a typology of contemporary party systems

A TYPOLOGY OF CONTEMPORARY PARTY SYSTEMS
Paper Presented at the 20th World Congress of the
International Political Science Association, Fukuoka, Japan,
9 — 13 July 2006
SS03.476
Global Theoretical Issues for Political Parties
Alan Siaroff
Department of Political Science
The University of Lethbridge
4401 University Drive
Lethbridge, Alberta, Canada
T1K 3M4
[email protected]
Introduction
The classic typology of party systems is that of Sartori (1976),
which
distinguishes
amongst
two
party
systems,
moderate
pluralism, polarized pluralism, and predominant-party systems.
However, as Mair (2002: 93) has argued, most every party system
is now one of moderate pluralism, with little remaining in the
other types. Consequently, in this paper I shall outline a new
typology, building on Blondel (1968) and Ware (1996), which
looks both at the number of parties and their relative size.
Following Ware (1996), a relevant party for this analysis will
be one with at least three percent of the seats in the (lower
house of the) legislature.
This typology will comprise
initially nine different party systems, as follows: imbalanced
two-party, competitive two-party, two-and-a-half-party, moderate
multiparty with one dominant party, moderate multiparty with two
main parties, moderate multiparty with a relative balance
amongst the parties, extreme multiparty with one dominant party,
extreme multiparty with two main parties, and extreme multiparty
with a relative balance amongst the parties.
Each of these
party systems will be defined and illustrated.
This typology
will then be used to classify the current party system pattern
of every electoral democracy in the world that has political
parties (through the end of May 2006). The consequent forms of
government for such systems will also be given. It will be seen
that some of these categories have very few cases, thus a
revised typology of six or seven categories will be proposed.
Hopefully this final typology should serve to disaggregate the
now-“overcrowded” category of moderate pluralism.
Earlier Classifications
The initial postwar classification of party systems relied in
whole or in part on quantitative (“counting”) measures of size
and relative size.
First there was Duverger’s (1954)
distinction between two party and multiparty systems. This was
made more elaborate by Blondel (1968: 184ff.), who suggested
1
four categories (based more on vote shares than seat shares):
two- party systems (two-party share of 90 percent or more),
three-party systems — which in fact are more strictly two-and-ahalf-party systems (two-party vote share between 75 and 80
percent), multiparty systems with a dominant party (dominant
party obtaining about 40 percent of the votes and generally
about twice as much as the second party), and multiparty systems
without a dominant party. That said, Blondel (1968: 184) notes
that Australia and Austria are ambiguous cases within the first,
two-party, group, as sometimes a smaller third party also
partakes in or at least supports a government. In my analysis,
the problem here — with Rae (1971: 93) making a similar “error”
— is using 90 percent as the cut-off for “two-partyness”.
A
cut-off of 95 percent is more definitive and will be used in
what follows.
Starting in the 1990s, an alternative way of examining party
systems – at least those in developing countries – arose, this
involving the notion of institutionalization.
These analyses
tended to be geographically limited to a given region, to wit
Mainwaring and Scully (1995) and the Inter-American Development
Bank (Payne et al. 2002) on Latin America, and Kuenzi and
Lambright (2001) on Africa.
However useful this may be for
developing countries and indeed some developed ones in postcommunist Europe (Estonia, Latvia, and Poland come to mind), it
does
not
seem
to
be
useful
globally
beyond
a
basic
institutionalized or not distinction.
In the 1990s, Ware (1996) provided a schema which returned to
using quantitative measures.
In examining the industrial
democracies,
Ware
(1996:
158ff.)
thus
has
four
overall
categories.
First, there is a predominant party system,
characterized by a continuous majority for a given party and by
a divided opposition to this (p. 159). Japan from 1960 through
the mid-1990s (the time of his analysis) is his sole example
here. Second, there is a two-party system, characterized by (a
median value of) two and only two relevant parties — with
relevance being defined by having at least three percent of the
2
seats. New Zealand and the United Kingdom from 1960 through the
mid-1990s are (were) examples here.
Next, there are party
systems with from three to five inclusive relevant parties.
These systems are further divided into three subtypes: two-anda-half-party systems (that is, two larger parties usually with
80 percent or more of the seats between them, often no single
party majority, and thus a smaller party holding the balance of
power), systems with one large party getting at least 45 percent
but not 50 percent of the seats (otherwise this would be a
predominant system), and even systems.
For the period up to
1990, Germany and Ireland were examples of Ware’s first subtype,
Norway and Sweden of the second, and Iceland and Luxembourg of
the third.
Finally, there are systems with more than five
relevant parties. These are further divided into two subtypes:
those with two large parties (with a combined seat share of 65
percent or more) and those which are evenly balanced.
Israel
and Italy are his (classic) examples of the first subtype here;
Finland and Switzerland of the second.
Some interesting differences between these scholars arise over
the issue of predominance. Whereas for Ware this presupposes a
one-party seat majority, for Sartori (1976: 196) this criterion
can be relaxed in those countries where single party governments
are normally formed on less than the absolute majority principle
(that is, in Scandinavia).
At the other extreme, Blondel as
noted does not ever require a majority, merely that the largest
party is double the size of the next largest.
There is also the issue of whether a predominant party system is
a category of its own, or whether predominance is but a factor
that can be combined with any type of party system. The latter
view is argues by Mair (2002: 106-107 [his endnote 3]):
Although the predominant-party system constitutes a
useful
category,
it
fits
rather
uneasily
into
Sartori’s framework, since it is defined by wholly
different criteria, and can by definition co-exist
with every possible category of party numbers (that
3
is, it can develop within a context of a two-party
system, a system of limited pluralism, and a system of
extreme pluralism) and, at least theoretically, with
every possible spread of ideological opinion.
We shall return to this point empirically later.
For now,
though, it is worth noting that Sartori’s criteria here include
duration — more so than for his other party systems. That is,
Sartori requires four consecutive absolute majorities for single
party predominance, or even just three absolute majorities (p.
