AR(1) sepcification in TSLS or 3SLS
by Trevor Zink
I'm estimating a simultaneous equations model using TSLS (or 3SLS). However,
my residuals are clearly serially correlated, so I would like to run it in
AR(1) specification. It's not apparent to me how to do that in gretl.
In EViews, one simply adds the regression term 'AR(1)' and it does some
.magic. I believe it just adds lags of the endogenous RHS and LHS variables
to the instruments list, but I'm unable to recreate eviews results with
gretl using the "lags" options in the TSLS dialogue.
Any help would be greatly appreciated. Thanks.
Trevor
12 years, 5 months
Error output of the switching algorithm
by artur tarassow
Dear gretl list,
I am trying to estimate a few restricted VECMs, but the estimation does
fail quite often using the default switching algorithm. Actually, I just
would like to know to what exactly does the following error message refer
to:
"Switching algorithm: failed
Exact or near collinearity encountered"
I could not find any hints in the C-sources. Are the restrictions collinear
or is it the high correlation between variables?
Thanks in advance,
Artur
12 years, 5 months
Unit root and structural breaks
by JOSE FRANCISCO PERLES RIBES
Hi:
Has anyone written any code to perform the Lumsdaine and Papell (1997) or Lee
and Strazicich (2003) unit root test with two structural breaks?
Thanks in advance.
José F. Perles
University of Alicante
(Spain)
12 years, 5 months
Compiling GRETL source in MAC
by Sanzad Siddique
Hi,
I wanted to use the libgretl and as suggested by group member, i was trying
compiling the gretl source in my MAC OSX lion. now its a pain. Whenever I
am running the configure file, it comes up with a different complain. I
have installed all necessary components using macport and fink but still
the configure is complaining on GMP and GLIB-2 whereas both are already
installed. Is there anyone who have compiled the source in MAC
successfully? If possible, can you please give me some hint, what should I
do?
Thanks in advance.
Regards,
Sanzad
12 years, 5 months
Restricting a sample to specific time series
by Jan Tille
Hello I am a new gretl user and have a question whether and how it is possible to run a loop only via a selection of time series that have minimum length of 24 months.
The series's have different start and end dates and I want to run a regression for each series that has a minimum of 24 data points.
Is there anybody, who can hint me on how to restrict or define the loop properly. On possibility is, to preselect the data that I import into gretl but this would be tedious, because I will also run a different model on all series that have a minimum of 12 data point.
Thanks in advance,
Jan
12 years, 5 months
ADF p-values difference
by JOSE FRANCISCO PERLES RIBES
Dear Gretl list
Finally I put my question on the differences observed in the p-value
in theGretl
ADF test, in Eviews and R package fUnitRoots forums (r-sig-finance).
>From Eviews team I have obtained the following answer, that I put on the
list (on the advice of Sven) if interest for the whole list
"Your reading of MacKinnon's comment about finite sample ADF values is
generally correct, though there is no evidence presented that they are
better or worse than the asymptotic values for the t-stat. I will point out
that the one case (z-stat) where MacKinnon strongly cautions against using
the finite sample values is not a test statistic that EVIews produces for
the ADF (though we do report related tests in the cointegration context --
perhaps in this case we shouldn't...). I think that the jury is still out
on whether the t-statistic finite sample or asymptotic values are better.
To provide some context, the basic idea is that that the finite sample
critical values are based on a set of simulations for which MacKinnon did
not employ ADF regressions. Were he to have run some with ADF corrections
he might very well have found that the finite sample DF results were closer
than the asymptotic results in some cases (but understandably he did not
run those simulations as the number of simulations that he did run is
already quite large and it is not clear the best way to set up the
correlation structure for evaluating the test statistics).
The most compelling argument for continuing to use the finite sample values
for the ADF is, I think, one of comparability. One concern with switching
over to the asymptotic values for ADF tests is that if you were to run a DF
test for a smallish sample and then add a single ADF lag, you are more
likely to get quite different results if we were to switch to using the
asymptotic values--and in the absence of simulation results it is not clear
whether this is a good or a bad thing. It would then be difficult to
evaluate whether the difference in results is the result of the
autocorrelation correction or the result of different critical values (or
both). With either the finite sample or asymptotic choice, one is in a bit
of a bind in the absence of finite sample simulations. By sticking with the
finite sample values we are at least holding one thing somewhat
constant...this may or may not be better...
As I write this, it occurs to me that one possibility would be to report
both values. That has it's own set of issues but would then allow users to
pick what they want to evaluate. I'll put it on a list of things to
consider."
>From the forum fUnitRoots still awaiting a response...
To be continued...
José F. Perles
University of Alicante (Spain)
12 years, 5 months
Unitroot ADF test p value
by JOSE FRANCISCO PERLES RIBES
I'm doing a unit root test ADF with Gretl on a series of tourism market
share of Spain specified with constant and trend.
