Re: [Gretl-devel] translation question
by Allin Cottrell
On Mon, 18 Oct 2010, Talha Yalta wrote:
> I have a translation question about "Choi meta-tests"
> Is meta here the same as in metadata (data about data). In other
> words, do meta-tests mean tests about tests?
Yes. See http://en.wikipedia.org/wiki/Meta-analysis
> Also, this is off topic but somewhat related: Is there also a
> panel-cointegration test available in gretl?
No, not at present.
Allin
14 years, 2 months
translation question
by Talha Yalta
Hi:
I have a translation question about "Choi meta-tests"
Is meta here the same as in metadata (data about data). In other
words, do meta-tests mean tests about tests?
Also, this is off topic but somewhat related: Is there also a
panel-cointegration test available in gretl?
Cheers
Talha
--
“An expert is a person who has made all the mistakes that can be made
in a very narrow field.” - Niels Bohr (1885-1962)
--
14 years, 2 months
Re: [Gretl-devel] Bug in adf --gls?
by Allin Cottrell
On Tue, 12 Oct 2010, Marcin B�^Bażejowski wrote:
[ script that compares adf results with and without the --gls
option ]
Fixed in CVS. The problem was related to something Sven mentioned
lately, namely the "reach-back" to pre-sample values. This was
definitely too liberal with adf --gls, for the regression that
estimates the GLS parameters.
One thing to note is that your commands to generate an AR(1)
process were not optimal:
series y = 10
series u = normal(0,1)
genr y = 0.1*y(-1) + u
This is rather slow, and requires a "lead in" period before the
steady state is attained (which is what tripped up adf). You're
better using the filter() function:
series u = normal(0,1)
series y = filter(u, 1, 0.1, 10)
Allin
14 years, 2 months
Re: [Gretl-devel] Bug in adf --gls?
by Allin Cottrell
On Tue, 12 Oct 2010, Marcin B�^Bażejowski wrote:
> simple script:
>
> <script>
> nulldata 1000
> setobs 1 1 --time-series
> set echo off
>
> scalar ADF = 0
> scalar ADFGLS = 0
>
> scalar N = 1000
>
> loop N --quiet
> smpl --full
>
> series y = 10
> series szum = normal(0,1)
>
> genr y = 0.1*y(-1) + szum #generating AR(1) with AR-coefficient 0.1
>
> smpl 200 1000
> adf 1 y --quiet
> ADF += ($pvalue>0.1?1:0)
> adf 1 y --quiet --gls
> ADFGLS += ($pvalue>0.1?1:0)
> end loop
>
> printf "\nPercentage of non-rejecting false H0:\n"
> printf "adf:\t\t%.1f\n", 100*ADF/N
> printf "adf-gls:\t%.1f\n", 100*ADFGLS/N
> </script>
>
> I've expected 0.0% two times, but percentage of non-rejecting H0 in case
> of "--gls" surprised me...
There is something funny going on here, yes. You can expect a
proper answer before long.
Allin
14 years, 2 months
Bug in adf --gls?
by Marcin Błażejowski
Hi,
simple script:
<script>
nulldata 1000
setobs 1 1 --time-series
set echo off
scalar ADF = 0
scalar ADFGLS = 0
scalar N = 1000
loop N --quiet
smpl --full
series y = 10
series szum = normal(0,1)
genr y = 0.1*y(-1) + szum #generating AR(1) with AR-coefficient 0.1
smpl 200 1000
adf 1 y --quiet
ADF += ($pvalue>0.1?1:0)
adf 1 y --quiet --gls
ADFGLS += ($pvalue>0.1?1:0)
end loop
printf "\nPercentage of non-rejecting false H0:\n"
printf "adf:\t\t%.1f\n", 100*ADF/N
printf "adf-gls:\t%.1f\n", 100*ADFGLS/N
</script>
I've expected 0.0% two times, but percentage of non-rejecting H0 in case
of "--gls" surprised me...
Marcin
--
Marcin Błażejowski
http://www.wrzosy.nsb.pl/~marcin/
GG# 203127
14 years, 2 months
Pooled mean group technique
by Kristyna Pokorna
Hi,
please is it possible to analyse panel data using the pooled mean group approach (i.e. method that incorporates that short run coefficients vary for different countries, however, the long run coefficients adjust to a more stable level irrespective of a country).
Thanks in advace for you help.
Regards,
Kristyna
14 years, 2 months
Re: [Gretl-devel] Pooled mean group technique
by Allin Cottrell
On Mon, 11 Oct 2010, Kristyna Pokorna wrote:
> please is it possible to analyse panel data using the pooled
> mean group approach (i.e. method that incorporates that short
> run coefficients vary for different countries, however, the long
> run coefficients adjust to a more stable level irrespective of a
> country).
This question would be more appropriate on the gretl-users list,
but anyway: No, this is not available as a built-in command in
gretl. No doubt it could be coded using gretl's scripting
language, but I suspect it would be a non-trivial exercise.
Allin Cottrell
14 years, 2 months
Pooled mean group technique
by Kristyna Pokorna
Hi,
please is it possible to analyse panel data using the pooled mean group approach (i.e. method that incorporates that short run coefficients vary for different countries, however, the long run coefficients adjust to a more stable level irrespective of a country).
Thanks in advace for you help.
Regards,
Kristyna
14 years, 2 months
Re: [Gretl-devel] Fed Database
by Allin Cottrell
On Wed, 6 Oct 2010, Henrique Andrade wrote:
> Thank you so much for your efforts in trying to make the upgrade of the
> databases easier to us! By the way, I'm almost finishing the database with
> the complete balance of payments of Brazil (with 255 series) and I would
> like to make it available for the Gretl community through the Gretl server.
> Is this possible?
Yes, of course. Just send me the database files when they are
ready.
Allin Cottrell
14 years, 2 months
Fed Database
by Allin Cottrell
In
http://lists.wfu.edu/pipermail/gretl-devel/2010-September/002930.html
Henrique asked if it was possible to update gretl's Fed database
independently.
I have now added in CVS, under the "utils" subdirectory, two new C
programs, one to build a gretl database from data from the St
Louis Fed (FRED), and one to build a database from Federal Reserve
Board (FRB) data. These correspond to the databases fedstl.bin and
fedbog.bin respectively.
In both cases we use data in XML format, and to compile and run
these programs you'll need a full installation of libxml2 (library
and header files).
The current fedbog data file must be downloaded manually before
running the processor (see FRB/README).
The fedstl program, fred2db, downloads the current data
non-interactively, and it requires libcurl. If you want to use
this program you should get yourself an "API key" for FRED -- see
http://api.stlouisfed.org/api_key.html . This should be placed in
a file named "api.key" in the same directory as the program
fred2db.
Allin Cottrell
14 years, 2 months
