Contributor
Posts: 61

# granger causality and simultaneous equations

[ Edited ]

I have a panel data for 5 years :

1. Set of endogenous variables: Y1it, Y1i(t-1), Y1i(t-2), Y1i(t-3), Y1i(t-4), X1it, X1i(t-1), X1i(t-2), X1i(t-3), X1i(t-4)
2. And exogenous variables that are also part of the model : X2it, X3it, Dummy1it, Dummy2it

Model:

Y1it= intercept + X1it + X1i(t-1) + X2it + X3it + Dummy1it + Dummy2it

X1it= intercept + Y1it + Y1i(t-1) + X1i(t-1) + Dummy1it + Dummy2it

I need to solve two problems :

1. It seems like I have a problem of simultaneity, that is, explained variable Y1it is influenced by X1it  , and vice versa X1it is influenced by Y1it. In order to overcome this endogeneity problem, I've solved a system of simultaneous equations (proc syslin).  The question is whether this procedure gives me BLUE estimators.
2. How I can infer whether Y1it can be predicted by X1it or vice versa, X1it is caused by Y1it. Or maybe I need another procedure in order to find a causal effect.                                                                                                                                               Maybe i should use proc varmax for checking granger causality. But, there, i check if group2 influences on group1, or group1 is Influenced by itself and it is not my case.                                                                                                                           Thank you
Discussion stats
• 0 replies
• 124 views
• 0 likes
• 1 in conversation