From: nmorin@weber.ucsd.edu (Norman Morin) To: "RATS Discussion List" Subject: Re: Problem with Nonlinear Equation System Date: Sun, 31 Mar 1996 11:19:34 -0800 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 v1.21 At 2:14 AM 3/31/96, Sukudhew Singh wrote: >* free parameters >NONLIN A0 A1 A2 A3 A4 B0 B1 B2 B3 L > >* equations >FRML DD Q = A0 + A1*P + A2*Y + A3*P*Z + A4*Z >FRML SS P = L*((-1*Q)/(A1 + A3*Z))+B0 + B1*Q + B2*W1 + B3*W2 > >* initial values >COMPUTE A0=A1=A2=A3=A4=0.0 >COMPUTE B0=B1=B2=B3=L=0.0 I think by initializing all your free parameters to zero might be the cause of your program crashing, since L*((-1*Q)/(A1 + A3*Z)) will then be the ratio of 0 and 0. That's from a quick look, though; I haven't tried executing your code. --Norm Norman J. Morin nmorin@weber.ucsd.edu ------------------------------------------------------------- Department of Economics * University of California, San Diego 9500 Gilman Drive * La Jolla, CA 92023-0508 ------------------------------------------------------------- ---------- End of message ---------- From: nmorin@weber.ucsd.edu (Norman Morin) To: "RATS Discussion List" Subject: Re: FEVD IN SVARs Date: Sun, 31 Mar 1996 16:11:00 -0800 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 v1.21 At 9:39 AM 3/28/96, F.Y.Kumah wrote: >Hello and G'day, > >Can someone out there give me a handy program for >estimating the forecast error variance decompositions >(and thier standard errors) for Structural VARs? > I have written a procedure that does this. It is quite large, though, as it is a "menu-driven" procedure which does about a dozen different VAR related things. Some of the code in it (e.g., that Bernanke.src procedure) is from other users' procedures, so it isn't 100% my own work, thus I am unsure about distribution. Anyway, if it's ok, I'd be happy to send it to you (it's about 88K). Does anyone have an answer for what is the proper etiquette about this sort of thing? I've written dozens of procedures to do various time series work, but since many contain bits and pieces of code from procedures from RATS or from other RATS users, I've been hesitant to share most of them, fearing some breach of "fair use". I'd be happy to upload them, if I knew it were not improper to do so. For example, inside the old public domain CATS for RATS 3.0(?) program, there's a nice little procedure to quickly build a larger matrix by adding rows or columns to a given matrix. I've used that procedure in a few of my own procedures which needed that sort of operation. But now (and especially since CATS is now commercialware), I am unsure about giving anyone a procedure I wrote containing that chunk of code. I'm happy to share anything I've written, but I don't feel like rewriting some code in a bunch of different procedures to do so. Thanks, --Norm Norman J. Morin nmorin@weber.ucsd.edu ------------------------------------------------------------- Department of Economics * University of California, San Diego 9500 Gilman Drive * La Jolla, CA 92023-0508 ------------------------------------------------------------- ---------- End of message ---------- From: "F.Y.Kumah" To: "RATS Discussion List" Subject: Re: Problem with Nonlinear Equation System Date: Mon, 1 Apr 1996 11:01:59 MET Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Tilburg University X-mailer: Pegasus Mail for Windows (v2.23) (via Mercury MTS v1.21) Dear Norman, I think you are damn correct. In fact I estimated some coefficients of an SVAR using MAXIMIZE and the error reported by Sukudhew keep coming up precisely because the initial values were zeros. So over to you Suk. just try other starting values and hopefully your problem will be over. > At 2:14 AM 3/31/96, Sukudhew Singh wrote: > > >* free parameters > >NONLIN A0 A1 A2 A3 A4 B0 B1 B2 B3 L > > > >* equations > >FRML DD Q = A0 + A1*P + A2*Y + A3*P*Z + A4*Z > >FRML SS P = L*((-1*Q)/(A1 + A3*Z))+B0 + B1*Q + B2*W1 + B3*W2 > > > >* initial values > >COMPUTE A0=A1=A2=A3=A4=0.0 > >COMPUTE B0=B1=B2=B3=L=0.0 > > > I think by initializing all your free parameters to zero might be the > cause of your program crashing, since > L*((-1*Q)/(A1 + A3*Z)) > will then be the ratio of 0 and 0. > > That's from a quick look, though; I haven't tried executing your code. > > --Norm > > Norman J. Morin > nmorin@weber.ucsd.edu > ------------------------------------------------------------- > Department of Economics * University of California, San Diego > 9500 Gilman Drive * La Jolla, CA 92023-0508 > ------------------------------------------------------------- > > > > ---------- End of message ---------- From: Sukudhew Singh To: "RATS Discussion List" Subject: SOLVED: Problem with Nonlinear Equation System Date: Mon, 01 Apr 1996 04:06:12 -0600 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-version: 1.0 X-Mailer: Windows Eudora Light Version 1.5.4b13 (32) (via Mercury MTS v1.21) Content-type: text/plain; charset="us-ascii" Content-transfer-encoding: 7BIT At 11:19 AM 3/31/96 -0800, Norman J. Morin wrote: > >I think by initializing all your free parameters to zero might be the >cause of your program crashing, since > L*((-1*Q)/(A1 + A3*Z)) >will then