Date: 2011-02-04 12:58 Sender: Arne HenningsenI think that even more comparisons are possible:
comparing 3SLS with 2SLS estimates: test if all exogenous variables (instrumental variables) are uncorrelated with all disturbance terms rather than that the exogenous variables (instrumental variables) of each single equation are uncorrelated with the corresponding disturbance term
comparing 2SLS with OLS estimates: test whether the exogenous variables (instrumental variables) of each single equation are uncorrelated with the corresponding disturbance term rather than that the regressors of each equation are uncorrelated with the corresponding disturbance term
comparing 3SLS with SUR estimates: test if all exogenous variables (instrumental variables) are uncorrelated with all disturbance terms rather than that all regressors are uncorrelated with all disturbance terms
comparing 3SLS with OLS estimates: test if all exogenous variables (instrumental variables) are uncorrelated with all disturbance terms rather than that the regressors of each equation are uncorrelated with the corresponding disturbance term
|