From: goux@univ-lyon2.fr To: "RATS Discussion List" Subject: GARCH Date: Fri, 31 May 1996 17:48:24 +0200 (MET DST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Mailer: Mercury MTS v1.21 I TRIED TO ESTIMATE AN ARCH(3) AND A GARCH(P,Q) (WITH VALUES AS SMALL AS 1) AS DESCRIBED IN THE RATS VERSION 4 USER'S MANUAL (IE WITH MAXIMISE (BHHH,ITERATIVE)) BUT THE MAXIMISATION STOPPED AFTER ONE ITERATION AND THE RESULTS ARE THE INITIAL VALUES OF THE PARAMETERS (I TRIED DIFFERENT VALUES WITH THE SAME RESULT). ANY HELP WILL BE THE WELCOME !!! ---------- End of message ---------- From: Scott Frame To: "RATS Discussion List" Subject: Penalty Functions Date: Tue, 4 Jun 1996 11:55:48 -0400 (EDT) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: ELM [version 2.4 PL23] (via Mercury MTS v1.21) Content-Type: text I am estimating a bivariate GARCH (1,1) model and would like to incorporate a penalty function, which would guarantee that my parameter estimates for the conditional variance equations will be nonnegative. I have set up my RATS program following Rob Trevor's article in a previous RATS newsletter. I believe that he mentions the use of such functions for the purpose stated above -- so I ahve reason to believe that someone has programed this. I am estimating my system using the maximize instruction. Any hints would be greatly appreciated. Thank you, Scott Frame University of Georgia ---------- End of message ---------- From: Rob Trevor To: "RATS Discussion List" Subject: RATS-L Status Date: Thu, 6 Jun 1996 10:32:21 +1000 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Mailer: Mercury MTS v1.21 Dear Colleagues As many of you were aware from my post to this list, I was away from mid-April to mid-May. Unfortunately, during that time this Schools system administrators upgraded both the hardware and the software which runs this list. In doing so they somehow managed to delete or misplace the list of subscribers. The net result was that most of you (including me, about 120 of us) suddenly were no longer part of RATS-L. The last message I received from the list was sent on 25 Apr 96 03:50:58 GMT+1000. The problem seems to have arisen sometime between then and May 2. Since then, only the twenty or so who had newly subscribed to the list have been getting messages. The noticeable effects this had were that (a) the list seemed rather quiet to most of us and (b) those who tried to post messages may have been told that they were not subscribed to the list, and hence unable to post. I was alerted to this problem while still in the US, and have been trying to get the system administrators to restore the old list from the backup tapes and merge it with the new one. Unfortunately, due to some staff difficulties, they only completed this task yesterday. (I have been tempted to shift the list administration to one of my own machines, but that would involve changing all the email addresses which you use to communicate with the list, which would only add to the current confusion. I hope that I don't regret this decision.) I hope that the list is now back to normal. Its possible that some subscribers have been lost, but unfortunately there is nothing I can do about that. Its also possible that some of you are now subscribed twice. (If you get two copies of this message, then I suggest that you try sending an 'unsub' message as outlined in the Welcome document you received when you joined the list.) For those of you, who like me missed the discussion since late April, I'll send the relevant parts of the archive in a separate message. Rob Trevor rats-l-owner@efs.mq.edu.au ----------------------------------------------------------------------- Rob Trevor Associate Professor Centre for Studies in Money, Banking and Finance Macquarie University, NSW, AUSTRALIA 2109 Internet: robt@efs.mq.edu.au VOICE: +61 (2) 850-8539 FAX: +61 (2) 850-7281 or +61 (2) 418-1847 ----------------------------------------------------------------------- ---------- End of message ---------- From: Rob Trevor To: "RATS Discussion List" Subject: RATS-L: Archives of Missed Messages Date: Thu, 6 Jun 1996 10:32:25 +1000 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Mailer: Mercury MTS v1.21 Here are the archives from the period when most of us were deleted from the list. Please read them if you didn't receive the originals. Rob rats-l-owner@efs.mq.edu.au ---------------------------------------------------- From: goux@univ-lyon2.fr To: "RATS Discussion List" Subject: invitation Date: Fri, 10 May 1996 11:54:58 +0200 (MET DST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Mailer: Mercury MTS v1.21 Hi, I would be very glad to invite an econometrician, specialist in RATS + money and finance, one or two months, during the next academic year 1996-1997. The wage could be of 15000-20000 french francs per month. Lectures : 16 hours / month Please could you send me a response as quick as possible with a CV. Jean-Francois GOUX University of LYON 2 Department "Monnaie-Finance-Banque" 16 quai claude Bernard 69365 Lyon Cedex 7 France ---------- End of message ---------- From: lmahens@vnet3.vub.ac.be (Luc M.A. Hens) To: "RATS Discussion List" Subject: Durbin-Watson test P-values Date: Wed, 15 May 1996 17:34:51 +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 v1.21 Dear members of the list, Is there a way in RATS to avoid the Durbin-Watson ***bounds*** test and get P-values for the Durbin-Watson statistic instead (like in SHAZAM)? Thanks in advance! Luc Hens ============================================= Luc M.A. Hens assistant professor of economics Vesalius College, Vrije Universiteit Brussel Pleinlaan 2, B-1050 Brussels Belgium phone +32-2-629 20 61 fax +32-2-629 20 60 e-mail lmahens@vnet3.vub.ac.be ============================================= ---------- End of message ---------- From: nikolaus@pc126.isdn.uni-konstanz.de (Nikolaus K.A. Laeufer) To: "RATS Discussion List" Subject: WIN-RATS and WABI Date: Fri, 17 May 1996 10:04:53 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Z-Mail (3.2.1 10apr95) (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Dear Rats-People, Has anyone of you tried to run the Windows-Version of Rats (WIN-RATS 16 bit-Version) using the WABI-Interface of Solaris 2.4 (e.g. on a sun-sparcstation 10)? Does it work or not? I would appreciate your response. Thanks Nikolaus ---------- End of message ---------- From: "L.Saldanha" To: "RATS Discussion List" Subject: Re: WIN-RATS and WABI Date: Fri, 17 May 96 09:49:20 BST Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mercury MTS v1.21 Dear Nikolaus, I am sorry , I have not used this version. Bye L. Saldanha. ---------- End of message ---------- From: "Christopher F. Doffing" To: "RATS Discussion List" Subject: MLE of Kalman Filter Hyperparameters in RATS 4.01 Date: Wed, 22 May 1996 11:14:19 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Version 1.4.4 (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" I've got this set of frml's to do maximum likelihood estimation of the hyperparameters of a simple random-coefficient Kalman Filter model. I've done this same problem in Excel, so I know it can be done. However, since the frml's depend on the value of a frml not specified until later in the structure, RATS can't handle it. For instance, "frml x10 = x1", but "frml x1 = x10 + kg * v1" doesn't come until later; they're "circular" in that way. So RATS gives me an error saying: ## SX11. Identifier X1 is Not Recognizable. Incorrect Option Field or Parameter Order? >>>>frml x10 = x1<<<< I can't figure out how to "get into the loop" here. (BTW, I need to keep the frml's in this general structure so that they can be added to later for the cases with exogenous variables and a transition matrix.) Can anyone help with this? all 0 121 open data d:\download\kalman\test.rat * "test.rat" is a one-column observation set of length 121. data(format=rats) nonlin s2 qq eval s2 = 0.08 * setting my initial values for these hyperparameters. eval qq = 0.04 smpl 2 121 frml x10 = x1 frml v1 = yt - x10 frml P10 = P1 + qq frml f1 = P10 + s2 frml kg = P10 / f1 frml P1 = (1 - kg) * P10 frml x1 = x10 + kg * v1 frml likely = -log(f1) - (v1**2)/f1 maximize(method=bhhh,trace,iterations=100) likely ---------- End of message ---------- From: "Christopher F. Doffing" To: "RATS Discussion List" Subject: MLE of Kalman Filter Hyperparameters in RATS 4.01 Date: Wed, 22 May 1996 11:24:41 -0500 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Version 1.4.4 (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" I've got this set of frml's to do maximum likelihood estimation of the hyperparameters of a simple random-coefficient Kalman Filter model. I've done this same problem in Excel, so I know