08 - Mainwaring - Presidentialism, Multipartism, and Democracy - The Difficult Combination
08 - Mainwaring - Presidentialism, Multipartism, and Democracy - The Difficult Combination
08 - Mainwaring - Presidentialism, Multipartism, and Democracy - The Difficult Combination
http://cps.sagepub.com/
Published by:
http://www.sagepublications.com
Additional services and information for Comparative Political Studies can be found at:
Subscriptions: http://cps.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
Citations: http://cps.sagepub.com/content/26/2/198.refs.html
What is This?
PRESIDENTIALISM,
MULTIPARTISM, AND DEMOCRACY
The Difficult Combination
SCOTIMAINWARING
University of Notre Dame
It is important to define how some key terms are used in this article. A
democracy must meet three criteria. First, democracies must have open,
competitive elections that determine who governs. Election results cannot be
determined by fraud, coercion, or major proscriptions, and legislative and
executive office must be decided on the basis of elections. Elections must, in
principle, afford the opportunity of alternation in power, even if, as in Japan,
actual alternation does not occur. Second, in the contemporary period, there
must be nearly universal adult suffrage. Until recent decades, this criterion
was debatable because some nations that were usually considered democra-
cies (e.g., Switzerland) excluded women, but this is no longer the case. Third,
there must be guarantees of traditional civil liberties such as freedom of
speech, freedom of the press, freedom of organization, due process of law,
and so on. 2
Even with a reasonably clear and operationalizable definition of democ-
racy, deciding what governments are democratic involves difficult judg-
ments. In constructing a list of democracies in the post-1945 period, I
tolerated some deviations from the three criteria that distinguish democratic
government. In Stephens' (1989) terms, I counted restricted democracies,
which are characterized by some limitations on participation, competition,
or civil liberties, as well as full democracies. Thus I included Switzerland
before 1971, Chile before 1958, Peru before 1980, Brazil between 1946 and
1964, and several other countries that had restrictions on participation.
Restrictions on participation in South Africa and Western Samoa, however,
are so extensive that they cannot be considered democracies, even though
competitive elections are regularly held and civil liberties are safeguarded
for the voting population. I also allowed for some restrictions on competition,
thereby including Argentina (1958-1962, 1963-1966), Colombia (1958-
1974), India (1975-1977), and several other cases. Finally, I tolerated some
restrictions on civil liberties.3
A presidential democracy has two distinguishing features. 4 First, the head
of government is essentially popularly elected; this includes the U.S., where
the electoral college has little autonomy with respect to the popular vote.
Legislative elections and postelection negotiations do not determine execu-
tive power. Wherever the head of government is selected by the legislature,
not as a second alternative where the popular vote does not produce a clear
winner, but as the fundamental process, the system is usually parliamentary5
and never presidential. Postelection negotiations that determine which par-
ties will govern and which will head the government are crucial in many
parliamentary regimes, so they indirectly determine who will be prime
minister. Such postelection negotiations are not part of the selection process
of chief executives in presidential systems.
In presidential systems, the president must be the head of government. In
sernipresidential systems (e.g., Finland, France), a popularly-elected presi-
dent is head of state but is not always the head of government. In Austria,
Iceland, and Ireland, a president is elected by direct popular vote but has only
3. Beginning in 1973, my list relies on The Freedom House annual publication Freedom
Around the World. I considered democratic those countries that Freedom House ranked as free.
For the period before 1973, sources consulted for classifying countries according to whether
they are democracies included Dahl (1971 ), Lijphart (1984), Powell (1982), Rustow (1967), and
Stephens (1989), as well as the annual publications The Europa World Year Book, The Political
s
Handbook of the World, and The Statesman Year-Book. For many purposes, the dichotomous
classification used here (democratic x nondemocratic) would be inadequate, and continuous
categories would be preferable. For present purposes however, it was neither necessary nor
possible to construct a continuous measure of democracy because of the difficulty of getting
adequate data on all countries for such a lengthy time period.
4. For related discussions of how presidentialism should be defined, see Linz (in press),
Lijphart (1984, pp. 68-74), Riggs (1988), and Shugart and Carey (1992, p. 18-27).
