The assignment of cabinet portfolios to parties has a profound effect on the policies adopted by governments. However, the relationship between portfolio allocation and public policy outcomes has not been adequately addressed in the literature. On the one hand, theories such as the portfolio-allocation model of government formation (Laver and Shepsle 1996) or the veto players model (Tsebelis 2002) rely on strong assumptions about how policy decisions within governments are made. On the other hand, bargaining models of government formation (c.f. Diermeier 2006) predict equilibrium policy outcomes, but without considering if these outcomes could be sustained as soon as parties turn to the day-to-day business of policy-making. In this paper, I present a spatial model of intra-cabinet decision-making that takes the strategic interaction between cabinet members into account. This model considers portfolio ministers as agenda-setters and allows both the minister of finance and the prime minister to either accept or reject proposals. In addition, the model considers that disagreement between the prime minister and finance minister may be costly, which limits players'' strategies when reviewing a proposal. The model yields results that run counter to the existing literature. It shows that the prime minister may benefit from sharing power with a strong finance minister. By constraining herself in this way, the prime minister can force portfolio ministers to propose policies that are closer to her ideal point than would be the case if the finance minister were weak. In addition, this strategy works particularly well if the cabinet includes ideologically extreme ministers. The implications of the model are tested with time-series cross-sectional data on cabinet composition and changes in government spending in various policy areas.
