iristoolbox Discussions Rss Feedhttps://iristoolbox.codeplex.com/discussionsiristoolbox Discussions Rss DescriptionNew Post: FEVD and sign restrictionshttp://iristoolbox.codeplex.com/discussions/661600<div style="line-height: normal;">I have a question regarding FEVD in sign-restricted SVAR. I identified only two shocks and then computed FEVD for 5 periods for closest-to-median draw, but it gave me a series of 4x4x5 matrices, like in the Cholesky SVAR.
<br />
<br />
How can I calculate FEVD for structural shocks identified by sign restrictions ? <br />
</div>michall23fullSun, 26 Feb 2017 21:43:05 GMTNew Post: FEVD and sign restrictions 20170226094305PNew Post: FEVD and Historical Decomposition After Sign Restrictionhttp://iristoolbox.codeplex.com/discussions/656074<div style="line-height: normal;">You can do it like that[Fa,Fr,VDabs,VDrel]=fevd(svar2(1:1),5); It takes only first draw from svar2 object. It was previously sorted by distance to median svar2=sort(svar,svardata,'S(:,:,1:16)')<br />
</div>michall23fullSun, 26 Feb 2017 21:37:29 GMTNew Post: FEVD and Historical Decomposition After Sign Restriction 20170226093729PNew Post: Estimating DSGE Models with Forward Guidancehttp://iristoolbox.codeplex.com/discussions/659573<div style="line-height: normal;">I found how to solve this problem by setting standard deviation to zero in J database to 'vary=' option for the entire period until 2008q3:<br />
<pre><code>J = struct;
for v=sprintfc('std_rm_sh%d',1:o.nant)
J.(v{1})=tseries(startHist:qq(2008,3),0);
end
filterOpt = {'relative=',false,'objRange=',startHist+2:endHist,'vary=',J};
optimSet = {'MaxFunEvals=',10000,'TolFun=',1e-16};
tic
[est,pos,C,H,mest] = estimate(m,d,startHist:endHist,E,'filter=',filterOpt,'optimSet=',optimSet,'sstate=',true,'nosolution=','penalty');
toc</code></pre>
</div>ikaribWed, 15 Feb 2017 19:28:33 GMTNew Post: Estimating DSGE Models with Forward Guidance 20170215072833PNew Post: Is IRIS still in development?http://iristoolbox.codeplex.com/discussions/659887<div style="line-height: normal;">Jaromir didn't have enough time this year, but he has big plans to make IRIS more flexible with plans to move to github by end of the year. Hopefully, he will implement my pull requests that I recently created at this fork: <a href="https://iristoolbox.codeplex.com/SourceControl/network/forks/ikarib/IRISdev" rel="nofollow">https://iristoolbox.codeplex.com/SourceControl/network/forks/ikarib/IRISdev</a><br />
</div>ikaribTue, 06 Dec 2016 05:24:35 GMTNew Post: Is IRIS still in development? 20161206052435ANew Post: Is IRIS still in development?http://iristoolbox.codeplex.com/discussions/659887<div style="line-height: normal;">Hi all,
<br />
<br />
Is IRIS still in development? Follow-up on discussion or issues have been very rare lately, and there have not been any new release for about a year.
<br />
<br />
Best,
<br />
Gabriel <br />
</div>gabrielbruneauFri, 02 Dec 2016 18:57:43 GMTNew Post: Is IRIS still in development? 20161202065743PNew Post: question about estimating time varying parameterhttps://iristoolbox.codeplex.com/discussions/659573<div style="line-height: normal;">I am replicating FRBNY model in IRIS and, in order to implement zero lower bound, I need to estimate the time-varying standard deviation of anticipated monetary policy shock after the financial crisis hit in 2008q4. I created J database where I set std_rm_sh1..6 = 0 before financial crisis (2008q3) and 0.2 after 2008q4. But since the 0.2 is hard coded as input to 'vary=' option in filterOpt cell array, I can not estimate this time varying parameter. Could you please help me?<br />
<br />
The files are available at<br />
<a href="http://karibzhanov.com/src/frbny.zip" rel="nofollow">http://karibzhanov.com/src/frbny.zip</a><br />
<br />
I created my IRIS fork to run estimate_params.m:<br />
<a href="https://iristoolbox.codeplex.com/SourceControl/network/forks/ikarib/IRISdev" rel="nofollow">https://iristoolbox.codeplex.com/SourceControl/network/forks/ikarib/IRISdev</a><br />
<br />
Thank you,<br />
Iskander<br />
</div>ikaribMon, 21 Nov 2016 06:04:36 GMTNew Post: question about estimating time varying parameter 20161121060436ANew Post: dynamic links and parameter updateshttp://iristoolbox.codeplex.com/discussions/659031<div style="line-height: normal;">what is the difference between dlinks and sstate_update equations?<br />
why do parameters on LHS in sstate_update equations have to have lead {1}?<br />
why do sstate_update equations get evaluated in simulate but not in sstate solver?<br />
</div>ikaribFri, 28 Oct 2016 03:14:53 GMTNew Post: dynamic links and parameter updates 20161028031453ANew Post: first iteration of Kalman filterhttp://iristoolbox.codeplex.com/discussions/658373<div style="line-height: normal;">why is there no correction in the first iteration of Kalman filter?
