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 JAA64854 for ; Tue, 1 Sep 1998 09:23:42 -0400 From: maiser@efs.mq.edu.au Received: (from root@localhost) by corot.bc.edu (8.8.7/8.8.7) with X.500 id JAA222180 for baum@mail1.bc.edu; Tue, 1 Sep 1998 09:23:42 -0400 Received: from baldrick.ocs.mq.edu.au (baldrick.ocs.mq.edu.au [137.111.1.12]) by corot.bc.edu (8.8.7/8.8.7) with ESMTP id JAA270246 for ; Tue, 1 Sep 1998 09:23:35 -0400 Received: from efs1.efs.mq.edu.au (efs1.efs.mq.edu.au [137.111.64.8]) by baldrick.ocs.mq.edu.au (8.9.0/8.9.0) with ESMTP id XAA14893 for ; Tue, 1 Sep 1998 23:23:29 +1000 (EST) Received: from EFS1/SpoolDir by efs1.efs.mq.edu.au (Mercury 1.40); 1 Sep 98 23:23:15 GMT+1000 Received: from SpoolDir by EFS1 (Mercury 1.40); 1 Sep 98 23:23:13 GMT+1000 To: baum@bc.edu Date: Tue, 1 Sep 98 23:23:12 GMT+1000 Subject: Re: Message-ID: <29177102F13@efs1.efs.mq.edu.au> From: WOO KAI YIN To: "RATS Discussion List" Subject: Re: bootstrap Date: Tue, 04 Aug 1998 10:01:49 +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, Anybody would like to tell me how to implement bootstrap to simulate Johansen's Likelihood ratio statistics (with the inclusion of many dummies in the multivariate cointegration model), both under the Ho:r=0 and Ho:r=1??? Regards, Kevin Senior Lecturer, HK SYC ---------- End of message ---------- From: "Simon.Van-Norden" To: "RATS Discussion List" Subject: Re: bootstrap Date: Mon, 03 Aug 1998 22:29:34 -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 Sure. The null (Ho) is very easy. - First-difference your data. - Estimate a VAR on it. - Simulate the VAR using bootstrapped errors (see p. 10-3 in the manual for an example.) - ACCumulate the resulting series. There are different ways to do the alternative H(r=1), depending on just what kind of cointegrating relationship you want to impose. A simple but less than totally general example would be to - estimate an ECM for one variable - do the VAR in 1st-differences to simulate all the other variables (as above, but with smaller dimension.) - bootstrap the ECM (with the simulated exogenous variables) to generate your last series. SvN WOO KAI YIN wrote: > Dear RATS users, > > Anybody would like to tell me how to implement bootstrap to simulate > Johansen's Likelihood ratio statistics (with the inclusion of many > dummies in the multivariate cointegration model), both under the Ho:r=0 > and Ho:r=1??? > > Regards, > > Kevin > Senior Lecturer, > HK SYC -- Simon van Norden Professeur invité simon.van-norden@hec.ca or svn@alum.mit.edu or http://www.hec.ca/~p280 (514)340-6781 or fax (514)340-5632 or cell (514)993-0466 Service de l'enseignement de la finance, Ecole des H.E.C. 3000 Cote-Sainte-Catherine, Montreal QC, CANADA H3T 2A7 ---------- End of message ---------- From: Giovanna Aguilar Andia To: "RATS Discussion List" Subject: Re: bootstrap Date: Mon, 3 Aug 1998 22:19:13 -0500 (PET) 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 would like to know if anybody have used VAR proceduce Written by Norman Morin to estimate a VAR whit a Blanchard & Quah decomposition. I need that somebody helps me to interprete the outcomes. Thanks in advance. +--------------------------------------------+ + Giovanna Aguilar A + + Departamento de Economia + + Pontificia Universidad Catolica del Peru + + Telf. (51 1) 4602870 (anexo 213), 9449067 + + Fax (51 1) 4601126 + + Apartado 1761 + + Lima 100 + + PERU + +--------------------------------------------+ ---------- End of message ---------- From: "Christian Schmitt" To: "RATS Discussion List" Subject: Stochastic Volatility Model Date: Tue, 4 Aug 1998 09:34:23 -0100 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Zentr. f. Europ. Wirtschaftsforschg MIME-Version: 1.0 Content-type: text/plain; charset=US-ASCII Content-transfer-encoding: 7BIT X-Mailer: Mercury MTS (Bindery) v1.40 Dear RATS users, I am looking for program code to estimate the univariate SV (stochastic volatility / stochastic variance) model by Taylor (1986), and Harvey, Shephard et al.. (199x). GAUSS code would also be appreciated. Regards, -- christian Christian Schmitt Research Fellow ZEW - Center for European Economic Research Internationale Finanzmaerkte (International Finance Dept.) P.O. Box 10 34 43 D - 68034 Mannheim, Germany Tel. (+49) 621 - 1235 - 146 Fax (+49) 621 - 1235 - 223 Email:schmitt@zew.de ---------- End of message ---------- From: Thomas Werner To: "RATS Discussion List" Subject: Re: Stochastic Volatility Model Date: Tue, 04 Aug 1998 13:22:46 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: TUD X-Mailer: Mozilla 4.03 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit Christian Schmitt wrote: > Dear RATS users, > > I am looking for program code to estimate the univariate > SV (stochastic volatility / stochastic variance) model > by Taylor (1986), and Harvey, Shephard et al.. (199x). > > GAUSS code would also be appreciated. > > There is a stochastic volatility package in ox written by Shephard. Look at http://www.nuff.ox.ac.uk/Users/Doornik/ Thomas **************************************************** Dipl. Volkswirt Thomas Werner Fachgebiet Wirtschaftstheorie (VWL1) Fachbereich 01 Technische Universität Darmstadt Residenzschloss D-64283 Darmstadt **************************************************** ---------- End of message ---------- From: Giovanna Aguilar Andia To: "RATS Discussion List" Subject: VAR procedure Date: Tue, 4 Aug 1998 09:39:53 -0500 (PET) 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 user, I would like to know if anybody have used the VAR procedure written by Norman Morin. I don't understand the outcomes from the Blanchard & Quah (B&Q) decomposition. If the innovations are not associated whith a especific variable, as the B&Q's paper proposes, Why do the Impulse Response Functions'grafics show an association between the shocks or innovations and the variables in the system? For exmaple, in a two variable system, with growth output rate and desemployment rate, I would expect a IRF to demand shocks or supply shocks for growth output rate and desemployment rate and not a IRF showing effects of a shock to growth rate or to unemployment rate. If somebody could help me to understant this outcomes, I would be greatful. +--------------------------------------------+ + Giovanna Aguilar A + + Departamento de Economia + + Pontificia Universidad Catolica del Peru + + Telf. (51 1) 4602870 (anexo 213), 9449067 + + Fax (51 1) 4601126 + + Apartado 1761 + + Lima 100 + + PERU + +--------------------------------------------+ ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: Re: bootstrap Date: Wed, 05 Aug 1998 01:43:41 +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=iso-8859-1 Content-Transfer-Encoding: quoted-printable Simon.Van-Norden wrote: >=20 > Sure. >=20 > The null (Ho) is very easy. > - First-difference your data. > - Estimate a VAR on it. > - Simulate the VAR using bootstrapped errors (see p. 10-3 in the manual= for > an example.) > - ACCumulate the resulting series. >=20 > There are different ways to do the alternative H(r=3D1), depending on j= ust > what kind of cointegrating relationship you want to impose. A simple b= ut > less than totally general example would be to > - estimate an ECM for one variable > - do the VAR in 1st-differences to simulate all the other variables (as > above, but with smaller dimension.) > - bootstrap the ECM (with the simulated exogenous variables) to generat= e > your last series. >=20 > SvN >=20 > WOO KAI YIN wrote: >=20 > > Dear RATS users, > > > > Anybody would like to tell me how to implement bootstrap to simulate > > Johansen's Likelihood ratio statistics (with the inclusion of many > > dummies in the multivariate cointegration model), both under the Ho:r= =3D0 > > and Ho:r=3D1??? > > > > Regards, > > > > Kevin > > Senior Lecturer, > > HK SYC >=20 > -- > Simon van Norden Professeur invit=E9 > simon.van-norden@hec.ca or svn@alum.mit.edu or http://www.hec.ca/~p280 > (514)340-6781 or fax (514)340-5632 or cell (514)993-0466 > Service de l'enseignement de la finance, Ecole des H.E.C. > 3000 Cote-Sainte-Catherine, Montreal QC, CANADA H3T 2A7 Dear DR. Norden, Thank you. Regards, KY Woo ---------- End of message ---------- From: wan To: "RATS Discussion List" Subject: memory problem! Date: Wed, 05 Aug 1998 16:06:04 +0800 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mozilla 4.03 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Hi All: When I use the command "dofor" for simulating,I encounter the problem for requesting additional memeory!!Can someone help me!!! The output of mine is following. ------------------------------------------------------------------- RATS386 4.20. Run on Aug 5 1998 (c) 1992-5 Thomas A. Doan. All rights reserved cal 1001 1 1 all 1072:1 source(noecho) gph.src compute a = 1 compute b = 30000 dofor i = a to b (01.0044) set u = %ran(1.0) (01.0062) set v = %ran(1.0) (01.0080) linreg(noprint) u / resid (01.0100) #v constant (01.0116) @gph(power=0.55) resid (01.0142) end dofor ## M4. A memory request for an additional 107414 bytes cannot be satisfied The Error Occurred At Location 0029 --------------------------------------------------------------------- If I set the paramenter b=20000,the problem will disappear!! Reagrds, Jer Yuh Wan ---------- End of message ---------- From: baumgt@wsr.ac.at (Josef Baumgartner ) To: "RATS Discussion List" Subject: Date: Wed, 5 Aug 1998 12:15:28 +0200 (METDST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Version 2.0.3 (via Mercury MTS (Bindery) v1.40) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Dear RATS-users, I hope you can help me in the following two cases: First, I want to apply Hylleberg et al. (1990) - HEGY - seasonal unit root test to MONTHLY data. Does anybody have or know about a procedure like HEGY.SRC - which is for quaterly data - for monthly data? Where can I find critical values for monthly frequency seasonal unit root tests? (Lit.: Hylleberg S., Engle R.F., Granger C.W.J., Yoo B.S., (1990), 'Seasonal integration and cointegration', Journal of Econometrics, 44, 215-238). Second, I want to implement the approximate band-pass filter of Baxter and King (1995).If anybody has or knows about such a procedure, please let me know. (Lit.: Baxter M., King R.G.,(1995), 'Measuring business cycles: Approximate Band-pass filters for economic time series', NBER Working Paper 5022, Feb. 1995). Thanks for your endeavor. With kind regards ----------------------------------------------------------------- Josef Baumgartner Austrian Institute of Phone: +431 7982601 230 Economic Research (WIFO) Fax: +431 7989386 P.O. Box 91 E-mail: baumgt@wifo.ac.at A-1103 Vienna, Austria http://wsrdb2.wsr.ac.at/ ---------- End of message ---------- From: Sang-sub Lee To: "RATS Discussion List" Subject: Re: Date: Wed, 05 Aug 1998 19:39:36 +0900 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: QUALCOMM Windows Eudora Light Version 3.0.2 (32) (via Mercury MTS (Bindery) v1.40) Mime-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable there is a rats program, written by Wouter Denhaan at UC San Diego, which implements various=20 low-frequency filtering procedures such as Hodrick and Prescott and Baxter and King. Check out his=20 home page at UCSD. If you can't find it at his webpage, let me know. =20 At 12:15 =BF=C0=C8=C4 98-08-05 +0200, you wrote: >Dear RATS-users, > >I hope you can help me in the following two cases: > >First, I want to apply Hylleberg et al. (1990) - HEGY - seasonal unit root >test to MONTHLY data. Does anybody have or know about a procedure like >HEGY.SRC - which is for quaterly data - for monthly data? Where can I find >critical values for monthly frequency seasonal unit root tests?