199) if the system is stable and clear.
Electoral Democracies, Party Politics, and Party Systems in the
World Today
As of the end of May 2006 there are by my assessment 117
electoral democracies in the world, not counting Serbia and
Montenegro which is in the process of dissolution.
Of these
117, six island states in the Pacific Ocean – Kiribati, the
Marshall Islands, Micronesia, Nauru, Palau, and Tuvalu – do not
have political parties for cultural-traditional reasons (Anckar
and Anckar 2000).
This leaves 111 electoral democracies with
political parties. Yet in two of these – Thailand and Venezuela
– the opposition boycotted the last elections, making analyses
thereof problematic. These two countries are thus excluded from
what follows.
Nevertheless, this still leaves 109 electoral
democracies with cross-national variations in party systems.
But for how long have they varied? This leads us to the “time
frame problem”: for how long (how many elections) does a party
system need to exist?
Sartori (1976) says four continuous
elections for his predominant party system. Does this mean that
a pattern (predominant or otherwise) of only two or three
elections is not a party system?
Presumably one election does
not make a party system, of course. There is indeed no accepted
standard of the number of elections or time length – nor any
agreement on what to call a pattern that does not last the
4
required length of time (transitional ? deviating ?).
The main “solution” to this problem has been to use an overall
long time frame of decades and then determine the standard
pattern therein for each country. Thus Blondel (1968) noted the
average pattern for the period 1945 to 1966.
Likewise, Ware
(1996: 158ff.) measured the standard national patterns for the
period 1958 to 1990.
Yet both Blondel and Ware limited their
analysis to the industrial democracies. Looking at a long time
frame (say, 1970 or 1980 to the present) is problematic
globally, since many of the current electoral democracies are
rather new.
Consequently, I shall thus indicate the party system pattern
based on the most recent election (through the end of May 2006)
for the 109 electoral democracies of analysis and then note
those countries where this reflects an actual stable pattern.
Again, though, the notion of how many elections for stability is
debatable, and is essentially beyond the scope of this paper.
The raw data for the following classifications (plus other
hopefully interesting data) is given in the Appendix Table at
the end, which is updated from Siaroff 2005, Table 7.3.
The Initial Classification
It is important to stress that the following classifications are
always based on the percentage, or ratio, of seats and not votes
— since it is seats which determines legislative power and (in
parliamentary systems) ultimately governmental control.
Brief
comments are given, though, on the electoral systems of the
various countries which demonstrate a given party pattern, and
likewise the resulting type(s) of governments — at least for
parliamentary systems — are noted (single party, coalition of
one large and one small party, multiparty coalition, et cetera).
Overall,
moderate
the first distinction is between two-party
multiparty systems, and extreme multiparty
5
systems,
systems.
Further criteria break these down into two, four, and three
categories respectively.
That is, in a simplified manner, we
can use the number of parties with at least three percent of the
seats (Ware’s criterion) to suggest three overall categories:
two-party systems (P3%S of two), moderate multiparty systems
(P3%S of three to five), and extreme multiparty systems (P3%S of
more than five).
The breakpoints here thus follow Sartori's
(1976) analysis of fragmentation.
However, with additional
factors we are able to thus expand these categories into more
precise party systems. In this analysis a party system is thus
considered initially to be one of nine different types,
depending on various quantitative factors. Of course, a polity
may go through, or shift amongst, various party systems over
time, however the main focus here is the contemporary situation.
That said, certainly a party system that is the output of only
one election cannot be presumed to be durable.
Thus I shall
distinguish between the single election party system pattern
(that is, the result of the most recent election) and those
countries which actually exhibit the party system (based on the
past few elections, or a historical series of elections).
Party system classification is determined by a combination of
the following factors: (i) the number of parties with at least
three percent of the seats — P3%S; (ii) the two-party seat
concentration value — 2PSC, with the key breakpoints here being
at 95 percent and 80 percent; (iii) the seat ratio between the
first and second party — SR1:2, with the key breakpoint here
being set quite high at 2.0 (reflecting Blondel’s notion of the
largest party being twice the size of the second largest,
although again the focus here is seats and not votes); and (iv)
the seat ratio between the second and third party — SR2:3, with
the breakpoint here being hypothesized at 2.5.
(1) Imbalanced (lopsided) two-party systems are those with a
2PSC of at least 95 percent, and a SR1:2 of 2.0 or more (that
is, double the seats of the opposition party). Usually the P3%S
value will be exactly two, and can never be more than three.
6
The dominant party will always have a comfortable majority of
seats — above 63 percent of these, given the parameters.
That
said, the 2PSC of at least 95 percent criterion can be relaxed
if there are but two parties yet also many independents
(currently the case in Georgia and Samoa). Also, the SR1:2 can
drop a bit below 2.0 if the long-term median value remains at or
above this.
Based on the last national election, such an imbalanced twoparty pattern exists in Barbados, Belize, Botswana, Georgia,
Samoa, Seychelles, Sierra Leone, St. Lucia, St. Vincent and the
Grenadines, and Turkey.
Of these ten countries, Botswana with
the dominance of the Botswana Democratic Party (BDP) since 1984
is the best (indeed the only) actual ongoing example of such an
imbalanced two-party system (as opposed to just an imbalanced
two-party pattern).
Two other historical examples of such a
system were Bahamas from the 1972 through the 1987 elections
inclusive (four elections) with the dominance of the Progressive
Liberal Party (PLP), and St. Vincent and the Grenadines from the
1984 through the 1994 elections inclusive (three elections) with
the dominance of the New Democratic Party (NDP).
In terms of electoral systems, half — but only half — of the
countries with this pattern use single member plurality (SMP).
Conversely, then, such a pattern is certainly possible under
other electoral systems.
(2) Competitive two-party systems are those with a 2PSC of at
least 95 percent, and a SR1:2 below 2.0. Usually the P3%S value
will be exactly two, and can never be more than three.