be the ratio of 0 and 0. > >That's from a quick look, though; I haven't tried executing your code. > >--Norm > You are exactly right Norman. Specifying non-zero initial values made the difference and the program works just fine now. My ordeal has ended. Many thanks for your insightful suggestion. Warmest regards, Sukhdave ---------- End of message ---------- From: Ranjini Sivakumar To: "RATS Discussion List" Subject: Re: FEVD IN SVARs Date: Mon, 1 Apr 1996 19:12:23 -0700 (MST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: ELM [version 2.4 PL23alpha2] (via Mercury MTS v1.21) MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Hello, Is there a way to incorporate inequality constraints with NLIN? Thanks, Ranjini ---------- End of message ---------- From: Tim Ng To: "RATS Discussion List" Subject: Vacation Mail for Tim Ng -Reply Date: Wed, 03 Apr 1996 09:14:51 +1200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Novell GroupWise 4.1 (via Mercury MTS v1.21) To all those users whose mailboxes I clogged up over the past week: please accept my sincere apology. Tim Ng ---------- End of message ---------- From: "Maureen Lage" To: "RATS Discussion List" Subject: Structural VARs Date: Wed, 3 Apr 1996 14:22:25 EST Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Miami University School of Business X-mailer: Pegasus Mail/Windows (v1.20) (via Mercury MTS v1.21) Does anyone know where I could get a copy of Professor Giannini's monograph on "Structural VAR Models"? Thanks in advance. Maureen Lage Miami University ---------- End of message ---------- From: "Ulrich Leuchtmann" To: "RATS Discussion List" Subject: Re: Structural VARs Date: Thu, 4 Apr 1996 17:58:34 GMT+0100 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Fak. WIWI Uni Bielefeld X-Mailer: Pegasus Mail for Windows (v2.01) (via Mercury MTS v1.21) Hi Maureen, > Does anyone know where I could get a copy of Professor Giannini's > monograph on "Structural VAR Models"? The book is: Carlo Giannini, "Topics in Structural VAR Econometrics." Springer Verlag: New York, Berlin, Heidelberg. 1992. ISBN 0-387-55262-6. Ulrich ============================================================= Ulrich Leuchtmann University of Bielefeld German Statistical Society Faculty of Economics Secretariat P.O.Box 10 01 31 Tel.: (+49)(0) 521-106-4870 33501 Bielefeld Fax: (+49)(0) 521-106-2994 Germany email: ULeucht@wiwi.uni-bielefeld.de ============================================================= ---------- End of message ---------- From: "Maureen Lage" To: "RATS Discussion List" Subject: Structural VARS Date: Mon, 8 Apr 1996 14:21:37 EST Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Miami University School of Business X-mailer: Pegasus Mail/Windows (v1.20) (via Mercury MTS v1.21) Thanks to those of you who replied to my request for information about Giannini's book on Structural VARS. While I wait for my copy, perhaps I can ask you one more question about them. It seems when working through his program that everthing hinges on the initial values for GAMMA given in the START vector. Does anyone know of any general "rule of thumb" to use here? Thanks again. Maureen Lage Miami University ---------- End of message ---------- From: nmorin@weber.ucsd.edu (Norman Morin) To: "RATS Discussion List" Subject: Re: Structural VARS Date: Mon, 8 Apr 1996 14:50:21 -0700 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 v1.21 At 2:21 PM 4/8/96, Maureen Lage wrote: >It seems when working through his program that everthing hinges on >the initial values for GAMMA given in the START vector. Does anyone >know of any general "rule of thumb" to use here? I don't know how kosher it is, here's what I do, and it generally seems to work (and I've tried it with simulated series, in which I know the "right" answer): Say it's a four variable VAR, x(1), x(2), x(3), x(4) with reduced form residuals r(1), r(2), r(3), r(4). And I think the 'pattern matrix' is 1 0 0 0 1 1 0 0 1 0 1 0 1 1 0 1 Then for my starup values, I'll just regress r(2) on r(1) to get an initial guess for the first nondiagonal entry, r(3) on r(1) for the second, and r(4) on r(1) and r(2) for the third and forth. Now, again, I haven't seen this done in papers and am not sure how many Time Series Commandments I am violating, but it seems to allow Bernanke-Sims or the SVAR program to converge quickly and produce values that make sense (with real data) and produce values quite close to the actual values when using a simulated multivariate VAR process. Your mileage my vary. --Norm Norman J. Morin nmorin@weber.ucsd.edu ------------------------------------------------------------- Department of Economics * University of California, San Diego 9500 Gilman Drive * La Jolla, CA 92023-0508 ------------------------------------------------------------- ---------- End of message ---------- From: Rob Trevor To: "RATS Discussion List" Subject: Absence Date: Thu, 11 Apr 1996 18:20:22 +1000 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 v1.21 Dear RATS-L Member I will be away until May 13. Since my email access during this period will be extremely limited, please address any *list* related problems