This paper offers an explanation for why policy makers stick to inefficient
policy decisions. I argue that repealing a policy is a bad signal to voter
s about the policy maker's competence if voters do not have complete knowle
dge about the effects of implemented policies. I derive the optimal policy
maker's decision on continuation of a policy, assuming that voters' beliefs
about the policy maker's competence are updated according to Bayes' rule.
I show that if the policy maker cares sufficiently about reelection, he wil
l never repeal a policy.