Given
the very high 2PSC, it is almost always the case that one of the
parties will have a majority of seats. However, the other party
should be in “striking” distance, at least with respect to the
next election. [Ideally, the two parties will win an equal
number of elections over the long haul (such as in St. Lucia
since universal suffrage), or at least a competitive number of
elections over the long haul.]
7
Based on the last national election, such a competitive twoparty pattern exists in Cape Verde, Chile (in terms of electoral
coalitions), Dominica, Ghana, Grenada, Italy (in terms of
electoral coalitions), Jamaica, Malta, Mauritius, Mozambique,
Nicaragua, Trinidad and Tobago, and the United States.
Some
clear durable examples of a competitive two-party system are
Chile (in terms of the centre-left Democratic Coalition and
rightist alliances), Jamaica (with the Jamaica Labour Party
(JLP) and the People’s National Party (PNP) as the two parties),
Malta (with the Malta Labour Party (MLP) and the Nationalist
Party (NP) as the two parties), and the United States (with the
Democrats and the Republicans as the two parties).
In terms of electoral systems, again almost half of the
countries with this pattern use SMP).
That said, of the
multimember electoral systems in these countries the district
magnitude is often small (5 in Malta) or very small (mostly 3 in
Mauritius, and just 2 in Chile).
(3) Moderate multiparty systems which are in fact two-and-ahalf-party systems are those with a P3%S of two to five
(although normally three to four), a 2PSC of between 80 and 95
percent, a SR1:2 below 2.0, and a SR2:3 of 2.5 or more (if not
indeed 3.0 or more). In other words, the gap between the second
and third parties is clearly greater than the gap between the
first and second parties.
Given these parameters, it is
certainly possible that one of the two main parties will have a
majority of the seats.
However, such a majority is not
necessarily probable – as Ball and Peters (2000: 112) state,
two-and-one-half-party systems are those “in which there are two
dominant parties but the larger parties rarely command an
overall majority and need the support of other parties to form a
government”.
Based on the last national election, such a two-and-a-half-party
pattern exists in Andorra, Australia, Austria, Greece, Guyana,
8
Honduras, Hungary, Kenya, South Korea, Liechtenstein, Moldova,
Mongolia, New Zealand, Portugal, São Tomé and Príncipe, Spain,
Sri Lanka, the United Kingdom, and Uruguay.
Several of these
countries have had a durable (longstanding) two-and-a-half-party
system, such as Australia (with the Liberal Party and the Labor
Party as the two main parties), Portugal (with the Social
Democratic Party and the Socialist Party as the two main
parties), Spain (with the Popular Party and the Socialist Party
as the two main parties), and the United Kingdom (with the
Conservative Party and the Labour Party as the two main
parties). One longstanding historical example of such a system
was Germany (with the Christian Democrats and the Social
Democrats as the two main parties) — indeed, this was definitely
Germany’s party system until its last elections.
The electoral systems used in the countries with this pattern
have involved a fair amount of bias in about half the cases;
that is, SMP, AV (Australia), or proportional representation
(PR) with small districts (Spain).
Government patterns under
this party system have involved single party majorities
(currently in Portugal and the United Kingdom), single party
minorities usually with formal or tacit support from a smaller
party (currently in Spain), or a formal two-party coalition
between one of the main parties and a smaller party (currently
in Australia and the historical norm in Germany).
(4) Moderate multiparty systems with one dominant party are
those with a P3%S of three to five, and a SR1:2 of 2.0 or more.
Given these parameters, it is certainly probable that the
dominant party will have a majority of the seats. In any case,
it will be clearly far ahead of each of the other parties.
Also, over a given time period the SR1:2 can drop a bit below
2.0 if the long-term median value remains at or above this.
Finally, for such a system to continue across time we must be
dealing with the same party of course.
9
Based on the last national election, such a moderate multiparty
with one dominant party pattern exists in Antigua and Barbuda,
Bahamas, Bangladesh, Burundi, the Dominican Republic, France,
Japan, Lesotho, Madagascar, Namibia, Nigeria, Senegal, South
Africa, St. Kitts and Nevis, and Timor-Leste.
Durable ongoing
examples of such a party system are those of Namibia (with the
dominance of the South West African People’s Organization
[SWAPO]) and South Africa (with the dominance of the African
National Congress [ANC]).
Historical examples here of such a moderate multiparty with one
dominant party system are Botswana from 1969 through 1979
inclusive of all said elections, Norway from 1949 through 1973
inclusive (with the dominance of the Labour Party), Japan from
1960 through 1993 inclusive (with the dominance of the Liberal
Democratic Party [LDP]), and Sweden from 1958 through 1970
inclusive and again from 1979 through arguably 1988 inclusive
(with the dominance of the Social Democratic Party [SAP]). Such
a pattern in Japan (again with LDP dominance) but not yet a
definite system was reestablished in its last election. As for
India, it moved in and out of this pattern from independence
through the mid-1970s, since its P3%S ranged from two to six.
Electoral systems in the countries which currently have this
pattern are, surprisingly, often PR, but many majoritarian ones
too (France, the SNTV system used historically in Japan). Thus
this party system type can be said to arise under varying
electoral systems.
The definitive pattern of governance under
this party system is single party rule, normally with a majority
(see below).
(5) Moderate multiparty systems with two main parties are those
with a P3%S of three to five (although normally four to five), a
SR1:2 below 2.0, and a SR2:3 of 2.5 or more.
Just as in twoand-a-half-party systems, here the gap between the second and
third parties is greater than the gap between the first and
second parties. However, it is unlikely that either of the two
main parties will have a majority of the seats.
10
Based on the last national election, such a moderate multiparty
system with two main parties pattern exists in Albania, Costa
Rica, Croatia, Germany, Mali, and Suriname. Albania is the most
longstanding and thus clear example of such a party system (with
the Democratic Party and the Socialist Party as the two main
parties).
