Robert Elgie
Semi-presidentialism, Cohabitation and
the Collapse of Electoral Democracies,
1990–2008goop_1303 29..49
SEMI-PRESIDENTIALISM IS THE SITUATION WHERE A CONSTITUTION
makes provision for both a directly elected fixed-term president and a
prime minister and cabinet who are responsible to the legislature.
Since 1990, semi-presidentialism has become the preferred constitutional
choice for new democracies. However, in academic terms, the
proliferation of semi-presidentialism seems worrying. The prevalent
view is that semi-presidentialism is a poor constitutional choice for new
democracies. There are various reasons why semi-presidentialism is
generally rejected by the academic community, but the most common
objection, and one that is specific to semi-presidentialism, is the
inherent potential for cohabitation. This is the situation where a
president from one party holds power at the same time as a prime
minister from an opposing party and where the president’s party is not
represented in the cabinet. The worry is that the president and prime
minister will be unwilling to share power. As a result, either the military
will intervene to assure effective decision-making at the expense of
both the president and the prime minister, or the president or the
prime minister will try to seize power unilaterally at the expense of the
other actor. Both scenarios are destructive of democracy.
In this article, we show that since 1990 only one semi-presidential
democracy has collapsed while experiencing cohabitation – Niger in
1996. However, we explore whether the mere threat of cohabitation
has been a factor in other collapses. If cohabitation is so problematic,
then actors may intervene to prevent it from happening, but in so
doing they may also precipitate the collapse of democracy. If that is
the case, we would not observe cohabitation, but it may still be a
factor in the collapse. In this article, we specify the conditions under
which cohabitation can occur. Having done so, we still find little
evidence to suggest that the threat of cohabitation has been associated
with collapse. Overall, we show that cohabitation refers to a very
Government and Opposition, Vol. 45, No. 1, pp. 29–49, 2010
doi:10.1111/j.1477-7053.2009.01303.x
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
Published by Blackwell Publishing, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main
Street, Malden, MA 02148, USA.
specific situation, that it occurs only under certain circumstances
and that these circumstances are unlikely to combine very frequently.
So, while cohabitation can be perilous for young semi-presidential
democracies, we suggest that cohabitation is less problematic than
the established wisdom would suggest. This does not mean that semipresidentialism
is a good constitutional choice for young democracies.
It simply means that if semi-presidentialism is a problematic
choice, then it is problematic for reasons other than the prospect of
cohabitation.
SEMI-PRESIDENTIALISM AND THE PROBLEM OF COHABITATION
The concept of semi-presidentialism was first introduced by Maurice
Duverger in the 1970s.1
His work focused solely on a small number of
West European countries as these were the only countries at the time
with semi-presidential constitutions. However, the wave of democratization
that began in the early 1990s greatly expanded the number of
semi-presidential countries and their geographical spread. Indeed,
one recent article identified over 55 countries with semi-presidential
constitutions.2
The spread of semi-presidentialism seems worrying. Following the
debate about the most appropriate institutional choices for new
democracies in the early 1990s, semi-presidentialism has generally
been rejected. There are various objections to semi-presidentialism,
but one that is common to all of this work is the problem of ‘cohabitation’.
First identified in France in the period 1986–88, cohabitation
is more than the situation where representatives from different coalition
parties hold the two main positions within the executive.
Instead, cohabitation is the situation where a president from one
party holds power at the same time as a prime minister from an
opposing party and where the president’s party is not represented in
the cabinet.
Cohabitation is said to be particularly dangerous for new democracies
because both the president and prime minister can legitimately
1
M. Duverger, ‘A New Political System Model: Semi-Presidential Government’,
European Journal of Political Research, 8 (1990), pp. 165–87.
2
R. Elgie, ‘What is Semi-presidentialism and Where is it Found?’, in R. Elgie and
S. Moestrup (eds), Semi-presidentialism Outside Europe, London, Routledge, 2007,
pp. 1–13.
