices: "Picture," "Bitmap," and >"Device Independent Bitmap." Of course, I don't have choice when I do just >Edit-Paste. I have just tested all combinations, and all of them give >funcky graphs in Word. I usually do OPEN COPY and then plot the graphs I >need by other means. > >This could be an exceptional case unique to the set up of my machine. >Nevermind. > >W. Lee > > Lee, I was about to complement Tom's comment on the subject: 1. COPY-PASTE has never worked for me (MS Word) 2. COPY-PASTE SPECIAL "CONVERT FROM BITMAP" has always worked ===================== for me faultlessly. The other two options on "convert from" do not work. I use Win95, if that matters. Hope this helps Krassimir Petrov P.O.Box 3703 Columbus, Oh 43210 cell. 614-579-7919 pager 614-665-2951 off. 614-292-9519 ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: infinite sum Date: Sat, 03 Oct 1998 11:03:23 +0800 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 conduct the infinite summation of the discounted value of Xt, ie. the so-called predicting geometric distributed leads' problems. infinite sum of ai Xt+i i = 0, 1, 2, .... infinity Do you know any RATS codes can deal with such problem, please? Thank you very much REgards, KY Senior Lecturer HK SYC ---------- End of message ---------- From: "Stephen A. DeLurgio" To: "RATS Discussion List" Subject: Re: Cutting and pasteing with office 97. Date: Sat, 03 Oct 1998 15:41:36 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: University of Missouri X-Mailer: Mozilla 4.02 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit I don't want to clutter the airwaves here, but Tom Maycock's suggestion worked perfectly when cutting and pasteing in a Windows 95, Office 97 environment.. Thanks, Steve DeLurgio Krassimir Petrov wrote: > >From: "Wai Lee" > >To: "RATS Discussion List" > >Subject: Re: Cutting and pasteing with office 97. > >Date: Fri, 2 Oct 1998 13:49:15 -0400 > >Reply-to: "RATS Discussion List" > > > >Tom's suggestions do not work for me. > > > >To be exact, I have Word version 7.0a for Windows 95 but running on NT. > >When I do Edit-Paste Special, I have 3 choices: "Picture," "Bitmap," > and > >"Device Independent Bitmap." Of course, I don't have choice when I do > just > >Edit-Paste. I have just tested all combinations, and all of them give > >funcky graphs in Word. I usually do OPEN COPY and then plot the graphs > I > >need by other means. > > > >This could be an exceptional case unique to the set up of my machine. > >Nevermind. > > > >W. Lee > > > > > > Lee, > > I was about to complement Tom's comment on the subject: > > 1. COPY-PASTE has never worked for me (MS Word) > 2. COPY-PASTE SPECIAL "CONVERT FROM BITMAP" has always worked > ===================== > for me faultlessly. The other two options on "convert from" > do not work. > > I use Win95, if that matters. > Hope this helps > > Krassimir Petrov > P.O.Box 3703 > Columbus, Oh 43210 > cell. 614-579-7919 > pager 614-665-2951 > off. 614-292-9519 > > ______________________________________________________ > Get Your Private, Free Email at http://www.hotmail.com ---------- End of message ---------- From: bertrand gordon To: "RATS Discussion List" Subject: Expert in Cats for rats Date: Sun, 4 Oct 1998 19:58:40 -0400 (EDT) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Light Version 1.5.2 (via Mercury MTS (Bindery) v1.40) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Dear Rats Users, In working through "uk.cat" example in the "cats in rats" Handbook I encountered a problem with the "matrix of restrictions" example on page 37. When I chose option 2 and input 1 and 1 for the different groups and number of restrictions respectively, the transpose H matrix showed up. I then filled in the matrix of restrictions as showned on page 37. However, when I press enter to execute that procedure I get the message " Alert, Bad Number " and the entire procedure starts all over again. I cannot continue woking through the example. The process seemed to break down. Can anyone tell me what the problem is? I greatly appreciate this help. Thanks bert. ---------- End of message ---------- From: Wolfgang Bauer To: "RATS Discussion List" Subject: VAR Date: Mon, 5 Oct 1998 10:22:25 +0200 (CEST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Mailer: Mercury MTS (Bindery) v1.40 Dear RATS users, I'm working on my first VAR estimation and have therefore some questions: 1) The determinant of my estimated covarince matrix of residuals is almost zeros; the eigenvalues are smaller than 1e-005. What does this mean? Is it because my time series are pretty highly auto- and cross correlated? Is the therefore negative AIC meaningful? 2) How can I store the results (of ESTIMATE) of a VAR? Only with separate LINREG? I am grateful for any advice, greetings Wolfgang Bauer ---------- End of message ---------- From: mark.astley@bankofengland.co.uk (Mark Astley) To: "RATS Discussion List" Subject: Problem with SVAR.src Date: Mon, 05 Oct 1998 17:15:17 +0100 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 Dear fellow Rats users. I have been trying to use the SVAR.src procedure to estimate a structural VAR. In particular, I've been using it to estimate a 5 variable system identified with the following 10 long-run restrictions (ie its an exactly identified C model): Shock 2 has no LR effects on variables 1,2 and 3. Shock 3 has no LR effect on variables 1 or 2 Shock 4 has no LR effect on variables 1,2 or 5 Shock 5 has no LR effect on variable 1 or 3. The problem is that SVAR.src informs me that this system is not identified (with identification Rank 1). This message surprised me somewhat as a visual inspection of the system suggests that it is identified (ie all the shocks can be distinguished from each other) & I've previously managed to estimate the system using a different program. I dont think that there is any problem with the code I am using to run SVAR.src - I've managed to use it to implement an overidentified system based upon the above (which adds four more restrictions - shock 1 has no LR effect on variable 2; shock 2 has no LR effect on variable 4; shock 3 has no LR effect on variable 4; shock 5 has no LR effect on variable 4) which works fine & I've implemented alternative exactly identified systems. I have tried reordering the endogenous variables/shocks - but this did not sort out the problem. Does anyone have any ide whats going on? (does anyone understand the identification test that SVAR.src running)? Many thanks in advance. ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: Cutting and pasteing with office 97. Date: Mon, 5 Oct 1998 13:43:08 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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) > > To be exact, I have Word version 7.0a for Windows 95 but running on NT. > When I do Edit-Paste Special, I have 3 choices: "Picture," "Bitmap," and > "Device Independent Bitmap." Of course, I don't have choice when I do just > Edit-Paste. I have just tested all combinations, and all of them give > funcky graphs in Word. I usually do OPEN COPY and then plot the graphs I > need by other means. > Which Version of RATS and what operating system are you using? 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: "Coleman, Mark" To: "RATS Discussion List" Subject: simple question Date: Mon, 5 Oct 1998 16:48:07 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2232.9) (via Mercury MTS (Bindery) v1.40) Content-Type: text/plain Hi, I'm (slightly) embarassed to ask, but is there a quick way (i.e., in a single line of RATS code) of counting the number of valid observations in a series. Normally I would use the STATISTICS command; but this fails if the series has no valid observations. Thanks Mark ---------- End of message ---------- From: "Coleman, Mark" To: "RATS Discussion List" Subject: Apology Date: Mon, 5 Oct 1998 16:55:40 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2232.9) (via Mercury MTS (Bindery) v1.40) Content-Type: text/plain Greetings: Please disregard my last posting. The answer is obvious. My apologies to the list. Regards, Mark ---------- End of message ---------- From: Barbara Jennes To: "RATS Discussion List" Subject: Re: Expert in Cats for rats Date: Tue, 06 Oct 1998 10:36:20 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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="iso-8859-1" Content-Transfer-Encoding: quoted-printable At 19:58 04.10.98 -0400, you wrote: >Dear Rats Users, > > In working through "uk.cat" example in the "cats in rats" Handbook I >encountered a problem with the "matrix of restrictions" example on page= 37. >When I chose option 2 and input 1 and 1 for the different groups and= number >of restrictions respectively, the transpose H matrix showed up. I then >filled in the matrix of restrictions as showned on page 37. However, when= I >press enter to execute that procedure I get the=20 > message " Alert, Bad Number " and the entire procedure >starts all over again. I cannot continue woking through the example. The >process seemed to break down.=20 > >Can anyone tell me what the problem is? I greatly appreciate this help. > > >Thanks bert. > =20 > =20 > > > > > Dear Bert, I=B4m not sure what the problem is, but I can tell you how to solve it: Don= =B4t press "Enter" to accept the first row of the matrix. Change all the entries of the transpose H matrix, move=20 the cursor to a field with a zero and then exit the procedure by clicking on the X at the upper right corner. You get the message "Save changes?". Click "Yes" and then you can continue working with the restrictions. =20 Good luck, Barbara =20 ---------- End of message ---------- From: Barbara Jennes To: "RATS Discussion List" Subject: Re: Problem with SVAR.src Date: Tue, 06 Oct 1998 10:59:31 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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" At 17:15 05.10.98 +0100, you wrote: >Dear fellow Rats users. > >I have been trying to use the SVAR.src procedure to estimate a >structural VAR. In particular, I've been using it to estimate a 5 >variable system identified with the following 10 long-run restrictions >(ie its an exactly identified C model): > >Shock 2 has no LR effects on variables 1,2 and 3. >Shock 3 has no LR effect on variables 1 or 2 >Shock 4 has no LR effect on variables 1,2 or 5 >Shock 5 has no LR effect on variable 1 or 3. > >The problem is that SVAR.src informs me that this system is not >identified >(with identification Rank 1). This message surprised me somewhat as a >visual inspection of the system suggests that it is identified (ie all >the shocks can be distinguished from each other) & I've previously >managed to estimate the system using a different program. > >I dont think that there is any problem with the code I am using to run >SVAR.src - I've managed to use it to implement an overidentified system >based upon the above (which adds four more restrictions - shock 1 has >no LR effect on variable 2; shock 2 has no LR effect on variable 4; >shock 3 has no LR effect on variable 4; shock 5 has no LR effect on >variable 4) which works fine & I've implemented alternative exactly >identified systems. I have tried reordering the endogenous >variables/shocks - but this did not sort out the problem. > >Does anyone have any ide whats going on? (does anyone understand the >identification test that SVAR.src running)? > >Many thanks in advance. > > > I mention a very similar problem in identifying a structural model from my estimated VAR. Imposing the correct number of restrictions is sometimes the nessesary, but not the sufficient condition for exact identification. There is an example in Enders (1995, p.324). I would think that the problem is the zero restrictions (3,2) and (2,3). My experience shows that this constellation leads to problems described in your mail. Maybe this could be a hint for you. Best regards, Barbara Jennes ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: NT Graphics problem (was Cutting and pasteing with office 97) Date: Tue, 6 Oct 1998 17:00:57 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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) Wai Lee had posted earlier that using Paste Special to select the standard "picture" or "metafile" format didn't eliminate his graphics problem. I'm cc'ing my response to the RATS