In this paper, we propose an integrated sparse Bayesian variable selection in regressions with a large number of possibly highly correlated macroeconomic predictors. The variable selection is performed through the stochastic search variable selection technique. We assign a sparse prior distribution on the regression parameters and a correlation prior distribution for the binary vector. The performance of the proposed variable selection method is illustrated in forecasting one major macroeconomic time series of the US economy. Empirical results show that in terms of absolute forecast error and log predictive likelihood, our proposed method performs better than other three methods.