30 GOVERNMENT AND OPPOSITION
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
claim to have the authority to speak on behalf of the people – the
president by virtue of direct election and the prime minister by virtue
of parliamentary support. Given that the prime minister is responsible
to the legislature under semi-presidentialism, the president has
either to accept the will of the legislature and co-exist with a political
opponent or, if the constitution allows, to defy the legislature and
dismiss the head of government in the knowledge that the legislature
may simply appoint as prime minister someone who is equally
opposed to the president. The prospect of ongoing intra-executive
conflict or a prime ministerial merry-go-round until the next presidential
or legislative election – and perhaps beyond – is said to be
particularly problematic in young democracies. It may lead to a gridlock
situation in which neither the president nor the prime minister
is willing to compromise and where decision-making comes to a halt.
In this case, the military may decide to intervene in order to restore
executive authority. Alternatively, it may lead one or other of the
executive actors, usually the president, to seize power themselves so
as to resolve the impasse.
In their work, Linz and Stepan are explicit about the dangers of
cohabitation for young democracies:
When supporters of one or the other component of semi-presidentialism feel
that the country would be better off if one branch of the democratically
legitimated structure of rule would disappear or be closed, the democratic
system is endangered and suffers an overall loss of legitimacy, since those
questioning one or the other will tend to consider the political system
undesirable as long as the side they favor does not prevail . . . [I]n a semipresidential
system, policy conflicts often express themselves as a conflict
between two branches of democracy.3
For Stepan and Suleiman, the ‘main theoretical and political worry
about semi-presidentialism, of course, is precisely the question of
deadlock and constitutional conflict between the dual executive. A
deadlock can become particularly dangerous if the president has
special authority over the security forces and some emergency
powers.’4
Fabbrini sums up the potential problem:
3
J. J. Linz and A. Stepan, Problems of Democratic Transition and Consolidation: Southern
Europe, South America, and Post-Communist Europe, Baltimore, Johns Hopkins University
Press, 1996, p. 286.
4
A. Stepan and E. N. Suleiman, ‘The French Fifth Republic: A Model for Import?
Reflections on Poland and Brazil’, in H. E. Cheibub and A. Stepan (eds), Politics,
Society, and Democracies. A Comparative Study, Boulder, CO, Westview Press, 1995, p. 399.
31SEMI-PRESIDENTIALISM, COHABITATION AND COLLAPSE
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
When the president is the leader of the party that controls the National
Assembly, the executive gaze rests on him. When a different party controls
the Assembly, the executive gaze focuses on the premier, with some conditions
imposed by the president. Herein lies the main weakness of semipresidentialism:
the possibility of a rift between the president with his popular
majority and the premier with his legislative majority. Such a split could
hamper or even paralyze the executive.5
The academic consensus against semi-presidentialism as a function
of the potential for cohabitation runs very deep. Indeed, the
supposed perils of cohabitation continue to influence the thinking of
those who are called upon to advise new democracies about constitutional
choices. For example, in a document by Barnett R. Rubin for
the Constitutional Drafting Commission of Afghanistan, the problem
of cohabitation was cited as a reason why the adoption of semipresidentialism
might be problematic: the semi-presidential system
‘risks creating two competing centers of power, which is probably not
healthy for a polarized society emerging from conflict’.6
The next
section examines the basic association between cohabitation and
democratic collapse.
COHABITATION AND THE COLLAPSE OF DEMOCRACY
The identification of countries with semi-presidential constitutions is
very straightforward. Whereas Duverger’s original definition of semipresidentialism
required a judgement call to be made as to whether
a president possessed ‘quite considerable powers’,7
we adopt a
slightly different definition of semi-presidentialism. Here, we make
no reference to presidential powers. Instead, a country is deemed
semi-presidential if the constitution makes provision for both a
directly elected fixed-term president and a prime minister and
cabinet who are responsible to the legislature. This definition has
5
S. Fabbrini, ‘Presidents, Parliaments, and Good Government’, Journal of Democracy,
6 (1995), p. 133.
