4:47 p.m.
Thanks Ignacio. I know that the single equation can be estimated by OLS. In my
application I have no stationary time series and I work using variables in
levels. I'd like use bootstrap method based on residuals, which are
uncorrelated. I have considered only the equation
1. ols Y const Y(-1) Y(-2) X(-1) X(-2) Z(-1) Z(-2)
I am interested in testing Granger causality from X to Y, so I test this null
hypothesis. Under this hypothesis I estimate the model 1 and save the
coefficients and the residuals u(t). In this way I obtain the pseudo-data Y*
given by
Y*(t) = a + b Y*(t-1) + c Y*(t-1) + d Z(t-1) + e Z(t-2) + u*(t)
where u*(t) are the bootstrap residuals and Z is the original regressor. The
p-
value is small. If I replace every equation of the VAR the p-value is very
large. In my opinion, considering only the first equation, the results are
better than those which replace the VAR because the statistics tests is
large.
In yuor opinion can I work considering only one equation of the VAR in my
bootstrap method?
Best regards
Alessandro
----Messaggio originale----
Da: gretl-users-request(a)lists.wfu.edu
Data: 23/02/2012 15.23
A: <gretl-users(a)lists.wfu.edu>
Ogg: Gretl-users Digest, Vol 61, Issue 18
Send Gretl-users mailing list submissions to
gretl-users(a)lists.wfu.edu
To subscribe or unsubscribe via the World Wide Web, visit
http://lists.wfu.edu/mailman/listinfo/gretl-users
or, via email, send a message with subject or body 'help' to
gretl-users-request(a)lists.wfu.edu
You can reach the person managing the list at
gretl-users-owner(a)lists.wfu.edu
When replying, please edit your Subject line so it is more specific
than "Re: Contents of Gretl-users digest..."
Today's Topics:
1. Re: CARR model MLE ( Dan B?so? )
2. Re: Problem with bootstrap result (Ignacio Diaz-Emparanza)
3. Re: CARR model MLE ( Dan B?so? )
----------------------------------------------------------------------
Message: 1
Date: Thu, 23 Feb 2012 12:11:05 +0100
From: " Dan B?so? " <eubie(a)centrum.cz>
Subject: Re: [Gretl-users] CARR model MLE
To: <gretl-users(a)lists.wfu.edu>
Message-ID: <20120223121105.924522A4(a)centrum.cz>
Content-Type: text/plain; charset=UTF-8
Thank you both a lot. I was out of my mind to overlook the +/- thing. Thank
you
also for the exp() trick, so far I thought the only way to ensure
positiveness is to formulate a model for log(Rng).
Best regards,
Daniel
______________________________________________________________
> Od: gretl-users-request(a)lists.wfu.edu
> Komu: <gretl-users(a)lists.wfu.edu>
> Datum: 22.02.2012 13:48
> P?edm?t: Gretl-users Digest, Vol 61, Issue 17
>
>Send Gretl-users mailing list submissions to
> gretl-users(a)lists.wfu.edu
>
>To subscribe or unsubscribe via the World Wide Web, visit
> http://lists.wfu.edu/mailman/listinfo/gretl-users
>or, via email, send a message with subject or body 'help' to
> gretl-users-request(a)lists.wfu.edu
>
>You can reach the person managing the list at
> gretl-users-owner(a)lists.wfu.edu
>
>When replying, please edit your Subject line so it is more specific
>than "Re: Contents of Gretl-users digest..."