Indeed, in the last Albania election most parties
outside of the two main ones did not bother to contest the
single member constituency seats, just the national PR ones.
Electoral systems here tend to be either pure proportional
representation
or
parallel
systems.
Governments
(in
parliamentary systems) are clearly coalitions, in the current
German case even involving the two main parties.
(6) Moderate multiparty systems with a relative balance amongst
the parties are those with a P3%S of three to five (although
normally four to five), a SR1:2 below 2.0, and a SR2:3 below
2.5.
There is thus not a massive gap between the first and
second parties, nor between the second and third parties.
Indeed, there may not even be a huge difference between the
first and third parties.
Based on the last national election (through May 2006), such a
balanced moderate multiparty pattern exists in Canada, Cyprus,
the Czech Republic, El Salvador, Iceland, Luxembourg, Macedonia,
Malawi, Mexico, Paraguay, the Philippines, Romania, San Marino,
Switzerland, Taiwan, and Ukraine. An actual moderate multiparty
system with a relative balance amongst the parties has existed
for quite some time in Cyprus, Luxembourg, and Switzerland.
Electoral systems here are normally ones of proportional
representation.
Likewise, governments are normally coalitions.
Canada is an exception on both points (but it is also new to
this pattern).
11
(7) Extreme multiparty systems with one dominant party are
those with a P3%S of more than five, and a SR1:2 of 2.0 or more.
Given such fragmentation, it is unlikely that the dominant party
will have a majority of the seats — and thus be as dominant as
in category (4). As such, the dominant party here could be kept
out of office; in fact, however, it tends to be in a strong
position for government formation, even if this requires an
executive or at least a legislative coalition.
Finally, as in
system (4), for party system (7) to continue across time we must
be dealing with the same party.
Based on the last national election, such an extreme multiparty
with one dominant party pattern exists in Argentina, Benin,
Ireland, Panama, and Sweden.
That said, in none of these
countries has such a durable party system existed for the past
few elections.
Rather, the key examples of this party system
are historical: first and foremost Israel from 1949 through the
1961 election (or even the 1969 election) inclusive (with the
dominance of the Labour Party), but also Denmark for the three
elections from 1977 through 1981 inclusive (with the dominance
of the Social Democrats).
[Note that for Sartori (1976: 154)
the Israeli Labour Party was never dominant since it was always
below 40 percent of the vote.]
Proportional representation electoral systems are found with
this party system / pattern. With this pattern, (parliamentary)
governments are indeed built around the dominant party, either
heading a coalition or forming a single party minority.
(8) Extreme multiparty systems with two main parties are those
with a P3%S of more than five, a SR1:2 below 2.0, and a SR2:3 of
2.5 or more.
Presumably, in such a system neither of the two
main parties will have a majority of the seats.
Based on the last national election, such a pattern exists only
in India (with the BJP and Congress as the two main parties).
Yet in India this is not just a pattern but a durable party
12
system, existing since 1996 (that is, for four elections thus
far). India uses single member plurality and has had coalitions
of one of the main parties with many smaller parties.
Israel from 1973 through 1996 inclusive (with Labour and Likud
as the two main parties) is the best historical example of such
a system, even if a couple of these elections fell short on one
of the criteria.
(9) Finally, extreme multiparty systems with a relative balance
amongst the parties are those with a P3%S of more than five, a
SR1:2 below 2.0, and a SR2:3 below 2.5.
There is thus not a
sizable gap between the top parties.
However, since the norm
here is to have a 2PSC of below 60 percent (and certainly below
65 percent), unless the top two parties combine, then at least
three parties would be necessary to achieve a majority of the
seats.
Based on the last national election (through May 2006), such a
balanced extreme multiparty pattern exists in Belgium, Bolivia,
Brazil, Bulgaria, Colombia, Denmark, Ecuador, Estonia, Finland,
Guatemala,
Indonesia,
Israel,
Latvia,
Lithuania,
the
Netherlands, Niger, Norway, Papua New Guinea, Peru, Poland,
Slovakia, Slovenia, the Solomon Islands, and Vanuatu. Of these
countries, one can note that Papua New Guinea and the Solomon
Islands barely have party politics at all (more so independents
and highly fluid parties).
Of the rest, a clear extreme
multiparty system with a relative balance amongst the parties
exists in Belgium, Brazil, Estonia, Finland, Latvia, and the
Netherlands, if not indeed elsewhere.
Outside of the cases in Oceania, the electoral system used in
countries with this pattern is almost always party list
proportional representation.
Moreover, governments tend not to
be just coalitions but multiparty coalitions.
13
Party Systems and a Majority Government (or Not)?
Of the many ways of measuring a party system, or even assessing
an election, a key if not indeed the key one is whether any
party actually wins or not (Siaroff 2003: 272), with “winning”
normally meaning a single party majority.
For the most recent
election in the countries under study, we can distinguish
between
(single
party)
majority
governments
and
hung
parliaments,
and
indeed
between
earned
majorities
and
manufactured ones.
Of the ten countries currently with an
imbalanced
two-party
system
pattern,
all
have
majority
governments, and indeed all but one of these are earned
majorities.
Of the 13 countries currently with a competitive
two-party system pattern, here as well all have majority
governments, although four of these reflect manufactured
majorities versus nine earned ones. The 19 countries currently
with a two-and-a-half-party system pattern provide the clearest
division: eight of these have majorities (three earned, five
manufactured) whereas 11 have hung parliaments.
This confirms
my previous finding (Siaroff 2003: 272) that this category does
not have a clear pattern with regard to majority governments or
not.
In contrast, a moderate multiparty system with one
dominant party is decisively majoritarian.
All 15 countries
currently with this pattern have majority governments (ten
earned, five manufactured).
[That said, such single party
majorities were not always true of the historical moderate
multiparty systems with one dominant party in Scandinavia,
especially Sweden.]
