We study how to efficiently motivate policy-makers to solve political multitask problems. Political multi-task problems typically have outcomes that are difficult to measure. Moreover, there are conflicts among citizens about optimal policies and the agents have the power to tax the citizens to invest in better outcomes of some tasks. We develop a political agency model with two tasks and only one measurable outcome. In such an environment, policy-makers choose socially inefficient public good levels and expropriate minorities. A judicious combination of constitutional limits on taxation and incentive pay for policy-makers is second-best. Incentive pay is conditional on the public good level.