Problems with instrumental variables estimation when the correlation Essay

Problems with instrumental variables estimation when the correlation between the instruments and the endogenous explanatory variable is weak - Essay Example To be able to understand the dynamics of a problem it is important to begin with determining the number of variables involved and the relationship between each variable and the problem. It is also important to understand how the involved variables affect each other or one another. The attempt of various researchers to understand the existing relationship has been unsuccessful under non-experimental scenery (Bound, Jaeger & Baker 443). The reason for the failure is the fact that some of the variables are endogenous. In other words, the variables are subject to the same factors that influence the research problem. The endogeniety trait of variables makes the ordinary least square biased and inconsistent therefore, making the research conclusion unreliable. Fortunately, there is a method of reducing the influence of variable endogeneity on research outcome has been determined. The method involves the use of an instrumental variable estimation. However, the method also experiences some problems. First, insufficient explanation of the variation of the endogenous explanatory variables. This could increase the error standard thus invalidates the research outcome. Secondly, the approximated limited samples are biased in the same magnitude as OLS when the coefficient of determination, (R2), between the instrument and the variables diminishes (Bound, Jaeger & Baker 443) To conclude, the problems with instrumental variable estimation are insufficient explanation in the deviation of the endogenous variables and the biases of finite samples when the R2 between the instrument and the variables decreases to zero. Bound, John, David A. Jaeger and Regina M. Baker. â€Å"Problems with Instrumental Variables Estimation when the Correlation between the Instruments and the Endogenous.† Journal of The American Statistical Association 90.430 (1995): 443.