>
>
>Today's Topics:
>
> 1. CARR model MLE ( Dan B?so? )
> 2. Re: CARR model MLE (Allin Cottrell)
> 3. Re: CARR model MLE (Riccardo (Jack) Lucchetti)
> 4. Re: Bootstrap VAR models (alexkakashi(a)libero.it)
> 5. Problem with bootstrap result (alexkakashi(a)libero.it)
>
>
>----------------------------------------------------------------------
>
>Message: 1
>Date: Tue, 21 Feb 2012 21:15:02 +0100
>From: " Dan B?so? " <eubie(a)centrum.cz>
>Subject: [Gretl-users] CARR model MLE
>To: <gretl-users(a)lists.wfu.edu>
>Message-ID: <20120221211502.4991D432(a)centrum.cz>
>Content-Type: text/plain; charset=UTF-8
>
>
>Dear fellow users,
>
>I am trying to write a script for MLE estimation of a CARR model, similar to
GARCH. It is a MEM model in daily ranges of some instrument, i,e,
>
>Eq. 1: lRng = lambda * epsilon # epsilon is exponentially distributed
here,
giving the shape of LL
>Eq. 2: lambda = a + b*Rng(-1) + c*lambda(-1)
>
>I tried the following code but the algorithm never converges. What should I
improve?
>Thanks in advance, Daniel
>
>
>
>scalar a = 0.1
>scalar b = 0.4
>scalar c = 0.4
>
>mle ll = -ln(lambda) + Rng/lambda
>
> series lambda = mean(Rng)
> series lambda = a + b*Rng(-1) + c*lambda(-1)
>
> params a b c
>end mle
>
>
>
>------------------------------
>
>Message: 2
>Date: Tue, 21 Feb 2012 15:57:19 -0500 (EST)
>From: Allin Cottrell <cottrell(a)wfu.edu>
>Subject: Re: [Gretl-users] CARR model MLE
>To: Gretl list <gretl-users(a)lists.wfu.edu>
>Message-ID:
> <alpine.LNX.2.01.1202211548120.22658(a)waverley.dhcp.wfu.edu>
>Content-Type: text/plain; charset="iso-8859-2"
>
>On Tue, 21 Feb 2012, Dan B?so? wrote:
>
>> I am trying to write a script for MLE estimation of a CARR model, similar
to GARCH. It is a MEM model in daily ranges of some instrument, i,e,
>>
>> Eq. 1: lRng = lambda * epsilon # epsilon is exponentially distributed
here, giving the shape of LL
>> Eq. 2: lambda = a + b*Rng(-1) + c*lambda(-1)
>>
>> I tried the following code but the algorithm never
>> converges. What should I improve?
>
>> scalar a = 0.1
>> scalar b = 0.4
>> scalar c = 0.4
>>
>> mle ll = -ln(lambda) + Rng/lambda
>
>Improve here: use the correct loglikelihood!
>
> mle ll = -ln(lambda) - Rng/lambda
>
>> series lambda = mean(Rng)
>> series lambda = a + b*Rng(-1) + c*lambda(-1)
>>
>> params a b c
>> end mle
>
>You may also have to constrain all the parameter values to be
>positive, if I'm reading the CARR literature correctly. In
>which case you could, for example, express lambda as
>exp(a) + exp(b)*Rng(-1) + exp(c)*lambda(-1).
>
>Allin Cottrell
>
>------------------------------
>
>Message: 3
>Date: Tue, 21 Feb 2012 22:25:59 +0100 (CET)
>From: "Riccardo (Jack) Lucchetti" <r.lucchetti(a)univpm.it>
>Subject: Re: [Gretl-users] CARR model MLE
>To: Gretl list <gretl-users(a)lists.wfu.edu>
>Message-ID: <alpine.DEB.2.02.1202212216510.14044(a)ec-4.econ.univpm.it>
>Content-Type: text/plain; charset="utf-8"
>
>On Tue, 21 Feb 2012, Allin Cottrell wrote:
>
>> On Tue, 21 Feb 2012, Dan B?so? wrote:
>>
>>> I am trying to write a script for MLE estimation of a CARR model,
similar
>>> to GARCH. It is a MEM model in daily ranges of some
instrument, i,e,
>>>
>>> Eq. 1: lRng = lambda * epsilon # epsilon is exponentially distributed
>>> here, giving the shape of LL
>>> Eq. 2: lambda = a + b*Rng(-1) + c*lambda(-1)
>>>
>>> I tried the following code but the algorithm never converges. What should
I
>>> improve?
>>
>>> scalar a = 0.1
>>> scalar b = 0.4
>>> scalar c = 0.4
>>>
>>> mle ll = -ln(lambda) + Rng/lambda
>>
>> Improve here: use the correct loglikelihood!