Thus it is useful to see this category as
distinctive and speaking to predominance, perhaps resulting from
Ware’s (1996: 159) point that a predominant party faces — and
presumably benefits from — a divided opposition.
As for the
remaining five categories, they are all cohesive in the opposite
way: With the single country exception of Panama (category 7),
none of the countries with these patterns has a single-party
majority, and thus all these categories can be said to reflect
hung parliaments.
14
Simplifying the Typology
One way to assess party systems generally may well be to focus
on the seat share of the largest party — 1PSC, since this (or
more precisely its inverse) has been suggested by Taagepera
(1999) as a key supplemental measure.
First off, Table One
shows the relationships between the number of relevant parties
(in Ware’s sense of having at least three percent of the seats,
P3%S) and the size of the largest party (the one-party seat
concentration, 1PSC).
We see here that from P3%S values of 2
through 8, the mean 1PSC drops continuously.
Conversely,
however, it is also true that the differences between the P3%S
values of 2 and 3 and also the values of 5 and 6 do not stand
out, limiting the argument of having three overall categories
based on P3%S (2, 3 to 5, 6 and more).
TABLE ONE: 1PSC by P3%S
P3%S
N
2
3
4
5
6
7
8
9
10
25
20
20
14
10
13
3
3
1
1PSC:
minimum
maximum
mean
46.9
31.6
35.0
27.5
17.4
8.0
16.7
17.7
24.2
82.4
74.2
65.3
76.4
51.3
41.3
22.1
39.4
24.2
63.2
55.8
49.8
42.3
34.2
28.9
19.3
27.0
24.2
More useful, however, is to assess 1PSC values in terms of the
aforementioned nine categories, as is shown in Table Two. (Note
that 1PSC is not part of any category definition.)
15
TABLE TWO: 1PSC by Party System Category
(Pattern Classification Based on Most Recent Election)
category
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
N
10
13
19
15
6
16
5
1
24
1PSC:
minimum
maximum
mean
66.0
53.3
41.3
53.9
36.8
27.5
37.3
26.7
8.0
82.4
64.0
59.5
76.4
45.8
50.0
51.3
26.7
41.6
73.4
56.8
50.0
65.4
42.2
39.1
43.6
26.7
26.5
We can possibly combine the first two categories and thus have a
single two-party system category.
Both of these categories
produce only majority governments (at least in the current
situations), both tend to use majoritarian electoral systems,
and there have been relatively few durable examples of
imbalanced
two-party
systems.
Conversely,
given
the
aforementioned factors the typical pattern in a two-party system
is for a given election to be lopsided and thus make the entire
system appear imbalanced (category 1), even though in most cases
such imbalance swings back and forth every election or two.
Alternatively, the category of imbalanced two-party system can
be retained on its own, but then one would need to be clear on
the minimum number of elections required to define this.
In any case, there are certainly a sufficient number of
countries with two-party systems to make it incorrect to claim
that
this
(combined)
category
has
undergone
a
“virtual
emptying”, as Mair (2002: 92) does. What has occurred, indeed,
is that there are now very limited examples of such a system in
the industrial democracies — just the United States and (in
terms of electoral alliances) Italy.
However, viewed globally
this is (still) a very extant system.
16
We can clearly combine categories 5 and 6, since there are few
patterns of category 5 and these two categories do not vary much
in terms of 1PSC values — and both differ greatly here from the
1PSC values of category 4.
The logical name for this new,
combined category would be “moderate multiparty systems without
a dominant party”. One can also note that for this new combined
category the mean 2PSC value is 69.9, in contrast to the mean
2PSC value for two-and-a-half-party systems of 88.4.
Even more definitely we can combine categories 8 and 9, since
category 8 only contains India, and each main party there has
needed to lead a broad coalition — as per category 9.
The
logical name for this new, combined category would seem to be
“extreme multiparty systems without a dominant party”.
That
said, the notion of dominance in the initial category 7 does not
match that in the initial category 4, since in the initial
category 7 (Panama excepted) hung parliaments remain the result.
Perhaps, then, what was category 4 should speak of predominance
and what was category 7 just of dominance?
In any case, in conclusion, the proposed simplified typology of
six or seven categories should serve well to break up and
separate clearly the hitherto “overcrowded” category of moderate
pluralism.
Conversely, almost all of the six or seven revised
categories contain lots of examples of electoral democracies
with the category pattern, and multiple cases of electoral
democracies with such a durable party system (the ongoing
category 7 admittedly less so).
In other words, the global
empirical reality is indeed spread out across these various
categories.
A Couple Additional Points
This paper has made a distinction between party systems and
party system patterns — with the latter indeed reflecting just
the current situation (that of the most recent national
17
election).
Ideally one would just want to analyse party
systems, or at least look for the most common system in each
country over a reasonably long time frame. Perhaps this can be
done by the end of this decade. By that time, one could analyse
the democratic world’s party systems since, say, 1990 — a time
period (by then of a couple of decades) long enough to involve
several elections in almost all electoral democracies, since
there have been only a small number of (16 or so) electoral
democracies (and/or sovereign nations) which have arisen since
1995 and thus for which there have not been several democratic
elections.
Such an approach would involve taking mean scores
over time for the variables P3%S, 2PSC, and 1PSC, and median
scores over (the same) time for the ratios SR1:2 and SR2:3.
Secondly, one issue that can be largely settled at this point is
the debate over predominance and whether it occurs in
combination with all numerical numbers of relevant parties. In
fact, predominance really is something found largely in moderate
multiparty systems, and can and should be seen primarily as a
subcategory within these.
References
Anckar, Dag, and Carsten Anckar (2000), “Democracies Without
Parties”, Comparative Political Studies 33: 2 (March), pp.
225-247.
Ball, Alan R., and B. Guy Peters (2000), Modern Politics and
Government, 6th edition (New York and London: Chatham
House).