6
B. R. Rubin, ‘Forms of Government and Electoral Systems. Summary of Briefing
Papers for the Constitutional Drafting Commission of Afghanistan’, Center on
International Cooperation: New York University, 2003 available at: http://
www.cic.nyu.edu/afghanistan/constitution.html#goverment (accessed 3 February
2009).
7
Duverger, ‘A New Political System Model’, p. 166.
32 GOVERNMENT AND OPPOSITION
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
been used by Elgie, Shugart and Skach.8
The advantage of this definition
is that there is little or no dispute as to which countries are
semi-presidential.
The identification of semi-presidential countries where democracy
has collapsed is more contested. There are various data sets that try to
capture cases of democratic collapse. The Przeworski et al. data set9
is widely used, but it is not helpful for the purposes of this article
because it covers only the period 1950–90. We are interested in the
period since 1990, when the spread of semi-presidentialism has been
most noticeable. The Polity data set is also widely used and now covers
the period up to and including 2007. However, Polity’s methodology
has been heavily criticized recently.10
Specifically, the Polity scale has
been accused of giving a false sense of precision. For example,
imagine we consider the collapse of democracy to be associated with
a move in a Polity score from Ն+1 to Յ0. The problem is that the
scores themselves are the aggregation of various indicators. Thus
countries may arrive at the same Polity score as the result of combination
of various individual codings. By the same token, a small shift
from above to below a particular threshold of democracy/nondemocracy
may be the result of factors unrelated to anything that
corresponds to any purported collapse. The same is true of Freedom
House scores. Indeed, Freedom House’s methodology has been the
subject of particular criticism over the years.11
In this way, neither
Polity nor Freedom House scores provide an entirely satisfactory basis
for identifying cases of democratic collapse. Instead, in this article
we use Freedom House’s notion of an electoral democracy as a
proxy for democratic collapse. Here, there are no problems with the
8
Elgie, ‘What is Semi-presidentialism and Where is it Found?’; M. S. Shugart,
‘Semi-presidential Systems: Dual Executive and Mixed Authority Patterns’, French Politics,
3 (2005), pp. 323–51; C. Skach, Borrowing Constitutional Designs: Constitutional Law
in Weimar Germany and the French Fifth Republic, Princeton, Princeton University Press,
2005.
9
A. Przeworski, M. E. Alvarez, J. A. Cheibub and F. Limongi, Democracy and
Development. Political Institutions and Well-being in the World. 1950–1990, Cambridge,
Cambridge University Press, 2000.
10
For example, S. Treier and S. Jackman, ‘Democracy as a Latent Variable’,
American Journal of Political Science, 52 (2008), pp. 201–17.
11
For example, Munck and Verkuilen state that Freedom House scores exemplify
the problems of ‘conceptualization, measurement, and aggregation’. G. Munck and
J. Verkuilen, ‘Conceptualizing and Measuring Democracy. Evaluating Alternative
Indices’, Comparative Political Studies, 35 (2002), p. 28.
33SEMI-PRESIDENTIALISM, COHABITATION AND COLLAPSE
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
aggregation of individual codings. This means that the loss of the
status of an electoral democracy will correspond to an explicitly
political event rather than any host of other factors that might have
changed during a given year. Moreover, there is no need to decide an
arbitrary score as a threshold for democracy/non-democracy. The
loss of the status of an electoral democracy is itself the threshold. So,
for the purposes of this article, we identify a young democracy as a
country with a semi-presidential constitution that has been classed as
an electoral democracy at any time from 1990 onwards. If a country
loses this status, we assume that a collapse has occurred.12
To identify periods of cohabitation, we take the list of semipresidential
countries that have achieved the status of electoral
democracy since 1990 inclusive, and identify all those cases where
www.worldstatesmen.org identifies a president from one party and a
prime minister from another. We then check these cases with secondary
literature to establish whether the president’s party was represented
in cabinet. If it was, then we do not include these periods as
examples of cohabitation. If it was not, then we do. If the president is
classed as non-partisan but the prime minister has a partisan affiliation,
we do not count this situation as a case of cohabitation.13
Table 1
lists the periods of cohabitation in semi-presidential countries
that have attained the status of an electoral democracy since 1990
inclusive.