=20 >(Lit.: Hylleberg S., Engle R.F., Granger C.W.J., Yoo B.S., (1990),= 'Seasonal >integration and cointegration', Journal of Econometrics, 44, 215-238). > >Second, I want to implement the approximate band-pass filter of Baxter and >King (1995).If anybody has or knows about such a procedure, please let me know. >(Lit.: Baxter M., King R.G.,(1995), 'Measuring business cycles: Approximate >Band-pass filters for economic time series', NBER Working Paper 5022, Feb. >1995). > >Thanks for your endeavor. > >With kind regards > >----------------------------------------------------------------- >Josef Baumgartner >Austrian Institute of Phone: +431 7982601 230 >Economic Research (WIFO) Fax: +431 7989386 >P.O. Box 91 E-mail: baumgt@wifo.ac.at=20 >A-1103 Vienna, Austria http://wsrdb2.wsr.ac.at/ > > Dr. Sang-Sub Lee Fellow Korea Institute of Public Finance 60 Yangjae-Dong, Seocho-Ku, Seoul, 137-130, Korea Tel: 82-2-3460-2220 Fax: 82-2-573-3698/574-9183 e-mail: slee@kipf.kipf.re,kr ---------- End of message ---------- From: "Mark Benfold" To: "RATS Discussion List" Subject: Mark Benfold/GB/GARTMORE is out of the office. Date: Thu, 6 Aug 1998 01:00:58 +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) v1.40 I will be out of the office from 04/08/98 until 17/08/98. I will respond to your message when I return. My colleague Claire Thomas should be able to help you in my absence. ___________________________________________________________________________ _____ ========================================================================== Gartmore Investment Management plc is an appointed representative of Gartmore Investment Ltd (GIL) which is regulated by IMRO and the Personal Investment Authority. GIL represents only the NatWest and Gartmore Marketing Group for life assurance, Pensions, unit trusts, other regulated collective investment schemes and investment services. CONFIDENTIALITY NOTICE This message is sent in confidence for the addressee only. The contents are not to be disclosed to anyone other than the addressee. Unauthorised recipients must preserve this confidentiality and should please advise the sender of any error in transmission. n+? ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE:present value model Date: Thu, 06 Aug 1998 12:11:40 +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=iso-8859-1 Content-Transfer-Encoding: quoted-printable Dear RATS users, Might I ask how to calculate the present value of the following equation by use of RATS commands: T-t-1 Pt =3D gamma * =A3U [gamma^i * Dt+i] + gamma^T-t * PT + constant i=3D0 which in fact is a present value model of stock price. T is a terminal price.=20 I would be grateful if you could give me sincere help! Regards, Kevin Senior Lecturer HKSYC ---------- End of message ---------- From: jean-Philip BELLOTTEAU CML/MK To: "RATS Discussion List" Subject: Re: Date: Fri, 07 Aug 1998 12:41:56 -0700 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mozilla 3.01 [fr] (Win16; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: multipart/mixed; boundary="------------70115CB937AD" Ce message est en plusieurs parties sous format MIME. --------------70115CB937AD Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by minos.noisy.ratp id MAA12946 Concerning the monthly HEGY procedure : There is an article by Beaulieu & Miron that derives the mechanics of the quaterly HEGY procedure for monthly data and computes critical values by Monte Carlo methods. Ref : "Seasonal unit roots in aggregate U.S. data" - J. Joseph Beaulieu & Jeffrey A. Miron - Journal of econometrics 55 (1993) 305-328 North-Holland I have written a file, "HEGY2.SRC", with two procedures "HEGY12" for monthly data and "HEGY4" (same OLS results than Suliman Al-Turki's procedure !) for quaterly data. Concerning "HEGY12" and the conclusion of the Beaulieu/Miron's article, I don't know how to connect the significance of the OLS t- and F- statistics to test the case I(0,0) I(1,0) I(0,1) I(1,1) ? Concerning "HEGY4" I don't have all the critical values computes by HEGY [1990] Any comments or code corrections would be appreciated ! **************************************************** Jean Philip BELLOTTEAU RATP/CML/MK 54 quai de la Rap=E9e 75012 Paris Tel : 01.44.68.35.29 Fax : 01.44.68.33.00 **************************************************** --------------70115CB937AD Content-Type: text/plain; charset=iso-8859-1; name="HEGY2.SRC" Content-Disposition: inline; filename="HEGY2.SRC" Content-Transfer-Encoding: quoted-printable X-MIME-Autoconverted: from 8bit to quoted-printable by minos.noisy.ratp id MAA12946 /* @HEGY12(lags=3D-1/[0]/..,det=3Dnone/constant/sd/trend/[strend],alpha=3D0.= 01/0.025/[0.05]/0.1,[notrace]/trace) SERIES START END Juin 1998 - 07/08/98 Jean-Philip BELLOTTEAU - RATP/CML/RRC Computes Beaulieu & Miron (1993) seasonal unit root tests for monthly time series, with and without intercept, seasonal dummies, and/or trend from the regression: a(B)Y13(t) =3D DET + PI1 Y1(t-1) + PI2 Y2(t-1) + PI3 Y3(t-1) + PI4 Y4(= t-1) + PI5 Y5(t-1) + PI6 Y6(t-1)=20 + PI7 Y7(t-1) + PI8 Y8(t-1) + PI9 Y9(t-1) + PI10 Y10(t= -1) + PI11 Y11(t-1) + PI12 Y12(t-1) + e(t) Please note: This procedure is written for use with monthly time series. see HEGY4 for quaterly data. The following OPTIONS are available: DET =3D NONE,CONS,SD,TREND,STREND Select the determinstic componant. NONE sets the determinstic componant to zero; CONS includes intercept only; SD includes CON= S and (11) seasonal dummies; TREND includes CONS and trend; STREND includes CONS, SD, and Trend LAGS =3D [0] Controls default lag length. Can be overridden for With the option LAGS=3D-1, the procedure determines the = number of lags by S-BIC criteria The procedure reports a Lagrange Multiplier of order 1-12 & a Ljung-Box = test for autocorrelation in the error term=20 REFERENCES: J.J. Beaulieu, J.A. Miron, "Seasonal unit roots in aggregate U.S. data" -= Journal of Econometrics [1993] monthly version by Jean-Philip Bellotteau Send comments and suggestions to: Jean-Philippe.belloteau@ratp.fr */ PROCEDURE HEGY12 SERIES START END TYPE SERIES SERIES TYPE INTEGER START END * OPTION INTEGER LAGS 0 OPTION CHOICE DET 5 NONE CONStant SD TREND STREND OPTION CHOICE alpha 3 0.01 0.025 0.05 0.1 OPTION switch trace 0 OPTION switch graph 0 * LOCAL INTEGER STARTL ENDL LOCAL VEC[SERIES] y LOCAL VEC[REAL] pi f LOCAL REC[REAL] cv LOCAL SERIES TREND SDUM et LOCAL REAL r3 r32 qlb plb qlm plm sbic minsbic LOCAL INT i j k l retard l12 nobs LOCAL VEC[INT] reg aux LOCAL VEC[VEC[INT]] regexc LOCAL equation hg lm hg2 LOCAL VEC[LABEL] testcv * if %defined(series) { INQUIRE(SERIES=3DSERIES) STARTL>>START ENDL>>END * dim y(13) pi(12) f(16) cv(6,20) regexc(16) testcv(16) do i=3D1,13 clear y(i) label y(i) # 'Y'+%string(i) endo i do i=3D1,16 com testcv(i) =3D ' ' endo i COM CV =3D || 0.01, 0.025, 0.05, 0.10, 0.01, 0.025, 0.05, = 0.10, 0.01, 0.025, 0.05, 0.10, 0.01, 0.025, 0.05, 0.10, = 0.9, 0.95, 0.975, 0.99 | $ -2.51, -2.18, -1.89, -1.58, -2.53, -2.16, -1.87, -= 1.57, -2.50, -2.16, -1.88, -1.55, -2.31, -1.95, -1.63, -1.27, 2= .34, 3.03, 3.71, 4.6 | $ -3.35, -3.06, -2.80, -2.51, -2.48, -2.15, -1.89, -= 1.57, -2.51, -2.16, -1.87, -1.54, -2.30, -1.93, -1.62, -1.27, 2= .32, 3.01, 3.68, 4.6 | $ -3.32, -3.02, -2.76, -2.47, -3.28, -3.01, -2.76, -= 2.48, -3.83, -3.51, -3.25, -2.95, -2.61, -2.21, -1.85, -1.45, 5= .27, 6.26, 7.19, 8.35 | $ -3.87, -3.58, -3.32, -3.06, -2.52, -2.18, -1.88, -= 1.55, -2.49, -2.16, -1.88, -1.54, -2.28, -1.93, -1.61, -1.25, 2= .30, 2.90, 3.64, 4.53 | $ -3.83, -3.54, -3.28, -2.99, -3.31, -3.02, -2.75, -= 2.47, -3.79, -3.50, -3.24, -2.95, -2.57, -2.18, -1.85, -1.45, 5= .25, 6.23, 7.14, 8.33 || com r32 =3D sqrt(3)/2.0, r3 =3D sqrt(3) if lags>0 com retard =3D lags else com retard =3D 0 SET Y(1) =3D SERIES + SERIES{1} + SERIES{2} + SERIES{3= } + SERIES{4} + SERIES{5} + SERIES{6} + SERIES{7} + SERIES{8} + SERIE= S{9} + SERIES{10} + SERIES{11} SET Y(2) =3D - SERIES + SERIES{1} - SERIES{2} + SERIES{3= } - SERIES{4} + SERIES{5} - SERIES{6} + SERIES{7} - SERIES{8} + SERIE= S{9} - SERIES{10} + SERIES{11} SET Y(3) =3D - SERIES{1} + SERIES{3= } - SERIES{5} + SERIES{7} - SERIE= S{9} + SERIES{11} SET Y(4) =3D - SERIES + SERIES{2} = - SERIES{4} + SERIES{6} - SERIES{8} = + SERIES{10} SET Y(5) =3D 0.5*(- SERIES - SERIES{1}+2*SERIES{2} - SERIES{3= } - SERIES{4}+2*SERIES{5} - SERIES{6} - SERIES{7}+2*SERIES{8} - SERIE= S{9} - SERIES{10}+2*SERIES{11} ) SET Y(6) =3D r32*( SERIES - SERIES{1} + SERIES{3= } - SERIES{4} + SERIES{6} - SERIES{7} + SERIE= S{9} - SERIES{10} ) SET Y(7) =3D 0.5*( SERIES - SERIES{1}-2*SERIES{2} - SERIES{3= } + SERIES{4}+2*SERIES{5} + SERIES{6} - SERIES{7}-2*SERIES{8} - SERIE= S{9} + SERIES{10}+2*SERIES{11} ) SET Y(8) =3D r32*(- SERIES - SERIES{1} + SERIES{3= } + SERIES{4} - SERIES{6} - SERIES{7} + SERIE= S{9} + SERIES{10} ) SET Y(9) =3D 0.5*(-r3*SERIES + SERIES{1} - SERIES{3= }+r3*SERIES{4}-2*SERIES{5}+r3*SERIES{6} - SERIES{7} + SERIE= S{9}-r3*SERIES{10}+2*SERIES{11} ) SET Y(10) =3D 0.5*( SERIES-r3*SERIES{1}+2*SERIES{2}-r3*SERIES{3= } + SERIES{4} - SERIES{6}+r3*SERIES{7}-2*SERIES{8}+r3*SERIE= S{9} - SERIES{10} ) SET Y(11) =3D 0.5*( r3*SERIES + SERIES{1} - SERIES{3= }-r3*SERIES{4}-2*SERIES{5}-r3*SERIES{6} - SERIES{7} + SERIE= S{9}+r3*SERIES{10}+2*SERIES{11} ) SET Y(12) =3D 0.5*(- SERIES-r3*SERIES{1}-2*SERIES{2}-r3*SERIES{3= } - SERIES{4} + SERIES{6}+r3*SERIES{7}+2*SERIES{8}+r3*SERIE= S{9} - SERIES{10} ) SET Y(13) =3D SERIES = = - SERIES{12} if graph { spgraph(hfields=3D3,vfields=3D5) graph 1 # series do i=3D1,13 graph 1 # y(i) enddo i spgraph(done) } clear et trend sdum SET TREND STARTL ENDL =3D T SEA SDUM * com nobs =3D endl-startl+1 DIS DIS @20 'Monthly seasonal