5. Switzerland is an exception. The executive (which is collegial) is selected by the
legislature, but the system is not parliamentary because the executive has a fixed term of office.
In Bolivia, as was also the case in Chile before 1973, when no presidential candidate wins an
absolute majority of the popular vote, congress elects the president. But there is a key difference
between these two cases. In Chile, the congress always selected the front-runnerin popular votes;
it did not broker the election but rather confirmed the popular winner, so it can be considered
presidential. If, however, congress plays the dominant role in selecting the president, as is the
case in Bolivia, then the system is not presidential. In Bolivia, the congress gave the presidency
to candidates who did not capture the most votes in 1979, 1985, and 1989. Legislative
negotiations became the primary mechanism for selecting the president. Consequently, the
system is not strictly presidential, but rather alternating; it is presidential when one candidate
obtains an absolute majority in the popular vote, but it is a hybrid when, as has been occurring
consistently, this is not the case. Because the president's term of office is fixed, the system is
not parliamentary.
Table l
Classifying Systems of Government
minor powers and, therefore, is not the head of government. In all three
countries, the system of government is parliamentary, notwithstanding the
existence of popular elections for president.6
The second distinguishing feature of presidential democracies is that the
president is elected for a fixed time period. Most presidential democracies
allow for impeachment, but this practice is rare and does not substantially
affect our definition because of its extraordinary character. The president
cannot be forced to resign because of a no-confidence vote by the legislature.
In contrast, in a parliamentary system, the head of government is selected by
the legislature and subsequently depends upon the ongoing confidence of the
legislature for remaining in office; thus the time period of the chief executive's
mandate is alterable.
In synthesis, following Lijphart ( 1984, pp. 68-74 ), I define presidentialism
according to two dimensions: whether the chief executive is elected by the
legislature and whether the term of office is fixed. Table 1 portrays these two
dimensions.
Some other arrangements normally are associated with either presidential
or parliamentary government but should not be included in a definition of
presidentialism. Some analysts associate the right of legislatures to interpel-
late cabinet members with parliamentary government, but this practice is also
found in several presidential systems in Latin America. In a few presidential
systems (e.g., Peru), the legislature can call for the dismissal of particular
ministers, another provision more frequently associated with parliamentary
systems. Many presidential systems allow congressional representatives to
become cabinet members; this feature, too, is sometimes considered an
element of parliamentary government.
6. Duverger (1980) argued that Austria, Iceland, and Ireland have semipresidential govern-
ments, but the presidents have only symbolic power in all three cases. What matters is whether
these offices are largely symbolic or, conversely, whether the office holders wield considerable
power.
Table 2
Stable Democracies, 1967-1992
Table3
Other Stable Democracies Since 1945"-
a. Countries that had continuous democracy for at least 25 years after 1945 but that experienced
democratic breakdowns.
b. Sri Lanka switched from a parliamentary to a hybrid system in 1977.
c. In the late 1960s, Uruguay moved to a three-party system.
Table4
Unstable Democracies Since 1945"-
Table 4 continued
Period of democracy System of government Party system
and Suarez (1981), Lijphart (in press), Linz (in press), Riggs (1988), and
Suarez (1982), have argued that presidentialism is less likely to promote
stable democracy. I do not share all of their criticisms,8 and most of the critics
have overlooked some strengths of presidential systems (Shugart & Carey,
1992; Ceaser, 1986). Nevertheless, I agree that presidentialism is generally
less favorable for democracy and that presidentialism has some distinct
liabilities, three of which are highlighted in the following discussion.
Perhaps the greatest comparative liability of presidential systems is their
difficulty in handling major crises. Presidential systems offer less flexibility
in crisis situations because attempts to depose the president easily shake the
8. Linz (in press) and Lijphart (in press) add a fourth liability: the supposedly majoritarian
bent of presidentialism. I disagree with this part of their argument. The most majoritarian
democracies are the Westminster style parliamentary systems with highly disciplined parties, in
which the winning party controls everything for a protracted period of time, possibly despite
winning well under 50% of the votes. Presidentialism is predicated upon a separation of powers,
so that an opposition party or coalition can control the legislature (or one house thereof), thereby
exercising some control over presidential initiatives even if it does not control the presidency. I
agree with Linz and Lijphart, however, that parliamentarism is more conducive to coalition
building. For a critical examination of Linz's seminal piece, see Mainwaring and Shugart (in
press).
