We claim that, in presidential democracies, the effect of increasing fragmentation on government spending should be conditional on polarization, defined as the ideological distance between the government's party and other parties in Congress. We build a model where this result follows from negotiations between the legislature and an independent government seeking the approval of its initiatives-as in presidential democracies. Using cross-country data over time, we test the empirical validity of our claim finding that, in presidential democracies, there is indeed a positive effect of fragmentation only when polarization is sufficiently high. The same is not true for parliamentary democracies.