mailing list, because I think it will be of general interest: Wai, et al: What you are running into here is an NT-specific problem that is entirely unrelated to the "metafile" vs. "enhanced metafile" issue. When running WinRATS 4.20 and 4.21 under NT, if you try to copy a graph to the clipboard or save a graph in WMF or PostScript formats and then try to paste that graph into another application, you will get poor results. Usually, the graph box itself will be very small while the labels, key, tick marks, etc. are exploded way out of their desired proportion. This problem occured because certain scaling calculations behaved differently in the NT environment. Fortunately, the problem was fixed with Version 4.3. So if you are running an older version under NT and want to export graphs, you'll want to update to the current version (which is just a $25 update from 4.2). Note that even with Version 4.3, you'll still need to use the standard metafile or picture format, rather than enhanced metafile format, when copying and pasting--as discussed in earlier posts. If anyone has questions on this or need update info, e-mail directly at estima@estima.com rather than the mailing list. 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" To: "RATS Discussion List" Subject: Unsubscibe Date: Wed, 7 Oct 1998 00:25:35 -0300 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 Content-Type: text/plain; Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Outlook Express 4.71.1712.3 (via Mercury MTS (Bindery) v1.40) ---------- End of message ---------- From: Alen Belullo To: "RATS Discussion List" Subject: rats graph into word with vlabel Date: Wed, 7 Oct 1998 12:07:44 +0200 (MET DST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Mailer: Mercury MTS (Bindery) v1.40 Dear RATS users I have problem when I try to copy a graph that has vertical label for inseting into Word document. When I try to copy the graph I must close application. Is it RATS or Windows problem? How can I put this kind of graph into Word Document? Sincerely, Alen Belullo ---------- End of message ---------- From: wobauer@studi.unizh.ch To: "RATS Discussion List" Subject: Syntax Date: Wed, 7 Oct 1998 17:20:10 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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've got three questions on how to write general procedures that can do with different number of time series. -I've defined a vector of series as regressors for LINREG. How can I inlcude {1 to lags} of each series in this vector? vector{1 to lags} does not work. -when using FORECAST n, I have to give n supplementary cards to store the forecasts. How can this be done if n is not known in beforehand? The series for storing than can't be written down explicitly. Thanks for your help Greetings Wolfgang Bauer ---------- End of message ---------- From: "Alejandro J. Santillan Iturres" To: "RATS Discussion List" Subject: Unsubscribe Date: Wed, 7 Oct 1998 12:30:37 -0300 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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) Anyone can tell me how to unsubscribe RATS? Thank you! Alejandro J. Santillan Iturres alela@netverk.com.ar www.netverk.com.ar/~alela/ ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: Syntax Date: Wed, 7 Oct 1998 10:57:09 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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've got three questions on how to write general procedures that can > do with different number of time series. > > -I've defined a vector of series as regressors for LINREG. How can I > inlcude {1 to lags} of each series in this vector? vector{1 to lags} > does not work. > If you want to write it out explictily, do: # v(1){1 to lags} v(2){1 to lags} v(3){1 to lags} etc. If you want to do it automatically, you'll need to use ENTER to build up a regressor list. For example, this is taken from the FPE.SRC procedure that ships with RATS (I added the DECLARE, because BUILDSUP is defined in FPE as a local procedure variable): declare vector[integer] buildsup ENTER(VARYING) BUILDSUP # DETVARS DO I=1,NUMVAR ENTER(VARYING) BUILDSUP # BUILDSUP VARS(I){0 TO MAXLAG} END DO I > -when using FORECAST n, I have to give n supplementary cards to > store the forecasts. How can this be done if n is not known in > beforehand? The series for storing than can't be written down > explicitly. > You can use the LIST and CARDS instructions for this. See, for example, the CNDITION.SRC proc. 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" To: "RATS Discussion List" Subject: Re: Syntax Date: Wed, 7 Oct 1998 11:01:38 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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) > If you want to do it automatically, you'll need to use ENTER to build > up a regressor list. For example, this is taken from the FPE.SRC > procedure that ships with RATS (I added the DECLARE, because > BUILDSUP is defined in FPE as a local procedure variable): > > declare vector[integer] buildsup > ENTER(VARYING) BUILDSUP > # DETVARS > DO I=1,NUMVAR > ENTER(VARYING) BUILDSUP > # BUILDSUP VARS(I){0 TO MAXLAG} > END DO I > > > I forgot to mention that the point of this is that you can then use BUILDSUP in the regressor list: linreg y # buildsup 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: wobauer@studi.unizh.ch To: "RATS Discussion List" Subject: LINREG for VAR Date: Thu, 8 Oct 1998 12:40:36 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 All, in the hanbook, p. 8-5 it says " you can use a set of LINREG instructions" on a VAR. What are the appropriate LINREG instructions to get the same results as with VAR? I've tried the following: instead of system 1 to n variables 1 to n lags 1 to lag det constant end(system) estimate start end I used (thanks to Tom Maycock!) DO I=1,N ENTER(VARYING) BUILDSUP # BUILDSUP ([series](I)){1 TO LAG} END DO I do i=1,n linreg i start end #buildsup end do i But this gives me n CONSTANTs in the LINREG instruction; what is the right way to do this? Thank you very much for your help Wolfgang Bauer ---------- End of message ---------- From: wobauer@studi.unizh.ch To: "RATS Discussion List" Subject: GMM without IV Date: Thu, 8 Oct 1998 12:40:36 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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, to do a GMM estimation, I've got moment restrictions of the form Kov(y(t), y(t+k))=function(k) where function(k) is a function of the parameters of interest. In the RATS hanbook it says that only moment restrictions in orthogonal form can be estimated with RATS; is there really no way to estimate with the restrictions above? Thank you very much for help Wolfgang Bauer ---------- End of message ---------- From: "Nico Valckx" To: "RATS Discussion List" Subject: Re: GMM without IV Date: Thu, 8 Oct 1998 13:12:30 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 Content-Type: text/plain; Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Outlook Express 4.72.2106.4 (via Mercury MTS (Bindery) v1.40) You wrote: To do a GMM estimation, I've got moment restrictions of the form Kov(y(t), y(t+k))=function(k) where function(k) is a function of the parameters of interest. In the RATS hanbook it says that only moment restrictions in orthogonal form can be estimated with RATS; is there really no way to estimate with the restrictions above? -------------------------------------- Dear Wolfgang, Can't you transform your expression to [ Kov()*inverse(function(k)) -1 ] * Z Z denoting the instruments This is the othogonalized form. GMM without IV is a contradictio in terminis, I would say. Cheers, Nico Valckx UFSIA-University of Antwerp, Belgium. ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: LINREG for VAR Date: Thu, 8 Oct 1998 09:20:55 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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) > > in the hanbook, p. 8-5 it says " you can use a set of LINREG > instructions" on a VAR. What are the appropriate LINREG > instructions to get the same results as with VAR? > I've tried the following: > > instead of > > system 1 to n > variables 1 to n > lags 1 to lag > det constant > end(system) > estimate start end > > I used (thanks to Tom Maycock!) > > DO I=1,N > ENTER(VARYING) BUILDSUP > # BUILDSUP ([series](I)){1 TO LAG} > END DO I > do i=1,n > linreg i start end > #buildsup > end do i > > But this gives me n CONSTANTs in the LINREG instruction; what is > the right way to do this? > No, probably not. You can certainly use ENTER to build up the regression list, but it's probably not worth it. If you simply want to estimate using LINREG rather than ESTIMATE, just do: system 1 to n variables 1 to n lags 1 to lag det constant end(system) to define the equations, and then estimate them using LINREG: LINREG(EQUATION=1) LINREG(EQUATION=2) etc. And you can always just spell out everything on the LINREG: LINREG x # constant x{1 to lags} y{1 to lags} z{1 to lags} 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: Simon.Van-Norden@hec.ca To: "RATS Discussion List" Subject: Testing Spectra Date: Sat, 10 Oct 1998 15:18:13 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 want to test whether the spectrum of an observed series is equal to some abitrary spectrum predicted by theory. Does anyone know of code for (or references to) formal statistical tests of this kind of hypothesis? Can this be extended to testing the co-spectrum (i.e. coherence and phase) of two observed series? Many thanks. SvN -- 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: "Christopher F Baum" To: "RATS Discussion List" Subject: Re: Testing Spectra Date: Sat, 10 Oct 1998 17:52:58 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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=iso-8859-1 Content-Transfer-Encoding: quoted-printable Isn't that what Durbin's cumulated periodogram test (as described in the manual) does against the null of a flat spectrum, i.e. white noise? This seems like the analog to the Chi-squared goodness of fit test, so I would think that it could be adapted to something other than a flat spectrum, using the same notion of the divergence from a theoretical base. Perhaps Durbin's original article (1969) would provide guidance along those lines. Kit Baum --On Sat, Oct 10, 1998 15:18 -0400 Simon.Van-Norden@hec.ca wrote: > I want to test whether the spectrum of an observed series is equal to > some abitrary spectrum predicted by theory. > > Does anyone know of code for (or references to) formal statistical tests > of this kind of hypothesis? > > Can this be extended to testing the co-spectrum (i.e. coherence and > phase) of two observed series? > > Many thanks. > > > SvN > -- > Simon van Norden Professeur invit=E9 > 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, =C9cole des H.E.C. > 3000 Cote-Sainte-Catherine, Montreal QC, CANADA H3T 2A7 > > ---------- End of message ---------- From: "Bao, Yuhua" To: "RATS Discussion List" Subject: simulation Date: Sat, 10 Oct 1998 17:50:33 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2232.9) (via Mercury MTS (Bindery) v1.40) Content-Type: text/plain; Dear friends on the list, I am currently running a four-variable VAR model with RATS and would like to do simulation on the model under the condition that one of the equations is revised according to some policy rule. Namely, the monetary base is a function of the deviation of actual value from desired value of essential macro indicators in the last period. As I am quite new to RATS, so far I haven't figured out how to do this kind of simulation. I tried FRML-GROUP-SIMULATE, with the VCV derived from VAR estimation as the covariance matrix to generate random shocks by, but it doesn't work. Further, I would like to compute a 95% confidence interval for the simulation results after, say, 500 sets of random shocks have been generated. But I simply don't know how to write the outcomes of the simulation and derive the up and lower bound. Please write me if you have some idea or comments on that. A sample program that is similar in nature would be especially helpful. Thanks a lot. Yuhua Bao ---------- End of message ---------- From: "Philippe PROTIN" To: "RATS Discussion List" Subject: KALMAN FILTER Date: Mon, 12 Oct 1998 18:01:41 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Ecole Superieure des Affaires MIME-Version: 1.0 Content-type: text/plain; charset=ISO-8859-1 Content-transfer-encoding: Quoted-printable X-mailer: Pegasus Mail for Windows (v2.54) (via Mercury MTS (Bindery) v1.40) Dear Rats users, I'd like to estimate a model using the Kalman Filter as discribed in manual pp. 13-1 to 13-6. Suppose we have two markets. Market n=B01 is the leading one : r1(t) =3D a(t).x1(t) + u1(t) V[u1(t)] =3D H1(t) a(t) =3D T1.a(t-1) + v1(t) V[v1(t)] =3D Q1(t) Market n=B02 is as follows : r2(t) =3D b0(t) + b1(t).r2(t-1) + b2(t).u1(t) + u2(t) V[u2(t)] =3D H2(t) bi(t) =3D Ti.bi(t-1) + zi(t) V[zi(t)] =3D Zi(t) How can I compute this in RATS ? in particular, how can I tell RATS to include the u1(t) terms in the r2(t) equation ? Thanks for help. Philippe \\\|||||/// \\ - - // ( @ @ ) +------oOOo-(_)-oOOo---- +----------------------------------------+ | | Philippe PROTIN | | Have a nice day ! | Ecole Sup=E9rieure des Affaires | | | Tel: (33) 04.76.82.57.48 | | | e-Mail: protin@esa.upmf-grenoble.fr | = +---------------Oooo-----+---------------------------------------------+ oooO ( ) ( ) ) / \ ( (_/ \_) ---------- End of message ---------- From: Sebastian.SCHICH@oecd.org To: "RATS Discussion List" Subject: Likelihood in "SUR" Date: Tue, 13 Oct 1998 12:05:44 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 X-Mailer: Internet Mail Service (5.5.2232.9) (via Mercury MTS (Bindery) v1.40) Content-Type: text/plain Dear RATS users, to see whether I should allow for contemporaneous covariance of the errors from two equations, I have estimated them both separately and jointly in order to compare the likelihoods. While the likelihoods from the separate equations seem to be fine (first two bold computations), the likelihood of the system (last bold computation) seems to be very high -- and is indeed much higher than the one obtained as standard output from SUR regressions with Microfit 4.0. i) Is the computation of the likelihood in the last equation correct (last bold row) and ii) how would I have to adjust for the numbers of regressors in each equation? Thanking you in advance for your thoughts. LINREG(NOPRINT,DEFINE=dlyusa_eq) dlyusa #constant dlyusa{1} dlyusa{2} dlqusa{9} dlqusa{10} d74 COMPUTE logl1 = - (%NOBS/2)*(1 + LOG(2*%PI*%RSS/%NOBS)) LINREG(NOPRINT, DEFINE=dlqusa_eq) dlqusa #constant dlyusa{1} dlyusa{2} dlqusa{1} dlqusa{2} COMPUTE logl2 = - (%NOBS/2)*(1 + LOG(2*%PI*%RSS/%NOBS)) SUR(NOPRINT, NOSIGMA) 2 # dlyusa_eq # dlqusa_eq COMPUTE logl3 = -(%NOBS*2/2)*LOG(2*%PI)-(%NOBS/2)* %LOGDET CDF CHISQUARE (2*(logl3-(logl1+logl2))) 1 ---------- End of message ---------- From: arturo jose galindo To: "RATS Discussion List" Subject: MSM Estimation Date: Tue, 13 Oct 1998 06:57:11 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 Dear Rats Users: Has any one done method of simulated moments estimation (as in Ingram and Lee, Journal of Econometrics 1991) using RATS? I am having trouble estimating the asymptotic standard errors of the estimated coeffcients. I would appreciate any help. Thanks, Arturo Galindo Department of Economics University of Illinois at Urbana-Champaign ---------- End of message ---------- From: To: "RATS Discussion List" Subject: Multivariate HP filter Date: Wed, 14 Oct 1998 09:06:08 +1200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-type: text/plain; charset=us-ascii X-Mailer: Mercury MTS (Bindery) v1.40 Dear RATS users, Do any of you have the RATS code to run the multivariate HP filter ala Laxton and Tetlow. I have contacted Doug Laxton and he says that it was coded into RATS a number of years ago but he doesn't have it anymore. I would appreciate any assistance. Rgds, Peter Redward ---------- End of message ---------- From: Simon.Van-Norden@hec.ca To: "RATS Discussion List" Subject: Re: Multivariate HP filter Date: Tue, 13 Oct 1998 17:50:44 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 Peter: I don't have code, but I have some experience with Laxton and Tetlow's filter. (I coauthored a Bank of Canada Technical Report that looked at its successor.) While not difficult to code, their filter doesn't make much sense and it doesn't do a good job of estimating current output gaps. If that's what you wanted to use it for, you might want to read some of the subsequent stuff that came out of the Bank of Canada after they left. SvN peter.redward@bankerstrust.co.nz wrote: > Dear RATS users, > > Do any of you have the RATS code to run the multivariate HP filter ala > Laxton and Tetlow. I have contacted Doug Laxton and he says that it was > coded into RATS a number of years ago but he doesn't have it anymore. I > would appreciate any assistance. > > Rgds, > > Peter Redward -- 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: KHL2288@UTARLG.UTA.EDU To: "RATS Discussion List" Subject: RESTRICT(create) in GARCH Date: Wed, 14 Oct 1998 13:27:47 -0500 (CDT) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-version: 1.0 Content-type: TEXT/PLAIN; CHARSET=US-ASCII X-Mailer: Mercury MTS (Bindery) v1.40 Dear RATS users, I'm trying to estimate a ARMA(1,1)-GARCH(1,1) model using Dr. Trevor's sample code. What I want to know is how I can estimate the coefficients of the restricted(restrictions on GARCH parameters; i.e. VA=VB=0) model. I used RESTRICT with create option, but the only out I got is test statitic with significance level. Many thanks in advance. Kyong Lee Kyong H. Lee Dept. of Economics University of Texas at Arlington ---------- End of message ---------- From: Norman Morin To: "RATS Discussion List" Subject: Re: RESTRICT(create) in GARCH Date: Wed, 14 Oct 1998 15:20:12 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 > Dear RATS users, > I'm trying to estimate a ARMA(1,1)-GARCH(1,1) model using Dr. Trevor's sample > code. What I want to know is how I can estimate the coefficients of the > restricted(restrictions on GARCH parameters; i.e. VA=VB=0) model. I used > RESTRICT with create option, but the only out I got is test statitic with > significance level. > > Kyong Lee I believe the vector %BETA will contain the restricted coefficient estimates if RESTRICT(create) is used. 