Why do some coalitions end before the constitutionally mandated election time, while others do not? Sometimes governments need to address issues that are not stated, or even contradict, the coalition agreement. Not all party members are willing to put the survival of the government above policy advocacy. Consequently, party leaders need to balance party activist’ demands versus the desire to prevent coalition breakdown. Meanwhile, party activists also want to be in government to achieve policy goals, and do not want to hurt the party’s reputation. I illustrate these dilemmas using a dynamic game-theoretic model. The model predicts that the larger parties are less likely to break a coalition given an exogenous policy shock. Moreover, counter-intuitively, coalitions with more ideologically divergent parties are more stable, and leaders of such parties are more likely to survive. Evidence from Belgian, Danish, Dutch, Finnish, and German coalitions offer empirical support for the model’s implications.