unit root by HEGY procedure (Beaulieu - M= iron [1993])' dis dis 'Series' %LABEL(SERIES) @-1 '.' %DATELABEL(startl) @-1 '-' @-1= %DATELABEL(endl) ',' nobs 'Obs.' dis enter(varying) reg # Y(1){1} Y(2){1} Y(3){1} Y(4){1} Y(5){1} Y(6){1} Y(7){1} Y(8){1} Y= (9){1} Y(10){1} Y(11){1} Y(12){1} com k =3D %rows(reg) do i=3D1,11 dim regexc(i)(k) do j=3D1,k com regexc(i)(j) =3D reg(j+%rows(reg)-k) enddo j com k =3D k - 4 enddo i com i =3D 12 do j=3D9,41,8 dim regexc(i)(8) do l=3D1,8 com regexc(i)(l) =3D reg(j+l-1) enddo l com i =3D i + 1 enddo j if det<>1 { if det=3D=3D2 enter(varying) aux # constant else if det=3D=3D3 enter(varying) aux # constant sdum{-10 to 0} else if det=3D=3D4 enter(varying) aux # constant trend else if det=3D=3D5 enter(varying) aux # constant sdum{-10 to 0} trend if retard=3D=3D0 { equation hg Y(13) # reg aux enter(varying) aux # reg aux equation lm et # et{1 to 12} reg aux } else { equation hg Y(13) # reg Y(13){1 to retard} aux enter(varying) aux # reg Y(13){1 to retard} aux equation lm et # et{1 to 12} reg Y(13){1 to retard} aux } } else { if retard=3D=3D0 { equation hg Y(13) # reg enter(varying) aux # reg equation lm et # et{1 to 12} reg } else { equation hg Y(13) # reg Y(13){1 to retard} enter(varying) aux # reg Y(13){1 to retard} equation lm et # et{1 to 12} reg Y(13){1 to retard} } } if lags=3D=3D-1 { dis dis 'Set of the number of lags by S-BIC criteria (and no serial = correlation). Sample' %DATELABEL(startl+12+l12) @-1 '-' @-1 %DATELABEL(en= dl) @-1 ',' endl-startl-12-l12+1 'Obs.' dis dis 'Lag S-BIC p(Q=B0=3D0)' dis '**************************' if 12*((0.01*nobs)**0.25) - fix(12*((0.01*nobs)**0.25))<0.5 com l12 =3D fix(12*((0.01*nobs)**0.25)) else com l12 =3D fix(12*((0.01*nobs)**0.25))+1 ; * Schwert informa= tion [1987] com minsbic =3D 1e8 do i=3D0,l12 if i>0 LINREG(noprint) Y(13) STARTL+12+l12 ENDL et # reg aux Y(13){1 to i} else LINREG(noprint) Y(13) STARTL+12+l12 ENDL et # reg aux com sbic =3D %nobs*log(%seesq*%nobs)+%nreg*log(%nobs) if (sbic < minsbic).and.(%qsignif>0.1) com minsbic =3D sbic, retard =3D i dis ## i @4 sbic #####.## %qsignif enddo i dis dis 'Optimal lag :' retard if retard>0 linreg(noprint,create,define=3Dhg) Y(13) # reg aux Y(13){1 to retard} } LINREG(print=3Dtrace,equation=3Dhg) Y(13) STARTL+12+retard ENDL e= t do i=3D1,12 COM pi(i) =3D %BETA(i)/SQRT(%SEESQ*%XX(i,i)) if i=3D=3D1 { if pi(i)>cv(det+1,alpha) com testcv(i) =3D '*' } else if i=3D=3D2 { if pi(i)>cv(det+1,alpha+4) com testcv(i) =3D '*' } else if (float(i)/2.0)<>fix(float(i)/2.0) { if pi(i)>cv(det+1,alpha+8) com testcv(i) =3D '*' } else { if pi(i)>cv(det+1,alpha+12) com testcv(i) =3D '*' } enddo i com qlb =3D %qstat , plb =3D %qsignif do i=3D1,16 exclude(noprint) # regexc(i) com f(i) =3D %CDSTAT if i>=3D12 { if f(i)>cv(det+1,21-alpha) com testcv(i) =3D '*' } enddo i LINREG(noprint,equation=3Dlm) et STARTL+retard+24 ENDL cdf(noprint) chisqr %trsq 12 com qlm =3D %cdstat, plm =3D %signif dis 'Freq. 0 pi -- pi/2 -- -- 2pi/3 - = -- pi/3 -- -- 5pi/6 - -- pi/6 --- pi/2 2pi/3 pi/3 5pi/6 = pi/6' dis 'Roots Lag PI1 PI2 PI3 PI4 PI5 PI6 = PI7 PI8 PI9 PI10 PI11 PI12 F[3-4] F[5-6] F[7-8] F[9-10] F[= 11-12]' dis '**************************************************************= *************************************************************************= ******' dis %label(series) @13 ## retard @18 ##.## pi(1) @-1 testcv(1) @25 = ##.## pi(2) @-1 testcv(2) @32 ##.## pi(3) @-1 testcv(3) @39 ##.## pi(4) @= -1 testcv(4) @46 ##.## pi(5) @-1 testcv(5) @53 ##.## pi(6) @-1 testcv(6) = $ @60 ##.## pi(7) @-1 testcv(7) @67 ##.## pi(8) @-1 testcv(8) @74= ##.## pi(9) @-1 testcv(9) @81 ##.## pi(10) @-1 testcv(10) @88 ##.## pi(1= 1) @-1 testcv(11) @95 ##.## pi(12) @-1 testcv(12) $ @102 ###.## f(12) @-1 testcv(12) @110 ###.## f(13) @-1 testcv(1= 3) @118 ###.## f(14) @-1 testcv(14) @126 ###.## f(15) @-1 testcv(15) @134= ###.## f(16) @-1 testcv(16) dis dis 'Ljung-Box test for serial correlation' @40 ##.## qlb ', signi= ficance level of' #.## plb dis 'LM test for serial correlation' @40 ##.## qlm ', significance= level of' #.## plm release y pi f cv reg aux regexc } else dis '@HEGY12(lags=3D-1/[0]/..,det=3Dnone/constant/sd/trend/[strend]= ,alpha=3D0.01/0.025/[0.05]/01,[notrace]/trace) SERIES START END' end /* @HEGY4 (lags=3D-1/[0]/..,det=3Dnone/constant/sd/[strend]/all,alpha=3D[0.0= 5]/0.1,[notrace]/trace) SERIES START END Juin 1998 - 07/08/98 Jean-Philip BELLOTTEAU - RATP/CML/RRC Computes HEGY (1993) seasonal unit root tests for quaterly time series, with and without intercept, trend and seasonal dummies from the regression: a(B)Y5(t) =3D DET + PI1 Y1(t-1) + PI2 Y2(t-1) + PI3 Y3(t-1) + PI4 Y4(t= -1) + e(t) Please note: This procedure is written for use with quaterly time series. see HEGY12 for quaterly data. The following OPTIONS are available: DET =3D NONE,CONS,STREND,ALL LAGS =3D [0] Controls default lag length. Can be overridden for With the option LAGS=3D-1, the procedure determines the = number of lags by S-BIC criteria The procedure reports a Lagrange Multiplier of order 1-4 & a Ljung-Box t= est for autocorrelation in the error term=20 quaterly version based on Suliman Al-Turki Send comments and suggestions to: Jean-Philippe.belloteau@ratp.fr */ PROCEDURE HEGY4 SERIES START END TYPE SERIES SERIES TYPE INTEGER START END * OPTION INTEGER LAGS 0 OPTION CHOICE DET 4 none constant sd strend all OPTION CHOICE alpha 1 0.05 0.10 OPTION switch trace 0 * LOCAL INTEGER STARTL ENDL LOCAL VEC[SERIES] y LOCAL VEC[REAL] pi f LOCAL REC[REAL] cv LOCAL SERIES TREND sd et LOCAL REAL qlb plb qlm plm sbic minsbic LOCAL INT i j k l m retard l4 nobs det2 det3 LOCAL VEC[INT] aux LOCAL equation hg lm hg2 LOCAL VEC[LABEL] testcv * if %defined(series) { INQUIRE(SERIES=3DSERIES) STARTL>>START ENDL>>END * dim y(5) pi(4) f(1) cv(5,6) testcv(5) do i=3D1,5 clear y(i) label y(i) # 'Y'+%string(i) endo i do i=3D1,3 com testcv(i) =3D ' ' endo i COM CV =3D || 0.05, 0.10, 0.05, 0.10, 0.05, 0.10 | $ -2.10, -2.30, -1.95, -1.80, 3.08, 2.85 | $ -2.10, -2.30, -1.95, -1.80, 3.08, 2.85 | $ -2.80, -2.47, -2.80, -2.48, 6.60, 5.47 | $ -3.37, -3.05, -2.80, -2.48, 6.57, 5.45 || if lags>0 com retard =3D lags else com retard =3D 0 SET Y(1) startl+4 endl =3D SERIES{1} + SERIES{2} + SERIES{3} + se= ries{4} SET Y(2) startl+4 endl =3D -SERIES{1} + SERIES{2} - SERIES{3} + se= ries{4} SET Y(3) startl+4 endl =3D -SERIES{1} + SERIES{3} SET Y(4) startl+4 endl =3D -SERIES{2} + SERIES{4} SET Y(5) startl+4 endl =3D SERIES - series{4} clear et trend sd SET TREND STARTL ENDL =3D T SEA sd * com nobs =3D endl-startl+1 DIS DIS @20 'Quaterly seasonal unit root by HEGY procedure ([1990])' dis dis 'Series' %LABEL(SERIES) @-1 '.' %DATELABEL(startl) @-1 '-' @-1= %DATELABEL(endl) ',' nobs 'Obs.' if det<>5 com det2 =3D det3 =3D det else com det2 =3D 1, det3 =3D 4 do m=3Ddet2,det3 if m=3D=3D2 { enter(varying) aux # constant dis 'Input : constant' } else if m=3D=3D3 { enter(varying) aux # constant sd{-2 to 0} dis 'Input : constant + seasonal dummies' } else if m=3D=3D4 { enter(varying) aux # constant sd{-2 to 0} trend dis 'Input : constant + seasonal dummies + trend' } else { dim aux(0) dis 'No input' } if lags=3D=3D-1 { com l4 =3D %imin(nobs/4,12) com minsbic =3D 1e8 if trace { dis dis 'Set of the number of lags by S-BIC criteria (and no s= erial correlation).' dis 'Sample' %DATELABEL(startl+4+l4) @-1 '-' @-1 %DATELABE= L(endl) @-1 ',' endl-startl-4-l4+1 'Obs.' dis dis 'Lag S-BIC p(Q=B0=3D0)' dis '**************************' } do i=3D0,l4 if i>0 LINREG(noprint) Y(5) STARTL+4+l4 ENDL et # Y(1) Y(2) Y(3) Y(4) aux Y(5){1 to i} else LINREG(noprint) Y(5) STARTL+4+l4 ENDL et # Y(1) Y(2) Y(3) Y(4) aux com sbic =3D %nobs*log(%seesq*%nobs)+%nreg*log(%nobs) if (sbic < minsbic).and.(%qsignif>0.1) com minsbic =3D sbic, retard =3D i if trace dis ## i @4 sbic #####.## %qsignif enddo i if trace { dis dis 'Optimal lag :' retard } } if retard=3D=3D0 { equation hg Y(5) # Y(1) Y(2) Y(3) Y(4) aux equation lm et # et{1 to 4} Y(1) Y(2) Y(3) Y(4) aux } else { equation hg Y(5) # Y(1) Y(2) Y(3) Y(4) aux Y(5){1 to retard} equation lm et # et{1 to 4} Y(1) Y(2) Y(3) Y(4) aux Y(5){1 to retard} } LINREG(print=3Dtrace,equation=3Dhg) Y(5) STARTL+4+retard ENDL = et do i=3D1,4 COM pi(i) =3D %BETA(i)/SQRT(%SEESQ*%XX(i,i)) enddo i if pi(1)>cv(m+1,alpha) com testcv(1) =3D '*' if pi(2)>cv(m+1,2+alpha) com testcv(2) =3D '*' com testcv(3) =3D '?', testcv(4) =3D '?' com qlb =3D %qstat , plb =3D %qsignif exclude(noprint) # y(3) y(4) com f(1) =3D %CDSTAT if f(1)>cv(m+1,4+alpha) com testcv(5) =3D '*' LINREG(noprint,equation=3Dlm) et STARTL+retard+8 ENDL cdf(noprint) chisqr %trsq 4 com qlm =3D %cdstat, plm =3D %signif dis dis 'Ang. Freq.(w) 0 pi -- pi/2 -- pi/2' dis 'Freq. (v) 0 1/2 --- 1/4 -- 1/4' dis 'P=E9riod (T) oo 2 ---- 4 --- 4' dis 'Roots Lag PI1 PI2 PI3 PI4 F[3-4]' dis '*****************************************************' dis %label(series) @13 ## retard @18 ##.## pi(1) @-1 testcv(1) @= 25 ##.## pi(2) @-1 testcv(2) @32 ##.## pi(3) @-1 testcv(3) @39 ##.## pi(4= ) @-1 testcv(4) @46 ###.## f(1) @-1 testcv(5) dis if (testcv(1)=3D=3D'*').and.(testcv(2)=3D=3D'*').and.(testcv(5)=3D= =3D'*') dis 'H0 : I(0,1) <-> One seasonal unit root' else if ((testcv(1)<>'*').and.(testcv(2)=3D=3D'*').and.(testcv(5= )<>'*')).or.((testcv(1)<>'*').and.(testcv(2)<>'*').and.(testcv(5)=3D=3D'*= ')) dis 'H0 : I(1,0) <-> One non seasonal unit root' else if ((testcv(1)=3D=3D'*').and.(testcv(2)=3D=3D'*').and.(test= cv(5)<>'*')).or.((testcv(1)=3D=3D'*').and.(testcv(2)<>'*').and.(testcv(5)= =3D=3D'*')) dis 'H0 : I(0,0) <-> No seasonal unit root' else if (testcv(1)<>'*').and.(testcv(2)=3D=3D'*').and.(testcv(5)= =3D=3D'*') dis 'H0 : I(1,1) <-> One non seasonal unit root and one seas= onal unit root' dis dis 'Ljung-Box test for serial correlation' @40 ##.## qlb ', si= gnificance level of' #.## plb dis 'LM test for serial correlation' @40 ##.## qlm ', significa= nce level of' #.