This readme file describes matlab application written by Stephanie Schmitt-Grohe and Martin Uribe for implementing the identifiability test proposed by Iskrev (JME, 2010). The background material for this software is ``Implementing Iskrev's Identifiability Test,'' (2012), by Stephanie Schmitt-Grohe and Martin Uribe. 

This application uses the matlab application for solving DSGE models up to first order written by Stephanie Schmitt-Grohe and Martin Uribe and available online at 
http://www.columbia.edu/~mu2166/2nd_order.htm

We explain the application using as an example ta specific model, namely, the neoclassical model.  But the programs can be applied to any DSGE model

Steps:
(1) Run iskrev_run.m this program puts the application in motion. You can change the first word in the name of this file (neoclassical) for one that fits your task.

1.a The string variable filename is model specific. In the present example it is se at `neoclassical'.

1.b iskrev_run calls the program neoclassical.m. This program contains the equilibrium contidions  of the DSGE model (equations e1, e2, etc.) You should modify these equations to fit your problem. But the name of this file and filename should match.  neoclassical.m must be run only if the file neoclassical.mat does not exist or if neoclassical.m has changed.  If neoclassical.m need not be run, you can percentage out this the command that calls it

1.c iskrev_run.m  also calls  iskrev_anal_deriv.m this program is NOT model dependent and should never be modified.  This program needs to be called only if neoclassical.mat has changed. Otherwise, you can percentage out the command that calls iskrev_anal_deriv.m 

1.d Two inputs in iskrev_run are 
param_identif is a symbolic vector containing the names of the parameters of the DSGE whose identifiability is to be tested. 

param_estim is a vector containing numerical values for all structural parameters of the DSGE model (not just the ones whose identifiability is being tested.). To determine the order of the parameters in this vector, see gx_hx_inputs.m

observables is a selection row vector indicating the rows of the matrix gx corresponding to the observables used to idenfify the parameters (in the example, we use consumption and ouput to identify the parametes of the model. These happen to be all of the rows of gx, but this need not be--try for instance settig observables = 2 or observables = 1)

1.e T the order of thecovariogram of the obserables used to chedk for identifiability. T=1 means that only the contemporaneous covariance of observables is used for identification. T=2 adds the fist-order autocovariance, and so on.

1.f iskrevrun.m also calls iskrev_test.m thsi program is not model dependent and therefore should not be modified (unless you want to modify the tolerance in computing the rank of the derivative of the covariogram with respect to the parameters involved in the test).

(2) Modify gx_hx_inputs.m this program is model specific. It solves for the steady-state of the model.  Do not alter the name of this program (i.e., alwasy use gx_hx_inputs.m).