The paper analyzes a widely neglected design of no-confidence votes in ostensibly parliamentary systems of government. This design involves two directly elected chambers of parliament, only one of which can terminate the cabinet in a no-confidence vote. Ganghof (EJPR, 2018) has conceptualized it as “semi-parliamentary” government, because only one of two directly elected agents of the voters becomes the principal of the government. The other agent transacts with the government as in a presidential system. Examples include Australian states such as New South Wales and Victoria.
The paper connects this discussion to the comparative patterns of investiture and no-confidence votes in parliamentary systems (e.g., Sieberer 2015, in Rasch et al. 2015, OUP; Cheibub et al., WEP, 2015). In doing so, the paper reflects on the use of the principal-agent model in positive political theory. It suggests to start with voters as the ultimate principal and to acknowledge that this principal does not have stable preferences in a one-dimensional issues space. Hence, institutional designers face a tension between the task of authorizing a particular government and representing voters in the (multidimensional) process of legislative coalition building.
Parliamentary systems allow for a limited number of responses to this tension. For example, in proportional representation systems, demanding investiture rules are conducive to unambiguous authorization, but they tend to exclude many voters from legislative coalition building. Conversely, rules that would be maximally conducive to the formation of stable and flexibly governing (one-party) minority cabinets (no investiture vote after election plus constructive vote of no-confidence) would create a severe authorization problem. Semi-parliamentarism is specific a response to this authorization-representation tradeoff, as the two functions can be allocated to the two parts of the legislature, only one of which can end the government in a no-confidence vote.