## plm enddo m release y pi f cv aux } else dis '@HEGY4(lags=3D-1/[0]/..,det=3Dconstant/sd/[strend]/all,alpha=3D= [0.05]/0.1,[notrace]/trace) SERIES START END' end --------------70115CB937AD-- ---------- End of message ---------- From: "¸U­õà±" To: "RATS Discussion List" Subject: Power Between GPH and ADF test!1 Date: Sat, 08 Aug 1998 21:51:56 +0800 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: 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 Hi: Is there any papers or articles comparing the power between GPH and ADF test on detecting unit root process and fractional integrated process!! Regards, Jer Yuh Wan ---------- End of message ---------- From: "jmcderm1" To: "RATS Discussion List" Subject: Simulation of an ARIMA(p,d,q) model Date: Sat, 8 Aug 1998 14:56:11 -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; Content-Transfer-Encoding: 7bit X-Mailer: Microsoft Outlook Express 4.71.1712.3 (via Mercury MTS (Bindery) v1.40) I am interested in generating a simulated series from an ARIMA (p,d,q) model. Does anyone know of a procedure that will accomplish this. Jim McDermott ---------- End of message ---------- From: Marco Hell To: "RATS Discussion List" Subject: Power of the ADF and PP Date: Sun, 09 Aug 1998 19:15:21 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Comcity X-Mailer: Mozilla 4.04 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit A test of the power of the ADF and PP is in U. Hassler, J. Wolters: On the power of unit root tests against fractional alternatives in Economic Letters(45), 1994, S.1-5 . Marco Hell. Hi: Is there any papers or articles comparing the power between GPH and ADF test on detecting unit root process and fractional integrated process!! Regards, Jer Yuh Wan -- \\\|||||/// \\ - - // ( @ @ ) +------oOOo-(_)-oOOo---- +----------------------------------------+ | | Marco Hell | | Have a nice day ! | | | | e-Mail: marco.hell@ki.comcity.de | +---------------Oooo-----+----------------------------------------+ oooO ( ) ( ) ) / \ ( (_/ \_) ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: Date: Mon, 10 Aug 1998 01:15:09 +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=iso-8859-1 Content-Transfer-Encoding: quoted-printable Dear RATS users, Might I ask how to calculate the present value of the following equation by use of RATS commands: T-t-1 Pt =3D gamma * =A3U [gamma^i * Dt+i] + gamma**T-t+1 * PT + constant i=3D0 which in fact is a present value model of stock price. T is a terminal price.=20 And, I did try to write a program to calculate: T-t-1 X=3Dgamma * =A3U [gamma**i * Dt+i] : i=3D0 cal 80 1 4 all 97:4 open data house.rat data(format=3Drats) / do i =3D 0,51 /* from the above, t=3D0, T=3D52, T-t-1=3D51 set 85:1+i 97:3 X=3D(0.9764*0.9764**i)*Dt end do i; I want to generate a new series xt.=20 However, I got an error message: # SR1. ALLOCATE Instruction Needed Before Series or Equations Can Be Used Might I ask how to modify my program, please???? I would be grateful if you could give me sincere assistance! Regards, Kevin Senior Lecturer HKSYC ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: Date: Mon, 10 Aug 1998 11:01:54 +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, Might I ask how to implement wald test using the Newey-West variant of White's heteroskedasticity and autocorrelation correction, please????? Thanks a lot!! Regards, KY Woo Senior Lecturer, HK SYC ---------- End of message ---------- From: Francisco Jose Climent Diranzo To: "RATS Discussion List" Subject: autocorrelation test Date: Mon, 10 Aug 1998 08:27:00 +0200 (METDST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Pro Version 3.0 (16) (via Mercury MTS (Bindery) v1.40) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Dear rats users: I want to test in a equation system if there are autocorrelation in the residuals but i don't know wich is the correct test, In Breusch y Pagan (1980) i think there a test consistent in the case of three equations: T[r(1,2)+r(1,3)+r(2,3)]-----chi squared wtih 3 dregrees of freedom r(1,2) is de correlation coeficient between residuals of eq. 1 and 2 calculates by indivual OLS, and T is de number of observations. can you help me if this test is correct or there are another? Thanks Francisco Jose Climent Diranzo E-Mail: fcliment@uv.es Profesor del Departamento de Economia Financiera y Matematica Tlf: +34 96 382 83 69 Fax: +34 96 382 83 70 Facultad de Ciencias Economicas y Empresariales Edificio departamental Oriental Avd. Els Tarongers s/n 46022 Valencia Spain ---------- End of message ---------- From: "Richard G. Anderson" To: "RATS Discussion List" Subject: Re[2]: histograms Date: Mon, 10 Aug 1998 17:09:11 +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) v1.40 [ADMIN: This message was delayed since it was not sent to the list address.] there are good examples in the Cats for Rats code... Richard Anderson FRB St Louis ______________________________ Reply Separator _________________________________ Subject: Re: histograms Author: rats-l-owner@efs.mq.edu.au at internet Date: 7/28/1998 9:01 PM MIME-Version: 1.0 Content-type: text/plain; charset=ISO-8859-1 Content-transfer-encoding: 7BIT Dear Rats users: Does anybody know a simple way to get frecuency histograms (bar-graphics) in RATS?. Regards, Content-type:MULTIPART/MIXED; BOUNDARY="0-1338571890" Date: Tue, 28 Jul 1998 21:01:35 -0400 Errors-to: rats-l-owner@efs.mq.edu.au From: szuniga@entelchile.net (SERGIO ZUNIGA JARA) Message-id: <42240C14EF8@efs1.efs.mq.edu.au> MIME-version: 1.0 Received: from a1pmdf.a1fw.frb.org by a1smtp.frb.org (PMDF V5.1-10 #29475) with ESMTP id <0EWU00O010WVCY@a1smtp.frb.org> for Richard.G.Anderson@stls.frb.org; Tue, 28 Jul 1998 21:17:20 -0400 (EDT) Received: from fed1.frb.org (fedora.a1fw.frb.org) by a1pmdf.a1fw.frb.org (PMDF V5.1-8 #29478) with SMTP id <0EWU0WS2I00E29@a1pmdf.a1fw.frb.org> for Richard.G.Anderson@stls.frb.org; Tue, 28 Jul 1998 21:17:17 -0400 (EDT) Received: by fed1.frb.org; id VAA08177; Tue, 28 Jul 1998 21:17:16 -0400 Received: from baldrick.ocs.mq.edu.au(137.111.1.12) by fedora.frb.org via smap (3.2) id xma008096; Tue, 28 Jul 1998 21:16:46 -0400 Received: from efs1.efs.mq.edu.au (efs1.efs.mq.edu.au [137.111.64.8]) by baldrick.ocs.mq.edu.au (8.9.0/8.9.0) with ESMTP id LAA18820; Wed, 29 Jul 1998 11:05:09 +1000 (EST) Received: from EFS1/SpoolDir by efs1.efs.mq.edu.au (Mercury 1.40) ; Wed, 29 Jul 1998 11:04:22 +0000 Received: from SpoolDir by EFS1 (Mercury 1.40); Wed, 29 Jul 1998 11:04:13 +0000 Reply-to: RATS Discussion List Sender: Maiser@efs1.efs.mq.edu.au Subject: RE: histograms To: RATS Discussion List X-Mailer: Microsoft Internet Mail 4.70.1161 (via Mercury MTS (Bindery) v1.40) X-Listname: ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: Date: Mon, 10 Aug 1998 15:34:34 +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=iso-8859-1 Content-Transfer-Encoding: quoted-printable Dear RATS users, Might I ask how to calculate the present value of the following equation by use of RATS commands: T-t-1 Pt =3D gamma * =A3U [gamma^i * Dt+i] + gamma**T-t+1 * PT + constant i=3D0 which in fact is a present value model of stock price. T is a terminal price.=20 And, I did try to write a program to calculate: T-t-1 X=3Dgamma * =A3U [gamma**i * Dt+i] : i=3D0 cal 80 1 4 all 97:4 open data house.rat data(format=3Drats) / do i =3D 0,51 /* from the above, t=3D0, T=3D52, T-t-1=3D51 set 85:1+i 97:3 X=3D(0.9764*0.9764**i)*Dt end do i; I want to generate a new series xt.=20 However, I got an error message: ## SX22. Expected Type Entry Value, Got REAL Matrix Instead >>>> x=3D0.9764**i*rara<<<< x is a new variable I want to create. rara is a data series in my rats file.=20 Might I ask how to modify my program, please???? I would be grateful if you could give me sincere assistance! Regards, Kevin Senior Lecturer HKSYC ---------- End of message ---------- From: Alejandro Arrieta To: "RATS Discussion List" Subject: Re: autocorrelation test Date: Mon, 10 Aug 1998 08:57:21 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Superintendencia de Banca y Seguros del Peru X-Mailer: Mozilla 3.03 (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit The Breusch y Pagan test, take the sum of square correlations, and have M(M-1)/2 degrees of freedom (M is the number of equations). There is another test in Greene (likelihood ratio statistic) and a simple explanation of SUR in his Econometric Analysis, Second edition, Ch. 7. Alejandro Arrieta SBS-Peru Francisco Jose Climent Diranzo wrote: > > Dear rats users: > > I want to test in a equation system if there are autocorrelation in the > residuals but i don't know wich is the correct test, > > In Breusch y Pagan (1980) i think there a test consistent in the case of > three equations: > > T[r(1,2)+r(1,3)+r(2,3)]-----chi squared wtih 3 dregrees of freedom > > r(1,2) is de correlation coeficient between residuals of eq. 1 and 2 > calculates by indivual OLS, and T is de number of observations. > > can you help me if this test is correct or there are another? > > Thanks > > Francisco Jose Climent Diranzo E-Mail: fcliment@uv.es > Profesor del Departamento de Economia Financiera y Matematica > Tlf: +34 96 382 83 69 > Fax: +34 96 382 83 70 > Facultad de Ciencias Economicas y Empresariales > Edificio departamental Oriental > Avd. Els Tarongers s/n > 46022 Valencia > Spain ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: Date: Tue, 11 Aug 1998 00:59:08 +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=iso-8859-1 Content-Transfer-Encoding: quoted-printable Dear RATS users, Might I ask how to calculate the present value of the following equation by use of RATS commands: T-t-1 Pt =3D gamma * =A3U [gamma^i * Dt+i] + gamma**T-t+1 * PT + constant i=3D0 which in fact is a present value model of stock price. T is a terminal price.=20 Or, anybody can give me hints to write a program for calculation, please??? I would be grateful if you could give me sincere assistance! Regards, Kevin Senior Lecturer HKSYC ---------- End of message ---------- From: Nesmith_T To: "RATS Discussion List" Subject: RE: Date: Mon, 10 Aug 1998 14:39:35 -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 (5.5.1960.3) (via Mercury MTS (Bindery) v1.40) Content-Type: text/plain; Content-Transfer-Encoding: quoted-printable Dear Kai Yin, Your code is mixing up the use of a set statement and a compute statement. As it stands the code would first be trying to set all the values from 85:1 to 97:3 equal to X=3D(0.9764*0.9764**1)*Dt on the = first run through the loop, then all the values from 85:2 to 97:3 to X=3D(0.9764*0.9764**2)*Dt on the second run through the loop, etc. = This is an inefficient way to code the problem, and although not wrong is probably causing RATS difficulty in resolving the reference to the series on the rhs of the equation. Also you are using i, an integer in the formula as a real. This is generally a bad idea because it relies on RATS converting integers to reals appropriately. There are two ways that you can fix the problems. First continue to = use the do loop construction as in the following: cal 80 1 4 all 97:4 open data house.rat data(format=3Drats) / set 85:1 93:4 x=3D0. compute j=3D0. do i=3D0,51 compute j=3Dj+1. compute(85:1+i) X(85:1+i)=3D(0.9764*0.9764**j)*D(85:1+i) end do i where D is the series (rara?) Alternatively, you can use a set statement and the intrinsically = defined variable for time as follows: cal 80 1 4 all 97:4 open data house.rat data(format=3Drats) / set 85:1 93:4 x=3D0. set 85:1 97:3 x=3D(0.9764*0.9764**(T-18.)