5:25 a.m.
To the Gretl community,
I am trying to do the long run Structural VAR using 9 variables. With the
(n^2-n)/2 formula, I am left with 36 constraints. I used the C-model and I
was able to restrict the template matrix but I cannot seem to generate the
SVAR result.
The following is the Hansl code shown:
Index value 0 is out of bounds.
*** error in function init_C, line 4
S = Ss [,1:k-1]
called by function SVAR_estimate
called by function GUI_SVAR
May I know what went wrong here? Your help is greatly appreciated.
9:16 a.m.
Am 24.06.2014 07:25, schrieb Deborah Sy:
The following is the Hansl code shown:
Index value 0 is out of bounds.
*** error in function init_C, line 4
> S = Ss [,1:k-1]
called by function SVAR_estimate
called by function GUI_SVAR
May I know what went wrong here? Your help is greatly appreciated.
Hi,
in hansl the matrix indices start at 1, not at zero, so that's the
error. Apparently here k=1, so k-1=0.
How can k=1 happen? Looking at the code (SVAR 0.994 shipped with my
gretl version) of 'init_C' shows that 'k=cols(Ss)', so the matrix
'Ss'
must have zero columns.
Now Ss is created by 'Ss = imp2exp(Rd)'. In terms of programming it's
probably an indication that some further input checking should be added
to the package. In terms of your application, it could be that
something's wrong with your restrictions. (Since 'Rd' seems to be
related internally to the restrictions, I think -- I'm not the author.)
So I suggest to double-check your restrictions. You can also show them
to us.
hth,
sven
11:35 a.m.
On Tue, 24 Jun 2014, Sven Schreiber wrote:
Am 24.06.2014 07:25, schrieb Deborah Sy:
> The following is the Hansl code shown:
> Index value 0 is out of bounds.
> *** error in function init_C, line 4
>> S = Ss [,1:k-1]
> called by function SVAR_estimate
> called by function GUI_SVAR
>
> May I know what went wrong here? Your help is greatly appreciated.
>
Hi,
in hansl the matrix indices start at 1, not at zero, so that's the
error. Apparently here k=1, so k-1=0.
How can k=1 happen? Looking at the code (SVAR 0.994 shipped with my
gretl version) of 'init_C' shows that 'k=cols(Ss)', so the matrix
'Ss'
must have zero columns.
Now Ss is created by 'Ss = imp2exp(Rd)'. In terms of programming it's
probably an indication that some further input checking should be added
to the package. In terms of your application, it could be that
something's wrong with your restrictions. (Since 'Rd' seems to be
related internally to the restrictions, I think -- I'm not the author.)
So I suggest to double-check your restrictions. You can also show them
to us.
Thanks for looking into this, Sven. I'll try to see what's wrong asap.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
1:40 p.m.
On Sat, 28 Jun 2014, Riccardo (Jack) Lucchetti wrote:
Thanks for looking into this, Sven. I'll try to see what's
wrong asap.
One thing is, the model is very probably under-identified. I _must_ find
some time to include into the SVAR package a few functions for checking
for identification before estimating the model.
However, SVAR, shouldn't give the error you saw. I debugged it a little
and it seems that we have a weird bug in hansl. I just uploaded a new
version of SVAR.gfn which contains a workaround, but still we ought to fix
this properly. The problem is exemplified by the following piece of code.
<hansl>
matrix C = { NA, 0; NA, NA }
print C
loop i = 1..2 -q
loop j = 1..2 -q
c = C[i,j]
if ok(c)
print i j
endif
endloop
endloop
</hansl>
I temporarily fixed this in CVS via the following change, but I'm still
puzzled and I'm not at all sure this was the proper way to do it; Allin,
could you please take a look?
--- lib/src/geneval.c 26 Jun 2014 13:59:50 -0000 1.892
+++ lib/src/geneval.c 28 Jun 2014 13:34:58 -0000
@@ -3522,7 +3522,7 @@
errno = 0;
- if (na(x)) {
+ if (xna(x)) {
switch (f) {
case F_MISSING:
return 1.0;
@@ -3530,7 +3530,7 @@
case F_MISSZERO:
return 0.0;
default:
- if (na(x)) {
+ if (xna(x)) {
return NADBL;
}
}
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
2:02 p.m.