By comparing the results with Eviews or R (package fUnitRoot) I get the same
t-statistic, but although both programs indicate that the critical values
are McKinnon (1996) MacKinnon, J. G. (1996) "Numerical distribution
functions for unit root and cointegration tests", Journal of Applied
Econometrics 11: 601-618.
p-values of the test are very different in either case .
Gretl: t = -3.62 p-value 0.02 asymptotic
Eviews t = -3.62 p-value (one-sided) = 0.04 which is the same value
obtained in R.
I know that obtain similar values with different software is very strange.
But this difference is normal?
At what may be due?
Thanks in advance and sorry for any inconvenience.
José F. Perles
University of Alicante
Spain
12 years, 5 months
Re: [Gretl-users] Data error with quantreg
by Ashley Dunstan
Sorry can this message please be removed from the list, I meant to send it just to Allin.
-----Original Message-----
From: Ashley Dunstan
Sent: Wednesday, 4 July 2012 12:38 p.m.
To: 'Gretl list'
Subject: RE: [Gretl-users] Data error with quantreg
Hi Allin,
Thanks for the offer. I've pasted my code below and attached a zip file containing the dataset. Let me know if the zip file doesn't make it due to size.
With this dataset I get a data error, although the code will run if you remove the variable free or change tau to something like 0.500000001. And it should run with ols.
Let me know if you need more detail.
Thanks,
Ash
CODE
# read in data
open c:\matlab_data\REINZ\gretltestNew.csv
# restrict sales to lifestyle
smpl PropertyClass = 2 --restrict --replace
# create variables for regression
discrete ReturnPeriod
list time = dummify(ReturnPeriod)
list region=dummify(RuralDistrictNa)
list sector=dummify(RuralType)
genr bare = (Bareland=2)
genr FreeholdStatus=misszero(FreeholdStatus)
genr free = (FreeholdStatus==2)
genr sqrtLandArea = LandArea^0.5
logs SalePrice
# remove missing values
smpl --no-missing l_SalePrice LandArea
# run lad regression for all non-lifestyle farms quantreg 0.5 l_SalePrice 0 LandArea sqrtLandArea region sector bare free time # construct price index based on coefficients on time dummies k=$ncoeff index= exp($coeff[27:k])
-----Original Message-----
From: gretl-users-bounces(a)lists.wfu.edu [mailto:gretl-users-bounces@lists.wfu.edu] On Behalf Of Allin Cottrell
Sent: Monday, 2 July 2012 8:29 p.m.
To: Gretl list
Subject: Re: [Gretl-users] Data error with quantreg
On Mon, 2 Jul 2012, Ashley Dunstan wrote:
> I want to estimate a lad regression on a large dataset with around
> 35,000 observations.
>
> The lad command crashes due to the large size of the dataset, however
> I have found the quantreg seems to use a different algorithm and looks
> more promising.
>
> It works most of the time (matches results from Matlab and does it
> very quickly).
>
> However, I have found it to be quite unstable and in some
> specifications it will give me a 'data error'. Unfortunately I can't
> work out how to dig into the source of the error. [...]
if you can send me your dataset and model specification(s) offlist I'll look into the problem.
Allin Cottrell
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
******************************************************************************
"This message (and any files transmitted with it) are confidential and
may be legally privileged. If you are not the intended recipient please
notify the sender immediately and delete this message from your system.
This message does not necessarily reflect the views of the
Reserve Bank of New Zealand. If the recipient has any concerns about
the content of this message they should seek alternative confirmation
from the Reserve Bank of New Zealand."
******************************************************************************
12 years, 5 months
Data error with quantreg
by Ashley Dunstan
I want to estimate a lad regression on a large dataset with around 35,000 observations.
The lad command crashes due to the large size of the dataset, however I have found the quantreg seems to use a different algorithm and looks more promising.
It works most of the time (matches results from Matlab and does it very quickly).
However, I have found it to be quite unstable and in some specifications it will give me a 'data error'. Unfortunately I can't work out how to dig into the source of the error. For example, quantreg will work with one of my dummy variables in, but doesn't work without it!
Interestingly, moving the 'tau' coefficient from 0.5 to 0.500000001 makes the code work?!?!
Any explanation for this strange behavior and how I could robustify my code?
p.s ols works fine on the dataset
******************************************************************************
"This message (and any files transmitted with it) are confidential and
may be legally privileged. If you are not the intended recipient please
notify the sender immediately and delete this message from your system.
This message does not necessarily reflect the views of the
Reserve Bank of New Zealand. If the recipient has any concerns about
the content of this message they should seek alternative confirmation
from the Reserve Bank of New Zealand."
******************************************************************************
12 years, 5 months