PRESIDENTIALISM,
MULTIPARTISM, AND STABLE DEMOCRACY
9. The Rae index (Fs) measures the probability that two randomly selected legislators belong
to different parties. It ranges from zero (all legislators are members of the same party) to one
(each legislator is the only representative of his/her party). It is derived by squaring each party's
share of seats and subtracting the sum of all these squares from 1.00.
N
Fs= 1-,... 2
L.Pi
where Fs is the index of fragmentation expressed in seats, N is the number of parties, and Pi is
the proportion of seats held by the ith party (Rae, 1967).
The Laakso/Taagepera (1979) effective number of parties (N,) measures how many parties
are in the party system, weighted according to siu. It is derived by squaring each party's share
of seats, adding all of these squares, and dividing 1.00 by this number.
1
Ns=-
I:p~
where Ns is the number of effective parties expressed in seats and P; is the fractional share of
seats of the ith party. If every party has an equal number of seats, the number of effective parties
is equal to the number of parties that have seats. If two roughly equal parties control a large
majority and a third has fewer seats, Ns is equal to some number between 2.0 and 3.0, reflecting
the domination of the two largest parties but the presence of a third, but smaller competitor. The
number of effective parties can be derived from the Rae Index through a simple algebraic
transformation. Both the Rae index and the Laakso/Taagepera index can be measured in seats
or votes, but in view of my predominant interest in executive/legislative relationships, I have
used seats.
Tables
Party System Fragmentation and Number of Effective Parties in Stable Democracies (Seats in
Lower Chamber)
Sourr:e: Various numbers of Electoral Studies, Keesing's Contemporary Archives, and The
Statesman's Year-Book.
Note: Indian data do not include 14 seats in Assam for which elections were postponed and 4
other contests that were postponed because of the deaths of candidates. In some cases, two
countries have the same Fs index but different values for Ns, or vice versa, because I calculated
Pl to four digits and then rounded.
The means for the 24 parliamentary democracies are an Fs of .60 and 3.1
effective parties; for the three remaining democracies the mean Fs is .73 and
the mean Ns is 4.6.
Table 6 gives the Fs andNs values for all presidential systems that enjoyed
at least 25 years of uninterrupted democracy, including three countries
(Chile, the Philippines, and Uruguay) that are not currently stable democra-
cies.10 Six of the seven stable presidential democracies have less than three
effective parties. Only Chile had a true multiparty system among the stable
presidential democracies, although Venezuela did so before 1973. The table
also suggests that a number of two-party presidential systems have worked
well enough to allow democracy to endure for at least 25 years.
This correlation between two-party systems and stable presidential de-
mocracy would be irrelevant if two-party systems were the norm in presi-
dential democracies, but this is not the case. Of the 31 presidential democra-
cies listed in Tables 2, 3, and 4, 15 had multiparty systems, compared to 10
two-party systems, 5 dominant-party systems, and 1 two-and-one-half party
system. Only 1 of the 15 multiparty presidential democracies endured for at
least 25 years, compared to 5 of 10 two-party presidential democracies and
to 11 of 21 multiparty parliamentary systems. Although they are not conclu-
sive, the data suggest that the problem may not be presidentialism or
multipartism so much as the combination.
Among stable presidential democracies, the virtual absence ofmultiparty
systems is striking. This observation, however, does not explain why multi-
party systems are less propitious to stable presidential democracy than
two-party systems. Without some logical explanation, it remains possible that
this is an accident or a spurious correlation. But there are reasons to believe
that the combination of presidentialism and multiparty systems makes it more
difficult to achieve stable democracy.
