Three issues need to be considered in selecting an estimation procedure. First, we want to allow for inertia in savings ratio that may arise from lagged effects of the explanatory variables on savings. Second, some regressors included in the equation such as real income growth and public savings may be jointly endogenous, that is, correlated with the error term. Third, unobserved time-and country-specific factors may be correlated with the explanatory variables producing biased and inconsistent estimates.
We address these problems by implementing a one- step-dynamic System GMM estimation (Arellano and bovver, 1995: Blundell and Bond, 1998). We consider fiscal variables as a predetermined regressor in our model, and the conflict measure as an endogenous variables, given the possibility of both reverse causality and simultaneity bias. Our instrument for the lagged dependent variable is its own first lag, while we instrument all other endogenous variables with their own second lags in the differenced equation. The results we report correspond to a specification where, say. GDP is considered exogenous, but the effect of conflict on fiscal capacity is not changed if inflation is considered as an endogenous variable: however, in the latter case instrument proliferation impedes an appropriation of the join exogeneity of instruments. It should
Three issues need to be considered in selecting an estimation procedure. First, we want to allow for inertia in savings ratio that may arise from lagged effects of the explanatory variables on savings. Second, some regressors included in the equation such as real income growth and public savings may be jointly endogenous, that is, correlated with the error term. Third, unobserved time-and country-specific factors may be correlated with the explanatory variables producing biased and inconsistent estimates. We address these problems by implementing a one- step-dynamic System GMM estimation (Arellano and bovver, 1995: Blundell and Bond, 1998). We consider fiscal variables as a predetermined regressor in our model, and the conflict measure as an endogenous variables, given the possibility of both reverse causality and simultaneity bias. Our instrument for the lagged dependent variable is its own first lag, while we instrument all other endogenous variables with their own second lags in the differenced equation. The results we report correspond to a specification where, say. GDP is considered exogenous, but the effect of conflict on fiscal capacity is not changed if inflation is considered as an endogenous variable: however, in the latter case instrument proliferation impedes an appropriation of the join exogeneity of instruments. It should
การแปล กรุณารอสักครู่..