Table 1 shows that there have been 12 cases where a country with
a semi-presidential constitution has lost the status of an electoral
democracy since 1991. However, in only one case did the loss of this
status coincide with a period of cohabitation – Niger 1 in 1996. This
case is well documented and constitutes a textbook example of the
perils of cohabitation.14
On 26 December 1992 Niger adopted a new
constitution following a national conference that had sovereign
12
The data on electoral democracies is available at: http://www.freedomhouse.
org/template.cfm?page=439 (accessed 3 February 2009).
13
To be clear, this means that the situation where the president is from one party
and the prime minister is from another party is not necessarily classed as a period of
cohabitation. It is only classed as such if the president’s party is not represented in the
cabinet.
14
For example, L. A. Villalón and A. Idrissa, ‘Repetitive Breakdowns and a Decade
of Experimentation. Institutional Choices and Unstable Democracy in Niger’, in
L. A. Villalón and P. VonDoepp (eds), The Fate of Africa’s Democratic Experiments. Elites
and Institutions, Bloomington, Indiana University Press, 2005, pp. 27–48.
34 GOVERNMENT AND OPPOSITION
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
Table 1
Cohabitation and the Collapse of Young Semi-presidential Democracies
Country Semi-presidentialism
and first year of
electoral democracy
(since 1990)
Year when electoral
democracy was lost
(to 2008 inc.)
Cohabitation (to end
2008 inc.)*
Armenia 1999 2003
Bulgaria 1991 – Jan 1995–Jan 1997
Jul 2001–Aug 2005
Cape Verde 1991 –
CAR 1 1993 2001
CAR 2 2005 2008
Congo-Brazzville 1992 1997
Croatia 1991 –
East Timor 2002 –
Georgia 2004 2008
Guinea-Bissau 1 1994 2003
Guinea-Bissau 2 2005 –
Haiti 1 1990 1991
Haiti 2 1994 2000
Haiti 3 2006 –
Kyrgyzstan 1995 2000
Lithuania 1991 – Nov 1996–Feb 1998
Feb 2003–Apr 2004
Macedonia 1992 – Nov 2002–May 2004
Aug 2006–
Madagascar 1993 –
Mali 1992 –
Mauritania 2007 2008
Moldova†
1995 –
Mongolia 1991 – June 1993–Jul 1996
June 1997–Jul 2000
Montenegro 2006 –
Mozambique 1994 –
Namibia 1990 –
Niger 1 1993 1996 Feb 1995–Jan 1996
Niger 2 1999 –
Poland 1990 – Dec 1991–Dec 1995
Oct 1997–Oct 2001
Nov 2007–
Romania 1992 – Apr 2007–Dec 2008
Russia 1993 2004
Sao Tome 1991 – Oct 1994–Sep 2001
March 2004–June 2005
Senegal 2000 –
Serbia 2006 –
Slovakia 1999 – June 2004–July 2006
Slovenia 1991 – Nov 2004–Jan 2006
Taiwan 1996 –
Ukraine 1994 –
Source: http://www.freedomhouse.org/template.cfm?page=439 and www.world-
statesmen.org.
*There are other cases of cohabitation, such as in France from 1997 to 2002, but
these occurred in countries that were classed as electoral democracies prior to 1990.
†
Moldova adopted a parliamentary constitution in 2000.
