The Effects of Proportional Seat Allocation Methods upon Legislative Fragmentation: Evidence from Russia

Abstract

This article examines the effects exerted by different seat allocation methods upon legislative fragmentation, conventionally operationalized as the effective number of parliamentary parties (ENPP), in proportional representation systems. A real-life data set of 191 sub-national elections held in Russia in 2001–13 is used to assess relatively restrictive methods, Imperiali and d'Hondt highest averages, against a more permissive method, Hare largest remainders. Statistical analysis demonstrates that if fragmentation in the electorate and the amount of wasted votes are properly controlled for, the effects of proportional seat allocation rules can be assessed with a high degree of precision. The Imperiali method reduces the ENPP by 0.34 in comparison with the Hare method, while the d'Hondt method makes it smaller by 0.12. Restrictive seat allocation rules suppress legislative fragmentation primarily because they lead to the over-representation of major parties.

Key words:

• Electoral systems
• seat allocation formulas
• proportionality
• effective number of parties
• legislative fragmentation
• Russia

Notes

1 The Hare method, also known as the Hare–Niemeyer method, belongs to the family of largest remainder methods of seat allocation. Under such methods, the numbers of votes for each party are divided by a quota representing the number of votes required for a seat (for the Hare method, it is defined as the quotient of the total number of valid votes and the total number of seats to be allocated), which usually yields an integer part and a fractional remainder. Unallocated seats are given to the parties with the largest remainders. In the highest averages systems, the votes that parties receive are divided by a series of numbers. Seats are allocated to the parties that have the highest average. As a series of divisors, the d'Hondt method uses 1, 2, 3, 4, etc., and the Imperiali method, 2, 3, 4, 5, etc. The latter is equivalent to 1, 1.5, 2.0, 2.5, etc.

2 All quantitative empirical evidence reported in this study are based on the author's calculations performed on the data from the official Internet sites of the Central Electoral Commission of Russia, http://www.cikrf.ru and http://www.izbirkom.ru, last accessed in January 2014, and from the previous version of these sites, http://www.fci.ru, at the time when it was available.