Arguments in favor of a strategic trade policy are based on the assump
tion that the government can credibly precommit to a policy that will
not be altered, even if it is suboptimal ex post. This paper examines
the implications of relaxing this assumption; to this end, a three sta
ge game is considered that accounts for the sunk costs associated with
capacity installment. It is found that-contrary to common belief-the
time-consistent optimal subsidy level is positive, though generally lo
wer than the optimal level with precommitment. This somewhat counterin
tuitive result is driven by the commitment value of capacity.