to David Rawling - BUT please don't ask David any RATS questions! Cheers Rob Trevor RATS-L-OWNER@efs.mq.edu.au ---------- End of message ---------- From: "Muhammad Alam" To: "RATS Discussion List" Subject: VARIANCE RATIO TEST Date: Thu, 18 Apr 1996 14:19:27 CST Errors-To: Reply-To: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-Listname: Organization: UNO, College of Business X-Mailer: Pegasus Mail for Windows (v2.23) (via Mercury MTS v1.21) Is there a RATS procedure for the VARIANCE RATIO TEST? I would really appreciate if someone could give me some info about this. Thanks, M. Alam [e-mail: malam@cbafaculty.unomaha.edu] ---------- End of message ---------- From: CROWLEY PATRICK # ECONOMICS To: "RATS Discussion List" Subject: Re: Multivariate Linear Rank Tests Date: Thu, 18 Apr 1996 22:05:50 +0000 (ADT4AST) Errors-To: Reply-To: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-Listname: Organization: Saint Mary's University X-Mailer: Pegasus Mail v3.22 (via Mercury MTS v1.21) Content-Transfer-Encoding: 7BIT Dear Everyone on the RATS discussion group, Does anyone know of the existance of a program to do Multivariate Linear Rank Tests (not Rank of a matrix, but Rank order statistics)? If not, I will develop one, but I was just interested if anyone out there has already written one. Regards, Patrick Crowley. =-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Patrick Crowley, email: pcrowley@SHARK.stmarys.ca Department of Economics, Tel: (902) 420-5675 Saint Mary's University, Fax: (902) 492-8758 Halifax, Home: (902) 4928673 Nova Scotia, B3H 3C3, CANADA. -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= ---------- End of message ---------- From: "HERBERT STOCKER" To: "RATS Discussion List" Subject: Principal Components Analysis Date: Mon, 22 Apr 1996 12:41:58 +0100 Errors-To: Reply-To: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-Listname: Organization: University of Innsbruck, Austria Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7BIT X-Mailer: Pegasus Mail v3.22 (via Mercury MTS v1.21) Estima offers a simple procedure for extracting principal components from a set of series on it's Web-site. However, this procedure seems to have some bugs. Is there someone who has a better procedure for this purpose? Thanx in advance! Herbert Stocker ---------- End of message ---------- From: branchr@bc.edu To: "RATS Discussion List" Subject: Impulse Responce Functions Date: Tue, 23 Apr 1996 17:49:13 +0000 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 v1.21 Hi allÉ I was wondering if it is possible to save the output from the IMPULSE command to an external file, so I can graph it in another package. I am currently using Rats 4.20 on a VAX/VMS system. Any help would be greatly appreciated. Richard Branch Boston College P.S. Actually this problem arose because I can't seem to figure out how to graph impulse responce functions in RATS. I tried what the manual suggests, but alasÉit doesn't seem to work to well! ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: Impulse Responce Functions Date: Wed, 24 Apr 1996 13:01:47 +0000 Errors-To: Reply-To: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-Listname: Organization: Estima X-Mailer: Pegasus Mail for Windows (v2.23) (via Mercury MTS v1.21) Richard: Regarding your question: > > I was wondering if it is possible to save the output from the IMPULSE >command to > an external file, so I can graph it in another package. I am currently using > Rats 4.20 on a VAX/VMS system. Any help would be greatly appreciated. > The IMPULSE command stores the responses in ordinary series, which you can certainly graph or write to a file. The most common way to write series to a file is to use the OPEN COPY and COPY instructions. > > P.S. Actually this problem arose because I can't seem to figure out how >to graph > impulse responce functions in RATS. I tried what the manual suggests, but > alasÉit doesn't seem to work to well! > Which section of the manual? The IMPULSES.PRG and MONTEVAR.PRG programs should work fine. These examples are a bit complicated, because they are designed to prodcue fairly fancy graphs of responses of all variables to shocks to all variables. If you're having problems, start with an easier task. Suppose you define and estimate a 2 eqn VAR. You could save and graph the responses to a shock to eqn 1 as follows: SYSTEM 1 2 VARIABLES X Y LAGS 1 TO 4 DET CONSTANT END(SYS) ESTIMATE(OUTSIGMA=V) IMPULSE 2 16 1 V # 1 XRESP # 2 YRESP GRAPH(KEY=UPRIGHT) 2 # XRESP # YRESP Hope this helps. Tom Maycock Estima -- +-----------------------------+-----------------------------------------+ | Estima | | | P.O. Box 1818 | Voice: (847) 864-8772 | | Evanston, IL 60204-1818 | Fax: (847) 864-6221 | | U.S.A | BBS: (847) 864-8816 | | e-mail: estima@estima.com | CompuServe: 73140,2202 | |-----------------------------------------------------------------------| | Web Site: http://homepage.interaccess.com/~estima/ | | RATS Internet Mailing List: New members can join by sending e-mail to | | MAISER@EFS.MQ.EDU.AU with the message: SUBSCRIBE RATS-L | +-----------------------------------------------------------------------+