>>
>> mle ll = -ln(lambda) - Rng/lambda
>>
>>> series lambda = mean(Rng)
>>> series lambda = a + b*Rng(-1) + c*lambda(-1)
>>>
>>> params a b c
>>> end mle
>>
>> You may also have to constrain all the parameter values to be positive,
if
>> I'm reading the CARR literature correctly. In which case
you could, for
>> example, express lambda as
>> exp(a) + exp(b)*Rng(-1) + exp(c)*lambda(-1).
>
>Alternatively, you can wrap everything up into a function so that checking
>is done automatically inside the function, as in
>
><hansl>
>function series CarrLoglik(series y, matrix theta)
> if minc(theta)<0
> series ret = NA
> else
> scalar a = theta[1]
> scalar b = theta[2]
> scalar c = theta[3]
> series lambda = mean(y)
> series lambda = a + b * y(-1) + c * lambda(-1)
> series ret = -ln(lambda) - y/lambda
> endif
>
> return ret
>end function
>
>param = {0.1;0.4;0.4}
>
>mle ll = CarrLoglik(Rng, theta)
> params theta
>end mle
></hansl>
>
>
>Riccardo (Jack) Lucchetti
>Dipartimento di Economia
>Universit? Politecnica delle Marche
>
>r.lucchetti(a)univpm.it
>http://www.econ.univpm.it/lucchetti
>
>------------------------------
>
>Message: 4
>Date: Wed, 22 Feb 2012 11:37:33 +0100 (CET)
>From: "alexkakashi(a)libero.it" <alexkakashi(a)libero.it>
>Subject: Re: [Gretl-users] Bootstrap VAR models
>To: <gretl-users(a)lists.wfu.edu>
>Message-ID:
> <9682170.1207401329907053657.JavaMail.defaultUser@defaultHost>
>Content-Type: text/plain;charset="UTF-8"
>
>Thanks Stev.
>
>I have used this procedure to study Granger causality from X to Y in a
>trivariate VAR:
>
>r=158 ## number of observation
>
>ysim=Y
>
>xsim=X
>
>Zsim=Z
>
>system method=sur
>equation Y const Y(-1) Y(-2) Z(-1) Z(-2)
>equation X const Y(-1) Y(-2) X(-1) X(-2) Z(-1) Z(-2)
>equation Z const Y(-1) Y(-2) X(-1) X(-2) Z(-1) Z(-2)
>end system
>
>genr residui=$uhat
>
>genr M=$coeff
>
>loop for i=3..r --quiet
>
>ysim[i]=M[1]+M[2]*ysim[i-1]+M[3]*ysim[i-2]+M[4]*zsim[i-1]+M[5]*zsim[i-2]
>
>xsim[i]=M[6]+M[7]*ysim[i-1]+M[8]*ysim[i-2]+M[9]*xsim[i-1]+M[10]*xsim[i-2]+M
[11]
>*zsim[i-1]+M[12]*zsim[i-2]
>
>zsim[i]=M[13]+M[14]*ysim[i-1]+M[15]*ysim[i-2]+M[16]*xsim[i-1]+M[17]*xsim[i-2]
+M
>[18]*zsim[i-1]+M[19]*zsim[i-2]
>
>loop replics --quiet ### BOOTSTRAP
>
>smpl full
>
>Ysim=ysim
>
>Xsim=xsim
>
>Zsim=zsim
>
>genr A=resample(residui)
>
>loop for i=3..r --quiet
>
>Ysim[i]=ysim[i]+A[i-2,1]
>
>Xsim[i]=ysim[i]+A[i-2,2]
>
>Zsim[i]=ysim[i]+A[i-2,3]
>
>endloop
>
>...................endloop
>
>There is a strange result when I apply the bootstrap procedure. In fact the
>statistic test based on observed data is very large, but the bootstrap p-
value
>is greater then 0.8 and I do not reject the null hypothesis of
Granger non-
>causality from X to Y. The results are different if I work considering only
the
>equation
>
>equation Y const Y(-1) Y(-2) X(-1) X(-2) Z(-1) Z(-2)
>
>Under the null hypothesis I replace the series Y. In this case the bootstrap
p-
>value is very small and I reject the null hypothesis of
no-causality. I
don't
>understant the different between these p-values.
