In parliamentary systems, the executive branch of government is not directly elected by the electorate. Instead, the executive is typically selected by a group of elected representatives. In line with Gamson's prominent hypothesis, empirical studies have documented a strong association between parties' contribution to the governing coalition and the share of executive leadership positions that they get. Yet, there rarely is an exact one-to-one mapping from parties' electoral performance to their representation in the executive branch of government. Formal theories of bargaining applied to political coalition formation often attribute this to the so-called formateur advantage, but this idea does not have strong support in data. We propose that parties' coalitional bargaining outcomes may depend not only on their seat shares but also on what kind of representatives they have. In particular, we argue that there is an incumbency bias in coalition formation. To evaluate this argument empirically, we construct a data set of political parties and governing coalitions in Finnish local governments. Using an instrumental variable strategy that hinges on within-party close elections between incumbents and non-incumbents, we find that, ceteris paribus, having more re-elected incumbents improves a party's coalitional bargaining outcomes. Incumbent representation is particularly useful when a party is in a disadvantaged position (e.g., ideologically distant from other parties) and when the bargaining environment is more complex (e.g., there are more parties). Incumbent representation also matters for political selection. Parties that have more re-elected incumbent representatives nominate more incumbents in the municipal executive.