Return-Path: Received: from jules.bc.edu (jules.bc.edu [136.167.2.173]) by monet.bc.edu (8.8.7/8.8.7) with ESMTP id PAA114382 for ; Sat, 1 Nov 1997 15:40:42 -0500 From: maiser@efs.mq.edu.au Received: (from root@localhost) by jules.bc.edu (8.8.7/8.8.7) with X.500 id PAA86378 for BAUM@MAIL1.BC.EDU; Sat, 1 Nov 1997 15:40:41 -0500 Received: from sunb.ocs.mq.edu.au (sunb.ocs.mq.edu.au [137.111.1.11]) by jules.bc.edu (8.8.7/8.8.7) with ESMTP id PAA50509 for ; Sat, 1 Nov 1997 15:40:34 -0500 Received: from efs1.efs.mq.edu.au (efs1.efs.mq.edu.au [137.111.64.8]) by sunb.ocs.mq.edu.au (8.8.5/8.8.5) with ESMTP id HAA27348 for ; Sun, 2 Nov 1997 07:40:30 +1100 (EST) Received: from EFS1/SpoolDir by efs1.efs.mq.edu.au (Mercury 1.21); 2 Nov 97 07:39:36 GMT+1000 Received: from SpoolDir by EFS1 (Mercury 1.30); 2 Nov 97 07:39:35 GMT+1000 To: baum@bc.edu Date: Sun, 2 Nov 97 7:39:35 GMT+1000 Subject: Re: Message-ID: <2CC07D45F60@efs1.efs.mq.edu.au> From: "N.A.Molinari" To: "RATS Discussion List" Subject: MARKOV SWITCHING Date: Mon, 06 Oct 1997 22:37:14 -0400 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) v1.30) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Does anyone know of code for multiple(i.e.N) state markov switching model? Any assistance would be appreciated. Thanks in advance, N.A.Molinari ---------- End of message ---------- From: M P Tomsett To: "RATS Discussion List" Subject: GMM Date: Tue, 7 Oct 1997 14:33:18 +0100 (BST) 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) v1.30 Has anyone had any luck estimating a large system of equations using GMM in Rats? I have an over-identified system of moment conditions which is not converging. I know this is very much dependent on the starting values but assuming I am in the right area does anyone have any other tips. Have people found the chi-square test very sensitive? Yours in hope, Mark. **************************************************************************** Mark Tomsett Department of Economics University of Durham 23-26 Old Elvet DURHAM CITY County Durham **************************************************************************** ---------- End of message ---------- From: Wai Lee To: "RATS Discussion List" Subject: Re: GMM Date: 7 Oct 97 10:52:16 EDT Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-Type: Text/Plain X-Mailer: Mercury MTS (Bindery) v1.30 I had no problem in estimating a seven equations system, with 5 common instrumental variables for all equations. But when each equation has different instruments, then the results are not very satisfactory (the way RATS handles this problem makes your weighting matrix even larger in this case). I used other matrix language software for that job. If the dimension of your weighting matrix is too large, numerical GMM may not be reliable anymore. Consider Zhou's (1996 or 1997, Review of Financial Studies) analytical GMM instead. W. Lee M.P.Tomsett @ durham.ac.uk (M P Tomsett) 10/07/97 02:33 PM To: RATS-L @ efs.mq.edu.au ("RATS Discussion List") @ SMTP cc: Subject: GMM Has anyone had any luck estimating a large system of equations using GMM in Rats? I have an over-identified system of moment conditions which is not converging. I know this is very much dependent on the starting values but assuming I am in the right area does anyone have any other tips. Have people found the chi-square test very sensitive? Yours in hope, Mark. **************************************************************************** Mark Tomsett Department of Economics University of Durham 23-26 Old Elvet DURHAM CITY County Durham **************************************************************************** ---------- End of message ---------- From: M P Tomsett To: "RATS Discussion List" Subject: Re: GMM Date: Tue, 7 Oct 1997 16:44:39 +0100 (BST) 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) v1.30 The system I have is twelve moment equations with nine free parameters. I am only using two observable variables whose characteristics make instrumental variable estimation unsuitable. Mark. **************************************************************************** Mark Tomsett Department of Economics University of Durham 23-26 Old Elvet DURHAM CITY County Durham **************************************************************************** ---------- End of message ---------- From: "Gregory, Richard" To: "RATS Discussion List" Subject: Date: Fri, 10 Oct 1997 14:08:46 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.995.52 (via Mercury MTS (Bindery) v1.30) MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit I am attepting to estimate a multivariate GARCH model similar to the one used by De Santis and Gerard in "International Asset Pricing and Portfolio Diversification with Time-Varying Risk" forthcoming in the Journal of Finance. The model coefficients are not converging despite several different choices of inital parameters Any suggestions would be appreciated. CALENDAR 1992 12 12 ALLOCATE 1996:10 OPEN DATA PAT2.RAT DATA(FORMAT=RATS) / LOCI CUSRATE MSCI DJIA SP NIKKEI HK SET RLOCI = LOG(LOCI) - LOG(LOCI{1}) SET RCUSRATE = LOG(CUSRATE) - LOG(CUSRATE{1}) SET RMSCI = LOG(MSCI) - LOG(MSCI{1}) SET RDJIA = LOG(DJIA) - LOG(DJIA{1}) SET RSP = LOG(SP) - LOG(SP{1}) SET RNIK = LOG(NIKKEI) - LOG(NIKKEI{1}) SET RHK = LOG(HK) - LOG(HK{1}) COMPUTE GSTART=1993:10, GEND=1996:10 LINREG(NOPRINT) RLOCI / R1 # CONSTANT RLOCI{1} LINREG(NOPRINT) RCUSRATE / R2 # CONSTANT LINREG(NOPRINT) RMSCI / R3 # constant RMSCI{1} LINREG(NOPRINT) RDJIA / R4 # CONSTANT LINREG(NOPRINT) RSP / R5 # CONSTANT LINREG(NOPRINT) RNIK / R6 # CONSTANT LINREG(NOPRINT) RHK / R7 # CONSTANT VCV(matrix=rr,noprint) # R1 R3 SET U1 = R1 SET U2 = R3 SET H11 = RR(1,1) SET H22 = RR(2,2) SET H12 = RR(1,2) DECLARE SYMETRIC H DECLARE VECTOR U NONLIN C11 C12 C22 A11 A12 A21 A22 B11 B12 B21 B22 D1 D2 NLPAR(CRITERION=COEFFS) FRML H11F = (VC=||C11,C12|0.0,C22||),$ (VA=||A11,A12|A21,A22||),$ (VB=||B11,B12|B21,B22||),$ (H=||H11{1}|H12{1},H22{1}||),$ (UB=||U1{1},U2{1}||*VB),$ (H=TR(VC)*VC+%MQFORM(H,VA)+TR(UB)*UB),$ H(1,1) FRML H12F = RR(1,2) FRML H22F = RR(2,2) FRML REGRESID1 = RLOCI-D1*H12{1} FRML REGRESID2 = RMSCI-D2*H22{1} COMPUTE CINIT = %DECOMP(RR) COMPUTE VC11 = CINIT(1,1),VC12 = CINIT(2,1),VC22 = CINIT(2,2) FRML ARCHLOGL = (H11(T)=H11F(T)),(H12(T)=H12F(T)),(H22(T)=H22F(T)),$ (U1(T)=REGRESID1(T)),(U2(T)=REGRESID2(T)),$ 0.5*((LOG(%DET(H)))+%QFORM(INV(H),UB)) STAT(NOPRINT) RLOCI COMPUTE MOY = %MEAN COMPUTE C11 = RR(1,1), A11 = 0.01, B11 = 0.1 STAT(NOPRINT) RMSCI COMPUTE D1 = 0.1 COMPUTE D2 = 0.1 cOMPUTE C22 = RR(2,2), A22= 0.01, B22 = 0.1 cOMPUTE C12 = 0.5 COMPUTE A12 = 0.01, A21 = 0.01, B12 = 0.1, B21 = 0.1 MAXIMIZE(METHOD=SIMPLEX,RECURSIVE,ITERATION=5) ARCHLOGL GSTART GEND MAXIMIZE(METHOD=bhhh,RECURSIVE,ITERATION=400) ARCHLOGL GSTART GEND NLPAR(SUBITERATIONS=400,CVCRIT=.01) ---------- End of message ---------- From: Aghdas Mirzaie To: "RATS Discussion List" Subject: Kpss Date: Fri, 10 Oct 1997 13:27:16 -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) v1.30 Hello, I am new in using rats. I would like to do the KPSS test with rats. I have downloaded the KPSS.src file from the web. Any idea how I can write the program to run this test. I appreciate your help. A. Mirzaie Aghdas Mirzaie - Economics Department University of Wisconsin-Milwaukee, Office: Bolton 813, Tel: 229-4375, Monday and Wednesday Marquette University, Office: 405, Tel: 288-7411, Tuesday and Thursday Email:mirzaie@csd.uwm.edu Homepage:http://www.uwm.edu/~mirzaie ---------- End of message ---------- From: Shuh-chyi Doong To: "RATS Discussion List" Subject: Can I do hypothesis testing inGARCH model? Date: Sun, 12 Oct 1997 14:49:51 +0800 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mozilla 3.0C (Win95; I) (via Mercury MTS (Bindery) v1.30) Mime-Version: 1.0 Content-Type: text/plain; charset=big5 Content-Transfer-Encoding: 7bit Dear sir: My name is Shuh-chyi Doong, associate prof. in Finance, Feng-chia U., Taiwan. Now,I am writing a paper which have to do one hypothesis testing in GARCH-Mmodel. I don't know how to do it. Could you help me to solve this problem? The command about the model is as follows. smpl 1987:01 1996:12 set u2 = 0.0 set v2 = 0.0 NONLIN B0 b1 b2 b3 b4 A0 A1 A2 FRML GARCHVAR = A0+A1*U2{1}**2+A2*V2{1} FRML REGRESID = rp11-B0-b1*stdinf-b2*stdyp-b3*stddex-b4*sqrt(garchvar(t)) FRML GARCHLOGL = (V2(T)=GARCHVAR(T)),(U2(T)=REGRESID(T)), $ -.5*(LOG(V2)+U2**2/V2) LINREG(noprint) rp11 # CONSTANT stdinf stdyp stddex COMPUTE B0=%BETA(1), b4=.05,a0=%SEESQ,A1=.05,a2=.05, $ b1=%BETA(2),b2=%BETA(3),b3=%BETA(4) MAXIMIZE(METHOD=BHHH,RECURSIVE,iterations=100) GARCHLOGL 3 * I want to test the hypothesis b1=b2=b3=0. How can I do? If you know the answer, please fax or email to me. My fax # is 886-4-4513796 and my email is scdoong@fcu.edu.tw. Thanks for your help. I am looking forward to hearing good news from you. Best regards, Shuh-chyi Doong, Ph.D. Associate Professor Feng-chia University Taichung, Taiwan Fax: 886-4-4513796 886-2-5581929 Email: scdoong@fcu.edu.tw ---------- End of message ---------- From: Hannu Kahra To: "RATS Discussion List" Subject: Re: Can I do hypothesis testing inGARCH model? Date: Sun, 12 Oct 1997 19:13:49 +0300 (EET 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) v1.30 On Sun, 12 Oct 1997, Shuh-chyi Doong wrote: > Dear sir: > > My name is Shuh-chyi Doong, associate prof. in Finance, Feng-chia U., > Taiwan. Now,I am writing a paper which have to do one hypothesis testing > in GARCH-Mmodel. I don't know how to do it. Could you help me to solve > this problem? > The command about the model is as follows. > > smpl 1987:01 1996:12 > set u2 = 0.0 > set v2 = 0.0 > NONLIN B0 b1 b2 b3 b4 A0 A1 A2 > FRML GARCHVAR = A0+A1*U2{1}**2+A2*V2{1} > FRML REGRESID = > rp11-B0-b1*stdinf-b2*stdyp-b3*stddex-b4*sqrt(garchvar(t)) > FRML GARCHLOGL = (V2(T)=GARCHVAR(T)),(U2(T)=REGRESID(T)), $ > -.5*(LOG(V2)+U2**2/V2) > LINREG(noprint) rp11 > # CONSTANT stdinf stdyp stddex > COMPUTE B0=%BETA(1), b4=.05,a0=%SEESQ,A1=.05,a2=.05, $ > b1=%BETA(2),b2=%BETA(3),b3=%BETA(4) > MAXIMIZE(METHOD=BHHH,RECURSIVE,iterations=100) GARCHLOGL 3 * > > I want to test the hypothesis b1=b2=b3=0. How can I do? If you know the > answer, please fax or email to me. My fax # is 886-4-4513796 and my > email is scdoong@fcu.edu.tw. Thanks for your help. I am looking forward > to hearing good news from you. You can use the TEST and RESTRICT instructions with the BFGS and BHHH options. Either TEST # 2 3 4 # 0 0 0 or RESTRICT 3 # 2 # 1 0 # 3 # 1 0 # 4 # 1 0 should do the job. See pages 14-252..253 in the Rats manual. Hannu Kahra +------------------------------------------------+ | University of Tampere, Department of Economics | | P.O. Box 607, FIN-33101 Tampere, Finland | | tel: +358-3-215 6821, fax: +358-3-215 7254 | | e-mail: kthaka@uta.fi, hannu.kahra@uta.fi | | http://www.uta.fi/~kthaka | +------------------------------------------------+ ---------- End of message ---------- From: Rob Trevor To: "RATS Discussion List" Subject: Re: GARCH Convergence Date: Mon, 13 Oct 1997 14:35:23 +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 (Bindery) v1.30 At 4:08 AM +1000 11/10/97, Gregory, Richard wrote: > I am attepting to estimate a multivariate GARCH model similar to the one > used by De Santis and Gerard in "International Asset Pricing and > Portfolio Diversification with Time-Varying Risk" forthcoming in the > Journal of Finance. > > The model coefficients are not converging despite several different > choices of inital parameters > > Any suggestions would be appreciated. > ... I haven't had time to have a good look at your code, but I'll try a few comments: * you don't have a lot of observations (360) for a bivariate BEKK estimation. It can be