This paper explores implications for optimal monetary policies under the dynamic time inconsistency problem. I develop a dynamic optimizing model calibrated to recent Korean data covering from 1998 to 2004. I investigate the consequences of alternative monetary policies with particular focus on the loss function. Policy simulations include variations on discretion, commitment, strict inflation targeting, and Taylor rule. The simulation results indicate that commitment remains the most preferred policy regime in terms of the loss function. The strict inflation targeting regime is worse than the Taylor rule regime in terms of the welfare loss. Moreover, the loss in the strict inflation targeting increases as the society puts higher weight on the output stability, the cost-push shock is less persistent, the price is more staggered, or the labor supply is more elastic.