“Centrists, extremists and the benefit of losing the popular vote: The case of Swiss referendums, 1978-2018” (with Valentin Schröder, University of Bremen)
Congruence of political elite actions and the preferences of the electorate is constitutive for democracy. Elections are crucial in this respect since they foster alignment between both, through sanctioning divergence of elite action ex post. An even closer alignment might be achieved by an additional means of citizen participation, namely, referendums. Under a logic of sanctioning then, elite divergence from citizens’ preferences is punished even ex ante, via reputation costs inflicted on parts of the elite for even trying to diverge from the wishes of a majority of voters. Consequently, a large literature deals with the advantages of direct democratic institutions and whether such institutions lead to policies more in line with citizens’ preferences. Instances of elite-voter incongruence are then typically studied by comparing attitudes of citizens with positions of political parties as expressed in surveys or by experts.
Referenda are often conceived of as fosteringelite-voter congruence since, by definition, their results mirror the majorityof voters. Specifically, decisions taken by a legislative majority that are notaligned with a majority in the politically active electorate can thus be“corrected”. Hence the normative appeal of referenda.
However, results of referenda do not only affect their subject matter. They also imply, for those on the losing side, a loss in reputation. We argue that political parties are affected from this loss to different degrees. Losses may be negligible for parties that compete only for voters at the ideological fringes of the electorate. Yet, the more centrist a party, the more sustained is the expectation of its policies representing a position amenable to a societal majority. We derive a series of hypotheses from this general line of argument.
We then assess empirical validity of our hypotheses using a novel dataset covering all national-level referenda held in Switzerlandin the period 1978-2018. We indeed find parties to be more likely to take a stand in favor of the (ultimately) losing side along the degree of their extremity.
What is more, Optimal Classification of positions taken by parties into positions in Euclidian space returns an alignment of parties very much in line with enduring findings on their positions on ageneral left-right dimension, with voters aligned at a centrist position. This corroborates our findings. Yet, it also indicates that referenda do make a difference when it comes to decisions tapping minor political dimensions. But their results very much fall in line with parliamentary majorities on the major one. This puts empirical realization of the above normative appeal into question.