On Sat, 28 Jun 2014, Riccardo (Jack) Lucchetti wrote:
On Sat, 28 Jun 2014, Riccardo (Jack) Lucchetti wrote:
> Thanks for looking into this, Sven. I'll try to see what's wrong asap.
One thing is, the model is very probably under-identified. I _must_ find some
time to include into the SVAR package a few functions for checking for
identification before estimating the model.
However, SVAR, shouldn't give the error you saw. I debugged it a little and
it seems that we have a weird bug in hansl. I just uploaded a new version of
SVAR.gfn which contains a workaround, but still we ought to fix this
properly. The problem is exemplified by the following piece of code.
<hansl>
matrix C = { NA, 0; NA, NA }
print C
loop i = 1..2 -q
loop j = 1..2 -q
c = C[i,j]
if ok(c)
print i j
endif
endloop
endloop
</hansl>
I temporarily fixed this in CVS via the following change, but I'm still
puzzled and I'm not at all sure this was the proper way to do it; Allin,
could you please take a look?
--- lib/src/geneval.c 26 Jun 2014 13:59:50 -0000 1.892
+++ lib/src/geneval.c 28 Jun 2014 13:34:58 -0000
@@ -3522,7 +3522,7 @@
errno = 0;
- if (na(x)) {
+ if (xna(x)) {
switch (f) {
case F_MISSING:
return 1.0;
@@ -3530,7 +3530,7 @@
case F_MISSZERO:
return 0.0;
default:
- if (na(x)) {
+ if (xna(x)) {
return NADBL;
}
}
This is to do with the fact that we handle missing/non-finite values
differently in matrices versus series and scalars: the actual "missing"
code, represented by "NA", is only allowed in series and scalars while
"nan" (Not-a-Number) is only allowed in matrices. These values get
converted deoending of the target type in a "genr" expression: NAs get
converted into nans when writing a matrix, and nans get converted to NAs
when writing series or scalars.
The trouble with the hansl code above is that the variable "c" is not
explicitly typed. If you specify either "scalar c" or "matrix c" then
the
results of the ok() test will be consistent and as expected.
Now arguably it is a gretl bug that the type-ambiguity of "c" leads to
inconsistent results but I'm not sure what the correct fix would be; I'm
not convinced the change to geneval.c above is right.
Allin
2:17 p.m.
On Sun, 29 Jun 2014, Allin Cottrell wrote:
Now arguably it is a gretl bug that the type-ambiguity of
"c" leads to
inconsistent results but I'm not sure what the correct fix would be; I'm not
convinced the change to geneval.c above is right.
Neither am I. However, in the past the example above ran as intented until
some moment I was unable to pinpoint precisely. The example I sent is a
simplified ripoff from the SVAR hansl code, which used to work ok and now
doesn't. I committed a change in the SVAR script in CVS and uploaded the
updated gfn file which works around the issue by doing exactly what you
said (forcing the result to a scalar), but the fact remains that, at some
point, we must have introduced a slight backwards incpmpatibility, which
is what my CVS commit to the C source was supposed to fix somehow.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
12:17 p.m.
Hi Sven Schreiber,
The template matrix looks like this:
Fint
FDI (phil)
Forex
MFG Index
RRP
Inflation
Invest
Loans
Stock Ret
Fint
nan
0
0
0
0
0
0
0
0
FDI
nan
nan
nan
nan
nan
nan
0
0
nan
Forex
nan
nan
nan
0
0
0
0
0
0
MFG Index
nan
nan
nan
nan
0
nan
0
nan
nan
RRP
nan
0
0
0
nan
nan
0
0
0
Inflation
0
nan
0
nan
nan
nan
0
nan
nan
Invest
nan
0
0
0
nan
nan
nan
nan
nan
Loans
nan
0
0
0
nan
nan
0
nan
nan
Stock Ret
nan
nan
0
nan
nan
0
nan
nan
nan
With Fint as an exogenous variable.
By the way, thank you very much for replying.
