We analyze optimal capital and labor taxes in a model where (i) the government makes noncontingent announcements about future policies and (ii) state-contingent deviations from these announcements are costly. With Full Commitment, optimal announcements coincide with expected future taxes. Costly state contingency dampens the response of both current and future capital taxes to government spending shocks and labor taxes play a major role in accommodating fiscal shocks. These features allow our quantitative model to account for the volatility of taxes in US data. In the absence of Full Commitment, optimal announcements are instead strategically biased, because governments have an incentive to partially constrain their successors. The cost of deviating from past announcements generates an endogenous degree of fiscal commitment, determining the average level of capital taxes.
Read full abstract