endogenise in simulate

[lex]

Hi Jaromir,

 

In the simulate command, is it possible to endogenise several shocks for only one variable which I exogenise?

 

If yes, how does IRIS choose the combination of shocks?

 

Thanks for your help,

lex

[Jaromir Benes]

Although still undocumented, this option is possible.

The combination of shocks is then chosen based on a minimisation criteria

min (e1/s1)^2 + (e2/s2)^2 + (e3/s3)^2 + ...

where e1, e2, e3 are the individual enodgenised shocks, and s1, s2, s3 are the weights you assign to them. These weights can be  interpreted as standard errors, in which case this is, in fact, a simple Kalman filtering exercise. In other words, the larger the weight s1, the more the shock e1 will be used in exogenising the specified variables.

The weights themselves are by default 1, but you can change them by using a fourth input argument in the function plan/endogenise:

p = plan(m,1:10);

p = exogenise(p,'x',1:4);

p = endogenise(p,'ey',1:4,1);

p = endogenise(p,'ez',1:4,2.5);

In this case, the variable x is exogenised in periods t=1, 2, 3, 4 using two types of shocks, ey and ez. In other words, we have a total of 4 exogenised data points and a total of 8 endogenised data points.

The shocks ez are assigned a higher weight (= "a higher std deviation") and hence will be allowed to deviate from zero more flexibly. If you, on the other hand, change the weight e.g. to zero this will effectively switch them off.

Remember that the assigned weights ("std deviations") have nothing to do with the actual std deviations of the shocks assigned in the model object.

This kind of simulation is a simplified version of what you can achieve with jforecast, that can combine exogenising and conditioning.

Hope this helps.

Best,

Jaromir

--
You received this message because you are subscribed to the Google Groups "iris-toolbox-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to iris-toolbox...@googlegroups.com.
To post to this group, send email to iris-t...@googlegroups.com.
Visit this group at http://groups.google.com/group/iris-toolbox?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[lex]

Thanks!

[aor...@nes.ru]

Hello, guys!

I was wondering whether the same syntax, i.e.

p = plan(m,1:10);

p = exogenise(p,'x',1:4);

p = endogenise(p,'ey',1:4,1);

p = endogenise(p,'ez',1:4,2.5);

would work with the jforecast function. I have tried to use differing weights (as in the example above) but it does not seem to make any difference - it works just like

p = endogenise(p,{'ey','ez'},1:4);

withoug specifying any weights.

Best,

Andrey

[Jaromir Benes]

Hi Andrey

The weights work only in simulate(), indeed. I will introduce the same into jforecast() at some point -- the jforecast() function is scheduled for a major overhaul anyway.

Note also, that you can achieve somewhat similar functionality by using conditioning (instead of exogenizing) and setting manually the std errors of shocks.

Best,

Jaromir

--

You received this message because you are subscribed to the Google Groups "iris-toolbox-discussion" group.
To unsubscribe from this group and stop receiving emails from it, send an email to iris-toolbox...@googlegroups.com.
To post to this group, send email to iris-t...@googlegroups.com.

Visit this group at http://groups.google.com/group/iris-toolbox.

[aor...@nes.ru]

Thank you, Mirek.

Is there is an option to select which shocks will be endogenised when using condition?

Best,

Andrey

[Jaromir Benes]

Yes. By setting their std deviations. The larger std dev the "more" the shock will be used. And shocks with zero will remain fixed. 

--

[aor...@nes.ru]

Thanks!

четверг, 18 апреля 2013 г., 15:26:24 UTC+4 пользователь lex написал: