Den 04/11/2011 kl. 14.16 skrev Sven Schreiber: No problem. Are attachments sustained in the archives? Best Andreas -- Andreas Noack Jensen Ph.d.-stipendiat Økonomisk Institut andreas.noack.jensen(a)econ.ku.dk Københavns Universitet http://www.econ.ku.dk/phdstudent/noack/ Øster Farimagsgade 5, bygning 26 Tlf.: 353 23094 1353 København K

On Fri, 4 Nov 2011, Ignacio Diaz-Emparanza wrote: Hmm, can you elaborate? This works fine here. <shell> waverley:~$ cat test.csv y x 1.2 3.4 1.3 4.5 1.4 5.6 waverley:~$ cat opencsv.inp open test.csv -q print -o waverley:~$ gretlcli -b opencsv.inp gretl version 1.9.6cvs Copyright [...] Current session: 2011-11-04 11:07 ? run opencsv.inp /home/cottrell/opencsv.inp ? open test.csv --quiet ? print --byobs y x 1 1.2 3.4 2 1.3 4.5 3 1.4 5.6 waverley:~$ LANG=es_ES.UTF8 gretlcli -b opencsv.inp gretl versión 1.9.6cvs copyright [...] Sesión actual: 2011-11-04 11:08 ? run opencsv.inp /home/cottrell/opencsv.inp ? open test.csv --quiet ? print --byobs y x 1 1,2 3,4 2 1,3 4,5 3 1,4 5,6 </shell> Allin

On 11/04/2011 06:16 PM, Allin Cottrell wrote: For those who cannot or don't want to run the example in Ox, could you tell us which of the results are different? For example, the broken trend variable is probably not identical across programs, because the normalizations of "time" and "Trend" are different. But I suppose that fact and the differences which it implies are not what you guys are referring to? thanks, sven

On Sat, 5 Nov 2011, Andreas Noack Jensen wrote: Yes, I came to the same conclusion, and I've just committed a change to gretl CVS that renders the results the same (from gretl and Ox) for your original script. You're suggesting that we shouldn't allow including unrestricted deterministic terms in cointegration testing (though we should allow them in VECM specifications, having determined cointegrating rank)? Allin Cottrell

On Sat, 5 Nov 2011, Sven Schreiber wrote: I would guess that in the vecm field the "industry standard" is pcgive/ox. If we're unsure on a convention, I'd rather follow Doornik. Sven, my man!!! Riccardo (Jack) Lucchetti Dipartimento di Economia Università Politecnica delle Marche r.lucchetti(a)univpm.it http://www.econ.univpm.it/lucchetti

On Sun, 6 Nov 2011, Andreas Noack Jensen wrote: Glad you like it. Allin here: I guess I'd say that one way or another there will be something at least slightly confusing here (I mean, either the restricted Xs will be treated differently from the unrestricted Xs, or they will be treated differently from the restricted Ys). Here I agree with Jack (pragmatism talking). Well, it seems that there _is_ a convention de facto. That's what gave rise to your original suggestion that there was something amiss in gretl's version of the analysis. Allin: No, far from it. (At least as far as I am concerned.) Hmm, the "default" is in a sense up to the user. It may be that it's unwise to test for cointegration while including unrestricted exogenous terms. But as you've illustrated, Ox is "quite happy" to do that if it's what the user requests. It would be be awkward (though of course not impossible) for gretl to say "Sorry, you can't do that!". Some (presumably rather extended) explanation would be required for why this was prohibited. Allin Cottrell

Den 06/11/2011 kl. 04.22 skrev Allin Cottrell: I have some new comments on this thread. First of all I would like hear about the status on the question on restricting the deterministic terms in cointegration testing. To state my view a bit differently: If there exist any defense for testing in the unrestricted models I would really like to know about it because I am only aware of recommendations to avoid those models. Next thing is about the restricted exogenous variables option. I think I have found a new bug. If you run the script you will see the problem. The short run parameters are problematic. Finally I have a comment regarding our discussion of the convention for restricted terms. I thinks the convention in Gretl right now is one of the better solutions. One could also think of a convention where the first differences of the restricted variables were added automatically and the level was lagged one period as it used to be. This would be more in line with CATS. But I think that these two solutions are the most sensible. However, the old convention i Gretl might come from the PcGive documention http://www.doornik.com/pcgive/ section 15.7.1. The funny thing is that I think that the documentation is wrong and that PcGive/PcFiml is doing the same thing as Gretl which the script should also show. It is a bit of a mess to keep track of. Best Andreas open Marriage.xls setobs 1 1947 --time-series genr time genr I76 = (t == 30) lags 1; I76 genr Sh76 = (t >= 30) lags 2; Sh76 genr Tr76 = time * Sh76 lags 2; Tr76 # With restricted broken trend vecm 1 2 Men Women; Sh76 I76; Tr76_2 --crt coint2 1 Men Women; Sh76 I76; Tr76 foreign language=Ox #include <oxstd.h> #import <pcfiml> main() { decl model = new PcFiml(); model.Load("Marriage.xls"); model.Deterministic(FALSE); model.Append(zeros(1976 - 1947, 1) | ones(2010 - 1976 + 1, 1), "Sh76"); model.Append(model.GetVar("Sh76") .* model.GetVar("Trend"), "Tr76"); model.Append(diff0(model.GetVar("Sh76"), 1), "I76"); // Model with restricted broken trend model.Select(Y_VAR, {"Men", 0, 1, "Women", 0, 1}); model.Select(X_VAR, {"Trend", 0, 0, "Tr76", 2, 2}); model.Select(U_VAR, {"Constant", 0, 0, "Sh76", 0, 0, "I76", 0, 0}); model.SetSelSample(-1, 1, -1, 1); model.Estimate(); model.Cointegration(); delete model; } end foreign -- Andreas Noack Jensen Ph.d.-stipendiat Økonomisk Institut andreas.noack.jensen(a)econ.ku.dk Københavns Universitet http://www.econ.ku.dk/phdstudent/noack/ Øster Farimagsgade 5, bygning 26 Tlf.: 353 23094 1353 København K

