Abstract: We consider a dynamic model in which two parties decide on a unidimensional policy in each period over an infinite horizon. The policy implemented in any period becomes the next status quo. Given the status quo, adjusting the policy involves a cost that may differ between the party in power and the party out of power due to favoritism. A second source of tension is ideological: the two parties may be polarized in their ideal policies. We show that efficiency requires no policy reversals and imposes bounds on the policy. We show that, depending on the parameter values, either the equilibrium is unique or its structure is unique. When polarization is high, the equilibrium is inefficient due to perpetual policy reversals. When polarization is not high but favoritism is, equilibrium can be inefficient due to “overshooting” or “undershooting” as a result of coordination failures. When neither polarization nor favoritism is high, equilibrium is always efficient after a short transition.