done, but 3 or 4 times the number of obs would be a lot more comfortable * when you get convergence problems on well tested code, think initial values, lets play around with SIMPLEX for a while and see what's going on. (Your 5 SIMPLEX may not be enough - do you shift from your initial values much, what's happening iteration to iteration. Use the TRACE option.) * however, before looking at the trees its always worth considering the forest - am I estimating something sensible? As I said I haven't time to have a good look at your code, but it does seem a little unusual. You seem to have something similar to BEKK, but then you seem to overwrite the conditional covariance entries with unconditional sample information from the VCV instruction (your RR matrix). * the best way to track specification problems down is to work out how *RATS* will evaluate your code. Start with the FRML which you have told it to maximise (ARCHLOGL in your case) and work out how RATS will calculate ARCHLOGL(1)... ARCHLOGL(T). That is, to calculate ARCHLOGL(1), it will first calculate H11(1)=H11F(1), but it will then need to calculate H11F(1)...etc. (See your FRML for details.) Remember, RATS won't calculate all the FRML's you give it - it will just calculate the ones it needs to compute ARCHLOGL(1)... ARCHLOGL(T). Because we tend to read code from the top of the page down, we can think that a model specification makes sense when in fact from RATS's point of view it doesn't. (I find it easy to fall foul of this one.) Hope this helps Rob > > > CALENDAR 1992 12 12 > ALLOCATE 1996:10 > OPEN DATA PAT2.RAT > DATA(FORMAT=RATS) / LOCI CUSRATE MSCI DJIA SP NIKKEI HK > SET RLOCI = LOG(LOCI) - LOG(LOCI{1}) > SET RCUSRATE = LOG(CUSRATE) - LOG(CUSRATE{1}) > SET RMSCI = LOG(MSCI) - LOG(MSCI{1}) > SET RDJIA = LOG(DJIA) - LOG(DJIA{1}) > SET RSP = LOG(SP) - LOG(SP{1}) > SET RNIK = LOG(NIKKEI) - LOG(NIKKEI{1}) > SET RHK = LOG(HK) - LOG(HK{1}) > > COMPUTE GSTART=1993:10, GEND=1996:10 > > > > LINREG(NOPRINT) RLOCI / R1 > # CONSTANT RLOCI{1} > LINREG(NOPRINT) RCUSRATE / R2 > # CONSTANT > LINREG(NOPRINT) RMSCI / R3 > # constant RMSCI{1} > LINREG(NOPRINT) RDJIA / R4 > # CONSTANT > LINREG(NOPRINT) RSP / R5 > # CONSTANT > LINREG(NOPRINT) RNIK / R6 > # CONSTANT > LINREG(NOPRINT) RHK / R7 > # CONSTANT > > > > > VCV(matrix=rr,noprint) > # R1 R3 > > SET U1 = R1 > SET U2 = R3 > SET H11 = RR(1,1) > SET H22 = RR(2,2) > SET H12 = RR(1,2) > > DECLARE SYMETRIC H > DECLARE VECTOR U > > NONLIN C11 C12 C22 A11 A12 A21 A22 B11 B12 B21 B22 D1 D2 > NLPAR(CRITERION=COEFFS) > > FRML H11F = (VC=||C11,C12|0.0,C22||),$ > (VA=||A11,A12|A21,A22||),$ > (VB=||B11,B12|B21,B22||),$ > (H=||H11{1}|H12{1},H22{1}||),$ > (UB=||U1{1},U2{1}||*VB),$ > (H=TR(VC)*VC+%MQFORM(H,VA)+TR(UB)*UB),$ > H(1,1) > > FRML H12F = RR(1,2) > FRML H22F = RR(2,2) > FRML REGRESID1 = RLOCI-D1*H12{1} > FRML REGRESID2 = RMSCI-D2*H22{1} > > COMPUTE CINIT = %DECOMP(RR) > COMPUTE VC11 = CINIT(1,1),VC12 = CINIT(2,1),VC22 = CINIT(2,2) > > FRML ARCHLOGL = (H11(T)=H11F(T)),(H12(T)=H12F(T)),(H22(T)=H22F(T)),$ > (U1(T)=REGRESID1(T)),(U2(T)=REGRESID2(T)),$ > 0.5*((LOG(%DET(H)))+%QFORM(INV(H),UB)) > > STAT(NOPRINT) RLOCI > COMPUTE MOY = %MEAN > COMPUTE C11 = RR(1,1), A11 = 0.01, B11 = 0.1 > > STAT(NOPRINT) RMSCI > COMPUTE D1 = 0.1 > COMPUTE D2 = 0.1 > cOMPUTE C22 = RR(2,2), A22= 0.01, B22 = 0.1 > cOMPUTE C12 = 0.5 > COMPUTE A12 = 0.01, A21 = 0.01, B12 = 0.1, B21 = 0.1 > > MAXIMIZE(METHOD=SIMPLEX,RECURSIVE,ITERATION=5) ARCHLOGL GSTART GEND > MAXIMIZE(METHOD=bhhh,RECURSIVE,ITERATION=400) ARCHLOGL GSTART GEND > NLPAR(SUBITERATIONS=400,CVCRIT=.01) ---------- End of message ---------- From: "Wai Lee" To: "RATS Discussion List" Subject: Re: Can I do hypothesis testing inGARCH model? Date: Mon, 13 Oct 1997 08:24:30 -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) v1.30 Are there any detailed technical information on how TEST, RESTRICT, and MRESTRICT work in MLE (p. 14-155 of RATS manual)? Is it a likelihood ratio, Lagrange Multiplier test, or something else? If it is an LM test, I would love to know how we can extract the information matrix from the test (possible or not?). I sort of assume that TEST, etc. only work for least square estimations. The way I will test b1=b2=b3=0 is, reestimate the model with the restrictions imposed, then do a likelihood ratio test: test statistic = 2 x (difference in likelihood values of unrestricted and restricted models) as a chi-square with deg of freedom = 3 in this case. If those TEST etc indeed work with MLE, it's going to save me a lot of work. W. Lee kthaka @ uta.fi (Hannu Kahra) on 10/12/97 07:13:49 PM To: RATS-L @ efs.mq.edu.au ("RATS Discussion List") @ SMTP cc: Subject: Re: Can I do hypothesis testing inGARCH model? On Sun, 12 Oct 1997, Shuh-chyi Doong wrote: > Dear sir: > > My name is Shuh-chyi Doong, associate prof. in Finance, Feng-chia U., > Taiwan. Now,I am writing a paper which have to do one hypothesis testing > in GARCH-Mmodel. I don't know how to do it. Could you help me to solve > this problem? > The command about the model is as follows. > > smpl 1987:01 1996:12 > set u2 = 0.0 > set v2 = 0.0 > NONLIN B0 b1 b2 b3 b4 A0 A1 A2 > FRML GARCHVAR = A0+A1*U2{1}**2+A2*V2{1} > FRML REGRESID = > rp11-B0-b1*stdinf-b2*stdyp-b3*stddex-b4*sqrt(garchvar(t)) > FRML GARCHLOGL = (V2(T)=GARCHVAR(T)),(U2(T)=REGRESID(T)), $ > -.5*(LOG(V2)+U2**2/V2) > LINREG(noprint) rp11 > # CONSTANT stdinf stdyp stddex > COMPUTE B0=%BETA(1), b4=.05,a0=%SEESQ,A1=.05,a2=.05, $ > b1=%BETA(2),b2=%BETA(3),b3=%BETA(4) > MAXIMIZE(METHOD=BHHH,RECURSIVE,iterations=100) GARCHLOGL 3 * > > I want to test the hypothesis b1=b2=b3=0. How can I do? If you know the > answer, please fax or email to me. My fax # is 886-4-4513796 and my > email is scdoong@fcu.edu.tw. Thanks for your help. I am looking forward > to hearing good news from you. You can use the TEST and RESTRICT instructions with the BFGS and BHHH options. Either TEST # 2 3 4 # 0 0 0 or RESTRICT 3 # 2 # 1 0 # 3 # 1 0 # 4 # 1 0 should do the job. See pages 14-252..253 in the Rats manual. Hannu Kahra +------------------------------------------------+ | University of Tampere, Department of Economics | | P.O. Box 607, FIN-33101 Tampere, Finland | | tel: +358-3-215 6821, fax: +358-3-215 7254 | | e-mail: kthaka@uta.fi, hannu.kahra@uta.fi | | http://www.uta.fi/~kthaka | +------------------------------------------------+ ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: Kpss Date: Mon, 13 Oct 1997 10:06:12 -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) v1.30) > Hello, > I am new in using rats. I would like to do the KPSS test with rats. I have > downloaded the KPSS.src file from the web. Any idea how I can write the > program to run this test. I appreciate your help. > If you haven't already done so, please take a look at the comments at the top of the KPSS.SRC file. These comment lines providing a general description of the procedure, the specific syntax for using the procedure, and references. Let me know if you have further questions. Sincerely, Tom Maycock -- +-----------------------------+-----------------------------------------+ | Estima | | | P.O. Box 1818 | Voice: (847) 864-8772 | | Evanston, IL 60204-1818 | Fax: (847) 864-6221 | | U.S.A | BBS: (847) 864-8816 | | e-mail: estima@estima.com | CompuServe: 73140,2202 | |-----------------------------------------------------------------------| | Web Site: http://www.estima.com | | RATS Internet Mailing List: New members can join by sending e-mail to | | MAISER@EFS.MQ.EDU.AU with the message: SUBSCRIBE RATS-L | +-----------------------------------------------------------------------+ ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: Can I do hypothesis testing inGARCH model? Date: Mon, 13 Oct 1997 10:06:11 -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) v1.30) > From: "Wai Lee" > To: "RATS Discussion List" > Subject: Re: Can I do hypothesis testing inGARCH model? > Date: Mon, 13 Oct 1997 08:24:30 -0400 > Reply-to: "RATS Discussion List" > > Are there any detailed technical information on how TEST, RESTRICT, and > MRESTRICT work in MLE (p. 14-155 of RATS manual)? We devote an entire chapter of the RATS manual (Chapter 6) to Hypothesis Testing. I would suggest you start there and then let us know if you have further questions. Sincerely, Tom Maycock -- +-----------------------------+-----------------------------------------+ | Estima | | | P.O. Box 1818 | Voice: (847) 864-8772 | | Evanston, IL 60204-1818 | Fax: (847) 864-6221 | | U.S.A | BBS: (847) 864-8816 | | e-mail: estima@estima.com | CompuServe: 73140,2202 | |-----------------------------------------------------------------------| | Web Site: http://www.estima.com | | RATS Internet Mailing List: New members can join by sending e-mail to | | MAISER@EFS.MQ.EDU.AU with the message: SUBSCRIBE RATS-L | +-----------------------------------------------------------------------+ ---------- End of message ---------- From: "N.A.Molinari" To: "RATS Discussion List" Subject: markov switching Date: Mon, 13 Oct 1997 21:56:21 -0400 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) v1.30) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, I know there is a routine for switching models in Rats, but I'm wondering if there is code for markov switching. Any help would be appreciated. Thanks in advance, N.A.Molinari ---------- End of message ---------- From: Tom Verbeke To: "RATS Discussion List" Subject: ## MAT13. Store into Out-of-Range Matrix or Series Element Date: Wed, 15 Oct 1997 20:05:46 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mozilla 3.0Gold (Win95; I) (via Mercury MTS (Bindery) v1.30) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi, When I'm running the following code, declare series U declare series H nonlin B0 VA VB VC FRML residu = retack-B0*retbel20 FRML HF = VC + VA*H{1} + VB*U{1}**2 FRML LOGL = (H(T)=HF(T)),(U(T)=RESIDU(T)),-.5*(LOG(H(T))+U(T)*U(T)/H(T)) linreg(noprint) retack / U #constant retbel20 COMPUTE VC = %SEESQ,VA=.05,VB=.05 SET H = %SEESQ compute B0=%BETA(1) MAXIMIZE(METHOD=BHHH,RECURSIVE,ITERS=250) LOGL 5 * I get the following error message: ## MAT13. Store into Out-of-Range Matrix or Series Element The two return series "retack" and "retbel20" are calculated from a series of prices so the first entry equals NA. All the other entries are somewhere between -0.1 and +0.1. The code works fine untill the maximize line is reached (i.e. the output from the linreg command looks good, the estimated parameters and residual graph). Any suggestions? Thanks Tom Verbeke Research Assistant FWO - Flanders Research Group General Economics University of Ghent tverbeke@innet.be ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: ## MAT13. Store into Out-of-Range Matrix or Series Element Date: Wed, 15 Oct 1997 15:11:51 -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) v1.30) > > When I'm running the following code, > > declare series U > declare series H > nonlin B0 VA VB VC > FRML residu = retack-B0*retbel20 > FRML HF = VC + VA*H{1} + VB*U{1}**2 > FRML LOGL = > (H(T)=HF(T)),(U(T)=RESIDU(T)),-.5*(LOG(H(T))+U(T)*U(T)/H(T)) > linreg(noprint) retack / U > #constant retbel20 > COMPUTE VC = %SEESQ,VA=.05,VB=.05 > SET H = %SEESQ > compute B0=%BETA(1) > MAXIMIZE(METHOD=BHHH,RECURSIVE,ITERS=250) LOGL 5 * > > I get the following error message: > > ## MAT13. Store into Out-of-Range Matrix or Series Element Tom: Are you sure this is exactly the code you are running? You mentioned that the output from the LINREG looked good, but the LINREG above has a NOPRINT option? Details matter here, so make sure that everything is done in the proper order. In any case, I suspect the problem is that RATS is trying to set an elements of U or H which haven't yet been