*D The T-18. factor is due to the starting date for the set statement not being the same a the first calendar date. Note that this is mixing integer and real numbers again. You might want to combine the two methods and use a do-loop to create a dummy series that contains a real arithmetic progression from 0. to 51. and use that in the set statement instead of (T-18.). I hope this helps and fixes the problem. Travis > ---------- > From: WOO KAI YIN[SMTP:hkwoo@hkabc.net] > Sent: Monday, August 10, 1998 3:34 AM > To: RATS Discussion List > Subject: RE: >=20 > Dear RATS users, >=20 > Might I ask how to calculate the present value of the following > equation > by use of RATS commands: >=20 >=20 > T-t-1 > Pt =3D gamma * =A3U [gamma^i * Dt+i] + gamma**T-t+1 * PT + constant > i=3D0 >=20 > which in fact is a present value model of stock price. T is a = terminal > price.=20 >=20 > And, I did try to write a program to calculate: >=20 > T-t-1 > X=3Dgamma * =A3U [gamma**i * Dt+i] : > i=3D0 >=20 > cal 80 1 4 > all 97:4 > open data house.rat > data(format=3Drats) / >=20 > do i =3D 0,51 /* from the above, t=3D0, T=3D52, T-t-1=3D51 >=20 > set 85:1+i 97:3 X=3D(0.9764*0.9764**i)*Dt >=20 > end do i; >=20 > I want to generate a new series xt.=20 >=20 > However, I got an error message: >=20 > ## SX22. Expected Type Entry Value, Got REAL Matrix Instead > >>>> x=3D0.9764**i*rara<<<< >=20 > x is a new variable I want to create. rara is a data series in my = rats > file.=20 >=20 > Might I ask how to modify my program, please???? >=20 > I would be grateful if you could give me sincere assistance! >=20 > Regards, >=20 > Kevin > Senior Lecturer > HKSYC >=20 ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: Date: Tue, 11 Aug 1998 22:12:30 +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 ran a program and got the answer such that: print / y ENTRY Y 85:01 -0.369721513027 ENTRY Y 85:02 -0.083721680418 ENTRY Y 85:03 0.199064068018 ENTRY Y 85:04 0.496813101470 Might I ask how to transfer the results for y into text file or winrats file, please??? I feel grateful if you could help me !!! Thanks a lot! Regards, Ky Woo Senior Lecturer, HK SYC ---------- End of message ---------- From: jtebeka@oddo.fr To: "RATS Discussion List" Subject: Hadamard product Date: Wed, 12 Aug 1998 10:39:26 +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 I have 2 matrices A and B of dimension N,N. I would like to calculate A*B, the hadamard product ie element by element product. Thanks for your help. Jacques. ---------- End of message ---------- From: jean-Philip BELLOTTEAU CML/MK To: "RATS Discussion List" Subject: Re: Simulation of an ARIMA(p,d,q) model Date: Wed, 12 Aug 1998 12:20:44 -0700 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mozilla 3.01 [fr] (Win16; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable Dear Jim, To simulate ARMA equations you could simply do : seed 20 clear e y set e 1 200 =3D %ran(1) set y 1 200 =3D 0 set y 1 200 =3D e ; * Yt =3D et <-> Y is a white noise * set y 3 200 =3D -0.7*y(1) + 0.3*y(2) + e ; * (1+0.7B-0.3B=B2)Yt =3D et <-> for a AR(2) com y(1) =3D y(2) =3D %na * set y 1 200 =3D e + 0.7*e(1) - 0.3*e(2) ; * Yt =3D (1+0.7B-0.3B=B2)et <-> for an MA(2) * set y 3 200 =3D -0.7*y(1) + 0.3*y(2) + e + 0.7*e(1) - 0.3*e(2) ; * (1+0.7B-0.3B=B2)Yt =3D (1+0.7B-0.3B=B2)et <-> for an ARMA(2,2) com y(1) =3D y(2) =3D %na * set y 4 200 =3D 0.3*y(1) + y(2) - 0.3*y(3) + e + 0.7*e(1) - 0.3*e(2) ; * (1+0.7B-0.3B=B2)(1-B)Yt =3D (1+0.7B-0.3B=B2)et <-> for an ARIMA(2,1,2) com y(1) =3D y(2) =3D y(3) =3D %na ---------- End of message ---------- From: Norman Morin To: "RATS Discussion List" Subject: Re: Hadamard product Date: Wed, 12 Aug 1998 08:28:32 -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 Here's a quick proc to do HADAMARD products. Warmest regards, Norm ******************************************************* * HADAMARD inmat1 inmat2 outmat * * HADAMARD does element-by-element multiplication * of two matrices. * * Inmat1 and inmat2 must be of the same dimension, * and outmat need not be defined beforehand. * ******************************************************* 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 -------------------------------------------------- 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: "VIJAYA SARATHI N." To: "RATS Discussion List" Subject: VAR with missing data Date: Wed, 12 Aug 1998 20:56:29 GMT+0530 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: IIM, Ahmedabad X-mailer: Pegasus Mail v3.31 (via Mercury MTS (Bindery) v1.40) Dear RATS users: I'm trying to model a set of interest rates (several points on the term structure) as a VAR system. but my data set contains a number of missing observations due to non-trading of some bonds. I need suggestions regarding the methods to handle this missing data, without distorting the term structure. I appreciate your help in this regard. Best regards, sarathi ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: Date: Thu, 13 Aug 1998 01:47:00 +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 run a regression by Least square method with consistent covariance matrix: linreq(robusterror,lag=4) y / # constant x1 x2 x3 x4 However, autocorrelated disturbances can still be found. Could you suggest any more RATS codes to allow for autocorrelated errors in my model, please??? Can I: linreq y / # constant x1 x2 x3 x4 ar{.} I would be grateful if you could help me! Regards, Kai-yin Woo Senior Lecturer, Hong KOng Shue Yan College ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: Date: Thu, 13 Aug 1998 01:53:09 +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, My model is such that: linreq(robusterror,lag=4) y / # constant x1 x2 x3 x4 Might I ask how can I modify the robust.prg in the manual shown below to implement robust estimation and compute the covariance matrix of my model in the above, please?? LINREG(ROBUSTERRORS) PCEXP / RESIDS # CONSTANT PCAID PCINC * COMPUTE C=SQRT(%SEESQ) DEC VECTOR BETA0 * SET SPREAD = 1.0 DO ITERS=1,10 COMPUTE BETA0=%BETA SET SPREAD = SQRT(C**2+RESIDS**2) LINREG(NOPRINT,SPREAD=SPREAD) PCEXP / RESIDS # CONSTANT PCAID PCINC IF %TESTDIFF(BETA0,%BETA)<.0001 BREAK END DO ITERS * * Compute the covariance matrix * SET F = RESIDS/SPREAD SET FPRIME = C**2/SPREAD**3 MCOV(MATRIX=B,LASTREG) / F MCOV(MATRIX=A,LASTREG,NOSQUARE) / FPRIME COMPUTE %XX=%MQFORM(B,INV(A)) LINREG(CREATE,NOSCALE,LASTREG) I feel grateful for your sincere help! REgards, Kevin HK SYC ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: Re: Simulation of an ARIMA(p,d,q) model Date: Wed, 12 Aug 1998 09:43:37 -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 Jim, > > To simulate ARMA equations you could simply do : > > seed 20 > clear e y > set e 1 200 = %ran(1) > set y 1 200 = 0 > set y 1 200 = e ; * > Yt = et <-> Y is a white noise > * > set y 3 200 = -0.7*y(1) + 0.3*y(2) + e ; * I must have missed the first message about ARMA simulations, but I had a couple of comments. First, regarding the method shown above--this will work, but note that you must use { and } for lag notation, and NOT parentheses. As currently written, the SET command above would set all entries of Y equal to -0.7 times the FIRST entry of Y plus 0.3 times the SECOND entry of Y. So, this should read: set y 3 200 = -0.7*y{1} + 0.3*y{2} + e In any case, you may find it easier to just use the SIMULATE command, which is designed to do exactly this sort of thing. See page 14-223 of the RATS manual for an example. 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: Durwood Marshall To: "RATS Discussion List" Subject: RE: Date: Wed, 12 Aug 1998 15:52:16 -0400 (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; charset=US-ASCII X-Mailer: Mercury MTS (Bindery) v1.40 On Thu, 13 Aug 1998, WOO KAI YIN wrote: > My model is such that: > linreq(robusterror,lag=4) y / > # constant x1 x2 x3 x4 > Might I ask how can I modify the robust.prg in the manual shown below to > implement robust estimation and compute the covariance matrix of my > model in the above, please?? HI Woo, You might find the following useful, but note that from Robust Statistics theory the cov(beta) is a function of the influence function and its derivative along with some scaling "factor". Different influence functions exists for different M-estimators, R-estimators and L-estimators. Hence their estimator of the Cov(beta) will be different. The example provided in Rats appears to be an approximation to the Huber M-estimator whose limit as c-->0 is an L-estimator(LAD). All of this may or may not be relevant depending on what you need to do. Check out: Robust Statistics, The Approach Based on Influence Functions, Hampel, Ronchetti, Rousseeuw, and Stahel, Wiley, Isbn: 0-471-82921-8 Good luck. Durwood Marshall-Tufts University Statistical and Research Computing Consulting Tufts Computer and Communications Services 617.628.5000 x2180 : Fax 617.627.3667 : E-Mail: dmarshal@tufts.edu ------------------------------------------------------------------ For all the X-Files fans: http://inetport.com/~one/mlufocvr.html ---------- * * Example ROBUST.PRG APPLIED TO STACKLOSS DATA SET * commented out below: *42 80 27 89 *37 80 27 88 *37 75 25 90 *28 62 24 87 *18 62 22 87 *18 62 23 87 *19 62 24 93 *20 62 24 93 *15 58 23 87 *14 58 18 80 *14 58 18 89 *13 58 17 88 *11 58 18 82 *12 58 19 93 *8 50 18 89 *7 50 18 86 *8 50 19 72 *8 50 19 79 *9 50 20 80 *15 56 20 82 *15 70 20 91 * ALLOCATE 0 21 OPEN DATA 'stackloss.dat' DATA(FORMAT=FREE,ORG=OBS) / LOSS AIR TEMP CONC * * Do a linear regression first * LINREG(ROBUSTERRORS) LOSS / RESIDS # CONSTANT AIR TEMP CONC * COMPUTE C=SQRT(%SEESQ) DEC VECTOR BETA0 * * Start with least squares (spread = constant) * Do 10 iterations or until convergence * SET SPREAD = 1.0 DO ITERS=1,10 COMPUTE BETA0=%BETA SET SPREAD = SQRT(C**2+RESIDS**2) LINREG(NOPRINT,SPREAD=SPREAD) LOSS / RESIDS # CONSTANT AIR TEMP CONC IF %TESTDIFF(BETA0,%BETA)<.0001 BREAK END DO ITERS * * Compute the covariance matrix * SET F = RESIDS/SPREAD SET FPRIME = C**2/SPREAD**3 MCOV(MATRIX=B,LASTREG) / F MCOV(MATRIX=A,LASTREG,NOSQUARE) / FPRIME COMPUTE %XX=%MQFORM(B,INV(A)) LINREG(CREATE,NOSCALE,LASTREG) WRITE 'COVAR_B' %XX ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: RE: present value model Date: Wed, 12 Aug 1998 16:04:34 -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) Just a note of clarification on this topic: The original error, shown below, suffered from a couple of syntax problems: > > set 85:1+i 97:3 X=(0.9764*0.9764**i)*Dt > > > > end do i; > > > > I want to generate a new series xt. > > > > However, I got an error message: > > > > ## SX22. Expected Type Entry Value, Got REAL Matrix Instead > > >>>> x=0.9764**i*rara<<<< Note the SET command: > > set 85:1+i 97:3 X=(0.9764*0.9764**i)*Dt The first problem here is that the start and end dates should come AFTER the series name, not before. For example: set x 85:1+i 97:3 = (etc.) What happened in this case is that RATS took the integer expression (85:1+i) as the series number--technically correct, but clearly not what was intended. It expected the start date next, and got 97:3, which was again syntactically correct. Finally, it was expecting an ending date, but got the real expression (X=(0.9764*0.9764**i)*Dt). That's why the error message indicated that it was expected an entry value (i.e. a date or entry number), but got a real matrix. Also, you must always have a space BEFORE the equal sign in a SET instruction. For example: set x=t is not valid. The reason is that using an assignment operator in an expression IS legal, so RATS interprets "x=t" as an expression returning an integer. As long as X hasn't already been declared as something other than an integer, this will work, and RATS will take the expression as a series number. However, you will get a syntax error, because there's nothing left on the line to parse as the = sign or the rhs expression. Aside from that, I think Travis' suggestions were quite sound. 