Shutdown methodology

[shimarats]

hi,
i have problem about shut down methodology. i do not know for this method which code use. i put for you one part of paper that i want get . that is , i send example figure of shut down methodology that you can help me.i need help me of which code i should get this figure on this paper.thanks .

[TomDoan]

The Ludvigson, Steindel, Lettau paper has several technical problems. First, their SVAR is not globally identified---it satisfies the simple counting rule, but doesn't meet the Rubio-Ramirez, Waggoner, and Zha(2010) rules, which would require 0, 1, 2, 3 and 4 zeros in the five rows in some order, while they have 1, 1, 1, 3 and 4. In fact, the interesting part of that model (the bottom 3 x 3 corner of the loading matrix) is well-known to have multiple identical likelihood peaks (see page UG-216 in the RATS v8 User's Guide). Second, you can't do the error bands on the SVAR as they describe (drawing directly from the asymptotic distribution), instead you either have to use that as an importance function, or do Gibbs sampling (as is done in the MONTESVAR.RPF program).

Once you get a correct SVAR and have a working program to do the error bands, the "shutdown methodology" is fairly simple---you just use %modelsetcoeffs to zero out a few coefficients in the VAR, zero out an element of the covariance matrix model and compute impulse responses.

[shimarats]

hi ,

i follow your guide.i can find impulse response but ,i need show some line into impulse response according to the attach paper in last file.for example : i need to baseline scenario and two line up and down as baseline +one standard error.please help me ,how i can figur out these lines in impulse response. thanks so much

[TomDoan]

That's in the SIMSIRF.RPF example in the Sims and Zha, Econometrica 1999 replication. That does two sets of bounds: both 16%-84% which are robust equivalents to 1 standard error and 2.5%-97.5%. You don't need the latter.

[shimarats]

hi

i am using svar methodology my project. i use of matrix A. but , i wrote matrix based on theory but i do not know ,give me error in result svar. i think have problem from identification structure of matrix please guide me ,what i should doing for to solve this problem. thanks.

[TomDoan]

hi

i am using svar methodology my project. i use of matrix A. but , i wrote matrix based on theory but i do not know ,give me error in result svar. i think have problem from identification structure of matrix please guide me ,what i should doing for to solve this problem. thanks.

The Rubio-Ramirez, Waggoner and Zha rules for identification are included in the (v8) User's Guide on page UG-216. Without more detail, there's nothing more that I can say,

[shimarats]

yeah. i used svar of sims,zha .i organized matrix A based on theories and follow some article such as sims , zha ,dugey pugan . when i want run this matrix , give me misleading. i donot which part of this matrix put mistake element.i chech this page of guide UG 216.but this example is for small matrix and different variables.thanks

[TomDoan]

You cannot seriously expect that the manual will already include your model. Yes, it's a small illustrative example. But the Rubio-Ramirez, et al counting rules aren't that difficult. And I already explained above how the paper you were citing had a problem with those.