defined. Note that the DECLARE instruction creates these series, but, as with arrays, doesn't actually define a range from them--in other words the series still have "zero" entries after the DECLARE instruction. As a first step, try printinging the U and H series along with your original variables, and start the print at entry 1 so you can verify that the series are defined for entries 5 and later: PRINT 1 * RETACK RETBEL20 U H Sincerely, Tom Maycock -- +-----------------------------+-----------------------------------------+ | Estima | | | P.O. Box 1818 | Voice: (847) 864-8772 | | Evanston, IL 60204-1818 | Fax: (847) 864-6221 | | U.S.A | BBS: (847) 864-8816 | | e-mail: estima@estima.com | CompuServe: 73140,2202 | |-----------------------------------------------------------------------| | Web Site: http://www.estima.com | | RATS Internet Mailing List: New members can join by sending e-mail to | | MAISER@EFS.MQ.EDU.AU with the message: SUBSCRIBE RATS-L | +-----------------------------------------------------------------------+ ---------- End of message ---------- From: Stefan Nydahl To: "RATS Discussion List" Subject: element by element muliplikation in RATS? Date: Sun, 19 Oct 1997 13:15:09 +0100 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) v1.30) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Hi, Does anyone know how to do element by element matrix multiplikation (the Hadamar product) in RATS? I know I can do it with the EWISE command but this time I need to use it with FRML in a Multivariate GARCH set up. Best Regards Stefan Nydahl Department of Economics,Uppsala University, Sweden ======================================================================== Nationalekonomiska institutionen / Department of Economics Uppsala University Phone: 018/18 25 00 P.O.-Box 513 Intl: +46-18 18 25 00 S-751 20 Uppsala, Sweden Fax: +46-18 18 14 78 ======================================================================= ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RATS Discussion List Date: Mon, 20 Oct 1997 01:14:48 +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) v1.30) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Dear Sir, When I'm running the following commands of ARCH modeling (from chapter 3 of the RATS Handbook for Econometric Time Series, by Dr. Enders) : allocate 100 open data a:\archreg.prn data(format=prn,org=obs) nonlin b ao a1 frml e = y - b*x frml v = ao + a1*e(t-1)**2 frml L = -0.5*(log(v)+e(t-1)**2/v) compute b = 0.5, ao = 2, a1 = 0.5 max L 2 100 Then, "log100:SING error" appears on my screen and the programme stops. Also, when I'm running the commands for ARCH_M modeling (from the same chapter) : all 100 open data a:arch_m.prn data(format=prn,org=obs) set u = 0.0 non bo b1 ao a1 frml var = ao + a1*u{1}**2 frml e = y - bo - b1*var(t) frml L = (u = e), -0.5*(log(var)+e**2/var) linreg(noprint) y # constant compute bo = %beta(1), b1 = 0.2 compute ao = %seesq, a1 = 0.3 max L 2 * Then, "pow:OVERFLOW error" appears. I would be very much grateful if anyone can give me a hand! Thank you. Best regards, Kai-yin (Kevin)Woo Economics Dept., Hong Kong Shue Yan College ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: RATS Discussion List Date: Mon, 20 Oct 1997 11:39:43 -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) v1.30) > > When I'm running the following commands of ARCH modeling (from chapter 3 > of the RATS Handbook for Econometric Time Series, by Dr. Enders) : > > allocate 100 > open data a:\archreg.prn > data(format=prn,org=obs) > nonlin b ao a1 > frml e = y - b*x > frml v = ao + a1*e(t-1)**2 > frml L = -0.5*(log(v)+e(t-1)**2/v) > compute b = 0.5, ao = 2, a1 = 0.5 > max L 2 100 > > Then, "log100:SING error" appears on my screen and the programme stops. Dear Mr. Woo: We are aware of this problem with the 16-bit version of WinRATS. For some reason, this particular problem with this set of initial values leads to an untrapped floating point error in the MAXIMIZE routine. We are currently investigating the nature of the problem. Fortunately, you can work around the problem easily by first doing a MAXIMIZE command with the SIMPLEX method to get some better starting values. For example: frml e = y - b*x frml v = ao + a1*e(t-1)**2 frml L = -0.5*(log(v)+e(t-1)**2/v) compute b = 0.5, ao = 2, a1 = 0.5 maximzie(method=simplex,iters=5) L 2 100 max L 2 100 In addition to being a workaround for this particular problem, doing a few iterations with the SIMPLEX method is generally good practice--it helps get the estimation going in the right direction, making it easier for the BFGS or BHHH methods to converge. > > Also, when I'm running the commands for ARCH_M modeling (from the same > chapter) : > > all 100 > open data a:arch_m.prn > data(format=prn,org=obs) > > set u = 0.0 > non bo b1 ao a1 > frml var = ao + a1*u{1}**2 > frml e = y - bo - b1*var(t) > frml L = (u = e), -0.5*(log(var)+e**2/var) > linreg(noprint) y > # constant > compute bo = %beta(1), b1 = 0.2 > compute ao = %seesq, a1 = 0.3 > max L 2 * > > Then, "pow:OVERFLOW error" appears. > Again, use the "MAX(METHOD=SIMPLEX,ITERS=5) L 2 * " command before the main MAXIMIZE instruction to avoid this problem. Also, note that the second section of the ARCH_M.PRG example program has a typo. The second Maximum Likelihood estimation should start with entry 4 (or later), not with entry 3. If you have further question, feel free to address them to me directly at estima@estima.com Sincerely, Tom Maycock Estima -- +-----------------------------+-----------------------------------------+ | Estima | | | P.O. Box 1818 | Voice: (847) 864-8772 | | Evanston, IL 60204-1818 | Fax: (847) 864-6221 | | U.S.A | BBS: (847) 864-8816 | | e-mail: estima@estima.com | CompuServe: 73140,2202 | |-----------------------------------------------------------------------| | Web Site: http://www.estima.com | | RATS Internet