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: Robert Scott To: "RATS Discussion List" Subject: GMM Estimation of Cox-Ingersoll-Ross Date: Wed, 12 Aug 1998 21:38:14 -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" Content-Transfer-Encoding: quoted-printable X-Mailer: Mercury MTS (Bindery) v1.40 I'm interested in estimating the coefficients for the Cox-Ingersoll-Ross = model of the term structure via a time series estimation. As I = understand, Generalized Method of Moments is the standard for estimating = this. (??) Has anyone run across some lines of code that specify and = estimate the model in this framework? =20 Thanks in advance. Rob Scott ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: Re: present value model Date: Thu, 13 Aug 1998 10:44:45 +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 Estima wrote: > > Just a note of clarification on this topic: > > The original error, shown below, suffered from a couple of syntax > problems: > > > > set 85:1+i 97:3 X=(0.9764*0.9764**i)*Dt > > > > > > end do i; > > > > > > I want to generate a new series xt. > > > > > > However, I got an error message: > > > > > > ## SX22. Expected Type Entry Value, Got REAL Matrix Instead > > > >>>> x=0.9764**i*rara<<<< > > Note the SET command: > > > > set 85:1+i 97:3 X=(0.9764*0.9764**i)*Dt > > The first problem here is that the start and end dates should come > AFTER the series name, not before. For example: > > set x 85:1+i 97:3 = (etc.) > > What happened in this case is that RATS took the integer expression > (85:1+i) as the series number--technically correct, but clearly not > what was intended. It expected the start date next, and got 97:3, > which was again syntactically correct. Finally, it was expecting an > ending date, but got the real expression (X=(0.9764*0.9764**i)*Dt). > That's why the error message indicated that it was expected an entry > value (i.e. a date or entry number), but got a real matrix. > > Also, you must always have a space BEFORE the equal sign in a SET > instruction. For example: > > set x=t > > is not valid. The reason is that using an assignment operator in an > expression IS legal, so RATS interprets "x=t" as an expression > returning an integer. As long as X hasn't already been declared as > something other than an integer, this will work, and RATS will take > the expression as a series number. However, you will get a syntax > error, because there's nothing left on the line to parse as the = > sign or the rhs expression. > > Aside from that, I think Travis' suggestions were quite sound. > > 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 | > ------------------------------------------------------------ Dear Tom, Thank you. Regards, KY Woo ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: robust estimation with AR Date: Thu, 13 Aug 1998 10:53:27 +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 encountered with a problem about robust estimation: linreq(robusterror) y / resid # constant x1 x2 x3 x4 y{1 to 7} I included lagged dependent variables in order to eliminate autocorrelated disturbances. But I found R-square to be 1.0000. Anybody can tell me the reasons and whether my RATS codes are wrong, please???? Regards, KY Woo Senior Lecturer, HK SYC ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE:GMM Date: Thu, 13 Aug 1998 11:04:18 +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, Might I ask the RATS codes for GMM estimation and the contruction of the newey-west covariance matrix , please? I cannot find a chapter in the manual to illustrate it. I feel grateful if you could give me sincere help! Regards, Kevin ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: another problem about robust estimation Date: Thu, 13 Aug 1998 15:01:50 +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 uses, I feel sorry I have another problem about robust estimation: linreq(robusterror) y / resid # constant x1 x2 x3 x4 Exclude # x1 x2 x3 x4 But ## X13. Redundant Restrictions. The number of degree of freedom is lower than the number of restriction. Might I ask what is wrong with my restrictions and how to correct it, please? Regards, KY Woo ---------- End of message ---------- From: Durwood Marshall To: "RATS Discussion List" Subject: RE: robust estimation with AR Date: Thu, 13 Aug 1998 07:42:41 -0400 (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; charset=US-ASCII X-Mailer: Mercury MTS (Bindery) v1.40 On Thu, 13 Aug 1998, WOO KAI YIN wrote: > I encountered with a problem about robust estimation: > linreq(robusterror) y / resid > # constant x1 x2 x3 x4 y{1 to 7} > I included lagged dependent variables in order to eliminate > autocorrelated disturbances. But I found R-square to be 1.0000. > Anybody can tell me the reasons and whether my RATS codes are wrong, > please???? Collinearity....most likely. Durwood Marshall-Tufts University Statistical and Research Computing Consulting Tufts Computer and Communications Services 617.628.5000 x2180 : Fax 617.627.3667 : E-Mail: dmarshal@tufts.edu ------------------------------------------------------------------ For all the X-Files fans: http://inetport.com/~one/mlufocvr.html ---------- End of message ---------- From: Nesmith_T To: "RATS Discussion List" Subject: RE: present value model Date: Thu, 13 Aug 1998 08:57:24 -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 (5.5.1960.3) (via Mercury MTS (Bindery) v1.40) Content-Type: text/plain Dear Tom, Thanks for fixing up my comments. I thought that the dates were reversed but was lazy and did not check. I have been programming Fortran lately and am rusty. I am glad the rest of my advice was o.k. > ---------- > From: Estima[SMTP:estima@estima.com] > Sent: Wednesday, August 12, 1998 6:04 PM > To: RATS Discussion List > Subject: RE: present value model > > Just a note of clarification on this topic: > > The original error, shown below, suffered from a couple of syntax > problems: > > > > set 85:1+i 97:3 X=(0.9764*0.9764**i)*Dt > > > > > > end do i; > > > > > > I want to generate a new series xt. > > > > > > However, I got an error message: > > > > > > ## SX22. Expected Type Entry Value, Got REAL Matrix Instead > > > >>>> x=0.9764**i*rara<<<< > > > Note the SET command: > > > > set 85:1+i 97:3 X=(0.9764*0.9764**i)*Dt > > The first problem here is that the start and end dates should come > AFTER the series name, not before. For example: > > set x 85:1+i 97:3 = (etc.) > > What happened in this case is that RATS took the integer expression > (85:1+i) as the series number--technically correct, but clearly not > what was intended. It expected the start date next, and got 97:3, > which was again syntactically correct. Finally, it was expecting an > ending date, but got the real expression (X=(0.9764*0.9764**i)*Dt). > That's why the error message indicated that it was expected an entry > value (i.e. a date or entry number), but got a real matrix. > > Also, you must always have a space BEFORE the equal sign in a SET > instruction. For example: > > set x=t > > is not valid. The reason is that using an assignment operator in an > expression IS legal, so RATS interprets "x=t" as an expression > returning an integer. As long as X hasn't already been declared as > something other than an integer, this will work, and RATS will take > the expression as a series number. However, you will get a syntax > error, because there's nothing left on the line to parse as the = > sign or the rhs expression. > > Aside from that, I think Travis' suggestions were quite sound. > > 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" To: "RATS Discussion List" Subject: Re: another problem about robust estimation Date: Thu, 13 Aug 1998 09:52:44 -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 WOO KAI YIN wrote: > Dear RATS uses, > > I feel sorry I have another problem about robust estimation: > > linreq(robusterror) y / resid > # constant x1 x2 x3 x4 > > Exclude > # x1 x2 x3 x4 > > But ## X13. Redundant Restrictions. > > The number of degree of freedom is lower than the number of > restriction. > > Might I ask what is wrong with my restrictions and how to correct it, > please? > > Regards, > > KY Woo Dear Kai Yin; Do you get the same problem if you omit the ROBUSTERRORS option? Are there any error messages after the LINREG command? In particular, are the standard errors present and non-zero for all the coefficients? How many Degrees of Freedom do you have in the regression? General suggestion; The most effective way to get help is to post the smallest self-contained program you can make that reproduces the problem. That way, anyone can see precisely what is going on. SvN -- Simon van Norden Professeur invité simon.van-norden@hec.ca or svn@alum.mit.edu or http://www.hec.ca/~p280 (514)340-6781 or fax (514)340-5632 or cell (514)993-0466 Service de l'enseignement de la finance, Ecole des H.E.C. 3000 Cote-Sainte-Catherine, Montreal QC, CANADA H3T 2A7 ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: Date: Fri, 14 Aug 1998 01:29:45 +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 Yesterday, I posted a problem about robust estimation but, following another RATS user's suggestion, I make clear my problem. I feel regretful for causing troubles. The following is my brief program about robust estimation (in order to use consistent covariance matrix to solve the problem of autocorrelated disturances, is it correct?) and wald test. However, I got a message: redundant restrictions. And, the no. of degree of freedom is different from the no. of restrictions. Also, the no. of degree of freedom assigned by RATS is different as the lags option in linreq, i.e. [linreq(robust, LAGS=??)] are different. I show two examples only: lags=7 and lags=6. cal 80 1 4 all 98:2 linreq(robusterrors,lags=7) ppia3 85:1 97:4 resid /*pay attention to the lags=7 / # constant drara drapa drara{1} drapa{1} drara{2} drapa{2} exclude # drara drapa drara{1} drara{2} drapa{1} drapa{2} /* the no. of restriction is 6/ Dependent Variable PPIA3 - Estimation by Least Squares Quarterly Data From 85:01 To 97:04 Usable Observations 52 Degrees of Freedom 45 Centered R**2 0.320281 R Bar **2 0.229652 Uncentered R**2 0.791473 T x R**2 41.157 Mean of Dependent Variable -2.074846098 Std Error of Dependent Variable 1.393749535 Standard Error of Estimate 1.223286540 Sum of Squared Residuals 67.339348199 Durbin-Watson Statistic 0.191068 Q(13-0) 170.682645 Significance Level of Q 0.00000000 Variable Coeff Std Error T-Stat Signif ************************************************************************ 1. Constant -2.09925515 0.37388121 -5.61477 0.00000002 2. DRARA -7.13197880 5.21360301 -1.36796 0.17132589 3. DRAPA -1.14745905 2.36398160 -0.48539 0.62739792 4. DRARA{1} -13.61128197 5.96549544 -2.28167 0.02250893 5. DRAPA{1} 2.57617176 1.37552255 1.87287 0.06108666 6. DRARA{2} -11.32890077 2.46350590 -4.59869 0.00000425 7. DRAPA{2} 3.88134966 2.82496305 1.37395 0.16945812 Correlations of Series RESID Quarterly Data From 85:01 To 97:04 Autocorrelations 1: 0.86112621 0.74725508 0.68068695 0.58918289 0.51314066 0.43671564 7: 0.34979235 0.26366242 0.23995330 0.23552661 0.21285911 0.14294342 13: 0.05060329 Null Hypothesis : The Following Coefficients Are Zero DRARA DRAPA DRARA Lag(s) 1 to 2 DRAPA Lag(s) 1 to 2 ## X13.Redundant Restrictions. Using 4 Degrees, not 6 /* wrong message / Chi-Squared(4)= 58.582784 with Significance Level 0.00000000 ANOTHER EXAMPLE: linreq(robusterrors,lags=6) ppia3 85:1 97:4 resid /*pay attention to the lags =6 / # constant drara drapa drara{1} drapa{1} drara{2} drapa{2} corr(dfc=%nreg) resid 85:1 97:4 exclude # drara drapa drara{1} drara{2} drapa{1} drapa{2} Dependent Variable PPIA3 - Estimation by Least Squares Quarterly Data From 85:01 To 97:04 Usable Observations 52 Degrees of Freedom 45 Centered R**2 0.320281 R Bar **2 0.229652 Uncentered R**2 0.791473 T x R**2 41.157 Mean of Dependent Variable -2.074846098 Std Error of Dependent Variable 1.393749535 Standard Error of Estimate 1.223286540 Sum of Squared Residuals 67.339348199 Durbin-Watson Statistic 0.191068 Q(13-0) 170.682645 Significance Level of Q 0.00000000 Variable Coeff Std Error T-Stat Signif ************************************************************************ 1. Constant 2.09925515 0.42158031 -4.97949 0.00000064 2. DRARA -7.13197880 6.51058224 -1.09544 0.27332213 3. DRAPA -1.14745905 1.70194695 -0.67420 0.50018177 4. DRARA{1} -13.61128197 6.03538529 -2.25525 0.02411786 5. DRAPA{1} 2.57617176 1.17307734 2.19608 0.02808621 6. DRARA{2} -11.32890077 3.87528991 -2.92337 0.00346266 7. DRAPA{2} 3.88134966 2.61896560 1.48202 0.13833596 Correlations of Series RESID Quarterly Data From 85:01 To 97:04 Autocorrelations 1: 0.86112621 0.74725508 0.68068695 0.58918289 0.51314066 0.43671564 7: 0.34979235 0.26366242 0.23995330 0.23552661 0.21285911 0.14294342 13: 0.05060329 Null Hypothesis : The Following Coefficients Are Zero DRARA DRAPA DRARA Lag(s) 1 to 2 DRAPA Lag(s) 1 to 2 ## X13. Redundant Restrictions. Using 3 Degrees, not 6 /* wrong message/ Chi-Squared(3)= 9.643357 with Significance Level 0.02185418 Might I ask any problem with the RATS program and how to modify it, please? [I'm running Durlauf and Hooker(1994)'s bubble and misspecification test, so the no. of restriction required from the model cannot be changed] Also, you can find that autocorrelation remain in the robust estimation. Might I ask how to solve the autocorrelation in my model? ignore robusterrors but add ARMA terms?? or use another methods to create another convariance matrix??? I would be grateful if anybody could give me sincere assistance!!! With best wishes, Kai-yin Woo Senior Lecturer, HK SYC ---------- End of message ---------- From: "Estima" To: "RATS Discussion List" Subject: RE: Date: Thu, 13 Aug 1998 15:12:13 -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) > Null Hypothesis : The Following Coefficients Are Zero > DRARA > DRAPA > DRARA Lag(s) 1 to 2 > DRAPA Lag(s) 1 to 2 > ## X13.Redundant Restrictions. Using 4 Degrees, not 6 /* wrong message / > Chi-Squared(4)= 58.582784 with Significance Level 0.00000000 > See page 6-4 in the manual for a brief explanation of this situation. It arises when the [R*inv(tr(x)*x)*tr(R)] matrix is not invertible. As suggested by the error message, this usually means that two or more of the restrictions you are imposing are, in fact, redundant. This is really a warning message, not an error--RATS is still doing the test, but is telling you that is dropping the redundant restrictions and adjusting the degrees of freedom accordingly. Sincerely, Tom Maycock Estima ------------------------------------------------------------ | Estima | Sales: (800) 822-8038 | | P.O. Box 1818 | Support: (847) 864-1910 | | Evanston, IL 60204-1818 | Fax: (847) 864-6221 | | USA | estima@estima.com | | | http://www.estima.com | ------------------------------------------------------------ ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE: problem re-examined Date: Fri, 14 Aug 1998 09:33:42 +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, Last time, I asked a question about unity of R-sqaure and the reply was collinearity given by a kind RATS user. I feel grateful for the sincere help. Since my question was not clear, I re-post my program and output here again: cal 80 1 4 all 98:2 diff rapa / drapa diff rara / drara linreq(robust,lags=4) ppia2 85:1 97:4 resid # constant drara drapa drara{1} drapa{1} ppia2{1 to 4} corr resid 85:1 97:4 exclude # drara drapa drara{1} drapa{1} Dependent Variable PPIA2 - Estimation by Least Squares Quarterly Data From 85:01 To 97:04 Usable Observations 48 Degrees of Freedom 39 Total Observations 52 Skipped/Missing 4 Centered R**2 1.000000 R Bar **2 1.000000 Uncentered R**2 1.000000 T x R**2 48.000 Mean of Dependent Variable -0.394956816 Std Error of Dependent Variable 0.791184965 Standard Error of Estimate 0.000000000 Sum of Squared Residuals 7.02819e-25 Durbin-Watson Statistic 1.134745 Q(13-0) 56.828783 Significance Level of Q 0.00000019 Variable Coeff Std Error T-Stat Signif ************************************************************************ 1. Constant -4.7850e-14 0.0000 0.00000 0.00000000 2. DRARA 6.2677e-14 0.0000 0.00000 0.00000000 3. DRAPA 1.0000 0.0000 0.00000 0.00000000 4. DRARA{1} 1.0000 0.0000 0.00000 0.00000000 5. DRAPA{1} -1.0427 0.0000 0.00000 0.00000000 6. PPIA2{1} 2.0427 0.0000 0.00000 0.00000000 7. PPIA2{2} -1.0427 1.4138e-12 -7.37562e+11 0.00000000 8. PPIA2{3} -1.6052e-12 1.0420e-12 -1.54046 0.12344927 9. PPIA2{4} 3.2924e-13 0.0000 0.00000 0.00000000 Correlations of Series RESID Quarterly Data From 85:01 To 97:04 Autocorrelations 1: 0.3692676 0.3964018 0.5322162 0.4209840 0.2540953 0.2336790 7: 0.2922258 0.0708404 0.0395721 0.0508002 0.0282164 0.0151864 13: -0.2574075 Null Hypothesis : The Following Coefficients Are Zero DRARA DRAPA DRARA Lag(s) 1 DRAPA Lag(s) 1 Chi-Squared(4)= 6.138654e+25 with Significance Level 0.00000000 PPia2 is the perfect foesight price of rapa. From the above results, standard errors converges to zero. But multi-collinearity causes wider confidence interval and high R-sqaure but insignificant t-ratio. It seems inconsistent with my present case. [The above problem appears as I added lagged dependent terms. Without lagged dependent terms, the results are quite normal with R-square less than unity] Anybody can show me the reasons for my problems and tell me how to modify the estimation procedure, please??? Thanks a lot!!! Regards, Kai-yin Woo Senior Lecturer, HK SYC ---------- End of message ---------- From: To: "RATS Discussion List" Subject: Codependence analysis Date: Fri, 14 Aug 1998 07:28:42 +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; charset=US-ASCII; name="Codependence" Content-Transfer-Encoding: 7bit X-Mailer: Mercury MTS (Bindery) v1.40 Dear RATS users, do you know if there are any RATS procedures available for codependence analysis in the spirit of Vahid & Engle , Engle & Kozicki. I am doing a cointegration/codepence analysis on quarterly gross domestic products. Regards Jani Koskinen ---------- End of message ---------- From: "Simon.Van-Norden" To: "RATS Discussion List" Subject: Re: problem re-examined Date: Fri, 14 Aug 1998 09:15:52 -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 Kai Yin; This output makes your problem clearer. The diagnosis of colinearity seems correct. In particular, it appears to the case that PPIA2 - 2* PPIA2{1} + PPIA2{2} = DRAPA + (DRARA{1}-DRAPA{1}) + 0.0427*(DPPIA2{1}-DRAPA{1}) To put it more simply, if we define DPPIA2 = PPIA2 - PPIA2{1} and DDPPIA2 = DPPIA2 - DPPIA2{1}, then DDPPIA2 = DRAPA + (DRARA{1}-DRAPA{1}) + 0.0427*(DPPIA2{1}-DRAPA{1}) I suspect this relationship is true by definition and comes from the way you constructed your variables. As for correcting this problem, would you have access to any other econometric software? It might be better suited to your needs than RATS, which seems to be giving you considerable trouble. SvN -- Simon van Norden Professeur invité simon.van-norden@hec.ca or svn@alum.mit.edu or http://www.hec.ca/~p280 (514)340-6781 or fax (514)340-5632 or cell (514)993-0466 Service de l'enseignement de la finance, Ecole des H.E.C. 3000 Cote-Sainte-Catherine, Montreal QC, CANADA H3T 2A7 ---------- End of message ---------- From: Durwood Marshall To: "RATS Discussion List" Subject: RE: problem re-examined Date: Fri, 14 Aug 1998 10:14:47 -0400 (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; charset=US-ASCII X-Mailer: Mercury MTS (Bindery) v1.40 On Fri, 14 Aug 1998, WOO KAI YIN wrote: > 8. PPIA2{3} -1.6052e-12 1.0420e-12 -1.54046 0.12344927 > >From the above results, standard errors converges to zero. But > multi-collinearity causes wider confidence interval and high R-sqaure > but insignificant t-ratio. It seems inconsistent with my present case. Hi again, As an aside, this may be true but is not automatic, unless PPIA2{3} is explicitly cited in a multicollinearity Variance-decompostition analysis. See: Belsley, Kuh and Welsch, Regression Diagnostics; Identifying Influential Data and Sources of Collinearity, Wiley, ISBN 0-471-05856-4 > [The above problem appears as I added