it can be done. However, since the frml's depend on the value of a frml not specified until later in the structure, RATS can't handle it. For instance, "frml x10 = x1", but "frml x1 = x10 + kg * v1" doesn't come until later; they're "circular" in that way. So RATS gives me an error saying: ## SX11. Identifier X1 is Not Recognizable. Incorrect Option Field or Parameter Order? >>>>frml x10 = x1<<<< I can't figure out how to "get into the loop" here. (BTW, I need to keep the frml's in this general structure so that they can be added to later for the cases with exogenous variables and a transition matrix.) Can anyone help with this? all 0 121 open data d:\download\kalman\test.rat * "test.rat" is a one-column observation set of length 121. data(format=rats) nonlin s2 qq eval s2 = 0.08 * setting my initial values for these hyperparameters. eval qq = 0.04 smpl 2 121 frml x10 = x1 frml v1 = yt - x10 frml P10 = P1 + qq frml f1 = P10 + s2 frml kg = P10 / f1 frml P1 = (1 - kg) * P10 frml x1 = x10 + kg * v1 frml likely = -log(f1) - (v1**2)/f1 maximize(method=bhhh,trace,iterations=100) likely ---------- End of message ---------- From: "Ward, Bert D" To: "RATS Discussion List" Subject: Quasi-MLE Date: Wed, 29 May 1996 14:10:25 +1200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Lincoln University X-Mailer: Pegasus Mail v3.31 (via Mercury MTS v1.21) Content-transfer-encoding: 7BIT If anyone has coded a RATS procedure for computing the robust standard errors for the parameters of a an ARCH model as discussed in pages 663-664 of Hamilton(1994), I'd be most grateful if they would contact me. Thanks very much. Kind regards, Bert D. Ward Senior Lecturer in Economics Dept. of Economics & Marketing PO Box 84 Lincoln University Canterbury NEW ZEALAND Telephone (64)(3) 325 2811 extn 8261 FAX (64)(3) 325 3847 ---------- End of message ---------- From: dwatt@bank-banque-canada.ca (David Watt) To: "RATS Discussion List" Subject: Gamma function for Rats Date: Thu, 30 May 1996 09:07:31 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mercury MTS v1.21 Hello Does anyone know of a gamma function for RATS? I am trying to do some work with a t-distribution but have been held up finding a gamma function. I am currently using Version 4.0 on a Unix Workstation under Solaris. Any suggestions would be greatly appreciated. David Watt Bank of Canada. dwatt@bank-banque-canada.ca ---------- End of message ---------- From: nikolaus@pc126.isdn.uni-konstanz.de (Nikolaus K.A. Laeufer) To: "RATS Discussion List" Subject: Re: Gamma function for Rats Date: Fri, 31 May 1996 11:12:14 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Z-Mail (3.2.1 10apr95) (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Hello David Watt, I am sorry that I cannot help you with the gamma function. However, I find it very interesting to discover in you someone who uses the same system as I (Unix worksation under Solaris). I have been interested in installing RATS on this system (Sparcstation 10) and have shied away from doing so using the Unix-Version of RATS. Instead I have been phantasizing about using the RATS Windows version under WABI. Unfortunately I could not find anyone who is able to tell me whether that would work at all. Even ESTIMA could not help me. So I have returned to the idea of using the Unix version. It would greatly help me in my decision to purchase the unix version of RATS if you could tell me shortly how you handle RATS graphics display unter Unix. (I do know the respective 'description' of ESTIMA in its adds but it was exactly that description which turned me off.) In case you have any further related experience that you consider worthwhile to communicate I would highly appreciate to obtain it. Since I have RATS installed on a Pentium PC I need good reasons for buying the UNIX version as an additional version of RATS. Thank you for reading this long message. I look forward to your response. Nikolaus ---------- End of message ---------- From: goux@univ-lyon2.fr To: "RATS Discussion List" Subject: GARCH Date: Fri, 31 May 1996 17:48:24 +0200 (MET DST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Mailer: Mercury MTS v1.21 I TRIED TO ESTIMATE AN ARCH(3) AND A GARCH(P,Q) (WITH VALUES AS SMALL AS 1) AS DESCRIBED IN THE RATS VERSION 4 USER'S MANUAL (IE WITH MAXIMISE (BHHH,ITERATIVE)) BUT THE MAXIMISATION STOPPED AFTER ONE