voter, and he and Laver applied this notion successfully in a number of empirical
cases. Laver and Shepsle took a model that Shepsle ( 1979 ) had originally developed
for the US Congress to describe how particular parties would form speciWc
coalitions based on policy positions, even when there was no dominant majority
outcome. Essentially, the coalition process strikes bargains in which party A is given
control over policy x, party B over policy y, etc. It would be a diVerent coalition
with diVerent policy outcome if party A controlled policy y and B policy x. They
and others applied their model extensively to explain governments that formed
(Laver and Shepsle 1994 ).
If theWrst question was which government formed, the second was how long
would it last. Again, this literature moved toward alignment between theory and
substance, but in this case, the literature unfolded in close dialogue between the two.
A short version of this is that Browne, Frerdreis, and Gleiber ( 1986 ) developed a
sophisticated statistical model of government duration that essentially showed how
governments could handle exogenous shocks (or collapse in the face of them). King,
Alt, Laver, and Burns ( 1990 ) developed this approach further. Lupia and Strom
( 1995 ) then began to develop a theoretical model that endogenized these events,
followed by an increasingly sophisticated series of game theoretic models by
Diermeier and associates (Diermeier and van Roozendaal 1998 ; Diermeier and
Stevenson 1999 ) that moved toward testable implications to pit against and
eventually extend the original statistical modeling of Browne, Frendreis, and Gleiber.
All of this increasingly precise, sophisticated, and empirically extensive research
treats the parliamentary party as the unit of analysis. Two developments have
moved towards treating the member of parliament as the unit of analysis. One
thrust was due to the study of new democracies and therefore the study of
the formation of parliaments and their practices, especially in Latin American
(Morganstern and Nacif 2001 ) and former Soviet Union and Warsaw Pact
nations. Smyth ( 2006 ), for example, examines early Russian Duma elections to
study conditions under which candidates in their mixed system would choose to
ally with a political party and when to run as an individual. Remington ( 2001 ) and
Remington and Smith ( 2001 ) examine the formation of the Duma in the new era,
looking at many of these same questions, while Andrews ( 2002 ) examines the
policy formation process (or its failure!) in the early years of the post-Soviet Duma,
Wnding precisely the kind of theoretical instability and policy chaos that underlies
much of the theoretical work noted above. This study shows that the apparent
stability of policy choices of most established legislatures, including the US
Congress and the archetypal European multiparty parliament, needs to be
derived—apparently from an established party system—rather than be assumed.
Party instability occurs even in established parliaments, however. Heller and
Mershon ( 2005 ) have examined theXuidity in MP partisan attachments after the
reforms of the Italian parliamentary system. Here, unlike the Russian case, there
seems to be reasonable policy stability within a great deal of partisan instability.
570 john h. aldrich