There are mixed findings on whether formal control will or will not improve performance. To address this paradox, we propose that types of formal control (enabling vs. coercive) influence performance separately because of the different costs they impose on the controlee. We conducted an experiment, and obtained unexpected, but interesting results: the controlee performs best for simple tasks in the no formal control condition, because costs of the control override its benefits. Our preliminary results suggest that there might be an interaction between types of formal control and degrees of task complexity. A difficult version of the same experimental task will be employed in our next experiment to test the interaction effect.