Duverger’s Law remains the canonical statement of the political consequence of
electoral systems, and one that informs the topic of electoral system change in
general.
Duverger’s Law notes that single-member simple plurality electoral systems are
associated with far fewer parties than are systems such as list PR (list proportional
representation) and oVers a causal explanation of why this is the case (see Cox 1999
for details and also Riker 1982 ). This statement of the relationship between electoral
system and number of parties has meant that an important agenda for electoral
systems research has been identifying who wins and who loses under the huge
variety of electoral rules (see, e.g., Rae 1971 ; Farrell 2001 ). Duverger is credited with
identifying one of the more prominent, if not the most prominent, eVects and
subsequent scholarship has searched for and established other such eVects in other
countries or with other systems, and with greater precision and detail. The electoral
systems literature is one of the more advanced within political science and a large
part of that advance has been due to ever better elaboration and generalization of
the kinds of eVects noted by Duverger.
In a practical political sense, because electoral systems make winners and losers,
the question of which electoral system is chosen to be used is an important one. It
did not take Duverger to realize that point. Electoral changes in Victorian Britain
make it plain that at least some politicians of the time understood the point. But
Duverger’s Law is an especially clear and focal statement of the argument.
Taken together, these two points mean that the study of electoral systems is one
that is closely allied to the study of institutions more generally and, in some sense,
represent an idealized type of what Tsebelis calls ‘‘distributional’’ institutions.
Tsebelis categorizes institutions as either ‘‘distributional’’ or ‘‘eYcient’’ (Tsebelis
1990 , 104 – 15 ). EYcient institutions are ones in which all or almost all people are
made better oV. Examples of these kinds of institutions might be the rules of the
road in which the rule is to stop at red traYc lights and go on green, or the decision
to drive on one side of the road rather than other. Others might range from the
adoption of a standardized system of weights and measures through (more
arguably) to a rule of law. This kind of institutional arrangement beneWts all or
most people. Distributional institutions, however, divide people into winners and
losers. Knight argues that all institutions have, at their base, some distributional
element. Even seemingly innocuous ones (say whether to use imperial or metric
systems of measurement) that make almost everyone better oVmay make some
people better oVstill, in which case there is scope for conXict among those many
who get better over who gets best oV(Knight 1997 ). It may be more accurate then to
talk of a continuum of institutions whose end points are deWned by idealized types
that are never fully realized in the real world.
But even with a more nuanced categorization of institutions, electoral systems
remain one of the clearest examples of distributional institutions. Not only do
electoral systems make winners and losers, this fact is common knowledge among
578 shaun bowler