Electoral institutional design

Citation

Shvetsova, Olga V. (1995) Electoral institutional design. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/7aee-hv33. https://resolver.caltech.edu/CaltechETD:etd-10222007-110128

Abstract

This dissertation addresses the role played by electoral institutions in the formation of systems of political parties. The introductory chapter contrasts two methodological approaches found in the studies of institutional influence: the approach of cross-country comparisons versus the detailed analysis of specific electoral systems.

In Chapter 2 a set of 25 democratic countries, yielding a total of 53 distinct electoral regimes, is analyzed in order to determine the effect of electoral institutional variables on the number and strengths of political parties. To resolve the argument in the literature of whether institutional influence is systematic, we change the traditional analytical setting and bring into the analysis variables characterizing the internal cohesiveness of the societies under consideration. In particular, we take district magnitude as our institutional variable, and ethnic, religious, and linguistic fragmentation of electorates as characteristics of internal cohesiveness. Our conclusion is that the role of electoral institutions is in mediating the impact of social factors on the formation of political landscapes, and that, therefore, institutional features should enter the analysis interactively with the parameters describing societies, i.e., institutional influence should not be viewed as absolute and independent of social context.

Chapter 3 contains theoretical analysis of a particular electoral system - the Single Non-Transferable Vote (SNTV) system and establishes some equilibrium properties of SNTV where the entry of new competitors is allowed. Among other things, we show that, in cohesive electorates, the SNTV system would generate an equilibrium in which there will be either no competitors in excess of the number of seats being filled, or just one extra competitor. Assumptions that lead us to this and other results are: (1) unidimensional policy space, (2) sincere Downsian voters with single-peaked preferences over policies, and (3) strategic candidates who care primarily about getting in office.

As our theoretical conclusions depend strongly on these assumptions, we offer in Chapter 4 empirical evidence in support of those conclusions. We analyze Japanese and Taiwanese district-level races (both countries used SNTV for their legislative elections, although Japan is now undergoing an electoral reform), and find that the equilibrium features that we predict theoretically under SNTV are, indeed, systematically present in elections there.

While Chapters 3 and 4 are devoted to candidate competition in multiseat elections, in Chapter 5 we look at the incentives and constraints that electoral rules create for political parties. Our analysis there applies to at-large electoral districts. We find that implicit or explicit thresholds for legislative representation seriously influence the number and strength of the parties that form. We also derive endogenous limits on the number of competing parties.

Because the analysis in Chapter 5 proceeds mainly within the same framework as Chapter 3, our resulting conclusions about the candidate-based and party-based electoral competition in the single-member as well as multimember districts are compatible.

Thesis Files