Rules versus Discretion without Common Knowledge

How should government design optimal policy with and without commitment when facing the trade-off between policy tool and endogenous outcome, such as inflation surprise and unemployment? We study this question allowing agents to face higher-order uncertainty about the economic conditions. With commitment, the government can overcome such trade-off and reach the first-best allocation. In contrast, with discretionary policy, the lack of common knowledge may improve or worsen the welfare. In an application where inflation is subject to cost-push shock, we show that the social cost reduces by over 50\% when calibrating the amount of higher-order uncertainty to the survey data on expectations.