Sincerely,
Deborah Sy
11:37 p.m.
Dear Mr. Sven Schreiber,
Thank you very much for your reply. Here is our template matrix.
Var1
Var2
Var3
Var4
Var5
Var6
Var7
Var8
Var9
Var1
nan
0
0
0
0
0
0
0
0
Var2
nan
nan
nan
nan
nan
nan
0
0
nan
Var3
nan
nan
nan
0
0
0
0
0
0
Var4
nan
nan
nan
nan
0
nan
0
nan
nan
Var5
nan
0
0
0
nan
nan
0
0
0
Var6
0
nan
0
nan
nan
nan
0
nan
nan
Var7
nan
0
0
0
nan
nan
nan
nan
nan
Var8
nan
0
0
0
nan
nan
0
nan
nan
Var9
nan
nan
0
nan
nan
0
nan
nan
nan
With Var1 as an endogenous variable.
Sincerely,
Deborah Sy
2:48 a.m.
Dear Mr. Sven Schreiber,
I would like to add some corrections regarding my previous message. Var1 is
not endogenous but rather, it is exogenous. Thank you!
Sincerely,
Deborah Sy
7 a.m.
Am 28.06.2014 04:48, schrieb Deborah Sy:
Dear Mr. Sven Schreiber,
I would like to add some corrections regarding my previous message. Var1
is not endogenous but rather, it is exogenous. Thank you!
Hi, it's a big matrix, and I don't have any concrete advice to offer,
sorry. In general I would just suggest to "play around a little" and
change the restrictions, reduce their number if possible (is this just
or over-identified?) until you get something that doesn't give you an
error. Then change your matrix element-by-element until the error
appears again, and that should hopefully give you an idea why it fails.
good luck,
sven
12:08 p.m.
Hi Mr. Sven Schreiber and Mr. Jack Lucchetti,
Thank you so much for your help!! I really appreciate it. In case you are
able to come up with some clue about the things that I have been doing
wrong, please feel free to contact me again anytime.
Sincerely yours,
Deborah Sy
On Sat, Jun 28, 2014 at 10:48 AM, Deborah Sy <deborahkimsy(a)gmail.com> wrote:
Dear Mr. Sven Schreiber,
I would like to add some corrections regarding my previous message. Var1
is not endogenous but rather, it is exogenous. Thank you!
Sincerely,
Deborah Sy
12:14 p.m.
Hi again, Mr. Sven Schreiber,
Pardon me for asking this, but can you take down all the template
restriction matrices I posted along with this message? Thank you so much!
This is for security purposes. I apologize for the inconvenience this might
cause you.
Sincerely yours,
Deborah Sy
On Sat, Jun 28, 2014 at 8:08 PM, Deborah Sy <deborahkimsy(a)gmail.com> wrote:
Hi Mr. Sven Schreiber and Mr. Jack Lucchetti,
Thank you so much for your help!! I really appreciate it. In case you are
able to come up with some clue about the things that I have been doing
wrong, please feel free to contact me again anytime.
Sincerely yours,
Deborah Sy
On Sat, Jun 28, 2014 at 10:48 AM, Deborah Sy <deborahkimsy(a)gmail.com>
wrote:
> Dear Mr. Sven Schreiber,
>
> I would like to add some corrections regarding my previous message. Var1
> is not endogenous but rather, it is exogenous. Thank you!
>
> Sincerely,
> Deborah Sy
>
4:11 p.m.
Am 28.06.2014 14:14, schrieb Deborah Sy:
Hi again, Mr. Sven Schreiber,
Pardon me for asking this, but can you take down all the template
restriction matrices I posted along with this message? Thank you so
much! This is for security purposes. I apologize for the inconvenience
this might cause you.
Deborah,
well I don't think your matrix constitutes a military secret, but apart
from that you have posted it on a public mailing list. Other list users
including myself cannot remove it.
sorry,
sven
6:07 a.m.
Hi guys,
I have installed the uploaded SVAR.gfn on Gretl version 1.9.12 but
unfortunately, the same error happens. How do I go about it?
Many thanks,
Deborah Sy
7:32 a.m.