Mailing List: New members can join by sending e-mail to | | MAISER@EFS.MQ.EDU.AU with the message: SUBSCRIBE RATS-L | +-----------------------------------------------------------------------+ ---------- End of message ---------- From: "Wai Lee" To: "RATS Discussion List" Subject: SIMULATE Date: Tue, 21 Oct 1997 10:36:53 -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) v1.30 Dear all: This is the first time I do simulation in RATS. I need to simulate a 3-equation system with variables x, y, and z. In another program, I got the coefficients and covariance matrix of residuals from regressions. These coeff are assigned into the FRML, and VCV matrix = the cov matrix of residuals. x, y, and z are initially set to their sample means from regression results. SET X = ..... SET Y = .... SET Z = ..... (etc) FRML EQ1 X = 0.030877241 + 0.789697638*Y{1} -0.840189131*Z{1} FRML EQ2 Y = 0.0012518 + 0.97742917*Y{1} FRML EQ3 Z = 0.0012128517 + 0.98488961*Z{1} COMPUTE [SYMMETRIC] V = || 0.00168 |-0.0001177, 0.0000169|-0.0000614, 0.000006, 0.0000207|| I then SIMULATE for 500 times: GROUP(VCV=V) MODEL1 EQ1>>X EQ2>>Y EQ3>>Z COMPUTE NDRAWS = 500 DECLARE VECTOR AVGX(NDRAWS) SDX(NDRAWS) AVGY(NDRAWS) SDY(NDRAWS) AVGZ(NDRAWS) SDZ(NDRAWS) * DO DRAWS = 1,NDRAWS SIMULATE(MODEL=MODEL1) 3 25000 251 SMPL 251 25000 STAT(NOPRINT) X EWISE AVGX(DRAWS) = %MEAN EWISE SDX(DRAWS) = SQRT(%VARIANCE) STAT(NOPRINT) Y EWISE AVGY(DRAWS) = %MEAN EWISE SDY(DRAWS) = SQRT(%VARIANCE) STAT(NOPRINT) Z EWISE AVGZ(DRAWS) = %MEAN EWISE SDZ(DRAWS) = SQRT(%VARIANCE) SMPL END DO DRAWS COMPUTE XMEAN = %SUM(AVGX)/NDRAWS, XSD = %SUM(SDX)/NDRAWS COMPUTE YMEAN = %SUM(AVGY)/NDRAWS, YSD = %SUM(SDY)/NDRAWS COMPUTE ZMEAN = %SUM(AVGZ)/NDRAWS, ZSD = %SUM(SDZ)/NDRAWS The means and standard deviations of the simulated series (in each DRAW as well as averages of all DRAWs) are quite far from their sample statistics. I suppose that the DO loop above changes the seed in each DRAW, and thus the results are not due to sampling errors. I also suppose that the set up of the model above already takes care of the unconditional versus conditional sample statistics. Thanks for any suggestions and help. Cheers. W. Lee * * * ---------- End of message ---------- From: "Gregory, Richard" To: "RATS Discussion List" Subject: RE: element by element muliplikation in RATS? Date: Tue, 21 Oct 1997 15:05:44 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.995.52 (via Mercury MTS (Bindery) v1.30) MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Hi, I wanted to know what a Hadamar product is. Could you refer me to a definition? Thanks. Richard Gregory, Ph.D. Senior Economist Virginia Economic Development Partnership rgregory@vedp.state.va.us >---------- >From: Stefan Nydahl[SMTP:Stefan.Nydahl@nek.uu.se] >Sent: Sunday, October 19, 1997 8:15 AM >To: RATS Discussion List >Subject: element by element muliplikation in RATS? > >Hi, >Does anyone know how to do element by element matrix multiplikation (the >Hadamar product) in RATS? I know I can do it with the EWISE command but this >time I need to use it with FRML in a Multivariate GARCH set up. > >Best Regards >Stefan Nydahl >Department of Economics,Uppsala University, Sweden > > >======================================================================== > > >Nationalekonomiska institutionen / Department of Economics >Uppsala University Phone: 018/18 25 00 >P.O.-Box 513 Intl: +46-18 18 25 00 >S-751 20 Uppsala, Sweden Fax: +46-18 18 14 78 >======================================================================= > > ---------- End of message ---------- From: Norman Morin To: "RATS Discussion List" Subject: Re: element by element muliplikation in RATS? Date: Tue, 21 Oct 1997 15:48:51 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: exmh version 1.6.9.4 5/23/97 (via Mercury MTS (Bindery) v1.30) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii > I wanted to know what a Hadamar product is. Could you refer me to a > definition? > > Richard Gregory, Ph.D. > Senior Economist > Virginia Economic Development Partnership > rgregory@vedp.state.va.us Richard, The Hadamard product of two mxn matrices is the mxn matrix whose i-jth element is the product of the i-jth elements of the two matricies. E.g., if A = 1 2 3 4 5 6 and B = 7 8 9 10 11 12 their Hadamard product is 7 16 27 40 55 72 Magnus and Neudecker (1988) discusses their properties. I think they are occasionally used in factor analysis. A quick way to do them in RATS would be using EWISE, I think. Here's a quick proc I wrote: Regards, Norm ************************************************************** PROCEDURE HADAMARD inmat1 inmat2 outmat OPTION SWITCH PRINT 1 TYPE rect inmat1 inmat2 *outmat LOCAL integer m n COMPUTE m = %rows(inmat1) COMPUTE n = %cols(inmat1) IF (%rows(inmat2).eq.m).and.(%cols(inmat2).eq.n) { DIM outmat(m,n) EWISE outmat(i,j) = inmat1(i,j)*inmat2(i,j) IF PRINT.eq.1 WRITE(noskip,format='(7F12.4)') outmat } ELSE DISPLAY DISPLAY 'Sorry, the input matrices must be of the same dimension' DISPLAY END HADAMARD PROCEDURE HADAMARD inmat1 inmat2 outmat OPTION CHOICE PRINT 1 TYPE rect inmat1 inmat2 *outmat LOCAL integer m n COMPUTE m = %rows(inmat1) COMPUTE n = %cols(inmat1) IF (%rows(inmat2).eq.m).and.