<br />
The Kalman gain should not be empty for computing prediction in the first period (line 175 of +kalman\ped.m).
<br />
<br />
<img src="http://karibzhanov.com/help/images/eq21.gif" alt="Image" /><br />
</div>ikaribFri, 30 Sep 2016 00:40:56 GMTNew Post: first iteration of Kalman filter 20160930124056ANew Post: Adjust the size of a table http://iristoolbox.codeplex.com/discussions/656297<div style="line-height: normal;">Hi guys
<br />
<br />
I am actually testing the Toolbox on Matlab R2016a and a table in my report now appears much larger than with Matalb R2014a, so that it exceeds the page size.
<br />
Is there an Option that allows to adjust the table size?
<br />
<br />
Bests,
<br />
Riccardo<br />
</div>RGambaTue, 12 Jul 2016 09:21:32 GMTNew Post: Adjust the size of a table 20160712092132ANew Post: FEVD and Historical Decomposition After Sign Restrictionhttp://iristoolbox.codeplex.com/discussions/656074<div style="line-height: normal;">How can I tell IRIS that I want to perform a FEVD or historical decomp after doing sign restrictions? In particular, how do I tell IRIS to use a particular draw (median or some other criteria) from my SVAR object?<br />
</div>macrowbSat, 02 Jul 2016 19:54:40 GMTNew Post: FEVD and Historical Decomposition After Sign Restriction 20160702075440PNew Post: robust Bayesian estimationhttp://iristoolbox.codeplex.com/discussions/655753<div style="line-height: normal;">Dear all,
<br />
<br />
I am new in Iris and have some problem in Bayesian estimation. I am using it to estimate some parameters from the model of Kalman filter for Output gap estimation( in other words it is MVHP filter) . I use quarterly data from 2005 to 2015. My posteriors are very sensitive to prior values, and if I expand the stds of prior distributions, it gives me the values close to 0. Do you think it is because of shortness of my time series?
<br />
<br />
Thank you in advance for your reply.<br />
</div>Mane_ManeTue, 21 Jun 2016 11:57:51 GMTNew Post: robust Bayesian estimation 20160621115751ANew Post: confidence bandshttp://iristoolbox.codeplex.com/discussions/655732<div style="line-height: normal;">hello,
<br />
I need to calculate & display the confidence bands for the calibrated linear model with some initial conditions (impulse response functions). Is the "resample" function the proper method how to do it? And if so, how should I define its input parameters?
<br />
Thanks in advance<br />
</div>BambiMon, 20 Jun 2016 17:18:47 GMTNew Post: confidence bands 20160620051847PNew Post: Scaling in a plot with two Y-axeshttp://iristoolbox.codeplex.com/discussions/655731<div style="line-height: normal;">Hello, guys!
<br />
<br />
Is it possible to set different scaling for two Y-axes an a two Y-axes plot?
<br />
In Matlab one would do this, e.g. as
<br />
<br />
ax.YAxis(1).yLim = ...
<br />
ax.YAxis(2).yLim = ...
<br />
<br />
But this does not seem to work in IRIS reporting when specifying styling through structures and 'style=' option.
<br />
<br />
Best,
<br />
Andrey<br />
</div>AndreyOMon, 20 Jun 2016 16:05:32 GMTNew Post: Scaling in a plot with two Y-axes 20160620040532PNew Post: Putting 3 graphs on one page in pdf reporthttp://iristoolbox.codeplex.com/discussions/655190<div style="line-height: normal;">Thank you, Sergey! Exactly what I was looking for.
<br />
<br />
Andrey<br />
</div>AndreyOFri, 03 Jun 2016 11:33:57 GMTNew Post: Putting 3 graphs on one page in pdf report 20160603113357ANew Post: Putting 3 graphs on one page in pdf reporthttps://iristoolbox.codeplex.com/discussions/655190<div style="line-height: normal;">Hi Andrey,
<br />
<br />
There are several possibilities, actually.
<br />
<br />
First,<br />
<pre><code>clear all;
close all;
r = report.new();
r.figure('','subplot',[2,2]);
cmd = ['set(get(H,''parent''),''units'',''normalized'');',...
'tmpPos = get(H,''position'');',...