Am 11.11.2011 17:44, schrieb Andreas Noack Jensen: First of all let me state that I am fully aware of the similarity issues in that context, which you already briefly alluded to. I also heard the recommendations you are following, and for some time I even tried to follow them. And there are cases where I fully agree, for example when I have to model the German unification break. To take the example of the unrestricted constant term, however, in the end I think there is often not a real problem. That is, if you have data which beyond any doubt have linear trend terms, and you are not ready to accept any long-run equilibrium which is trend stationary instead of mean stationary. (Because after all, how do you really interpret economically an equilibrium which is diverging?) Well then you say: no problem, just kick the trend term out of the CI space after determining the rank. However with typical small-sample data often this just doesn't work; in my real-world experience first you often get quite strange rank test results when you allow for linear trends in the CI space, and secondly afterwards often the restriction test wouldn't allow you to set the trend coefficient to zero. The alternative is to not allow the trend terms right from the start, because you really have no use for them. That's what we often do in econometrics: we make some specification assumptions to gain efficiency, power and so on. And sure there is the danger that those assumptions could be wrong, and so you really have to think hard about them. But I don't see why a fixed rule like "thou shalt not test for cointegration with unrestricted deterministic terms" would be the best solution of the dilemma. BTW, it could be argued that since I am holding small sample sizes responsible for the dilemma, we should always use the newly available small-sample tools like bootstrapping and Bartlett corrections. I have some experience with that and indeed, maybe that would resolve some of the otherwise strange test results. This may be a case to push for these tools to be more widely available. Thanks for initiating an interesting discussion, Sven

On Sun, 13 Nov 2011, Andreas Noack Jensen wrote: I've put my view on unrestricted deterministic terms briefly. I have a working paper that's somewhat relevant. Tomorrow I'll dig that out and post a link. Some months ago Prof Johansen was good enough to send me his RATS code for the Bartlett correction, and I translated it to gretl in a function package. Since there seems to be interest I'll dust that off and make it available. Wow! I'll definitely take a look at that. Sorry, that's a documentation issue (which we should fix before the next release). In fact, both the "var" and "vecm" commands accept the options --quiet and --silent. The --quiet option suppresses printing of the individual equations but allows printing of the system-level information, and the --silent option does what you'd expect. I'll have to take a look at what you have done. But there's no built-in way to "batch process" a matrix into p series, at present you'd have to use a loop. Something like (untested): <hansl> string sname = "Mcol" loop i=1..cols(M) series @sname_$i = M[,i] endloop </hansl> Allin

On Mon, 14 Nov 2011, Andreas Noack Jensen wrote: Another trick you may like is <hansl> outfile null --write ... do stuff ... outfile null --close </hansl> The effect is to suppress ALL output. Riccardo (Jack) Lucchetti Dipartimento di Economia Università Politecnica delle Marche r.lucchetti(a)univpm.it http://www.econ.univpm.it/lucchetti

Den 16/11/2011 kl. 22.49 skrev Allin Cottrell: Thank you for the link. Turner's article seems a bit odd to me so it is good that you object to his conclusions. Without really noting it he exercises the problems of using the unrestricted model (on data without a trend). Problems which are described in Johansens book and many other places. I wonder about the relation between the editor of the journal, the author of the preferred software package, the author of the article with the preferred critical values and the acceptance of his article. Thank you. I'll take a look at it. -- Andreas Noack Jensen Ph.d.-stipendiat Økonomisk Institut andreas.noack.jensen(a)econ.ku.dk Københavns Universitet http://www.econ.ku.dk/phdstudent/noack/ Øster Farimagsgade 5, bygning 26 Tlf.: 353 23094 1353 København K

On Sun, 13 Nov 2011, Allin Cottrell wrote: Coincidence: I met Giuseppe Cavaliere no longer than 10 days ago and we agreed to work on bootstrap-based cointegration rank tests next month. Riccardo (Jack) Lucchetti Dipartimento di Economia Università Politecnica delle Marche r.lucchetti(a)univpm.it http://www.econ.univpm.it/lucchetti