Dynamic Panel Data for Financial Markets

Dear all,

I am trying to simultaneously estimate three equations using dynamic panel data. In each equation, the dependent variable (say, Y(3)) is a function of the two other dependent variables from that same period (say, X(3) and Z(3)), as well as the lagged values of the same dependent variable (say, Y(2) and Y(1)).

To deal with endogeneity issues and unobserved heterogeneity, some papers suggest using the GMM framework (e.g. xtabond or in STATA). However, GMM is optimal for datasets with large N and small T. My dataset has large T (1500) and small N (150), and as such the GMM method may cause issues due to a large number of instrumental variables.

Which is the most effective way to develop the model? I have looked at SEM and GMM, but am unable to find an example where dynamic panel data models have been employed.

Thank you for any help in advance,