On Wed, 2 Jul 2014, Deborah Sy wrote:
Hi guys,
I have installed the uploaded SVAR.gfn on Gretl version 1.9.12 but
unfortunately, the same error happens. How do I go about it?
The GUI interface needs some massaging, sorry :)
I'll see what I can do later today or tomorrow. By the way, I'm finally in
the process of adding the code for the identification checking procedure I
devoloped a few years ago, so the package will probably undergo a few
changes in the next few days. In the meantime, you should use the script
interface, which is the recommended method anyway.
Thanks for your help with debugging SVAR.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
8:06 a.m.
Hi all,
Yes, I'll be reading onto that. Another question I have is that I've been
using a 9x9 matrix with a lot of constraints (here, it is 36) which is not
random (i.e. they were backed up by theoretical assumptions). In such case,
given any possible combinations of the 0's, the determinant will most
likely be zero and therefore, no solution will exist (unless it is a
Cholesky-decomposition or if there really exists a combination that
provides a solution in this relatively large matrix). Does it follow that I
will not be able to estimate the SVAR because no such solution exists? This
has been bugging me for days already. Any ideas?
Truly grateful for your responses,
Deborah Sy
8:19 a.m.
Am 02.07.2014 10:06, schrieb Deborah Sy:
Hi all,
Yes, I'll be reading onto that. Another question I have is that I've
been using a 9x9 matrix with a lot of constraints (here, it is 36) which
is not random (i.e. they were backed up by theoretical assumptions). In
such case, given any possible combinations of the 0's, the determinant
will most likely be zero and therefore, no solution will exist (unless
it is a Cholesky-decomposition or if there really exists a combination
that provides a solution in this relatively large matrix). Does it
follow that I will not be able to estimate the SVAR because no such
solution exists? This has been bugging me for days already. Any ideas?
Ok, so 36 constraints in a 9x9 matrix sounds like it's just-identified
(at best). I guess you could check a rank deficiency of the template
matrix numerically by first substituting out the non-restricted elements
with ones, then multiply it (elementwise) with a random matrix (and then
use the rank() function).
Of course, using Jack's identification check algorithm which he will
publish soon apparently will be much better...
hth,
sven
9 a.m.
On Wed, 2 Jul 2014, Sven Schreiber wrote:
Ok, so 36 constraints in a 9x9 matrix sounds like it's
just-identified
(at best). I guess you could check a rank deficiency of the template
matrix numerically by first substituting out the non-restricted elements
with ones, then multiply it (elementwise) with a random matrix (and then
use the rank() function).
Of course, using Jack's identification check algorithm which he will
publish soon apparently will be much better...
Actually, my algorithm has been out for a long time:
http://ideas.repec.org/a/cup/etheor/v22y2006i02p235-257_06.html
I had originally coded the above in Ox, and now I'm porting the whole
thing in hansl and incorporating it into the SVAR package.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
9:14 a.m.
On Wed, 2 Jul 2014, Deborah Sy wrote:
Hi all,
Yes, I'll be reading onto that. Another question I have is that I've been
using a 9x9 matrix with a lot of constraints (here, it is 36) which is not
random (i.e. they were backed up by theoretical assumptions). In such case,
given any possible combinations of the 0's, the determinant will most
likely be zero and therefore, no solution will exist (unless it is a
Cholesky-decomposition or if there really exists a combination that
provides a solution in this relatively large matrix). Does it follow that I
will not be able to estimate the SVAR because no such solution exists? This
has been bugging me for days already. Any ideas?
The problem is unlikely to be the rank of your contraint matrix. If
you fill a 9x9 matrix with 36 zeros at random, the probability if it being
singular is zero (it shouldn't be difficult to prove this formally)
although of course the event is not impossible (nice example of an event
with 0 Lebesgue measure).
Assuming your matrix is nonsingular, the requirement of having 36
constrained elements is necessary, but not sufficient: it's an order
condition. If you look at my 2006 ET article I mentioned earlier, you'll
see that there is another condition that needs to be satisfied (I call it
the "structure" condition). Automtic checking of both is part of the code
I'm working on now. When I'm done, I'll post a message to the list.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
10:57 a.m.