Blondel, Jean (1968), “Party Systems and Patterns of Government
in Western Democracies”, Canadian Journal of Political
Science 1: 2 (June), pp. 180-203.
18
Duverger, Maurice (1954), Political Parties: Their Organization
and Activity in the Modern State (New York: John Wiley &
Sons).
Kuenzi, Michelle, and Gina Lambright (2001), “Party System
Institutionalization
in
30
African
Countries”,
Party
Politics 7: 4 (July), pp. 437-468.
Mainwaring, Scott, and Timothy R. Scully (1995), “Introduction:
Party Systems in Latin America”, pp. 1-34 in Mainwaring and
Scully, eds., Building Democratic Institutions: Party
Systems in Latin America (Stanford: Stanford University
Press).
Mair, Peter (2002), “Comparing Party Systems”, pp. 88-107 in
Lawrence LeDuc, Richard G. Niemi, and Pippa Norris, eds.,
Comparing Democracies 2: New Challenges in the Study of
Elections and Voting (London: Sage).
Payne, J. Mark, et al. (2002), Democracies in Development:
Politics and Reform in Latin America (Washington, DC:
Inter-American Development Bank).
Rae, Douglas W. (1971), The Political Consequences of Electoral
Laws, revised edition (New Haven, CT and London: Yale
University Press).
Sartori, Giovanni (1976), Parties and Party Systems: A Framework
for Analysis (New York: Cambridge University Press).
Siaroff, Alan (2003), “Two-and-a-Half-Party Systems and the
Comparative Role of the ‘Half’”, Party Politics 9: 3 (May),
pp. 267-290.
Siaroff, Alan (2005), Comparing Political Regimes: A Thematic
Introduction to Comparative Politics (Peterborough, ON:
Broadview Press).
19
Taagepera, Rein (1999), “Supplementing the Effective Number of
Parties”, Electoral Studies 18: 4 (December), pp. 497-504.
Ware, Alan (1996), Political Parties and Party Systems (Oxford:
Oxford University Press).
20
IPSA APPENDIX TABLE: DATA ON MOST RECENT ELECTION FOR ALL ELECTORAL DEMOCRACIES
[through end of May 2006]
YEAR = Year of Election
PFRG = Parliamentary Fragmentation
ENPP = Effective Number of Parliamentary Parties
P3%S = Parties With Three Percent or More of the Seats
2PSC = Two-Party Seat Concentration (combined seat percentage of the top two parties)
2PVC = Two-Party Vote Concentration (combined vote percentage of the top two parties)
1PSC = One-Party Seat Concentration (seat percentage of the top party)
1PVC = One-Party Vote Concentration (vote percentage of the top party)
ED = Electoral Decisiveness (EM = earned majority, MM = manufactured majority, HP = hung parliament)
SR 1 : 2 = Ratio of Seats Between the Largest Party and the Second-Largest Party
SR 2 : 3 = Ratio of Seats Between the Second-Largest Party and the Third-Largest Party
TVOL = Total Volatility (Vote Shares) Between the Given Election and the Previous One [Loosemore-Hanby
Index]
DISP = Disproportionality Between Seat Percentages and Vote Percentages (for all parties) [LoosemoreHanby Index]
Albania
Andorra
Antigua and Barbuda
Argentina
Australia
Austria
Bahamas
Bangladesh
Barbados
Belgium
Belize
Benin
Bolivia
Botswana
YEAR
PFRG
ENPP
P3%S
2PSC
2005
2005
2004
2005
2004
2002
2002
2001
2003
2003
2003
2003
2002
2004
0.737
0.589
0.443
0.815
0.583
0.653
0.444
0.484
0.358
0.858
0.366
0.791
0.799
0.359
3.81
2.43
1.80
5.41
2.40
2.88
1.80
1.94
1.56
7.03
1.58
4.79
4.96
1.56
5
4
3
9
3
4
2
4
2
8
2
6
7
2
70.0
89.3
94.1
47.2
90.0
80.9
90.0
85.0
100.0
33.3
100.0
55.4
48.5
98.2
21
2PVC
79.3
96.9
38.8
78.5
79.2
92.7
81.1
99.9
28.4
98.8
38.8
77.8
1PSC
40.0
50.0
70.6
39.4
50.0
43.2
72.5
64.3
76.7
16.7
75.9
37.3
27.7
77.2
1PVC
41.2
55.2
29.9
40.8
42.3
51.8
41.0
55.8
15.4
53.2
26.9
51.7
ED
SR
1:2
SR
2:3
HP
HP
EM
HP
HP
HP
EM
MM
EM
HP
EM
HP
HP
EM
1.33
1.27
3.00
5.00
1.25
1.14
4.14
3.11
3.29
1.00
3.14
2.07
1.33
3.67
3.82
5.50
4.00
1.11
5.00
3.63
7.00
3.65
∞
1.04
∞
1.36
1.04
12.00
TVOL
DISP
28.8
11.5
8.5
12.8
6.5
10.0
20.3
18.5
13.6
1.9
23.4
25.0
20.9
11.1
22.7
29.9
6.6
12.3
25.5
8.2
20.9
16.1
Brazil
Bulgaria
Burundi
Canada
Cape Verde
Chile *
Colombia
Costa Rica
Croatia
Cyprus
Czech Republic
Denmark
Dominica
Dominican Republic
Ecuador
El Salvador
Estonia
Finland
France
Georgia
Germany
Ghana
Greece
Grenada
Guatemala
Guyana
Honduras
Hungary
Iceland
India
Indonesia
Ireland
Israel
Italy *