'set(H,''position'',[tmpPos(1:2),tmpPos(3)*2+0.1,tmpPos(4)])'];
r.graph('','postProcess=',cmd);
r.series('',tseries(qq(2000,1:16),@randn));
r.empty();
r.graph('');
r.series('',tseries(qq(2000,1:16),@randn));
r.graph('');
r.series('',tseries(qq(2000,1:16),@randn));
r.publish('ccc.pdf');</code></pre>
Second,<br />
<pre><code>figure('visible','off');
subplot(2,2,1:2);
plot(tseries(qq(2000,1:16),@randn));
subplot(2,2,3);
plot(tseries(qq(2000,1:16),@randn));
subplot(2,2,4);
plot(tseries(qq(2000,1:16),@randn));
r.userfigure('',gcf);
r.publish('ccc.pdf');</code></pre>
Once issue <a href="https://iristoolbox.codeplex.com/workitem/164" rel="nofollow">issue 164</a> is fixed you should also be able to do it like this,<br />
<pre><code>clear all;
close all;
r = report.new();
r.figure('','subplot',[2,2]);
r.graph('','postProcess=','grfun.movetosubplot(H,2,2,1:2)');
r.series('',tseries(qq(2000,1:16),@randn));
r.empty();
r.graph('');
r.series('',tseries(qq(2000,1:16),@randn));
r.graph('');
r.series('',tseries(qq(2000,1:16),@randn));
r.publish('ccc.pdf');</code></pre>
</div>sergeyplotnikovFri, 03 Jun 2016 09:53:15 GMTNew Post: Putting 3 graphs on one page in pdf report 20160603095315ANew Post: Putting 3 graphs on one page in pdf reporthttp://iristoolbox.codeplex.com/discussions/655190<div style="line-height: normal;">Hello, guys!
<br />
<br />
Is it possible to put 3 graphs on one page in a pdf report, so that one of the graphs is twice the width of two others, i.e. there would be no blank space left for the possible forth graph.
<br />
<br />
Best,
<br />
Andrey<br />
</div>AndreyOFri, 03 Jun 2016 07:43:05 GMTNew Post: Putting 3 graphs on one page in pdf report 20160603074305ANew Post: A SVAR Problemhttp://iristoolbox.codeplex.com/discussions/654153<div style="line-height: normal;">According to iris_doc, to build a SVAR we can choose any method as follows,<br />
<br />
'method=' [ 'chol' | 'householder' | 'qr' | 'svd' ] - Method that will be used to identify<br />
structural VAR and structural shocks.<br />
<br />
but i am wondering if there is any paper discussing them, for i don't really no householder, qr and svd. And if I need to have restriction on matrices B, only could the householder method be used? I apply it to a 6 variables-SVAR model and it is really slow sometimes impossible to compute, since I place 30 restriction on it. By the way, 0 restriction is placed as abs(S(i,j,1))<0.01 and none-zeros >0.01. How can i speed is up?<br />
<br />
At the same time, the restriction on matrices B such as S(1,2,1) seem to work but it is supposed to be contemporaneous shock by the A-B model thoery of SVAR, not in perioid 1. Am I misunderstanding something?<br />
</div>chinabeeFri, 29 Apr 2016 08:05:44 GMTNew Post: A SVAR Problem 20160429080544ANew Post: Panel VAR modificationshttp://iristoolbox.codeplex.com/discussions/653242<div style="line-height: normal;">Hi Jaromir,
<br />
<br />
I have a panel with T=60 and N=5 with 3 to 5 variables, and I would need to estimate a panel VAR on it.
<br />
<br />
I was looking at your code, which looks very similar to what I'm doing, but I think I would need to make some modifications to it and not quite sure how -
<br />
<br />
1) I will need to add dummy variables for different time points in the sample. is this possible in the framework?
<br />
<br />
2) you say that the model is fixed effects. I thought fixed effects are homogenous so would produce the same IRFs for the N groups? but your results show these as being different across groups - why is this?
<br />
<br />
thanks a lot in advance!<br />
</div>MM72Tue, 05 Apr 2016 15:43:46 GMTNew Post: Panel VAR modifications 20160405034346PNew Post: Estimate annual VAR frequency errorhttp://iristoolbox.codeplex.com/discussions/652387<div style="line-height: normal;">Dear all,
<br />
<br />
I try to estimate a VAR(1) model with annual data. However, I get the following error message, even when I follow your instructions in this discussion: <a href="https://iristoolbox.codeplex.com/discussions/647473" rel="nofollow">https://iristoolbox.codeplex.com/discussions/647473</a><br />
<blockquote>
Warning: IRIS Toolbox Warning @ dbase:db2array.<br />
*** This database entry does not match the date frequency requested: GNPPC<br />
*** This database entry does not match the date frequency requested: CPI<br />
*** This database entry does not match the date frequency requested: ER<br />
*** This database entry does not match the date frequency requested: ST<br />
*** This database entry does not match the date frequency requested: LT <br />
</blockquote>
How do I solve this problem?
<br />
<br />
Thanks in advance<br />
</div>Mrtn3003Thu, 10 Mar 2016 13:18:44 GMTNew Post: Estimate annual VAR frequency error 20160310011844PNew Post: SVARhttp://iristoolbox.codeplex.com/discussions/651871<div style="line-height: normal;">Relating to the above: is it possible to impose less restrictions in the contemporaneous matrix than what is needed for full identification? When I am only interested in identifying one particular shock and do not care about the rest, then maybe it is better to avoid making more restrictions by a full Cholesky (lower triangular matrix). Can this be done? This relates to the above question whether we can set the restrictions by hand.
<br />
<br />
Thanks!<br />
</div>motyo6Tue, 08 Mar 2016 17:47:37 GMTNew Post: SVAR 20160308054737P