Am 02.07.2014 11:14, schrieb Riccardo (Jack) Lucchetti:
On Wed, 2 Jul 2014, Deborah Sy wrote:
> Hi all,
>
> Yes, I'll be reading onto that. Another question I have is that I've been
> using a 9x9 matrix with a lot of constraints (here, it is 36) which is
> not
> random (i.e. they were backed up by theoretical assumptions). In such
> case,
> given any possible combinations of the 0's, the determinant will most
> likely be zero and therefore, no solution will exist (unless it is a
> Cholesky-decomposition or if there really exists a combination that
> provides a solution in this relatively large matrix). Does it follow
> that I
> will not be able to estimate the SVAR because no such solution exists?
> This
> has been bugging me for days already. Any ideas?
The problem is unlikely to be the rank of your contraint matrix. If you
fill a 9x9 matrix with 36 zeros at random, the probability if it being
singular is zero (it shouldn't be difficult to prove this formally)
although of course the event is not impossible (nice example of an event
with 0 Lebesgue measure).
Jack, I don't think this is true, for example the probability of getting
a column with all (9) zeros by chance may be small, but certainly not
zero if you allocate 36 zeros randomly. If I were good in combinatorics
I could give you the exact numbers, but I'm not.
Apart from that the constraints are not chosen at random. A line has
measure zero in a plane, but still I can pick exactly points on the line.
So I'm not saying the matrix is singular, but without further
investigation/thinking I cannot rule out that it is.
cheers,
sven
11:20 a.m.
On Wed, 2 Jul 2014, Sven Schreiber wrote:
> The problem is unlikely to be the rank of your contraint matrix.
If you
> fill a 9x9 matrix with 36 zeros at random, the probability if it being
> singular is zero (it shouldn't be difficult to prove this formally)
> although of course the event is not impossible (nice example of an event
> with 0 Lebesgue measure).
Jack, I don't think this is true, for example the probability of getting
a column with all (9) zeros by chance may be small, but certainly not
zero if you allocate 36 zeros randomly. If I were good in combinatorics
I could give you the exact numbers, but I'm not.
Apart from that the constraints are not chosen at random. A line has
measure zero in a plane, but still I can pick exactly points on the line.
So I'm not saying the matrix is singular, but without further
investigation/thinking I cannot rule out that it is.
You're absolutely right, my friend. After all, the number of ways you can
put n*(n-1)/2 zeros into a square matrix with n rows is finite, and some
of those do give rise to singular matrices, so yes, it isn't a Lebesgue 0.
I stand corrected.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
2:10 p.m.
On Wed, 2 Jul 2014, Riccardo (Jack) Lucchetti wrote:
On Wed, 2 Jul 2014, Sven Schreiber wrote:
>> The problem is unlikely to be the rank of your contraint matrix. If you
>> fill a 9x9 matrix with 36 zeros at random, the probability if it being
>> singular is zero (it shouldn't be difficult to prove this formally)
>> although of course the event is not impossible (nice example of an event
>> with 0 Lebesgue measure).
>
> Jack, I don't think this is true, for example the probability of getting
> a column with all (9) zeros by chance may be small, but certainly not
> zero if you allocate 36 zeros randomly. If I were good in combinatorics
> I could give you the exact numbers, but I'm not.
>
> Apart from that the constraints are not chosen at random. A line has
> measure zero in a plane, but still I can pick exactly points on the line.
>
> So I'm not saying the matrix is singular, but without further
> investigation/thinking I cannot rule out that it is.
You're absolutely right, my friend. After all, the number of ways you can put
n*(n-1)/2 zeros into a square matrix with n rows is finite, and some of those
do give rise to singular matrices, so yes, it isn't a Lebesgue 0. I stand
corrected.
I believe Prob(one column is all zeros) = 9 * (1/9)^9 = 2.3e-08
when inserting 9 zeros at random; it would be higher for 36 zeros.
Allin
2:24 p.m.
On Wed, 2 Jul 2014, Allin Cottrell wrote:
>>> The problem is unlikely to be the rank of your contraint
matrix. If you
>>> fill a 9x9 matrix with 36 zeros at random, the probability if it being
>>> singular is zero (it shouldn't be difficult to prove this formally)
>>> although of course the event is not impossible (nice example of an event
>>> with 0 Lebesgue measure).