Jamaica
Japan
Kenya
Kiribati
2002
2005
2005
2006
2006
2005
2006
2006
2003
2006
2002
2005
2005
2006
2002
2003
2003
2003
2002
2004
2005
2004
2004
2004
2003
2001
2005
2006
2003
2004
2004
2002
2006
2006
2002
2005
2002
2003
0.882
0.792
0.577
0.690
0.513
0.504
0.861
0.691
0.687
0.744
0.727
0.795
0.472
0.603
0.866
0.717
0.786
0.797
0.558
0.531
0.709
0.523
0.543
0.498
0.784
0.552
0.578
0.594
0.731
0.847
0.859
0.706
0.873
0.508
0.491
0.559
0.548
1.000
8.48
4.80
2.37
3.22
2.05
2.02
7.19
3.23
3.19
3.90
3.67
4.89
1.89
2.52
7.46
3.54
4.67
4.92
2.26
2.13
3.44
2.10
2.19
1.99
4.64
2.23
2.37
2.47
3.71
6.53
7.07
3.41
7.84
2.03
1.97
2.27
2.21
9
34.3
7
56.3
4
84.0
4
73.7
2
97.2
2
99.2
8
40.5
4
75.4
5
75.7
5
64.3
4
64.0
7
56.6
2
100.0
3
77.5
9
40.0
5
69.0
6
55.4
7
54.0
4
85.8
2
80.0
5
73.0
2
96.5
3
94.0
2
100.0
6
57.0
3
93.8
4
91.4
3
89.1
5
66.7
6
52.1
7
43.1
6
67.5
10
40.0
2
98.7
2
100.0
3
85.2
3
90.0
No Parties
22
34.3
50.9
80.2
66.5
95.0
90.8
34.8
62.2
56.5
61.4
54.7
54.8
96.7
76.1
65.8
50.0
49.2
57.2
75.3
69.5
86.0
90.8
45.9
94.8
82.6
64.7
48.9
40.1
64.0
36.9
96.5
99.5
69.2
91.5
17.7
34.2
59.0
40.3
56.9
54.2
22.1
43.9
45.8
32.1
35.0
29.7
61.9
53.9
24.0
36.9
27.7
27.5
61.5
67.2
36.8
55.7
55.0
53.3
31.0
52.3
48.4
49.2
34.9
26.7
23.3
48.8
24.2
54.1
56.7
61.7
59.5
17.7
31.0
58.6
36.3
52.3
51.8
19.0
36.4
33.9
31.1
30.2
29.0
53.1
52.4
33.8
25.4
24.7
33.3
67.6
35.2
45.4
46.7
25.6
53.1
43.2
33.7
26.7
21.6
41.5
21.9
48.3
52.3
38.2
61.0
HP
HP
EM
HP
EM
EM
HP
HP
HP
HP
HP
HP
EM
EM
HP
HP
HP
HP
MM
EM
HP
EM
MM
MM
HP
EM
HP
HP
HP
HP
HP
HP
HP
MM
EM
MM
EM
1.07
1.55
2.36
1.20
1.41
1.20
1.20
1.39
1.53
1.00
1.21
1.11
1.63
2.29
1.50
1.15
1.00
1.04
2.54
5.27
1.02
1.36
1.41
1.14
1.20
1.26
1.13
1.23
1.10
1.05
1.17
2.61
1.53
1.21
1.31
2.62
1.95
1.15
1.56
2.50
2.02
14.50
54.00
1.03
3.00
3.91
1.64
1.41
1.96
∞
1.05
1.07
1.69
1.47
1.33
4.83
∞
3.64
23.50
9.75
∞
1.37
13.50
11.00
7.70
1.67
3.21
1.88
1.48
1.58
281.0
∞
3.65
4.57
15.0
33.3
8.7
5.0
6.3
22.6
14.4
8.3
8.6
7.5
10.5
23.3
11.8
34.1
6.1
23.0
8.0
0.2
8.8
6.5
13.6
4.6
8.4
10.3
15.6
19.2
4.7
12.5
3.1
8.8
2.6
9.2
6.6
6.3
28.9
25.1
3.9
3.2
23.9
9.6
9.2
13.1
2.0
4.1
8.1
8.6
24.8
6.2
40.2
1.6
8.1
8.3
6.5
3.3
16.4
12.2
10.0
6.5
7.1
4.4
23.5
0.9
Korea, South
Latvia
Lesotho
Liechtenstein
Lithuania
Luxembourg
Macedonia
Madagascar
Malawi
Mali
Malta
Marshall Islands
Mauritius
Mexico
Micronesia
Moldova
Mongolia
Mozambique
Namibia
Nauru
Netherlands
New Zealand
Nicaragua
Niger
Nigeria
Norway
Palau
Panama
Papua New Guinea
Paraguay
Peru
Philippines
Poland
Portugal
Romania
Samoa
2004
2002
2002
2005
2004
2004
2002
2002
2004
2002
2003
2003
2005
2003
2003
2005
2004
2004
2004
2004
2003
2005
2001
2004
2003
2005
2004
2004
2002
2003
2001
2004
2005
2005
2004
2006
0.576
0.801
0.537
0.595
0.837
0.737
0.653
0.565
0.809
0.689
0.497
1.000
0.497
0.668
1.000
0.567
0.551
0.461
0.405
1.000
0.789
0.664
0.497
0.731
0.538
0.781
1.000
0.670
0.941
0.685
0.794
0.713
0.765
0.609
0.703
0.505
2.36
5.04
2.16
2.47
6.13
3.81
2.88
2.30
5.24
3.21
1.99
3.03
17.02
3.18
4.85
3.48
4.26
2.56
3.36
2.02
4
6
4
3
7
5
4
3
3
4
2
No
3
3
No
3
2
2
5
No
7
5
2
6
3
7
No
6
6
4
7
4
6
5
4
2
San Marino
2001
0.716
3.52
5
1.99
3.01
2.31
2.23
1.85
1.68
4.74
2.98
1.99
3.72
2.16
4.56
91.3
50.0
83.1
88.0
45.4
63.3
77.5
78.1
57.8
78.1
100.0
Parties
96.8
75.6
Parties
89.1
94.6
100.0
83.3
Parties
57.3
81.0
98.9
63.7
89.0
58.6
Parties
73.1
28.4
72.5
58.3
69.3
62.6
85.2
73.5
87.8
66.7
23
74.1
42.8
77.3
87.0
43.2
59.5
64.9
43.1
38.3
23.9
54.9
48.7
28.6
36.1
40.5
34.3
99.3
50.8
26.0
65.3
48.0
27.7
40.0
50.0
64.4
31.6
41.3
53.8
91.4
69.9
74.5
91.2
81.9
55.9
80.2
95.3
81.9
54.8
57.0
61.0
46.0
51.1
76.0
67.9
65.7
51.8
MM
HP
EM
HP
HP
HP
HP
MM
HP
HP
EM
1.26
1.08
3.67
1.20
1.56
1.71
1.82
4.68
1.20
1.12
1.17
12.10
1.14
4.20
3.33
1.19
1.40
2.06
4.40
1.81
5.90
∞
61.3
44.6
48.8
38.1
MM
HP
1.73
1.44
11.00
1.61
55.4
48.6
64.0
76.4
46.0
46.5
MM
HP
EM
EM
1.65
1.06
1.78
11.0
3.09
34.00
∞
1.25
8.6
16.4
3.9
8.7
3.4
29.3
41.3
57.6
41.6
61.6
36.1
28.6
41.1
53.2
HP
HP
EM
HP
EM
HP
1.05
1.04
1.39
1.88
2.24
1.61
1.50
6.86
38.00
1.14
3.06
1.65
15.8
20.1
12.8
2.4
2.2
4.4
18.8
7.1
3.9
51.3
17.4
46.3
34.2
43.9
33.7
52.6
39.8
67.3
37.8
MM
HP
HP
HP
HP
HP
MM
HP
EM
?
