Discussions about the legitimacy and welfare consequences of paternalistic interventions usually begin with the assumption that regulators are both benevolent and competent. We present experimental evidence that neither need be the case. In our experiment, individuals choose whether to restrict the choice of another participant and we see that regulation, on average, decreases choice efficiency. While more competent regulators are more likely to restrict choice sets in order to improve welfare for subjects when they use their regulatory privilege, selection into being an active regulator is unrelated to competence. The propensity for kind regulation is increasing in own competence, while the propensity for unkind regulation is both negatively related to own competence and positively related to the competence of the subject.