>
>Best regards.
>
>Alessandro
>
>
>
>>----Messaggio originale----
>>Da: gretl-users-request(a)lists.wfu.edu
>>Data: 16/02/2012 18.00
>>A: <gretl-users(a)lists.wfu.edu>
>>Ogg: Gretl-users Digest, Vol 61, Issue 11
>>
>>Send Gretl-users mailing list submissions to
>> gretl-users(a)lists.wfu.edu
>>
>>To subscribe or unsubscribe via the World Wide Web, visit
>> http://lists.wfu.edu/mailman/listinfo/gretl-users
>>or, via email, send a message with subject or body 'help' to
>> gretl-users-request(a)lists.wfu.edu
>>
>>You can reach the person managing the list at
>> gretl-users-owner(a)lists.wfu.edu
>>
>>When replying, please edit your Subject line so it is more specific
>>than "Re: Contents of Gretl-users digest..."
>>
>>
>>Today's Topics:
>>
>> 1. Bootstrap VAR models (alexkakashi(a)libero.it)
>> 2. Re: Bootstrap VAR models (Sven Schreiber)
>> 3. Gretl error/crash ((s) Simon Grenville-Wood)
>> 4. Re: Gretl error/crash (Riccardo (Jack) Lucchetti)
>> 5. Re: Gretl error/crash ((s) Simon Grenville-Wood)
>> 6. Random-effects panel-data (Helgi Tomasson)
>>
>>
>>----------------------------------------------------------------------
>>
>>Message: 1
>>Date: Thu, 16 Feb 2012 09:34:37 +0100 (CET)
>>From: "alexkakashi(a)libero.it" <alexkakashi(a)libero.it>
>>Subject: [Gretl-users] Bootstrap VAR models
>>To: gretl-users(a)lists.wfu.edu
>>Message-ID: <25986877.1523161329381277575.JavaMail.root@wmail22>
>>Content-Type: text/plain;charset="UTF-8"
>>
>>Hi,
>>
>>I have the following question. Let us consider a trivariate VAR model. The
>>parameters of the model are estimated using:
>>
>>system method=sur
>>equation Y const Y(-1) X(-1) Z(-1)
>>equation X const Y(-1) X(-1) Z(-1)
>>equation Z const Y(-1) X(-1) Z(-1)
>>end system
>>
>>I'd like study Granger causality from X to Y and the critical values are
>>calculated using bootstrap method based on residuals. Is this procedure
>>implemented in gretl?
>>
>>Best regards.
>>
>>Alessandro
>>
>>
>>------------------------------
>>
>>Message: 2
>>Date: Thu, 16 Feb 2012 10:18:44 +0100
>>From: Sven Schreiber <svetosch(a)gmx.net>
>>Subject: Re: [Gretl-users] Bootstrap VAR models
>>To: gretl-users(a)lists.wfu.edu
>>Message-ID: <4F3CC9F4.4050602(a)gmx.net>
>>Content-Type: text/plain; charset=ISO-8859-1
>>
>>On 02/16/2012 09:34 AM, alexkakashi(a)libero.it wrote:
>>> Hi,
>>>
>>> I have the following question. Let us consider a trivariate VAR model.
The
>>> parameters of the model are estimated using:
>>>
>>> system method=sur
>>> equation Y const Y(-1) X(-1) Z(-1)
>>> equation X const Y(-1) X(-1) Z(-1)
>>> equation Z const Y(-1) X(-1) Z(-1)
>>> end system
>>>
>>> I'd like study Granger causality from X to Y and the critical values are
>>> calculated using bootstrap method based on residuals. Is this procedure
>>> implemented in gretl?
>>>
>>
>>
>>Not that I know of, but have a look at the varsimul() and resample()
>>functions, with those it's not too difficult to program it.