Two-party systems, in and of themselves, are not necessarily a desidera-
tum. They constrict the breadth of opinion represented, and they hinder the
building of coalition governments, making it difficult to establish consocia-
tional forms of democracy (Lijphart, in press). As Sartori (1976, pp. 191-192)
observed, two-party systems become less functional and less viable as the
spread of opinion becomes greater. Nevertheless, in presidential systems a
two-party format seems more favorable to stable democracy. The question is
why multipartism and presidentialism make a difficult combination, why a
two-party system ameliorates the problems of presidentialism, and why
parliamentarism mitigates the difficulties of multipartism.
The answer to these questions, I submit, is threefold. In presidential
systems, multipartism increases the likelihood of executiveflegislative dead-
10. The inclusion of Uruguay as a presidential system is debatable fortbe 1951-1967 period,
but the collegial executive met the two distinguishing criteria of presidentialism discussed above.
Table6
Party System Fragmentation (Fs), Number of Effective Parties (Ns), and Mean Share of
Legislative Seats Controlled by President's Party in Stable Presidential Democracies (Seats in
Lower Chamber)
bilities of, the presidency has grown in recent years as a result of severe
economic crises. The economic crises have led to closed, technocratic
decision making within the executive branch, and congresses have been
excluded from economic policy-making. But presidential capacity to handle
these crises has eroded. In the past, the weaknesses of democratically elected
presidents helped justify coups that led to the extreme hypertrophy of the
executive and the emasculation or abolition of congress.
Two-party presidential systems also face institutional rigidity when
executive/legislative deadlock occurs but, as noted earlier, presidents are
more likely to have stable support in congress. In contrast to presidential
systems, parliamentary systems have an institutionalized mechanism for
overcoming deadlocks when they arise. A vote of no-confidence can topple
the government, leading to new elections that may change the balance of
power and help resolve the crisis. This provision allows for replacing, with
less institutional strain, unpopular or inept executives. Frequent recourse to
dismissing governments can breed instability, but this problem can be miti-
gated by measures such as the West German or Spanish constructive vote of
no-confidence. Conversely, if a prime minister is frustrated because of the
difficulty of effecting policy in the face of opposition control of the legisla-
ture, in most parliamentary systems, he/she can call new elections in an effort
to achieve a majority. In either case, there are means of changing the
government without threatening the regime.
PRESIDENTIALISM, MULTIPARTISM,
AND IDEOLOGICAL POLARIZATION
PRESIDENTIALISM,
MULTIPARTISM, AND PARTY COALITIONS
Thus far the argument has focused on why multipartism exacerbates the
problems of presidentialism. Now I look at why presidentialism makes it
difficult for multipartism to function well, focusing on problems of coalition
building in presidential systems. In multiparty systems, interparty coalition
building is essential for attaining a legislative majority. While the need for
such coalition building exists in both presidential and parliamentary
multiparty systems, three factors make building stable interparty legisla-
tive coalitions more difficult in presidential democracies than in parliamen-
tary systems.
First, party support for the government tends to be more secure in
parliamentary systems because of the way executive power is formed and
dissolved. In a coalition parliamentary government, the parties forming the
government choose the cabinet and the prime minister. Executive power is
formed through post-election agreements among parties and is divided
among several parties. The parties themselves are corresponsible for govern-
ing and are committed to supporting government policy. When they cease
supporting the government, there is a chance that new elections will be called.
The coalition that brings the parties together is binding for the postelection
period. These arrangements help ensure that there will either be legislative
support for the executive or a means of toppling the government.
In presidential systems the president (not the parties) has the responsibility
of putting together a cabinet. The president may make prior deals with the
parties that support him or her, but these deals are not as binding as they are
in a parliamentary system. Presidents are freer to dismiss ministers and to
rearrange the cabinet than prime ministers in a coalition government are.
Changes in cabinets are usually the president's decision and are not brought
about by party decisions. Presidential autonomy is naming a caabinet is part
of a generally looser institutional arrangement that can easily lead to a lack
of stable congressional support, for just as presidents are less bound to the
parties, so are the parties less bound to the presidents. Coalition partners
are more likely to remain faithful allies if they themselves have negotiated
the terms of the coalition than if they have agreed to terms established by the
president.