>>
>> Jack, I don't think this is true, for example the probability of getting
>> a column with all (9) zeros by chance may be small, but certainly not
>> zero if you allocate 36 zeros randomly. If I were good in combinatorics
>> I could give you the exact numbers, but I'm not.
>>
>> Apart from that the constraints are not chosen at random. A line has
>> measure zero in a plane, but still I can pick exactly points on the line.
>>
>> So I'm not saying the matrix is singular, but without further
>> investigation/thinking I cannot rule out that it is.
>
> You're absolutely right, my friend. After all, the number of ways you can
> put n*(n-1)/2 zeros into a square matrix with n rows is finite, and some of
> those do give rise to singular matrices, so yes, it isn't a Lebesgue 0. I
> stand corrected.
I believe Prob(one column is all zeros) = 9 * (1/9)^9 = 2.3e-08
when inserting 9 zeros at random; it would be higher for 36 zeros.
Now _this_ is fun:
<hansl>
set echo off
set messages off
set seed 20140702
function matrix C(scalar n)
/* build a random nxn matrix with
n*(n-1)/2 zeros in random locations
*/
scalar n2 = n*n
scalar m = round((n2 - n)/2)
matrix a = mnormal(n2,1)
matrix e = msortby(mnormal(n2,1) ~ seq(1,n2)', 1)
matrix e = e[1:m,2]
a[e] = 0
return mshape(a, n, n)
end function
function scalar sing_freq(scalar n, scalar repli)
x = 0
loop repli --quiet
matrix a = C(n)
if rank(a) < n
x++
endif
endloop
return x/repli
end function
# --- main -------------------------------------
scalar repli = 5000
loop i=2..9 --quiet
scalar p = sing_freq(i, repli)
printf "%2d: %10.6f\n", i, p
endloop
</hansl>
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
10:48 a.m.
Hi Jack,
Thank you for your explanation. It was very enlightening. Hope to hear from
you asap. :)
Regards,
Deborah
2:04 p.m.
On Wed, 2 Jul 2014, Deborah Sy wrote:
Hi Jack,
Thank you for your explanation. It was very enlightening. Hope to hear from
you asap. :)
Ok, I uploaded an updated version of the package (v. 0.997) which includes
the identification code. All the relevant changes are in CVS too; people
who want to try out the new features will probably need the pdf
documentation as well (especially section 5).
One way to get it is to download the tarball of the doc directory
(http://gretl.cvs.sourceforge.net/viewvc/gretl/gretl-addons/SVAR/doc/?view...)
and compile it by yourselves. If you prefer not to bake your own cake,
you'll find a copy here:
http://www2.econ.univpm.it/servizi/hpp/lucchetti/gretl/SVAR.pdf
Comments welcome.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
4:58 p.m.
Hi Jack,
It is nice that the Blanchard-Quah model was given an example in the
revised manual. However, as I try it I stumbled upon an error which I
think, requires my matrix to be positive definite. The following is the
script code shown:
Matrix is not positive definite
*** error in function imp2exp
matrix s = R'invpd(R*R')*d
called by function init_C
called by function SVAR_estimate
Error executing script: halting
SVAR_estimate(&Mod)
Any feedbacks about this?
Many thanks,
Deborah
7:08 p.m.
On Thu, 3 Jul 2014, Deborah Sy wrote:
Hi Jack,
It is nice that the Blanchard-Quah model was given an example in the
revised manual. However, as I try it I stumbled upon an error which I
think, requires my matrix to be positive definite. The following is the
script code shown:
Matrix is not positive definite
*** error in function imp2exp
> matrix s = R'invpd(R*R')*d
called by function init_C
called by function SVAR_estimate
Error executing script: halting
> SVAR_estimate(&Mod)
Any feedbacks about this?
Ah, this is interesting. Is the structure of your constraints the same as
the one you postred a few days ago?
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
6:44 a.m.
Hi Jack,
Yes, the one I used is the one posted few days ago. I also tried 2 other
combinations of the matrices but none of them seems to work. The error
shown for the three trials is the same as above.
Thanks,
Deborah
5:51 a.m.
On Thu, 3 Jul 2014, Deborah Sy wrote:
Hi Jack,
Yes, the one I used is the one posted few days ago. I also tried 2 other
combinations of the matrices but none of them seems to work. The error
shown for the three trials is the same as above.
I believe that the model you posted is under-identified. The funny thing
is, my checking procedures don't spot it as such. It may be a problem with
my code or, possibly, a problem with the algorithm itself. I'm working on
this. For the moment, your best bet is to increase the number of
restrictions and try to achieve identification.
By the way, however, your model seems to imply that Var1 and Var6 must be
incorrelated, as they share no structural shocks. Are you sure this is
intended?
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
8:45 a.m.
Hello Jack,
I'll try to increase my restrictions. Yes, I intended for such to happen.
Why? Is there anything wrong?
I'm just curious. After some time, will the GUI of the SVAR change provided
that you are able to finish the codes?
Anyway, I tried to replicate the Section 5 in the manual as to identify for
restrictions in my software. The following is the output shown:
gretl version 1.9.12
Current session: 2014-07-04 16:35
? set echo off
SVAR 0.991, 2013-03-13 (Riccardo "Jack" Lucchetti)
Read datafile C:\Program Files (x86)\gretl\data\misc\sw_ch14.gdt
periodicity: 4, maxobs: 164
observations range: 1959:1 to 1999:4
Listing 3 variables:
0) const 1) LHUR 2) PUNEW
scalar is_identified = SVAR_ident(
The symbol
'SVAR_ident' is undefined
Error executing script: halting
scalar is_identified = SVAR_ident(&Mod)
Seeing "SVAR 0.991, 2013-03-13" above, I also tried to do this by using
SVAR 0.997 but it didn't work as well. The same thing happened except that
"SVAR 0.991, 2013-03-13 (Riccardo "Jack" Lucchetti)" was ommitted in
the
output.
Many thanks,
Deborah
9:02 a.m.
Hi again Jack,
I tried to replicate also your Blquah example in my software. It generated
the output as expected but this was shown in the end:
Error executing script: halting
bfail = SVAR_boot(&BQModel, 1024, 0.9)
Also, I'm wondering as to how my results won't look like this format:
Variable DY cumulated
Unconstrained Sigma:
0.82634 -0.16843
-0.16843 0.08711
Long-run matrix:
0.478 2.176
-3.992 1.463
Long-run Sigma:
0.251 0.044
0.044 15.322
My output looks chaotic. It's like this:
Variable fdi cumulated
Unconstrained Sigma:
0.01642126148410.59459 -0.00047 -0.02944 0.00319
0.01337212265108.80190243388876.04952 5.25354
126148410.5945945412015205400175000.00000-462725304.40788-2988188750.4771043940484.06802195691726.63337-8472739510774676500.0000019765878389136777000.0000029098094183.22789
-0.00047-462725304.40788 0.39197 0.46347 0.01623
0.02631407239650.33888-2190288069.93774 -20.03538
-0.02944-2988188750.47710 0.46347 39.40833 0.12328
0.27657-5470335015.2043172384855090.01930 115.02088
0.0031943940484.06802 0.01623 0.12328 0.01314
0.00146-533396824.97609569561794.54559 -1.43007
0.01337195691726.63337 0.02631 0.27657 0.00146
0.19816910242881.24939628329430.90083 -2.01899
212265108.80190-8472739510774676500.00000407239650.33888-5470335015.20431-533396824.97609910242881.249391892663732226180000000.00000-389286941941788770000.00000373688994691.24390
243388876.0495219765878389136777000.00000-2190288069.9377472384855090.01930569561794.54559628329430.90083-389286941941788770000.000002029131226762805200000.00000335520049658.56036
5.2535429098094183.22789 -20.03538 115.02088 -1.43007
-2.01899373688994691.24390335520049658.56036 19746.18549
Long-run matrix:
3.873 -0.000 5.454 0.095 -57.958 4.241 0.000 0.000 0.028
21618696687.792
