Estimation question

[shr]

Hi,

I am stuck between a rock and a hard place trying to estimate a log-linearized model. On the one hand, I am using fsolve to find my steady state, which means I would like to supply a steady state file instead of using pound operators inside the model block. At the same time, however, I use some of my steady state values in my model equations, e.g. in the linearized budget and resource constraints, which means Dynare needs them either as parameters or pound-expressions. What can I do to get around this problem? Thanks in advance for any help!

[jpfeifer]

I don't understand your problem. You can use a steady state file to update parameter values. The NK_baseline.mod in the Dynare examples folder does exactly that. You do not need model-local variables for that (pound operator)

[shr]

Sorry if my question was unclear. My problem is the following: Let's call output in my model yhat. Since the model is log-linearized, the steady state value of yhat is zero. But my market clearing equation reads "yhat = (Cs/Ys)*chat + (Is/Ys)*ihat", where Ys (which enters the model as a parameter) is not zero. If I could compute Ys on closed form, I would just use a pound-expression to make sure that Ys is updated in each step when I estimate my model (because Ys depends on some of the parameters I estimate). But since I can only find Ys using an fsolve-function, I cannot use a pound-expression. My question is then; if I write a steadystate-file in which I include not only the steady state values (which are zero), but also the parameter Ys (calling my fsolve function), will Ys then also be updated in each step, as it should? Thanks in advance!

[jpfeifer]

As I said, please take a look at the NK_baseline.mod in the Dynare examples folder. It uses fsolve to solve for an endogenous variables and resets a parameter accordingly. That is exactly what you are trying to achieve.