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: "Estima" To: "RATS Discussion List" Subject: Re: RESTRICT(create) in GARCH Date: Wed, 14 Oct 1998 16:14:09 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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'm trying to estimate a ARMA(1,1)-GARCH(1,1) model using Dr. Trevor's sample > > code. What I want to know is how I can estimate the coefficients of the > > restricted(restrictions on GARCH parameters; i.e. VA=VB=0) model. I used > > RESTRICT with create option, but the only out I got is test statitic with > > significance level. > > > > Kyong Lee > > I believe the vector %BETA will contain the restricted > coefficient estimates if RESTRICT(create) is used. > This is true. However, be aware that doing RESTRICT(CREATE) is not the same thing as actually estimating the restricted model with MAXIMIZE (i.e. if you were to set VA and VB equal to zero, use NONLIN(DROP) to drop them from the list of free parameters, and then re-estimate using MAXIMIZE). Both methods are valid, depending on the intended use. RESTRICT(CREATE) is doing a straightforward numerical computation involving the existing covariance and coefficient matrices and the matrix of restrictions (see page 6-4 in the RATS manual). If the restrictions are "true", then the resulting coefficients will be very close to the results you would get by actually estimating the restricted model. The results will not be as close if the restrictions are not "true". Also, note that the imposed restrictions may not hold exactly when applied to non-linear estimators--for example, if you use RESTRICT(CREATE) to impose the restriction that B1==0, the resulting value of B1 may not be EXACTLY zero, although it should be very close. This usually isn't something to be concerned with, just something to be aware of depending on your intended purpose. 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: bertrand gordon To: "RATS Discussion List" Subject: Expert in Cats for rats Date: Sun, 18 Oct 1998 13:24:54 -0400 (EDT) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Light Version 1.5.2 (via Mercury MTS (Bindery) v1.40) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Dear Rats Users, Thanks to many of your excellent answers I was able to solve my problem dealing with "Matrix restrictions" in the cats for rats handbook. There is another issue I will like to discuss. In implementing various example of matrix restrictions on a given H-matrix whose rank is 4 I always get a likelihood ratio statistic and p-value as The LR test, CHISQ2) = 00, p-value = NA Can any comment on this result. I will greatly appreciate various comment on this issue. Thanks bert. ---------- End of message ---------- From: khl2288@UTARLG.UTA.EDU To: "RATS Discussion List" Subject: Convergence problem in GARCH Date: Sun, 18 Oct 1998 17:08:45 -0500 (CDT) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII X-Mailer: Mercury MTS (Bindery) v1.40 Dear RATS users, I have a convergence problem in estimating multivariate GARCH model. The output says that I need to set higher iterations limit using NLPAR, so that I increased the subiteration limit from 50 to 100, 200, and 300. But I still got the same problem. Even though the iteration limit in MAXIMIZE was 500, the number of iterations taken till it failed to converge was around 60. The following is the part of the output: NLPAR(SUBITER=300) MAXIMIZE(METHOD=BFGS,ROBUSTERRORS,RECURSIVE,ITERS=500) LOGL GSTART GEND Estimation by BFGS NO CONVERGENCE IN 68 ITERATIONS LAST CRITERION WAS 0.0000000 SUBITERATIONS LIMIT EXCEEDED. USE NLPAR TO SET HIGHER LIMIT Should I increase the subiterations limit even more, or is it O.K. to scale the variables by multiplication and to run the same program? If I scale the variables equally, I think the only parameter affected in scale is constant parameters in equations for mean and variance. Is it right? I greatly appreciate any comments. Best regards, Kyong H. Lee Dept. of Economics University of Texas at Arlington ---------- End of message ---------- From: "Wai Lee" To: "RATS Discussion List" Subject: Re: Convergence problem in GARCH Date: Mon, 19 Oct 1998 08:48:02 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-type: text/plain; charset=us-ascii X-Mailer: Mercury MTS (Bindery) v1.40 I guess most RATS users will suggest using SIMPLEX for a few iterations first. If problem persists, use different starting values. The likelihood surface may have many flat regions. I usually do a small grid search for starting value to check both convergence and whether my estimates are global optimum, but this takes longer, of course. Rescaling of some variables (NOT all) to make coefficients of similar magnitudes sometimes help in my experience. My last set of alternatives is to play with different options in NLPAR, but in doing so, I always need additional reference on optimization in additional to RATS manual. W. Lee khl2288 @ UTARLG.UTA.EDU on 10/18/98 06:08:45 PM Please respond to "RATS Discussion List" To: "RATS Discussion List" cc: (bcc: Wai Lee) Subject: Convergence problem in GARCH Dear RATS users, I have a convergence problem in estimating multivariate GARCH model. The output says that I need to set higher iterations limit using NLPAR, so that I increased the subiteration limit from 50 to 100, 200, and 300. But I still got the same problem. Even though the iteration limit in MAXIMIZE was 500, the number of iterations taken till it failed to converge was around 60. The following is the part of the output: NLPAR(SUBITER=300) MAXIMIZE(METHOD=BFGS,ROBUSTERRORS,RECURSIVE,ITERS=500) LOGL GSTART GEND Estimation by BFGS NO CONVERGENCE IN 68 ITERATIONS LAST CRITERION WAS 0.0000000 SUBITERATIONS LIMIT EXCEEDED. USE NLPAR TO SET HIGHER LIMIT Should I increase the subiterations limit even more, or is it O.K. to scale the variables by multiplication and to run the same program? If I scale the variables equally, I think the only parameter affected in scale is constant parameters in equations for mean and variance. Is it right? I greatly appreciate any comments. Best regards, Kyong H. Lee Dept. of Economics University of Texas at Arlington ---------- End of message ---------- From: khl2288@UTARLG.UTA.EDU To: "RATS Discussion List" Subject: Re: RESTRICT(create) in GARCH Date: Mon, 19 Oct 1998 14:27:19 -0500 (CDT) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-version: 1.0 Content-type: TEXT/PLAIN; charset=US-ASCII X-Mailer: Mercury MTS (Bindery) v1.40 Let me recompose the question I have. I am trying to estimate a version of CAPM model. In the first model, the conditional variance is assumed to be constant. In the second model, the conditional variance is assumed to follow GARCH(1,1) process. Ri(t) = Ai + Bi*Hii(t) + Ci*Ui(t-1) + Ui(t). Hii(t) = VCi + VAi*(Ui(t-1)**2) + VBi*Hii(t-1). I'd like to conduct the LB tests for the residuals from both models. Can I still use the procedure suggested by Tom? I mean, is it OK to estimate the first model by using MAXIMIZE with NONLIN(DROP) VA VB? Thank you so much for your help. Best regards, Kyong H. Lee Dept. of Economics University of Texas at Arlington On Wed, 14 Oct 1998, Estima wrote: > > > I'm trying to estimate a ARMA(1,1)-GARCH(1,1) model using Dr. Trevor's sample > > > code. What I want to know is how I can estimate the coefficients of the > > > restricted(restrictions on GARCH parameters; i.e. VA=VB=0) model. I used > > > RESTRICT with create option, but the only out I got is test statitic with > > > significance level. > > > > > > Kyong Lee > > > > I believe the vector %BETA will contain the restricted > > coefficient estimates if RESTRICT(create) is used. > > > > This is true. > > However, be aware that doing RESTRICT(CREATE) is not the same thing > as actually estimating the restricted model with MAXIMIZE (i.e. if > you were to set VA and VB equal to zero, use NONLIN(DROP) to drop > them from the list of free parameters, and then re-estimate using > MAXIMIZE). Both methods are valid, depending on the intended use. > > RESTRICT(CREATE) is doing a straightforward numerical computation > involving the existing covariance and coefficient matrices and the > matrix of restrictions (see page 6-4 in the RATS manual). > > If the restrictions are "true", then the resulting coefficients will > be very close to the results you would get by actually estimating > the restricted model. The results will not be as close if the > restrictions are not "true". > > Also, note that the imposed restrictions may not hold exactly > when applied to non-linear estimators--for example, if you use > RESTRICT(CREATE) to impose the restriction that B1==0, the resulting > value of B1 may not be EXACTLY zero, although it should be very > close. This usually isn't something to be concerned with, just > something to be aware of depending on your intended purpose. > > 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: "=?iso-8859-1?Q?Luis_Fernando_Mej=EDa_Alzate?=" To: "RATS Discussion List" Subject: Need help with a simple exercise!!! Date: Tue, 20 Oct 1998 19:49:53 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 Content-Type: multipart/alternative; X-Mailer: Microsoft Outlook Express 4.71.1712.3 (via Mercury MTS (Bindery) v1.40) This is a multi-part message in MIME format. ------=_NextPart_000_001E_01BDFC62.CE595F00 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable Sorry to use this list to ask such a "simple" question but I need your = advice on this problem: I need to simulate 100 observations of the = following models: Yt =3D 2.0 + 0.6 Yt-1 + Et Yt =3D 1.0 + 0.7 Yt-1 + 0.4 Yt-2 + Et Yt =3D 3.0 + Et + 1.8 Et-1 Yt =3D 3.0 + Et + 0.8 Et-1 - 0.3 Et-3 Yt =3D 0.2 + 0.6 Yt-1 + Et + 0.4 Et-1 Here is what I've done: ALL 1 100 SET E =3D %RAN(0.5) SET Y1 1 1 =3D 0.0 SET Y2 1 1 =3D 0.0 SET Y3 1 1 =3D 0.0 SET Y4 1 1 =3D 0.0 SET Y5 1 1 =3D 0.0 EQU(CONSTANT) EQY1 Y1 1 0 ASS(RESIDS=3DE) EQY1 #2.0 0.6 EQU(CONSTANT) EQY2 Y2 2 0 ASS(RESIDS=3DE) EQY2 #1.0 0.7 0.4 EQU(CONSTANT) EQY3 Y3 0 1 ASS(RESIDS=3DE) EQY3 #3.0 1.8 EQU(CONSTANT,MA=3DINPUT) EQY4 Y4 0 2 #1 3 ASS(RESIDS=3DE) EQY4 #3.0 0.8 -0.3 EQU(CONSTANT) EQY5 Y5 1 1 ASS(RESIDS=3DE) EQY5 #0.2 0.6 0.4 SIMULATE 1 100 2 #EQY1 Y1 SIMULATE 1 100 2 #EQY2 Y2 SIMULATE 1 100 2 #EQY3 Y3 SIMULATE 1 100 2 #EQY4 Y4 SIMULATE 1 100 2 #EQY5 Y5 The graph of the ACF and PACF of these simulated series (not to mention = the graphic of the series itself) is not what I expected them to be. Can = anyone tell me what I'm doing wrong? Please, I need your advice AS SOON = AS POSSIBLE. Best regards, Luis Fernando. P.D: I know this list should only used by post-graduate and advanced = undergraduate students and government and commercial researchers. Again, = I'll have to apologize to everyone who feels ofended by this rather = "stupid" question. ------=_NextPart_000_001E_01BDFC62.CE595F00 Content-Type: text/html; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable
Sorry to use this list to ask such a "simple" question = but I need=20 your advice on this problem: I need to simulate 100 observations of the=20 following models:

Yt =3D 2.0 + 0.6 Yt-1 + Et
Yt =3D 1.0 + 0.7 = Yt-1 + 0.4=20 Yt-2 + Et
Yt =3D 3.0 + Et + 1.8 Et-1
Yt =3D 3.0 + Et + 0.8 Et-1 - = 0.3=20 Et-3
Yt =3D 0.2 + 0.6 Yt-1 + Et + 0.4 Et-1

Here is what I've=20 done:

ALL 1 100
SET E =3D %RAN(0.5)
SET Y1 1 1 =3D = 0.0
SET Y2 1 1 =3D=20 0.0
SET Y3 1 1 =3D 0.0
SET Y4 1 1 =3D 0.0
SET Y5 1 1 =3D = 0.0
EQU(CONSTANT)=20 EQY1 Y1 1 0
ASS(RESIDS=3DE) EQY1
#2.0 0.6
EQU(CONSTANT) EQY2 Y2 = 2=20 0
ASS(RESIDS=3DE) EQY2
#1.0 0.7 0.4
EQU(CONSTANT) EQY3 Y3 0=20 1
ASS(RESIDS=3DE) EQY3
#3.0 1.8
EQU(CONSTANT,MA=3DINPUT) EQY4 = Y4 0 2
#1=20 3
ASS(RESIDS=3DE) EQY4
#3.0 0.8 -0.3
EQU(CONSTANT) EQY5 Y5 1=20 1
ASS(RESIDS=3DE) EQY5
#0.2 0.6 0.4

SIMULATE 1 100 = 2
#EQY1=20 Y1
SIMULATE 1 100 2
#EQY2 Y2
SIMULATE 1 100 2
#EQY3 = Y3
SIMULATE 1=20 100 2
#EQY4 Y4
SIMULATE 1 100 2
#EQY5 Y5

The graph of = the ACF=20 and PACF of these simulated series (not to mention the graphic of the = series=20 itself) is not what I expected them to be. Can anyone tell me what I'm = doing=20 wrong? Please, I need your advice AS SOON AS POSSIBLE.

Best=20 regards,

Luis Fernando.
 
P.D: I know this list should only used by post-graduate and = advanced=20 undergraduate students and government and commercial researchers. Again, = I'll=20 have to apologize to everyone who feels ofended by this rather=20 "stupid" question. 