lagged dependent terms. Without > lagged dependent terms, the results are quite normal with R-square less > than unity] > > Anybody can show me the reasons for my problems and tell me how to > modify the estimation procedure, please??? IMHO, It seems that if theory demands an a priori preference for this model and inference is your goal, then you may be somewhat in a bind. If you are only concerned with prediction and not inference per se and can tolerate alternate models and error stuructures, try some standard approaches like: Standardize each column of X about it's mean to ensure approximate prior independence, or removing columns, or shrinkage estimation and ridge regression methods, or transformations of one or more r.h.s. variables. Most transformational approaches work best after one knows the nature and degree of collinearity via a multicollinearity Variance-decompostition analysis. Your options depend somewhat on the goals. Good luck. Durwood Marshall-Tufts University Statistical and Research Computing Consulting Tufts Computer and Communications Services 617.628.5000 x2180 : Fax 617.627.3667 : E-Mail: dmarshal@tufts.edu ------------------------------------------------------------------ For all the X-Files fans: http://inetport.com/~one/mlufocvr.html ---------- End of message ---------- From: "VIJAYA SARATHI N." To: "RATS Discussion List" Subject: VAR with missing data Date: Fri, 14 Aug 1998 20:27:19 GMT+0530 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: IIM, Ahmedabad X-mailer: Pegasus Mail v3.31 (via Mercury MTS (Bindery) v1.40) Hello ! I'm trying to estimate a VAR with several points on the term structure. But i've missing points in my data set due to non-availability of trading data for some of the interest rates. I want to fill in the missing values by infering the information available from the other points on the term structure for the same observation. is there a code or an algorithm which will help me do this? I appreciate your help in this regard. Best Regards, Sarathi Doctoral Candidate Indian Institute of Management Ahmedabad, India. ---------- End of message ---------- From: Marco Hell To: "RATS Discussion List" Subject: LM-Test from Agiakloglou and Newbold(1994) Date: Fri, 21 Aug 1998 12:03:28 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Comcity X-Mailer: Mozilla 4.04 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Dear all, Does anyone have a procedure which allows RATS to estimate the fractional integration parameter d with the lagrange multiplier test, see e.g. Agiakloglou and Newbold(1994) ? Thanks, Marco. -- \\\|||||/// \\ - - // ( @ @ ) +------oOOo-(_)-oOOo---- +----------------------------------------+ | | Marco Hell | | Have a nice day ! | Tel: 0049 (0)431/80.67.66 | | | e-Mail: marco.hell@ki.comcity.de | +---------------Oooo-----+----------------------------------------+ oooO ( ) ( ) ) / \ ( (_/ \_) ---------- End of message ---------- From: szuniga@entelchile.net (SERGIO ZUNIGA JARA) To: "RATS Discussion List" Subject: Re: GMM Estimation of Cox-Ingersoll-Ross Date: Fri, 21 Aug 1998 21:23:05 -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; Content-Transfer-Encoding: 8bit X-Mailer: Microsoft Outlook Express 4.72.3110.5 (via Mercury MTS (Bindery) v1.40) -----Original Message----- From: Robert Scott To: RATS Discussion List Date: Miércoles 12 de Agosto de 1998 11:34 PM Subject: GMM Estimation of Cox-Ingersoll-Ross I'm interested in estimating the coefficients for the Cox-Ingersoll-Ross model of the term structure via a time series estimation. As I understand, Generalized Method of Moments is the standard for estimating this. (??) Has anyone run across some lines of code that specify and estimate the model in this framework? Thanks in advance. Rob Scott In estimate the CIR model you can use Maximun Likelihood (Brown and Dybvig, JF july 1986), or Generalized Method of Moments (Chan et al. JF, july 1992). I am working in some codes for do this. Regards, \\ - - // ( @ @ ) +--oOOo-(_)-oOOo-- +------------------------------------------+ | Sergio Zuniga | U. Catolica de Norte - Chile | | Tel: 56-51-327248 | szuniga@entelchile.net | +------------------------------+----------------------------------------+ ---------- End of message ---------- From: WOO KAI YIN To: "RATS Discussion List" Subject: RE:maximal eigenvalue Date: Sun, 23 Aug 1998 01:37:22 +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 CATS users, Might I ask how to get the Maximal eigenvalue statistics at 5% significance level? For my bi-variate model with a constant restricted into the cointegration space (without seasonal and non-seasonal dummies), L-max90 is always 10.29 for Ho:r=0 and 7.5 for Ho:r=1. I cannot find them in Johansen and Juselius(1990) and Osterwald-Lenum(1992) since no Dt is included in my model. DisCo can only simulate trace statistics. Regards, KY Woo ---------- End of message ---------- From: Marco Hell To: "RATS Discussion List" Subject: Fractional Integration Date: Mon, 24 Aug 1998 21:24:17 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Comcity X-Mailer: Mozilla 4.04 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Dear all, unfortunately the popular test of fractional intgration, like GDP, RMM or LM, have no power against shifts in the mean. Is there a test for fractional integration in the presence of structural changes? Marco Hell. CAU-Kiel, Germany -- \\\|||||/// \\ - - // ( @ @ ) +------oOOo-(_)-oOOo---- +----------------------------------------+ | | Marco Hell | | Have a nice day ! | Tel: 0049 (0)431/80.67.66 | | | e-Mail: marco.hell@ki.comcity.de | +---------------Oooo-----+----------------------------------------+ oooO ( ) ( ) ) / \ ( (_/ \_) ---------- End of message ---------- From: "clecourt" To: "RATS Discussion List" Subject: The Pearson goodness-of-fit test Date: Wed, 26 Aug 1998 18:06:46 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Pro Version 3.0.1 (32) [F] (via Mercury MTS (Bindery) v1.40) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Dear all, Does anyone have a procedure which allows RATS to estimate the Pearson goodness-of-fit test or a similar test? Thanks, Christelle. \\\|||||/// \\ - - // ( @ @ ) +------oOOo-(_)-oOOo---- +----------------------------------------+ | Christelle Lecourt | Centres de Recherche: MEDE ET CADRE | | Economiste | Tel: 0032 (0)85/25.04.06 | | UNIVERSITE DE LILLE II | e-Mail: clecourt@mailsc.univ-lille2.fr | +---------------Oooo-----+----------------------------------------+ oooO ( ) ( ) ) / \ ( (_/ \_) ---------- End of message ---------- From: Ying-Man Chung To: "RATS Discussion List" Subject: overshooting Date: Wed, 26 Aug 1998 19:12:10 -0700 (PDT) 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 CATS users, I implemented a partial vector error correction model: delta Y = A * lagged delta Y + B* delta X + C * lagged delta X + error-correction term + e But I found coefficient(short-run impact multiplier) of the delta X is greater than one. Is it a normal situation? can I name it as overshooting? Thank you for anybody who is eager to give me reply!!! With best wishes, Sarah _________________________________________________________ DO YOU YAHOO!? Get your free @yahoo.com address at http://mail.yahoo.com ---------- End of message ---------- From: laurent ferrara CML RRC To: "RATS Discussion List" Subject: arccosine Date: Fri, 28 Aug 1998 16:25:16 -0700 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mozilla 3.01 [fr] (Win16; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: quoted-printable Dear RATS users, Does anyone have a procedure that will compute arccos(x) or arcsin(x) for -1<=3Dx<=3D1. Thanks, Laurent Ferrara RATP, D=E9partement Commercial. ---------- End of message ---------- From: "Philip Hsu" To: "RATS Discussion List" Subject: Expanded EGARCH model Date: Sat, 29 Aug 1998 19:48:46 +0800 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Microsoft Internet Mail 4.70.1155 (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=BIG5 Content-Transfer-Encoding: 7bit Dear RATS users, I'm going to use the expanded EGARCH model of Pericli and Koutmos (The Journal of Futures Market, 17(8),1997), but I don't know if I can do it by using the GARCH procedure in GARCHN.SRC. In the GARCH procedure we can add for the conditional variance with up to thirteen exogenous variables: H(t)=EXP(c+q1*G(t-1)+p1*log(H(t-1))+v1*vx1(t)+...+v13*vx13(t)) But the expended EGARCH model fo Pericli and Koutmos is: H(t)=EXP(c+(q1+alpha1*Dummy(t))*G(t-1)+(p1+phi1*Dummy(t))*log(H(t-1))) Can I direct use the vx1(t) as the Dummy(t) in the case of Pericli and Koutmos? Any assistance would be appreciated. Thanks in advance, Philip Hsu ---------- End of message ---------- From: Marco Hell To: "RATS Discussion List" Subject: CATS with fractional correction Date: Fri, 28 Aug 1998 19:00:26 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Comcity X-Mailer: Mozilla 4.04 [en] (Win95; I) (via Mercury MTS (Bindery) v1.40) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Dear all, I found some papers about multivariate fractional cointegration e.g. Lyhagen (1998), ML-estimation of the multivariate fractional cointegration model, No.233, Stockholm School of Economics. Does anyone know if there exists a program like CATS adjusted for fractional cointegration? Thank you. Marco Hell CAU-Kiel -- \\\|||||/// \\ - - // ( @ @ ) +------oOOo-(_)-oOOo---- +----------------------------------------+ | | Marco Hell | | Have a nice day ! | Tel: 0049 (0)431/80.67.66 | | | e-Mail: marco.hell@ki.comcity.de | +---------------Oooo-----+----------------------------------------+ oooO ( ) ( ) ) / \ ( (_/ \_) ---------- End of message ---------- From: Rumi Masih To: "RATS Discussion List" Subject: Logit as a SUR Date: Mon, 31 Aug 1998 14:51:55 +0100 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" Dear RATs Users: I have 2 separate questions: (1): Is there any way I can estimate a set of logit regressions in RATs and then set it up as a SUR just for sake of estimation? RATs will not allow me to, for example, use: LGT(equation=EQ) LY as this only allows LINREG and then one may set up a SUR using SUR(VCV) 2 # EQ etc. (2): Does RATS allow the estimation of unbalanced panels (either fixed or random effects) Thanks for your assistance. Rumi *************************************************************** Rumi Masih, Faculty of Economics, University of Cambridge, England Address for Correspondence: Rumi Masih, Wolfson College, Cambridge CB3 9BB, U.K. Fax: +44-1223-335-908; E-Mail: rm212@cus.cam.ac.uk Web Page: http://www.ozemail.com.au/~masih/rumi/ *************************************************************** ---------- End of message ----------