Am 08.04.2014 11:29, schrieb valentina colombo: Which kind of SVAR, i.e. are you using the built-in VAR with Choleski identification, or the SVAR add-on package with more general ID? -sven

Am 08.04.2014 16:18, schrieb valentina colombo: The help file (pdf) of SVAR tells you what is stored from the bootstrap runs: "Once the bootstrap is done, its results are stored into the bundle for later use: upon successful completion, the model bundle will contain a matrix called bootdata, internally organised as follows: the rst row contains some information on the bootstrap details; from the second row down you get 3 submatrices in which each column is one of the n2 IRFs, and the rows contain 1. the lower limit of the condence interval in the rst block of h + 1 rows 2. the upper limit of the condence interval in the second block of h+1 rows 3. the medians in the second block of h + 1 rows." What do you mean by "betahat"? If you're actually talking about a cointegrated system (SVECM), then the cointegration matrix beta is not estimated by SVAR according to the help doc (p. 24). hth (until jack the author of SVAR comes along), sven