ITERATION AND THE RESULTS ARE THE INITIAL VALUES OF THE PARAMETERS (I TRIED DIFFERENT VALUES WITH THE SAME RESULT). ANY HELP WILL BE THE WELCOME !!! ---------- End of message ---------- From: Scott Frame To: "RATS Discussion List" Subject: Penalty Functions Date: Tue, 4 Jun 1996 11:55:48 -0400 (EDT) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: ELM [version 2.4 PL23] (via Mercury MTS v1.21) Content-Type: text I am estimating a bivariate GARCH (1,1) model and would like to incorporate a penalty function, which would guarantee that my parameter estimates for the conditional variance equations will be nonnegative. I have set up my RATS program following Rob Trevor's article in a previous RATS newsletter. I believe that he mentions the use of such functions for the purpose stated above -- so I ahve reason to believe that someone has programed this. I am estimating my system using the maximize instruction. Any hints would be greatly appreciated. Thank you, Scott Frame University of Georgia ---------- End of message ---------- ---------- End of message ---------- From: nikolaus@pc126.isdn.uni-konstanz.de (Nikolaus K.A. Laeufer) To: "RATS Discussion List" Subject: WIN-RATS-WABI Date: Thu, 6 Jun 1996 11:18:56 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Z-Mail (3.2.1 10apr95) (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Dear Rats-People, Has anyone of you tried to run the Windows-Version of Rats (WIN-RATS 16 bit-Version) using the WABI-Interface of Solaris 2.4 (e.g. on a sun-sparcstation 10)? Does it work or not? I would appreciate your response. Thanks Nikolaus P.S. I had posted this message before. Given the recent problems with the discussion list I consider it appropriate to repost it. Excuse me for making you perhaps read something twice. ---------- End of message ---------- From: David Prieul <101566.244@CompuServe.COM> To: "RATS Discussion List" Subject: Re: MLE of Kalman Filter Date: 06 Jun 96 05:53:34 EDT Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mercury MTS v1.21 The reason is the same as for ARCH-GARCH estimation, you should first DEFINE as a SERIES your FRML where you will store previous values, then write your FRML, and, when writing the ML equation, define the FRML equal to the SERIES such as: DEFINE SERIES H FRML� FRML� FRML HETERO = ALPHA + BETA*H{1} FRML MMXL = (H=HETERO), AND THE LOG-LIKELIHOOD MAXIMIZE(RECURSIVE,�.) ---------- End of message ---------- From: David Prieul <101566.244@CompuServe.COM> To: "RATS Discussion List" Subject: Re:Gamma function in RATS Date: 06 Jun 96 06:02:07 EDT Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mercury MTS v1.21 As far as I know, the %LNGAMMA function preprogrammed in RATS 4.01 can do the job for you. I had anyway the same problem as you have for estimating GARCH-class models with a conditional Student-t distribution, and my only solution was to take my Abramowitz and Stegun (handbook of Mathematical Functions) and to write an approximation to this function. David Prieul ---------- End of message ---------- From: David Prieul <101566.244@CompuServe.COM> To: "RATS Discussion List" Subject: Re:GARCH Date: 06 Jun 96 06:05:31 EDT Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mercury MTS v1.21 Very standard problem: - Increase the criterion value in NLPAR - Refine initial values - Refine initial values with MAXIMIZE(SIMPLEX) before using BHHH or BFGS - There could be some problem in your code I hope it will help David Prieul ---------- End of message ---------- From: wlee@pop.hbs.edu To: "RATS Discussion List" Subject: Re: RATS-L: Archives of Missed Messages Date: Thu, 6 Jun 1996 09:18:27 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Pro Version 2.1.2 (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Dear Scott: I have been discussing the problem of restricting the parameters in the GARCH equation to be positive with several others. We cannot figure out an easy way to do that in RATS neither. However, you may follow the suggestion of Engle and Kroner (in their BEKK multivariate GARCH paper published in 95 or 96 issue of Econometric Theory) in dealing this problem, i.e., estimating the square roots of the parameters: h(t) = sqrt(a0) + sqrt(a1)*h(t-1) + sqrt(a2)*e(t-1)^2 One drawback of this trick is the fact that we can only get the standard error of the parameter of ai's. I hope this will