Whereas in parliamentary systems, party coalitions generally take place
after the election and are binding, in presidential systems, they often take
place before the election and are not binding passt election day. Executive
power is not formed through postelection agreements among parties and is
not divided among several parties that are corresponsible for governing, even
though members of several parties often participate in cabinets. Governing
coalitions in presidential systems can differ markedly from electoral coali-
tions, whereas in parliamentary systems the same coalition responsible for
creating the government is also responsible for governing. In contrast to the
situation in coalition governments in parliamentary systems, a simple plural-
ity can confer virtually absolute control of executive power (Lijphart, 1990,
in press).
Given the separation of powers, an agreement among parties may pertain
only to congressional matters, with no binding implication for relations
between the parties and the president. Several parties may support the
president during the electoral campaign, but this does not ensure their support
once he or she assumes office. Even though members of several parties often
participate in cabinets, the parties are not responsible for the government.
Parties or individual legislators can join the opposition without bringing
down the government. A president can end his or her term with little support
in congress.
Second, in presidential systems, the commitment of individual legislators
to support an agreement negotiated by the party leadership is often less
secure. The extension of a cabinet portfolio does not necessarily imply party
support for the president, as it does in a parliamentary system. The commit-
ment of individual legislators to vote the party line varies a great deal, ranging
from the extremely cohesive congressional parties in Venezuela to the
CONCLUSIONS
tide of this discussion changed in the 1960s and 1970s, with the contributions
of Lijphart (1968, 1977) being particularly significant. Noting that many
multiparty democracies had achieved stability for a long period of time in
the smaller European democracies, Lijphart argued that in plural societies,
with sharp cultural, ethnic, religious, or linguistic cleavages, multiparty
systems could be more propitious in promoting stable democracy. With a
two-party system, significant minorities might be permanent "outs," a situa-
tion that could reduce their willingness to abide by the rules of the game. A
multiparty system could enable these minorities to attain meaningful repre-
sentation and to participate in governing coalitions. This issue is not resolved,
but most recent analysts (e.g., Sani & Sartori, 1983; Sanders & Herman)
1977) have sided with Lijphart.
This debate has overlooked the difference between parliamentary and
presidential systems. Multipartism may not adversely affect prospects for
democracy in parliamentary systems, but it appears to with presidentialism.
Institutional combinations make a difference. Some problems typical of
presidential systems-especially conflict between the executive and legisla-
ture resulting in immobilism-are exacerbated by multipartism. Conversely,
with parliamentary governments but not with presidential governments,
multiparty systems seem as capable of sustaining democratic regimes as
two-party systems. Parliamentary regimes have more coalition-building
mechanisms that facilitate multiparty democracy.
This is not to say that the combination of presidentialism and multipartism
makes it impossible for a democracy to function well. The Chilean case from
the 1930s until the late 1960s shows that presidentialism, multipartism, and
stable democracy can go together, even in an ideologically polarized polity.
However, with this institutional combination, democratic stability hinges
largely on the desire of elites and citizens to compromise and create enduring
democratic institutions. Optimally, political systems should have institu-
tional mechanisms that reinforce such dispositions.
I also do not intend to suggest that institutional issues are always the main
factor in determining whether democracy succeeds or fails. Some societies
face conflicts that are irresolvable in the short run, regardless of institutional
structures. Social, cultural, and economic conditions also affect prospects for
democracy. However, some institutional combinations facilitate and others
obstruct the management of social, economic, and political problems.
The two-party system and exceptionally limited ideological polarization
have contributed to making the U.S. a viable presidential democracy (Riggs,
1988), but these conditions have been difficult to reproduce. Among presi-
dential democracies outside the United States, only Colombia and Costa Rica
have consistently approximated two-party systems. Venezuela has also come
REFERENCES
Archer, R. P., & Chernick, M. (1989). El Presidente frente a las instituciones nacionales. In
P. Vasquez, (Ed.), La democracia en blanco y negro: Colombia en los aiios 80 (pp.31-79).
Bogotii: Fondo Editorial CEREC.
