Dear professor:
When the observed variable is less than the number of shocks, there may be a parameter identification problem. I added the identification command to the model estimation, but apart from the graph, where is the specific output saved (I saw in an article that the author reported the rank test results of the matrix H and matrix J in a table)
The article is Chinese. I have translated the relevant statement on this issue as follows:
In this paper, the method proposed by Iskrv (2010) is used to test the parameter identification. Iskrv (2010) provides two test statistics. The necessary condition for parameter recognition is that the rank of matrix H is full rank, which is used to test whether the model setting can ensure that the parameters are identifiable; The sufficient condition for parameter recognition is that the rank of matrix J is full rank, and comprehensively test whether the model settings and sample data can ensure that the parameters can be identified (see Isskrv (2010) for details). By performing Monte Carlo simulation in the prior distribution space of all parameters, the parameter identification test results obtained in this paper are shown in Table 2. Firstly, the rank of the matrix H is full rank, indicating that the model setting in this paper can ensure that all parameters can be identified; Second, in the rank test of matrix J, only a few parameters fail the test, but the number of failed tests is a very low proportion of all Monte Carlo simulations. The results of this test show that the model settings and sample data in this paper can ensure that all the parameters to be evaluated, including expected and unpredictable shocks, can be identified.
So I would like to know where the identification results mentioned in the article are stored, or what results should I report in my paper better~
The identification-subfolder has a _identif.mat file. It contains for example an IDE_MOMENTS structure, which has a subfield ino that stores whether there is an identification problem.