help. If anyone has a better alternative approach, please post. > >I am estimating a bivariate GARCH (1,1) model and would like to >incorporate a penalty function, which would guarantee that my >parameter estimates for the conditional variance equations will be >nonnegative. I have set up my RATS program following Rob Trevor's >article in a previous RATS newsletter. I believe that he mentions >the use of such functions for the purpose stated above -- so I ahve >reason to believe that someone has programed this. I am estimating >my system using the maximize instruction. Any hints would be greatly >appreciated. > >Thank you, > >Scott Frame >University of Georgia > >---------- End of message ---------- > > > > **************************************************** Wai Lee (WLEE@HBS.EDU) Morgan Hall 362C Phone: 617-495-6368 Harvard Business School Fax: 617-496-6592 Boston, MA 02163 **************************************************** ---------- End of message ---------- From: Rob Trevor To: "RATS Discussion List" Subject: RATS-L Problem - Again Date: Fri, 7 Jun 1996 11:39:12 +1000 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Mailer: Mercury MTS v1.21 Oh dear�it happened again! Sorry about this folks, but the subscription list to RATS-L disappeared again about midnight local time here. I've just been chewing the admin's ear, and I hope that the problem is now under control. (It appears that the system's backup logs were tying up all the disk space. Then when somebody tries the sub or unsub to the list, the system can't write the modified list - and it doesn't keep the previous one. Bingo - no subscribers. At least that's the hypothesis.) You only need to repost if your post was rejected. (We caught it this time before anyone had re-subbed to the list.) My apologies Rob Trevor rats-l-owner@efs.mq.edu.au ---------- End of message ---------- From: Rob Trevor To: "RATS Discussion List" Subject: Re: GARCH Constraints Date: Fri, 7 Jun 1996 11:39:27 +1000 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Mailer: Mercury MTS v1.21 Well, I'm probably going to regret sending this off without testing it (or having my morning coffee), but here goes� When I need to impose positive parameters I've always used the the square of the parameter - a1*a1*h(t-1) in the example below. While RATS doesn't give you the right standard error, you can compute the correct one using the Jacobian of the parameter transformation. (This also works if you use the method below.) HOWEVER, it now seems obvious that it might be worth combining my method with the one below - eg, something like sqrt(a1*a1)*h(t-1) or sqrt(a1)*sqrt(a1)*h(t-1). If this works, RATS should then give the correct standard error for the coefficient of h(t-1). Let me know if this works. As for the penalty functions, the idea is to modify the objective function with extra terms so that if the parameters go off in the wrong direction, the modified objective function also goes off in the wrong direction in a very fast manner. However, you need to be careful how you do this since you want to make sure that (a) you don't add something which is non-differentiable and (b) that when the parameters are in the 'right' region, the modified objective function is still the log-likelihood function. (You might have to be careful about the BHHH standard errors also.) Rob At 11:18 PM 6/6/96, wlee@spica.hbs.edu wrote: > >I have been discussing the problem of restricting the parameters in the >GARCH equation to be positive with several others. We cannot figure out an >easy way to do that in RATS neither. However, you may follow the suggestion >of Engle and Kroner (in their BEKK multivariate GARCH paper published in 95 >or 96 issue of Econometric Theory) in dealing this problem, i.e., estimating >the square roots of the parameters: > > h(t) = sqrt(a0) + sqrt(a1)*h(t-1) + sqrt(a2)*e(t-1)^2 > >One drawback of this trick is the fact that we can only get the standard >error of the parameter of ai's. I hope this will help. > >If anyone has a better alternative approach, please post. > >> >>I am estimating a bivariate GARCH (1,1) model and would like to >>incorporate a penalty function, which would guarantee that my >>parameter estimates for the conditional variance equations will be >>nonnegative. I have set up my RATS program following Rob Trevor's >>article in a previous RATS