0.7775422141563.020-301906065.548-39492833829.5819589688103.383 0.006
0.0815704926.063
-72.686 -0.000 -5.740 0.658 -67.261 -11.849 0.000 0.000 0.060
159.550 0.000 43.007 -0.751 373.301 48.060 -0.000 -0.000 -0.101
-14.705 -0.000 0.051 0.214 -42.939 -1.210 0.000 0.000 0.029
-6.982 0.000 3.107 -0.097 40.300 7.383 -0.000 -0.000 0.026
2880733118478.700
53.919577971098501.013-47300383935.44512141498082074.432955674169700.129
-29.740 -55.771-5237104403.800
4260342936164.078
62.380527013290549.206-52129622222.77611547585114408.1581284829507715.273
-34.634 -58.338-6973263074.691
13361.884 0.0002622.961-139.30625549.6072525.218 -0.000 -0.000 -9.112
Long-run Sigma:
163.44763274288799.312 262.655-1023.376 153.231
-59.393-34668832129413.941-37079568366827.719-71658.113
63274288799.31295531399329602830000.000-10144224947.681-57500484911.35729636288979.946-8365827154.848-5869992106664872900000.000-4586490758418398200000.000-4180800718734.184
262.655-10144224947.681 805.286-3134.494
355.657-192.614-89128730506594.281-97421372665469.781-211574.335
-1023.376-57500484911.357-3134.49413195.873-1399.302
916.970365016156747478.500390723124062996.060867816.301
153.23129636288979.946 355.657-1399.302 175.989
-84.901-42419664179726.031-45885699014244.812-95263.905
-59.393-8365827154.848-192.614 916.970 -84.901
85.49924011847813818.51624838692527584.27757947.124
-34668832129413.941-5869992106664872900000.000-89128730506594.281365016156747478.500-42419664179726.03124011847813818.51610647257220661315000000000.00011430556765385453000000000.00024427885094137932.000
-37079568366827.719-4586490758418398200000.000-97421372665469.781390723124062996.060-45885699014244.81224838692527584.27711430556765385453000000000.00012368169130824001000000000.00026261480021310868.000
Is there any way to change it? Please note that I am unable to transform my
variables because it might affect the long run effects.
Thanks,
Deborah
12:25 p.m.
On Fri, 4 Jul 2014, Deborah Sy wrote:
My output looks chaotic. It's like this:
Variable fdi cumulated
[...]
Uh, that's probably the effect of some problem with the data. I don't have
time foir this right now, but if you send me your data and your script I
can have a closer look at what's actually happening.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
12:23 p.m.
On Fri, 4 Jul 2014, Deborah Sy wrote:
Hello Jack,
I'll try to increase my restrictions. Yes, I intended for such to happen.
Why? Is there anything wrong?
I'm just curious. After some time, will the GUI of the SVAR change provided
that you are able to finish the codes?
Anyway, I tried to replicate the Section 5 in the manual as to identify for
restrictions in my software. The following is the output shown:
gretl version 1.9.12
Current session: 2014-07-04 16:35
? set echo off
SVAR 0.991, 2013-03-13 (Riccardo "Jack" Lucchetti)
Read datafile C:\Program Files (x86)\gretl\data\misc\sw_ch14.gdt
periodicity: 4, maxobs: 164
observations range: 1959:1 to 1999:4
Listing 3 variables:
0) const 1) LHUR 2) PUNEW
> scalar is_identified = SVAR_ident(
The symbol 'SVAR_ident' is undefined
Error executing script: halting
> scalar is_identified = SVAR_ident(&Mod)
Seeing "SVAR 0.991, 2013-03-13" above, I also tried to do this by using
SVAR 0.997 but it didn't work as well. The same thing happened except that
"SVAR 0.991, 2013-03-13 (Riccardo "Jack" Lucchetti)" was ommitted in
the
output.
It look as if you have both versions installed and, for some reason, gretl
is using the old one. I couldn't say why.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
3823
days inactive
3833
days old
33 comments
4 participants
participants (4)
-
Allin Cottrell -
Deborah Sy -
Riccardo (Jack) Lucchetti -
Sven Schreiber