Hi,
I would like to know what is the advantage of using GMM for estimation? Why don’t the researchers simply choose parameters to minimize sum of squares error (SSE) of the difference between the real data and simulated data? I think the latter is better in that the estimation result does not depend on the moments we pick.
My second question is, is the estimation result of minimizing SSE equivalent to the MLE?
Thanks a lot!
I am not sure what you mean. Being able to compare data realizations and simulation realizations would require knowing the exact shocks that happened at each point in time.
Hi Prof.,
I typed my problem out, and I am glad if you can tell me if there is any problem of this approach. Thanks!
draft.pdf (56.9 KB)
As I said above, your simulation data realizations will be very different from the actual data as you don’t know the underlying shocks. That is what is why we use moments, i.e. averages over simulations.