>>
>>hth,
>>sven
>>
>>
>>------------------------------
>>
>>Message: 3
>>Date: Thu, 16 Feb 2012 14:50:11 +0000
>>From: "(s) Simon Grenville-Wood"
>> <simon.grenville-wood(a)students.plymouth.ac.uk>
>>Subject: [Gretl-users] Gretl error/crash
>>To: "gretl-users(a)lists.wfu.edu" <gretl-users(a)lists.wfu.edu>
>>Message-ID:
>> <A775801493FD6643B49F9B292B0376FB2F63D3CC(a)AMSPRD0302MB113.eurprd03.prod.
>outlook.com>
>>
>>Content-Type: text/plain; charset="iso-8859-1"
>>
>>Hi,
>>
>>I'm currently attempting a multinomial logit model on a large sample of
panel
>data.
>>I'm having a problem where I need to be able to create dummies for each
>individual in the sample to control for the individual fixed effects.
>>
>>In order to do so while using the logit model, I need to dummify the
person
>ID (PID) in the sample.
>>I've opened the gretl console and I enter the following commands:
>>
>>discrete PID <----transform the variable, 19,421 values, to be
discrete
>>dummify PID <----create dummies for each individual to
control for
fixed
>effects
>>
>>The last command causes gretl to crash, giving a libcairo_32.dll error.
This
>>happens on whichever computer I am using. Occasionally, if I use the menus
>>instead of the console, it will give an 'out of memory' error. My PC is
>>definitely sufficiently powerful.
>>>
>>>Does anyone have any ideas as to how I can get around this issue?
>>>
>>>Thanks,
>>>
>>>Simon
>>>
7:34 p.m.
New subject: Problem with bootstrap result
On Thu, 23 Feb 2012, alexkakashi(a)libero.it wrote:
I know that the single equation can be estimated by OLS. In
my application I have no stationary time series and I work
using variables in levels. I'd like use bootstrap method
based on residuals [...]
On a point of etiquette, please do not dump 1700 lines of
gretl-users digest onto the end of mail to the list.
But as for the bootstrap test of a restriction on a VAR, this
is quite an interesting question. At present you can't use
gretl's "restrict" command on VARs, but you can use "restrict"
if you formulate the model as an SUR system. Simulating a VAR
"by hand" can be quite tricky, but gretl offers the varsimul()
function to help you get it right; this can be applied using
the results from SUR estimation.
I'm attaching an example script which bootstraps a Granger
non-causality test (using a data file from the gretl package).
If Jack Lucchetti is interested I bet he can come up with a
more elegant version, but this one seems to do the job. On
this example, at least, the bootstrapped p-value is similar to
that obtained from a standard Wald test.
Allin Cottrell
10:08 p.m.
New subject: Problem with bootstrap result
Allin,
I'm attaching an example script which bootstraps a Granger
non-causality test (using a data file from the gretl package).
For some reason, I get an error message when running your example script:
Data types not conformable for operation
> series y1 = S[,1]
This seems odd because a) it's your script, and b) I can generate a series from a
matrix just fine in the console. The following works fine (after getting the error
above):
T = $nobs
Doh = mnormal(T,2)
Duh = Doh[,1]
I'm running the latest snapshot on Windows.
PS
10:41 p.m.
New subject: Problem with bootstrap result
On Thu, 23 Feb 2012, Summers, Peter wrote:
Allin,
> I'm attaching an example script which bootstraps a Granger
> non-causality test (using a data file from the gretl
> package).
For some reason, I get an error message when running your example script:
Data types not conformable for operation
>> series y1 = S[,1]
Oops, you're right, I forgot: I made a change to the gretl
code so that the matrix produced by varsimul() preserves the
dates information (if any) attached to the input matrices.
Without that modification you'd get an error unless you reset
the sample appropriately before converting from matrix columns
to series.
(BTW, the resample() function as applied to a matrix now also
preserves dates information.)
That change is now in CVS and snapshots.
Allin
4686
days inactive
4686
days old
3 comments
3 participants
participants (3)
-
alexkakashi＠libero.it -
Allin Cottrell -
Summers, Peter