gretl crashed once again, but for a different reason this time.
It seems that whenever I tried to do a search for files within GUI after “file->Open data->User file...” or “file->Session files->Open session…”, it crashes right after it showed a list of files.
I can’t remember a time when I did a search like this without gretl crashing.
I suspect it has something to do with my character encoding. Even though I use standard English, I have Chinese on my system also.
This is more like a FYI, and it is not urgent, as I know how to get around it easily.
Attached is the crash report if it helps.
Dear gretl users,
especially those working with so-called Structural Time Series Models,
please note that the two packages BSMestim(.gfn) and LLTestim(.gfn) will
be removed from the package server soon. These packages covered the
"Basic Structural Model" and the "Local Linear Trend" model, and this
functionality is now included (has been for a while) in the newer
Background: The author of those packages is gretl veteran Ignacio
Díaz-Emparanza who did a pioneering job years ago programming the
necessary Kalman filter "manually" in gretl's language hansl. But for
some years now gretl has had a native Kalman filter engine for
state-space modeling (see ch. 33 of the user guide). This made it
possible to develop the more general package StrucTiSM (for the current
package version you need at least gretl 2017d).
Ignacio's original implementation will still be an interesting and
inspiring example of programming with hansl, so the code will be turned
into two additional example scripts that are shipped with gretl. (See
File/Script files/Practice scripts...)
If for some reason you rely on these two packages, please keep local
copies for yourself on your machine.
> Am 26.06.2019 um 17:01 schrieb Marcelo Duarte:
>> I have tried upgrading and downgrading gretl and the same keeps
>> When I run the Chow test, a new window opens with the augmented model,
>> but in neither appears the statistics and p-value.
>> The model I'm working on contains lagged independent variables, and,
>> just to get that idea out of my head, I've tried to run it without the
>> laggs: It appeared.
> If you can reproduce it, please specify a concrete example, ideally with
> one of the example data files, or if it is a small dataset you can also
> attach it (I think). Be very specific what model you estimate.
> (And I guess you mean lagged dependent, not independent? I _thought_ I
> just saw a case where I had this problem, but then I couldn't reproduce
I'm sorry but I can't seem to find a database somewhat similar to mine on
data file examples.
I'm attaching the *.gdt file were the problem occurs.
The regression is as follows:
panel FDIstock 0 lnGDP(-1) XR(-1) BIT Border PT --pooled --robust
chow Eurozone --dummy
chow IberoAm --dummy
Hope you can help,
Date: Mon, 24 Jun 2019 10:43:04 -0400 (EDT)
From: Allin Cottrell <cottrell(a)wfu.edu>
Subject: [Gretl-users] Re: Chow test p-value
To: Gretl list <gretl-users(a)gretlml.univpm.it>
Content-Type: text/plain; charset=US-ASCII
On Mon, 24 Jun 2019, Sven Schreiber wrote:
> Am 24.06.2019 um 15:01 schrieb Marcelo Duarte:
> > In previous versions of gretl, if I ran a chow test (along a dummy
> > variable) from the tests menu it would print the test statistics and
> > respective p-value.
> > As of now, only an augumented model is returned, and no statistics and
> > p-value.
> > Is there a way to get those?
> > I'm working on win7 64bit with gretl 2019c-git (2019-06-01).
> I'm not seeing this problem on gretl 2019b (from May 21st), so if it is
> a bug, it must have been introduced in the ten days after the release of
> 2019b. (Or it's specific to Win7, which I somehow doubt, however.)
> Perhaps you should update to a brand new snapshot, or downgrade to 2019b.
>I just checked an example with current git, and the test statistic and
>p-value were shown OK. They appear underneath the augmented model
>output so in some cases it may be necessary to scroll down to see
I have tried upgrading and downgrading gretl and the same keeps happening.
When I run the Chow test, a new window opens with the augmented model, but
in neither appears the statistics and p-value.
The model I'm working on contains lagged independent variables, and, just
to get that idea out of my head, I've tried to run it without the laggs: It
My question may have an obvious answer, but I do not know: Is the chow test
sensitive to laggs in independent variables?
Thank you once more,
I would like to make a suggestion regarding the FGLS/WLS estimator implementation in Gretl.
When we have a model where the errors present autocorrelation (for example, an AR(1) process), and we try to get the estimated model this way: Model->Time Series->AR errors (GLS) -> AR(1), some post-estimation tests/results… are disabled. In particular, no autocorrelation and/or heteroskedasticity tests can be implemented (Tests->Heteroskedasticity or Tests->Autocorrelation). My question is: why does this happen? I think that it would be very useful if these options were active. Otherwise, after running FGLS there is no possibility to check whether the autocorrelation has been “corrected”, unless we run the regressions "by hand".
Something similar happens with WLS estimator for the case in which errors show heteroskedasticity. Furthermore, I think that in this case the Total Sum of Squares (TSS) based on the weighted data is not properly calculated, because if we run an OLS regression on the weighted variables all the statistics depending on the TSS (R-squared, Adjusted R-squared, F value and its p-value…) are different (however, the sum of the squared residuals, the S.E. of the regression, etc. are right). I suspect that Gretl is mixing the original dependent variable and the weighted results. Why? Because when FGLS estimator is applied to deal with the autocorrelation problem (for example using Cochrane-Orcutt), statistics based on the rho-differenced data give the mean (and the standard deviation) of the original dependent variable, whereas all the other statistics are based on the weighted data.
Finally, in the WLS estimator option the residuals are the originals, and not the weighted ones. This should be also changed (with the FGLS estimator this does not happen, being the residuals the weighted ones).
for a package I need to know of what type the loaded data set is
(time-series, cross-sectional, panel). Does gretl have a built-in
accessor for it? I couldn't find one. Or what's a robust way to
Users of gretl may wish to explore some features that are new -- or
in some cases not totally new but hitherto undocumented -- in
current gretl 2019c git and snapshots. These include the functions
fzero(), conv2d(), flatten() and msplitby()
along with enhancement of mread().
I'm experiencing some difficulties in obtaining the Chow test statistics
In previous versions of gretl, if I ran a chow test (along a dummy
variable) from the tests menu it would print the test statistics and
As of now, only an augumented model is returned, and no statistics and
Is there a way to get those?
I'm working on win7 64bit with gretl 2019c-git (2019-06-01).
On Thu, 20 Jun 2019, Logan Kelly wrote:
> flush did not seem to work inside an mpi block, but I will try again.
No, I'm pretty sure that won't work. "flush" is a GUI-specific thing,
but inside an MPI block it's actually the command-line program
gretlmpi that's doing the work. The output is collected and shown in a
GUI window only when the gretlmpi instances have terminated.
As things stand, I think you can view real-time MPI progress only by
running gretlmpi yourself (in a terminal/console). Maybe we can figure
out some clever means of grabbing and showing interim output but I
suspect it won't be trivial.