Regression is termed as spurious regression if the regression outputs shows very significant relationship between two independent series. It occurs due to two major reasons, (i) nonstationarity and (ii) omitted relevant variables. The solution for omitted variable bias is to include all the relevant variables into the model. But sometimes this method becomes quite difficult to handle due to very large number of potential explanatory variables and it becomes very difficult to avoid missing variable bias. The recently introduced Weighted Average Least Square (WALS) technique which is a Bayesian combination of frequentist estimators can handle the case of large number of regressors. But does it really reduce the probability of spurious regression, no answer to the question can be found in literature. Therefore, this study evaluates the performance of WALS estimator to avoid the problem of spurious regression and compare the forecast performance of WALS with the classical estimator. The size and power for WALS and OLS estimates has been calculated because size depicts the probability of spurious relationship between consumption and GDP of two different countries whereas, the power calculation refers the probability of true relationship between Consumption and GDP of the same countries. Thirty countries belongs to low and lower middle income group have selected for this study. The estimated results suggested that WALS and OLS have same power and forecast performance but other than this, WALS is much better than OLS because it reduces the probability of spurious regression form 99.7% to 20.6% under 5% while 8.4% under 1% nominal size respectively. So, from these it is suggested that WALS can avoid the problem of spurious regression. Supervisor:- Dr. Atiq Ur Rehman