HP
2.35
1.58
1.76
1.41
1.72
1.17
1.61
1.18
3.30
1.89
1.50
2.10
1.93
1.59
2.38
5.36
2.33
10.00
1.67
1.25
41.7
74.7
54.5
32.7
35.3
26.3
27.0
45.1
36.6
41.5
51.3
42.9
15.7
4.2
50.3
10.4
27.4
17.9
16.0
10.5
1.8
12.0
4.5
14.7
36.8
0.5
2.0
9.7
14.9
7.5
16.1
37.1
24.0
11.5
13.2
34.1
13.1
19.4
13.1
9.2
10.1
5.8
1.1
São Tomé and Príncipe
Senegal
Seychelles
Sierra Leone
Slovakia
Slovenia
Solomon Islands
South Africa
Spain
Sri Lanka
St. Kitts and Nevis
St. Lucia
St. Vincent and the
Grenadines
Suriname
Sweden
Switzerland
Taiwan
Timor-Leste
Trinidad and Tobago
Turkey
Tuvalu
Ukraine
United Kingdom
United States
Uruguay
Vanuatu
2002
2001
2002
2002
2002
2004
2006
2004
2004
2004
2004
2001
2005
0.614
0.433
0.438
0.392
0.837
0.787
0.975
0.491
0.600
0.638
0.545
0.291
0.320
2.59
1.76
1.78
1.65
6.12
4.69
40.32
1.97
2.50
2.76
2.20
1.41
1.47
3
3
2
2
7
7
7
3
2
4
4
2
2
85.5
83.3
100.0
98.2
42.7
59.1
16.0
82.3
89.1
83.1
81.8
100.0
100.0
79.0
65.7
96.9
89.7
34.6
51.9
13.2
82.1
80.3
83.4
59.4
90.8
99.9
43.6
74.2
67.6
74.1
24.0
33.0
8.0
69.8
46.9
46.7
63.6
82.4
80.0
39.6
49.6
54.3
69.9
19.5
29.1
6.9
69.7
42.7
45.6
50.6
54.2
55.3
HP
MM
EM
EM
HP
HP
HP
EM
HP
HP
EM
EM
EM
1.04
8.09
2.09
3.07
1.29
1.26
1.00
5.58
1.11
1.28
3.50
4.67
4.00
2.86
1.10
∞
13.50
1.12
2.30
1.33
1.79
14.80
3.73
2.00
∞
∞
23.4
43.3
16.5
48.1
37.5
21.6
2005
2002
2003
2004
2001
2002
2002
2002
2006
2005
2004
2004
2004
0.687
0.763
0.800
0.694
0.591
0.494
0.460
1.000
0.707
0.686
0.500
0.585
0.893
3.19
4.23
4.99
3.27
2.44
1.98
1.85
5
74.5
7
57.0
5
53.5
4
74.7
4
70.5
2
100.0
2
98.4
No Parties
5
70.0
3
85.7
2
99.8
3
87.9
8
36.5
64.3
55.0
49.9
72.9
66.1
97.3
53.7
45.1
41.3
27.5
39.6
62.5
55.6
66.0
41.5
39.8
26.6
38.0
57.4
50.7
34.3
HP
HP
HP
HP
EM
EM
MM
1.53
2.62
1.06
1.13
7.86
1.25
2.04
2.50
1.15
1.44
2.32
1.17
∞
∞
35.7
13.9
8.0
8.5
4.2
41.4
11.1
3.0
4.8
4.2
6.4
4.8
46.3
54.4
67.5
95.8
84.7
41.3
55.2
53.3
53.5
19.2
32.1
35.2
49.3
50.4
HP
MM
MM
EM
HP
1.44
1.81
1.15
1.56
1.11
1.59
3.18
∞
3.40
1.13
35.3
6.4
1.5
25.2
22.2
20.8
4.0
3.4
3.41
3.18
2.00
2.41
9.36
8.3
10.1
1.9
5.9
4.0
6.5
25.0
13.3
8.7
18.2
9.7
4.9
0.9
9.3
5.1
24.0
28.2
24.7
* calculated by electoral alliances
Note: Serbia and Montenegro excluded due to imminent dissolution; Thailand and Venezuela excluded due to
opposition boycott of last election.
24