newsletter. I believe that he mentions >>the use of such functions for the purpose stated above -- so I ahve >>reason to believe that someone has programed this. I am estimating >>my system using the maximize instruction. Any hints would be greatly >>appreciated. >> >>Thank you, >> >>Scott Frame >>University of Georgia >> >>---------- End of message ---------- >> >> >> >> >**************************************************** >Wai Lee (WLEE@HBS.EDU) >Morgan Hall 362C Phone: 617-495-6368 >Harvard Business School Fax: 617-496-6592 >Boston, MA 02163 >**************************************************** ----------------------------------------------------------------------- Rob Trevor Associate Professor Centre for Studies in Money, Banking and Finance Macquarie University, NSW, AUSTRALIA 2109 Internet: robt@efs.mq.edu.au VOICE: +61 (2) 850-8539 FAX: +61 (2) 850-7281 or +61 (2) 418-1847 ----------------------------------------------------------------------- ---------- End of message ---------- From: nikolaus@pc126.isdn.uni-konstanz.de (Nikolaus K.A. Laeufer) To: "RATS Discussion List" Subject: subscribe Date: Fri, 7 Jun 1996 08:31:06 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Z-Mail (3.2.1 10apr95) (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii subscribe ---------- End of message ---------- From: Mats Marcusson To: "RATS Discussion List" Subject: THEILS' U Date: Wed, 12 Jun 96 08:02:59 -0700 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Nutek X-Mailer: Mozilla 1.1 (Windows; I; 16bit) (via Mercury MTS v1.21) MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Type: text/plain; charset=us-ascii Hi there, does anyone know how to decompose Theil's U into the proportions of inequality;the bias proportion, the variance proportion and the covariance proportion. My efforts so far have not been very successful. Marcus ---------- End of message ---------- From: David Montgomery To: "RATS Discussion List" Subject: Date: Wed, 12 Jun 1996 00:05:17 -1000 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Light Version 1.5.2 (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Hi, I am looking for anybody that has written a program that performs the iterative outlier procedure described by Tsay 1988 or Chen and Liu 1993. I have written these programs myself and am now converting it into a source code. I am interested in anyone elses program for comparison and also because I can't duplicate the results of Tsay 1988. However I do duplicate the results of Balke 1993 and Balke and Fomby 1993. Of more desperate neds, I am looking for a code that performs maximum likelihood estimation of a fractional differencing parameter based on Sowell 1992. I have partially completed the procedure but I have little doubt that I well ever figure out the rest in this lifetime. Anyone interested in this area of work, don't hesitate to correspond to me. When I am finished with these programs I will readilly make them available to all whose interested. Thanks David Montgomery Dept of Economics University of Hawaii at Manoa ---------- End of message ---------- From: Alejandro Fonseca Ramirez To: "RATS Discussion List" Subject: e-garch models Date: Mon, 17 Jun 1996 16:37:57 -0500 (CDT) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII X-Mailer: Mercury MTS v1.21 Hello, I recently find this list and I'd like to know if you have programs for egarch models. thanks ---------- End of message ---------- From: Rob Trevor To: "RATS Discussion List" Subject: Test Message: Please Ignore Date: Tue, 18 Jun 1996 13:06:00 +1000 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Mailer: Mercury MTS v1.21 Test of reconfigured RATS list. >Rob, > >The RATS list bombed out again. I contacted the author and found the >problem was that the mail spool was relocated to another volume by >David Rawling about March / May leaving the Rats list and the mail >spool on different volumes, which apparantly causes a bug in the mail >handler software to raise its ugly head. > >Therefore, I have relocated your list to the spool: volume along with >the other mail stuff. Could you please sent a test message to the >list and see if all is well? > >Regards, > >Tony ---------- End of message ---------- From: Mats Marcusson To: "RATS Discussion List" Subject: THEIL�S U Date: Tue, 18 Jun 96 07:58:37 -0700 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Nutek X-Mailer: Mozilla 1.1 (Windows; I; 16bit) (via Mercury MTS v1.21) MIME-Version: 1.0 Content-Transfer-Encoding: 8bit Content-Type: text/plain; charset=iso-8859-1 Hi there: As it seems that the RATSLIST has been infected by bugs again, I�ll send my question once more. Does anyone know how to decompose Theil�s U into the bias proportion, the covariance proportion and variance proportion respectively? ---------- End of message ---------- From: Professor Patricia Fraser To: "RATS Discussion List" Subject: Re: Test Message: Please Ignore Date: Tue, 18 Jun 1996 08:37:25 +0100 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Organization: Dept of Accountancy, University of Aberdeen. X-Mailer: Mozilla 2.0 (Win16; I) (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Rob Trevor wrote: > > Test of reconfigured RATS list. > > >Rob, > > > >The RATS list bombed out again. I contacted the author and found the > >problem was that the mail spool was relocated to another volume by > >David Rawling about March / May leaving the Rats list and the mail > >spool on different volumes, which apparantly causes a bug in the mail > >handler software to raise its ugly head. > > > >Therefore, I have relocated your list to the spool: volume along with > >the other mail stuff. Could you please sent a test message to the > >list and see if all is well? > > > >Regards, > > > >Tonytest from prof.p.fraser@abdn.ac.uk ---------- End of message ---------- From: jjunttil@sun3.oulu.fi (Juha Junttila) To: "RATS Discussion List" Subject: Help with Tsay-procedure Date: Wed, 19 Jun 1996 10:23:46 +0300 (EET DST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" X-Mailer: (via Mercury MTS v1.21) Hi everyone, Are there any RATS-users, who are familiar with the Tsay's procedure for detcting outliers, level shifts and variance changes in a univariate time series (ARIMA) model and how to do it in RATS? The key reference to the procedure is Tsay, R. S. (1988) 'Outliers, Level Shifts, and Variance Changes in Time Series', Journal of Forecasting 7, 1-20. Juha Junttila University of Oulu Department of Economics tel. + 358 81 553 2945 fax + 358 81 553 2906 e-mail jjunttil@cc.oulu.fi ---------- End of message ---------- From: David Montgomery To: "RATS Discussion List" Subject: Re: Help with Tsay-procedure Date: Tue, 18 Jun 1996 23:51:40 -1000 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Light Version 1.5.2 (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" The procedure is on my home page at http://www2.hawaii.edu/~dmontgom If you have any questions, feel free to e-mail me. You will also see the Chen and Liu procedure on there but at the time it is not relaible enough to run David Montgomery At 09:23 PM 6/18/96 -1000, you wrote: >Hi everyone, > >Are there any RATS-users, who are familiar with the Tsay's procedure for >detcting outliers, level shifts and variance changes in a univariate time >series (ARIMA) model and how to do it in RATS? > >The key reference to the procedure is Tsay, R. S. (1988) 'Outliers, Level >Shifts, and Variance Changes in Time Series', Journal of Forecasting 7, 1-20. > Juha Junttila > University of Oulu > Department of Economics > tel. + 358 81 553 2945 > fax + 358 81 553 2906 > e-mail jjunttil@cc.oulu.fi > > ---------- End of message ---------- From: David Montgomery To: "RATS Discussion List" Subject: Re: Help with Tsay-procedure Date: Tue, 18 Jun 1996 23:53:27 -1000 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Light Version 1.5.2 (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" One more thing, if you get any other responses from people that have programs, please let me know, thank you. David Montgomery At 09:23 PM 6/18/96 -1000, you wrote: >Hi everyone, > >Are there any RATS-users, who are familiar with the Tsay's procedure for >detcting outliers, level shifts and variance changes in a univariate time >series (ARIMA) model and how to do it in RATS? > >The key reference to the procedure is Tsay, R. S. (1988) 'Outliers, Level >Shifts, and Variance Changes in Time Series', Journal of Forecasting 7, 1-20. > Juha Junttila > University of Oulu > Department of Economics > tel. + 358 81 553 2945 > fax + 358 81 553 2906 > e-mail jjunttil@cc.oulu.fi > > ---------- End of message ---------- From: Sean Collins To: "RATS Discussion List" Subject: Date: Thu, 20 Jun 1996 09:22:40 +1200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Novell GroupWise 4.1 (via Mercury MTS v1.21) Anybody out there have RATS code for Stock and Watson dynamic OLS procedures? Thanks much, Sean Collins Reserve Bank of New Zealand email: collins@rbnz.govt.nz ---------- End of message ---------- From: nikolaus@pc126.isdn.uni-konstanz.de (Nikolaus K.A. Laeufer) To: "RATS Discussion List" Subject: FM-OLS and FM-VAR Date: Fri, 21 Jun 1996 17:59:23 +0200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Z-Mail (3.2.1 10apr95) (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Hi everyone, Has anyone produced RATS-code to apply the methods of "Fully Modified Least Squares and Vector Autoregression" , Econometrica, Vol 63, No. 5, Sept. 1995 by Peter C.B. Phillips? I would appreciate if you would contact me. Nikolaus K.A. Laufer ---------- End of message ---------- From: Sean Collins To: "RATS Discussion List" Subject: FM-OLS and FM-VAR -Reply Date: Mon, 24 Jun 1996 09:35:55 +1200 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Novell GroupWise 4.1 (via Mercury MTS v1.21) I would be interested in code for Fully-Modified OLS, as well. Thanks, Sean Collins ---------- End of message ---------- From: "Gilles, Martin" To: "RATS Discussion List" Subject: FM-OLS and FM-VAR -Reply Date: Mon, 24 Jun 96 09:56:21 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mercury MTS v1.21 I would be interested in code for Fully-Modified OLS, as well. Thanks, Martin Gilles, Frankfurt ---------- End of message ---------- From: John Frain To: "RATS Discussion List" Subject: Re: FM-OLS and FM-VAR Date: Mon, 24 Jun 1996 13:40:49 +0100 (BST) Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Windows Eudora Pro Version 2.1.2 (via Mercury MTS v1.21) Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" At 17:59 21/06/96 +0200, you wrote: > >Hi everyone, >Has anyone produced RATS-code to apply the methods of "Fully Modified Least >Squares and Vector Autoregression" , Econometrica, Vol 63, No. 5, Sept. 1995 >by Peter C.B. Phillips? >I would appreciate if you would contact me. >Nikolaus K.A. Laufer > > > > A timely request which is of interest to many of us. -- John Frain Phone (+3531)6716666 Head of Research Fax (+3531)6716561 Economics (EARP) Department E-mail cbres@iol.ie Central Bank of Ireland POB 559 Dame St., Dublin 2, Ireland ---------- End of message ---------- From: dprieul@lehman.com (David Prieul) To: "RATS Discussion List" Subject: Structural VARS - Blanchard et Quah Date: Tue, 25 Jun 96 09:49:52 BST Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mercury MTS v1.21 Does anyone write a procedure using the structural VAR methodology as presented in the paper by Pierre St-Amant (Bank of Canada Working Paper 96-2)? If Pierre is on the RATS users'list, I would be delighted to get a copy of his code Thanks. David Prieul PS: I would obviously be happy to get any insight on the FM-OLS and FM-VAR but after three messages on the subject, I thought than mine could have been interpreted as the last straw that brakes the camel's back! ---------- End of message ---------- From: Rob Scott To: "RATS Discussion List" Subject: Stock Watson Date: Tue, 25 Jun 1996 10:59:38 -0400 Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mozilla 3.0b3 (WinNT; I) (via Mercury MTS v1.21) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Someone had posted recently requesting some code which performs a Stock-Watson regression. If anyone happens to have it and is willing to distribute it, I would greatly appreciate a copy. Thanks in advance. Rob Scott --------- ---------- End of message ---------- From: dprieul@lehman.com (David Prieul) To: "RATS Discussion List" Subject: STRUCTURAL VARS Date: Wed, 26 Jun 96 19:24:57 BST Errors-to: Reply-to: "RATS Discussion List" Sender: Maiser@efs1.efs.mq.edu.au X-listname: X-Mailer: Mercury MTS v1.21 Does anyone write a procedure using the structural VAR methodology as presented in the paper by Pierre St-Amant (Bank of Canada Working Paper 96-2)? If Pierre is on the RATS users'list, I would be delighted to get a copy of his code Thanks. David Prieul PS: I would obviously be happy to get any insight on the FM-OLS and FM-VAR but after three messages on the subject, I thought than mine could have been interpreted as the last straw that brakes the camel's back! ---------- End of message ----------