6:47 a.m.
Hi,
this is not directly gretl-related I think, but who knows, maybe there
are some possible workarounds in the API:
Recently I (or actually, a student) have noticed that in some cases in
db.nomics the information about the multiplication factor (million,
billion...) for a series seems to be missing. We saw that for example in
the balance of payments statistics the units are just given as "US
dollars" or "local currency", but the numbers are far too small to be
only single dollars (as opposed to bn$ or something like that).
Of course it's not too difficult to match the magnitudes with some known
level or based on plausibility, but it requires manual intervention,
which is kind of beside the point for the whole db.nomics thing.
Does anybody have an idea how to solve that? Is gretl perhaps not
grabbing all the given information?
thanks
sven
8:04 p.m.
On Mon, 13 May 2019, Sven Schreiber wrote:
Hi,
this is not directly gretl-related I think, but who knows, maybe there
are some possible workarounds in the API:
Recently I (or actually, a student) have noticed that in some cases in
db.nomics the information about the multiplication factor (million,
billion...) for a series seems to be missing. We saw that for example in
the balance of payments statistics the units are just given as "US
dollars" or "local currency", but the numbers are far too small to be
only single dollars (as opposed to bn$ or something like that).
Please give a reference to a specific dbnomics series ID, then we
can investigate.
Allin
12:39 p.m.
Am 13.05.2019 um 22:04 schrieb Allin Cottrell:
On Mon, 13 May 2019, Sven Schreiber wrote:
> Recently I (or actually, a student) have noticed that in some cases in
> db.nomics the information about the multiplication factor (million,
> billion...) for a series seems to be missing. We saw that for example in
> the balance of payments statistics the units are just given as "US
> dollars" or "local currency", but the numbers are far too small to be
> only single dollars (as opposed to bn$ or something like that).
Please give a reference to a specific dbnomics series ID, then we can
investigate.
Dear Allin, of course you're right.
Consider the following series for France:
-------------------------
Series: A.FR.BACK_BP6_EUR
Provider: IMF
Dataset: BOP (Balance of Payments (BOP))
Identifier: IMF/BOP/A.FR.BACK_BP6_EUR
Name: Annual – France – Net Lending (+) / Net Borrowing (-) (Balance
from Current and Capital Account), Euros
Dimensions, showing JSON mask form:
Frequency: Annual, {"FREQ":["A"]}
Reference Area: France, {"REF_AREA":["FR"]}
Indicator: Net Lending (+) / Net Borrowing (-) (Balance from Current
and Capital Account), Euros, {"INDICATOR":["BACK_BP6_EUR"]}
-------------------------
pd = 1; 20 observations, 1999 - 2018
The values range from -25286 to 48810, surely these cannot be pure
Euros, instead probably million Euros such that the maximum is 48.8bn
(in the American sense, bn=10^9). But I don't see that scaling information.
thanks
sven
3:25 p.m.
On Tue, 14 May 2019, Sven Schreiber wrote:
Am 13.05.2019 um 22:04 schrieb Allin Cottrell:
> On Mon, 13 May 2019, Sven Schreiber wrote:
>
> > Recently I (or actually, a student) have noticed that in some cases in
> > db.nomics the information about the multiplication factor (million,
> > billion...) for a series seems to be missing. We saw that for example in
> > the balance of payments statistics the units are just given as "US
> > dollars" or "local currency", but the numbers are far too small
to be
> > only single dollars (as opposed to bn$ or something like that).
>
> Please give a reference to a specific dbnomics series ID, then we can
> investigate.
>
Dear Allin, of course you're right.
Consider the following series for France:
-------------------------
Series: A.FR.BACK_BP6_EUR
Provider: IMF
Dataset: BOP (Balance of Payments (BOP))
Identifier: IMF/BOP/A.FR.BACK_BP6_EUR
Name: Annual – France – Net Lending (+) / Net Borrowing (-) (Balance
from Current and Capital Account), Euros
Dimensions, showing JSON mask form:
Frequency: Annual, {"FREQ":["A"]}
Reference Area: France, {"REF_AREA":["FR"]}
Indicator: Net Lending (+) / Net Borrowing (-) (Balance from Current
and Capital Account), Euros, {"INDICATOR":["BACK_BP6_EUR"]}
-------------------------
pd = 1; 20 observations, 1999 - 2018
The values range from -25286 to 48810, surely these cannot be pure
Euros, instead probably million Euros such that the maximum is 48.8bn
(in the American sense, bn=10^9). But I don't see that scaling information.
I see what you mean. It turns out that gretl isn't missing anything in
the JSON retrieved from dbnomics -- the scale is simply not stated. I
checked further and this is also the case with IMF BoP data for
countries other than for France and for currency other than Euros.
But if you go directly to the IMF you see that the numbers are indeed
in millions (Don't know if this URL is portable but I'm looking at,
e.g., http://data.imf.org/regular.aspx?key=60961513 )
So something is not being communicated from the IMF to dbnomics; I've
posted a comment on their forum.
Allin
12:14 p.m.
New subject: About Linear Restrictions on the short term coefficients of VECM
Am 16.05.2019 um 09:07 schrieb Reynaldo Senra:
Dear all,
I read the User Guide of Gretl and I only have found explanation on
how to set linear restrictions on alpha and beta (jointly or not).
However, I couldn't find how to set linear restrictions on the
coefficients of the lagged differences of the endogenous variables.
It would be great if you suggest me any pdf or advise about how to do
this.
Hi,
you can restrict certain lags to zero in a 'var' command (see the doc
there), but I'm not sure if it also works in a VECM context.
If you need more general restrictions, you would have to take a two-step
approach AFAIK: First estimate the cointegration coefficients (beta) and
save the error-correction term(s) as one or multiple time series, then
set up a simultaneous equation system in differences and throw in the
ECTs. In that system you should be able to apply restrictions on the
dynamics. Caveat: I guess it will be difficult to get gretl to recognize
the ECTS as lagged endogenous terms, so things like forecasting will not
work correctly out of the box.
cheers
sven
2050
days inactive
2053
days old
5 comments
3 participants
participants (3)
-
Allin Cottrell -
Reynaldo Senra -
Sven Schreiber