------=_NextPart_000_001E_01BDFC62.CE595F00-- ---------- End of message ---------- From: "=?iso-8859-1?Q?Luis_Fernando_Mej=EDa_Alzate?=" To: "RATS Discussion List" Subject: Need help with a simple exercise!!! Date: Tue, 20 Oct 1998 21:11:17 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 Content-Type: text/plain; Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Outlook Express 4.71.1712.3 (via Mercury MTS (Bindery) v1.40) Sorry to use this list to ask such a "simple" question but I need your advice on this problem: I need to simulate 100 observations of the following models: Yt = 2.0 + 0.6 Yt-1 + Et Yt = 1.0 + 0.7 Yt-1 + 0.4 Yt-2 + Et Yt = 3.0 + Et + 1.8 Et-1 Yt = 3.0 + Et + 0.8 Et-1 - 0.3 Et-3 Yt = 0.2 + 0.6 Yt-1 + Et + 0.4 Et-1 Here is what I've done: ALL 1 100 SET E = %RAN(0.5) SET Y1 1 1 = 0.0 SET Y2 1 1 = 0.0 SET Y3 1 1 = 0.0 SET Y4 1 1 = 0.0 SET Y5 1 1 = 0.0 EQU(CONSTANT) EQY1 Y1 1 0 ASS(RESIDS=E) EQY1 #2.0 0.6 EQU(CONSTANT) EQY2 Y2 2 0 ASS(RESIDS=E) EQY2 #1.0 0.7 0.4 EQU(CONSTANT) EQY3 Y3 0 1 ASS(RESIDS=E) EQY3 #3.0 1.8 EQU(CONSTANT,MA=INPUT) EQY4 Y4 0 2 #1 3 ASS(RESIDS=E) EQY4 #3.0 0.8 -0.3 EQU(CONSTANT) EQY5 Y5 1 1 ASS(RESIDS=E) EQY5 #0.2 0.6 0.4 SIMULATE 1 100 2 #EQY1 Y1 SIMULATE 1 100 2 #EQY2 Y2 SIMULATE 1 100 2 #EQY3 Y3 SIMULATE 1 100 2 #EQY4 Y4 SIMULATE 1 100 2 #EQY5 Y5 The graph of the ACF and PACF of these simulated series (not to mention the graphic of the series itself) is not what I expected them to be. Can anyone tell me what I'm doing wrong? Please, I need your advice AS SOON AS POSSIBLE. Best regards, Luis Fernando. P.D: I know this list should only used by post-graduate and advanced undergraduate students and government and commercial researchers. Again, I'll have to apologize to everyone who feels ofended by this rather "stupid" question. ---------- End of message ---------- From: Ricardo Chaves Lima To: "RATS Discussion List" Subject: question on Kalman Filter Date: Wed, 21 Oct 1998 08:00:20 -0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 have a simple and easy to replicate good example of time-varying (Kalman) program on RATS. Thanks for any info on this. Ricardo Chaves Lima rcl@netpe.com.br rlima@npd.ufpe.br ---------- End of message ---------- From: "Wai Lee" To: "RATS Discussion List" Subject: SPREAD in VCV Date: Wed, 21 Oct 1998 17:00:22 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-type: text/plain; charset=us-ascii X-Mailer: Mercury MTS (Bindery) v1.40 I understand how to use the option SPREAD in LINREG to do weighted least square, but I am not quite sure how the option SPREAD works in VCV. If I have a series w which is an exponential function of time from t = 1 to t = T, and I do VCV(SPREAD=w) # x y z is it computing an exponential smoothing covariance matrix of x, y, and z for me, like Riskmetrics? Thanks for any help. W. Lee ---------- End of message ---------- From: KHL2288@UTARLG.UTA.EDU To: "RATS Discussion List" Subject: subscribing discussion list Date: Fri, 30 Oct 1998 09:48:03 -0600 (CST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-version: 1.0 Content-type: TEXT/PLAIN; CHARSET=US-ASCII X-Mailer: Mercury MTS (Bindery) v1.40 Am I still subscribing the discussion list? I havn't received any mails lately. If my connection is closed, how can i reopen it? Thank you so much. Kyong H. Lee Dept. of Economics University of Texas at Arlington ---------- End of message ---------- From: "Timberlake Consultants" To: "RATS Discussion List" Subject: Re: subscribing discussion list Date: Fri, 30 Oct 1998 16:40:36 -0000 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Microsoft Internet Mail 4.70.1161 (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: quoted-printable There have not been any e-mails since the 14th of October, if you did not receive this please re-subscribe. Please do not hesitate to contact me should you have any further queries. SPECIAL OFFERS WHEN BUNDLING WITH YOUR PURCHASE: Interactive MathDictionary =A359.50 & Scientific Notebook (both powered b= y Maple) =A359.50 - Both =A3105 Regards, Teresa Timberlake ___________________________________________ For and on behalf of Timberlake Consultants A leading distributor of statistical software: Stata, PcGive, EViews, TSP Pc,=20 Gauss, Ox, RATS, Stamp, Statgraphics, Mathematica and others 47 Hartfield Crescent, West Wickham, Kent BR4 9DW, UK Tel: +44 (0) 181 4620495/0093 - Fax: +44 (0) 181 462 0493 E-Mail: ana@timberlake.co.uk URL: www.timberlake.co.uk ---------- > From: KHL2288@UTARLG.UTA.EDU > To: RATS Discussion List > Subject: subscribing discussion list > Date: 30 October 1998 15:48 >=20 > Am I still subscribing the discussion list? I havn't received any mails lately. > If my connection is closed, how can i reopen it? >=20 > Thank you so much. >=20 > Kyong H. Lee > Dept. of Economics > University of Texas at Arlington ---------- End of message ---------- From: "Wai Lee" To: "RATS Discussion List" Subject: Re: subscribing discussion list Date: Fri, 30 Oct 1998 11:06:49 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-type: text/plain; charset=us-ascii X-Mailer: Mercury MTS (Bindery) v1.40 I have noticed that the number of messages posed in RATS-L is highly positively autocorrelated. It takes time to mean revert. The last email I received is the question on SPREAD of VCV posed by myself. No reply yet. We are still on. W. Lee KHL2288 @ utarlg.uta.edu on 10/30/98 10:48:03 AM Please respond to RATS-L@efs.mq.edu.au To: RATS-L @ efs.mq.edu.au cc: (bcc: Wai Lee) Subject: subscribing discussion list Am I still subscribing the discussion list? I havn't received any mails lately. If my connection is closed, how can i reopen it? Thank you so much. Kyong H. Lee Dept. of Economics University of Texas at Arlington ---------- End of message ---------- From: Colin McLatchie To: "RATS Discussion List" Subject: Re: subscribing discussion list (off-topic) Date: Sat, 31 Oct 1998 06:53:56 +0000 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mozilla 4.07 (Macintosh; I; PPC) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit There is weak evidence that periods of high volatility are also periods of high expected return. That, combined with the empirical result that postings are positively correlated, means my sending this message could be to the list's benefit. Right? cheers, Colin Wai Lee wrote: > > I have noticed that the number of messages posed in RATS-L is highly > positively autocorrelated. It takes time to mean revert. > > The last email I received is the question on SPREAD of VCV posed by myself. > No reply yet. > > We are still on. > > W. Lee > > KHL2288 @ utarlg.uta.edu on 10/30/98 10:48:03 AM > > Please respond to RATS-L@efs.mq.edu.au > > To: RATS-L @ efs.mq.edu.au > cc: (bcc: Wai Lee) > Subject: subscribing discussion list > > Am I still subscribing the discussion list? I havn't received any mails > lately. > If my connection is closed, how can i reopen it? > > Thank you so much. > > Kyong H. Lee > Dept. of Economics > University of Texas at Arlington ---------- End of message ----------