During the past 5 decades a number of econometric techniques were developed and applied to a variety ofrneconometric relationships to deal with the problem of single equation estimation as well as simultaneousrnequations bias. These days, such methods have very wide applications especially in more developedrncountries. However, there has been very little attempt to apply these techniques to empirical relationshipsrndescribing the macroeconomic sector of developing countries in general and Ethiopia in particular. In thisrnstudy, a small macro econometric model of Ethiopia is used to identify the best estimation techniques that willrnproduce accurate forecast of the economy of Ethiopia. In this study six econometric methods were considered.rnThe prediction accuracy of these estimators of economic model was examined using time series data coveringrnthe period 1970 to 2004. The results indicated that considerable gain in forecasting accuracy can be achievedrnby using 2SLSAUT01 and 2SLSAUT02 than simple ordinary least squares or two stage least squares tornestimate macroeconomic models. Moreover, the results do indicate that series attempts should be made tornestimate models by techniques other than ordinary least squares or two-stage least squares