Discounting of Party Positions. Some Preconditions and Comparative Evidence Across Heterogeneous Democracies

Up to date, the Grofman discounting model has not seen sufficient theoretical and empirical research. Discounting models operate within an euclidean political space, but extend the basic logic of the Downsian model in one crucial dimension: voters do not exclusively evaluate the proximity of their personal ideal points to spatial positions laid out by the party alternatives, but also consider the degree to which political parties, once elected, are able to move the existing status quo towards their ideological or policy positions. When voters, instead of taking candidate or party positions at face value, discount the ability of political actors to actually deliver their advertised positions, voters face incentive to "overshoot the mark" so as to get their preferred positions enacted thereby inducing potential patterns of centrifugal electoral behavior and/or party strategy. Discounting models envisage some fairly demanding conditions. Voters not only, as in the conventional proximity model, need to be able to locate themselves and the parties competing in the political space. Moreover, they need to be aware of the status quo and about the parties' options to shift it to one or the other direction. Our paper thus hypothesizes that the explanatory value of discounting models is structured by context at the individual and at the individual level: (1) Discounting at the individual level is conditional, and voters need sufficient levels of political sophisitication and information to cast a discounting vote. Voters with higher levels of education, political information and sophistication are therefore expected to be more likely to cast a discounting vote. (2) At the electoral and national levels, we discriminate among institutional setups which enable easy vs. costly policy shifts away from the status quo. Strictly majoritarian democracies are expected to enable discounting; consensus democracies and institutional setups with many veto players tend to prevent discounting. Our examination of these key hypotheses builds upon a large comparative dataset which integrates national election studies across a wide variety of diverse political and institutional settings. Building upon a vast comparative dataset derived from the first four modules of the CSES project, we set up Bayesian discrete choice models which explore the added explanatory value of the discounting model. We consider voter sophistication and information on the micro level and also consider the heterogeneous contextual effects provided by majoritarian vs. consensus democracies or systems with many and with few institutional veto players.