19. I cannot delve into this complex question in detail here. See the interesting debate
between Linz (in press), who is skeptical of the hybrid formulas, and Sartori (in press), who
advocates them on the grounds that parliamentary government would not work in most of Latin
America See also Carey and Shugart (1992, pp. 55-75) and Duverger (1980).
Arriagada, G. (1984). El sistema politico chileno (Una exploraci6n del futuro). Colecci6n
Estudios CIEPIAN, Estudio No. 99, pp. 171-202.
Blonde!, J., & Suarez, W. (1981) Las limitaciones institucionales del sistema presidencialista.
Criterio, No. 1853-1854, 57-70.
Ceaser, J. C. (1986). In defense of separation of powers. In R. Goldwin & A. Kaufman (Eds.),
Separation of powers-Does it still work? (pp. 168-193). Washington, DC: American
Enterprise Institute.
Coppedge, M. (1988). Strong parties and lame ducks: A study of the quality and stability of
Venezuelan democracy. Unpublished doctoral dissertation, Yale University.
Dahl, R. (1971). Polyarchy. New Haven: Yale University Press.
Downs, A. (1957). An economic theory of democracy. New York: Harper.
Duverger, M. (1954). Political parties: Their organization and activity in the modem state.
London: Methuen.
Duverger, M. (1980). A new political system model: Semipresidential government. European
Journal of Political Research, 8, 165-187.
Edelmann, A. (1969). The rise and demise of Uruguay's second plural executive. Journal of
Politics, 31, 119-139.
Epstein, L. (1964). A comparative study of Canadian parties.American Political Science Review,
58,46-59.
Epstein, L. (1967). Political parties in western democracies. New York: Praeger.
Hartlyn, J. (in press). Presidentialism and Colombian politics. In J. J. Linz & A. Valenzuela
(Eds.), Presidential or parliamentary democracy: Does it make a difference? Baltimore:
Johns Hopkins University Press.
Hermens, F. (1941). Democracy or anarchy? A study of proportional representation. Notre
Dame: Review of Politics.
Laakso, M., & Taagepera, R. (1979). The "effective" number of parties: A measure with
application to Western Europe. Comparative Political Studies, 12, 3-28.
Lambert, J. (1969). Latin America: Social structure and political institutions. Berkeley and Los
Angeles: University of California Press.
Levine, D. (1973). Conflict andpolitical change in Venezuela. Princeton: Princeton University Press.
Lijphart, A. (1968). Typologies of democratic systems. Comparative Political Studies, I, 3-44.
Lijphart, A. (1977). Democracy in plural societies: A comparative exploration. New Haven, CT:
Yale University Press.
Lijphart, A. (1984). Democracies. New Haven, CT: Yale University Press.
Lijphart, A. (in press). Presidentialism and majoritarian democracy: Theoretical observations.
In J. J. Linz & A. Valenzuela (Eds.), Presidential or parliamentary democracy: Does it make
a difference? Baltimore: Johns Hopkins University Press.
Linz, J. J. (1990). The perils ofpresidentialism. The Journal of Democracy, I, 51-69.
Linz, J. J. (in press). Democracy: Presidential or parliamentary. Does it make a difference? In
J. J. Linz & A. Valenzuela (Eds.), Presidential or parliamentary democracy: Does it make
a difference? Baltimore: Johns Hopkins University Press.
Loewenstein, K. (1949). The presidency outside the United States: A study in comparative
political institutions. Journal of Politics, II, 447-496.
Mainwaring, S. (1990). Presidentialism in Latin America. Latin American Research Review, 25,
157-179.
Mainwaring, S., & Shugart, M. (in press). Juan Linz, presidentialism, and democracy: A critical
appraisal. In A. Valenzuela (Ed.), Politics, society, and democracy: Latin America. New
York: Praeger.
Mayhew, D. (1991). Divided we govern. New Haven, CT: Yale University Press.
Church and Politics in Brazil, 1916-1985 (Stanford, 1986), The Progressive Church in
Latin America (coedited, Notre Dame, ]989), Issues in Democratic Consolidation
(coedited, Notre Dame, 1992), and Building Democratic Institutions: Party Systems in
Latin America (Stanford University Press, in press). He has published articles in journals
in the United States, Europe, and Latin America.