(%cols(inmat2).eq.n) { DIM outmat(m,n) EWISE outmat(i,j) = inmat1(i,j)*inmat2(i,j) IF PRINT.eq.1 WRITE(noskip,format='(7F12.4)') outmat } ELSE DISPLAY 'The input matrices must be of the same dimension' END HADAMARD ***************************************************************** -- Norman J. Morin nmorin@frb.gov or m1njm00@frb.gov --------------------------------------------------------------- Industrial Output Section * Division of Research and Statistics Board of Governors of the Federal Reserve System * Mail Stop 82 Washington, D.C. 20551 * (202) 452-2476 --------------------------------------------------------------- ---------- End of message ---------- From: "Gregory, Richard" To: "RATS Discussion List" Subject: RE: element by element muliplikation in RATS? Date: Wed, 22 Oct 1997 08:13:06 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Microsoft Exchange Server Internet Mail Connector Version 4.0.995.52 (via Mercury MTS (Bindery) v1.30) MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Thank you for your help! Richard Gregory, Ph.D. Senior Economist Virginia Economic Development Partnership rgregory@vedp.state.va.us >---------- >From: Norman Morin[SMTP:m1njm00@FRB.GOV] >Sent: Tuesday, October 21, 1997 3:48 PM >To: RATS Discussion List >Subject: Re: element by element muliplikation in RATS? > >> I wanted to know what a Hadamar product is. Could you refer me to a >> definition? >> >> Richard Gregory, Ph.D. >> Senior Economist >> Virginia Economic Development Partnership >> rgregory@vedp.state.va.us > >Richard, > >The Hadamard product of two mxn matrices is the mxn matrix whose i-jth >element is the product of the i-jth elements of the two matricies. > >E.g., if A = 1 2 > 3 4 > 5 6 >and > B = 7 8 > 9 10 > 11 12 > >their Hadamard product is 7 16 > 27 40 > 55 72 > >Magnus and Neudecker (1988) discusses their properties. I think >they are occasionally used in factor analysis. > >A quick way to do them in RATS would be using EWISE, I think. >Here's a quick proc I wrote: > >Regards, > >Norm > >************************************************************** >PROCEDURE HADAMARD inmat1 inmat2 outmat > >OPTION SWITCH PRINT 1 > >TYPE rect inmat1 inmat2 *outmat > >LOCAL integer m n > >COMPUTE m = %rows(inmat1) >COMPUTE n = %cols(inmat1) > >IF (%rows(inmat2).eq.m).and.(%cols(inmat2).eq.n) > { > DIM outmat(m,n) > EWISE outmat(i,j) = inmat1(i,j)*inmat2(i,j) > IF PRINT.eq.1 > WRITE(noskip,format='(7F12.4)') outmat > } > >ELSE > DISPLAY > DISPLAY 'Sorry, the input matrices must be of the same dimension' > DISPLAY > >END HADAMARD >PROCEDURE HADAMARD inmat1 inmat2 outmat > >OPTION CHOICE PRINT 1 > >TYPE rect inmat1 inmat2 *outmat > >LOCAL integer m n > >COMPUTE m = %rows(inmat1) >COMPUTE n = %cols(inmat1) > >IF (%rows(inmat2).eq.m).and.(%cols(inmat2).eq.n) > { > DIM outmat(m,n) > EWISE outmat(i,j) = inmat1(i,j)*inmat2(i,j) > IF PRINT.eq.1 > WRITE(noskip,format='(7F12.4)') outmat > } > >ELSE > DISPLAY 'The input matrices must be of the same dimension' > > >END HADAMARD >***************************************************************** > >-- > > Norman J. Morin > nmorin@frb.gov or m1njm00@frb.gov >--------------------------------------------------------------- >Industrial Output Section * Division of Research and Statistics >Board of Governors of the Federal Reserve System * Mail Stop 82 > Washington, D.C. 20551 * (202) 452-2476 >--------------------------------------------------------------- > > > ---------- End of message ---------- From: Norman Morin To: "RATS Discussion List" Subject: Re: element by element muliplikation in RATS? Date: Wed, 22 Oct 1997 08:18:43 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: exmh version 1.6.9.4 5/23/97 (via Mercury MTS (Bindery) v1.30) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii I made a mistake in the Hadamard proc I sent yesterday. I guess I should run the proc before sending it out. Anyway, here's the corrected version. Sorry. Regards, Norm ********************************************************** PROCEDURE HADAMARD inmat1 inmat2 outmat OPTION SWITCH PRINT 1 TYPE rect inmat1 inmat2 *outmat LOCAL integer m n COMPUTE m = %rows(inmat1) COMPUTE n = %cols(inmat1) IF (%rows(inmat2).eq.m).and.(%cols(inmat2).eq.n) { DIM outmat(m,n) EWISE outmat(i,j) = inmat1(i,j)*inmat2(i,j) IF PRINT WRITE(noskip,format='(7F12.4)') outmat } ELSE { DISPLAY DISPLAY 'The input matrices must be of the same dimension' DISPLAY } END HADAMARD ************************************************************* -- Norman J. Morin nmorin@frb.gov or m1njm00@frb.gov --------------------------------------------------------------- Industrial Output Section * Division of Research and Statistics Board of Governors of the Federal Reserve System * Mail Stop 82 Washington, D.C. 20551 * (202) 452-2476 --------------------------------------------------------------- ---------- End of message ---------- From: To: "RATS Discussion List" Subject: Vol forecasts Date: Wed, 22 Oct 1997 14:36:58 +0100 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) v1.30 Hi! At the moment I am calculating GARCH volatrility forecast by looping through the volatility forecast recursion. I would like to know if there is an more efficient way of doing it. Use of Forecast or an equation that uses the recursion relation, without looping ?! Any comments would be welcome. Regards, Anders Pelli anpe08@handelsbanken.se ---------- End of message ---------- From: GZhang@bank-banque-canada.ca (Guang-Jia Zhang) To: "RATS Discussion List" Subject: GMM estimation Date: Wed, 22 Oct 1997 09:27:49 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mercury MTS (Bindery) v1.30 Hi there, I am estimating the unknown parameters appeared in a dynamic general equilibrium problem by using Hansen-Singleton 's GMM technique. When I ran my RATS program, I got an error message: ## SR10 Missing Values and/or SMPL options leave no usable data Guang Zhang Economist Bank of Canada ---------- End of message ---------- From: "Philippe PROTIN" To: "RATS Discussion List" Subject: Re: element by element muliplikation in RATS? Date: Wed, 22 Oct 1997 15:54:50 +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=US-ASCII Content-transfer-encoding: 7BIT X-mailer: Pegasus Mail for Windows (v2.54) (via Mercury MTS (Bindery) v1.30) Hi, is it possible to execute the EWISE instruction for Hadamard matrix product in a frml ??? Thanks for help. Philippe PROTIN ESA-CERAG BP 47X 38040 GRENOBLE CEDEX 09 FRANCE 04.76.82.57.48. protin@esa.upmf-grenoble.fr ---------- End of message ---------- From: GZhang@bank-banque-canada.ca (Guang-Jia Zhang) To: "RATS Discussion List" Subject: Date: Wed, 22 Oct 1997 15:05:34 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mercury MTS (Bindery) v1.30 Hi, The problem about GMM estimation I post this morning has been solved. Thank you for your attention. Guang-Jia Zhang Bank of Canada ---------- End of message ---------- From: Frederick Bourgoin To: "RATS Discussion List" Subject: Quadratic Programming problem Date: Fri, 24 Oct 1997 10:52:19 +0100 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: MillenniuM Global Investments Ltd. X-Mailer: Mozilla 4.02 [en] (Win95; I) (via Mercury MTS (Bindery) v1.30) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit hi everyone, I'm trying to solve a quadratic programming problem but i have a problem somewhere can someone give a help? * * Fill The Correlation Matrix * Compute [Rectangular[Real]] Q = ||0.949986,0.70636,0.382141,0.279188,0.628126|$ 0.70636,1.201844,0.375144,0.335944,0.683257|0.382141,0.375144,0.60818,0.210983,0.498619|$ 0.279188,0.335944,0.210983,1.282713,0.278315|0.628126,0.683257,0.498619,0.278315,1.166197|| Display Display 'Correlation Matrix:' Write Q Compute Q = 2 * Q * * Fill The Constraints Matrix * Compute [Rectangular[Real]] A = ||14.59232,13.11304,11.29401,-2.60595,8.29158|1.0,1.0,1.0,1.0,1.0|| Display Display 'Constraints' Write A Compute [Vector[Real]] B = ||10,1|| Display Display 'Constraints:' Write B * * Setup the Linear part of the optimizations to Zero * Compute [Vector[Real]] C = ||0,0,0,0,0|| Display Display 'Linear Part' Write C LQPROG(Equalities=2) Results C A B Q This gives wrong solution compare to the excel solver. Frederick Bourgoin MillenniuM London ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RATS discussion List Date: Mon, 27 Oct 1997 02:22:03 +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) v1.30) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Dear Sir, I still don't know how to "unzip" the I(2) program after I have consulted your web page. Would you like to tell me the details? Also, may you tell me where the file I2PAPER.ZIP and PKZ204G.EXE should be downloaded, CATS or Word or others( I can't read the files in CATS and Word)? Thank you for your help! Kai-yin Woo Economics Dept., Hong Kong Shue Yan College ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: RATS discussion List Date: Mon, 27 Oct 1997 12:03:06 -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) v1.30) > > I still don't know how to "unzip" the I(2) program after I have > consulted your web page. Would you like to tell me the details? Also, > may you tell me where the file I2PAPER.ZIP and PKZ204G.EXE should be > downloaded, CATS or Word or others( I can't read the files in CATS and > Word)? > Dear RATS-list readers: I have already responded to this request privately. I just wanted to mention that basic technical support questions regarding RATS, questions about our web site, product info, etc., should be sent directly to us at estima@estima.com, not to the RATS mailing list. For more information on the general purpose of the RATS Mailing List, please see http://www.estima.com/maillist.htm. Sincerely, Tom Maycock Estima -- +-----------------------------+-----------------------------------------+ | Estima | | | P.O. Box 1818 | Voice: (847) 864-8772 | | Evanston, IL 60204-1818 | Fax: (847) 864-6221 | | U.S.A | BBS: (847) 864-8816 | | e-mail: estima@estima.com | CompuServe: 73140,2202 | |-----------------------------------------------------------------------| | Web Site: http://www.estima.com | | RATS Internet Mailing List: New members can join by sending e-mail to | | MAISER@EFS.MQ.EDU.AU with the message: SUBSCRIBE RATS-L | +-----------------------------------------------------------------------+ ---------- End of message ---------- From: M P Tomsett To: "RATS Discussion List" Subject: Date: Tue, 28 Oct 1997 15:49:08 +0000 (GMT) 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) v1.30 Dear All, After estimating a model rather than descriptive stats or coefficient estimates all I obtained was '-nam'. Could anyone explain what is happening here? Cheers, Mark. **************************************************************************** Mark Tomsett Department of Economics University of Durham 23-26 Old Elvet DURHAM CITY County Durham **************************************************************************** ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: Quadratic Programming problem Date: Tue, 28 Oct 1997 14:41:29 -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) v1.30) > From: Frederick Bourgoin > To: "RATS Discussion List" > Subject: Quadratic Programming problem > Date: Fri, 24 Oct 1997 10:52:19 +0100 > Reply-to: "RATS Discussion List" > Organization: MillenniuM Global Investments Ltd. > hi everyone, > > I'm trying to solve a quadratic programming problem but i have a problem > somewhere can someone give a help? > Something's clearly going wrong here, because the answers don't even meet the constraints. We'll look into it. In the meantime, you might want to try the QPROG procedure (stored in the LPQPPROC.SRC file)--it gives the right solution. Sincerely, Tom Maycock Estima -- +-----------------------------+-----------------------------------------+ | Estima | | | P.O. Box 1818 | Voice: (847) 864-8772 | | Evanston, IL 60204-1818 | Fax: (847) 864-6221 | | U.S.A | BBS: (847) 864-8816 | | e-mail: estima@estima.com | CompuServe: 73140,2202 | |-----------------------------------------------------------------------| | Web Site: http://www.estima.com | | RATS Internet Mailing List: New members can join by sending e-mail to | | MAISER@EFS.MQ.EDU.AU with the message: SUBSCRIBE RATS-L | +-----------------------------------------------------------------------+ ---------- End of message ----------