35SEMI-PRESIDENTIALISM, COHABITATION AND COLLAPSE
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
decision-making authority.15
The February 1993 elections to the
National Assembly did not provide a majority for any party. In March
1993 Mahamane Ousmane, the candidate of the Convention
Démocratique et Sociale (CDS), was elected as president, winning
nearly 55 per cent of the vote at the second ballot. In April 1993
Ousmane appointed Mahamadou Issoufou of the Parti Nigérien
pour la Démocratie et le Socialisme (PNDS) as prime minister. The
PNDS had supported Ousmane at the second ballot of the presidential
election. However, in September 1994 the PNDS withdrew from
the coalition and joined forces with the Mouvement National pour
la Société de Développement (MNSD), the former ruling party. As a
result, in October President Ousmane dissolved the National Assembly
and a new election took place in January 1995. The elections
returned a majority opposed to the incumbent president and in
February 1995 the National Assembly accepted the appointment of
Hama Amadou of the MNSD as prime minister. There were no
representatives of the president’s party in the government. A period
of cohabitation began. This period was marked by an ‘institutional
crisis’.16
There was a stand-off between the two parts of the exective:
‘As both president and prime minister went “on strike”, refusing to
carry out duties prescribed by the constitution for the normal functioning
of the government, a near-total breakdown in constitutional
procedures resulted.’17
On 27 January 1996, the day when it was
rumoured that the president was going to dissolve the legislature for
a second time, the military stepped in and Niger’s first experiment
with electoral democracy came to an end.
The Niger case may well be a textbook example of the perils of
cohabitation, but it is also the only case among 12 where the collapse
of electoral democracy has coincided with cohabitation. Moreover,
ten countries have experienced cohabitation from 1990 to 2008, and
some have experienced multiple periods of cohabitation, yet Niger is
the only one to have collapsed during a period of cohabitation. For
example, Mongolia experienced seven years of cohabitation in the
first ten years of its history as an electoral democracy, but democracy
survived. Poland experienced six years of cohabitation in the same
15
J.-J. Raynal, Les institutions politiques du Niger, Saint-Maur, Sépia, 1993.
16
B. Issa Abdourhamane, Crise institutionnelle et démocratisation au Niger, Bordeaux,
Centre d’étude d’Afrique noire, 1996.
17
Villalón and Idrissa, ‘Repetitive Breakdowns and a Decade of Experimentation’,
p. 38.
36 GOVERNMENT AND OPPOSITION
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
period and it too survived. Sao Tome also experienced a long period
of cohabitation in the early years of its democratic process and survived.
In this article, there is insufficient room to explore why these
young democracies were able to avoid collapse, specifically whether
or not they survived despite cohabitation. It is quite possible that
cohabitation caused extreme executive conflict, as the literature predicts,
but that other factors cancelled out such conflict and allowed
these countries to survive. That said, it is worth recording that of the
14 cases where a period of cohabitation came to end between 1990
and 2008 inclusive (and excluding Niger), it did so by way of an
irregular political crisis on only three occasions.18
Overall, there is
little prima facie evidence to support the association between
cohabitation and the collapse of young semi-presidential electoral
democracies.
THE THREAT OF COHABITATION AND THE COLLAPSE OF
ELECTORAL DEMOCRACY
In the previous section we showed that there were 12 cases where a
country with a semi-presidential constitution lost the status of an
electoral democracy since 1991, but that in only one case did the
collapse of democracy coincide with a period of cohabitation. If
cohabitation is such a problem, though, then it is at least possible that
the mere threat of it has been enough to provoke a collapse. In the
context where political actors expect cohabitation to occur, they may
take pre-emptive action that successfully prevents cohabitation, but
that also results in the collapse of democracy. If so, we would not
observe cohabitation, but we could still say that there was a link
between cohabitation and collapse. To see whether there is a link
between the threat of cohabitation and the other 11 cases of collapse,
18
The cases of cohabitation ending irregularly are Bulgaria in 1997 (coalition
collapses), Lithuania in 2004 (president impeached), and Sao Tome in 2005
(president/prime minister conflict). The other cases mainly ended by way of a regular
election: Bulgaria in 2005 (parliamentary), Lithuania in 1998 (presidential), Macedonia
in 2004 (the president died just before the end of his term), Mongolia in 1996
and 2000 (both parliamentary), Poland in 1995 (presidential) and 2001 (parliamentary),
Romania in 2008 (parliamentary), Sao Tome in 2001 (presidential), Slovakia in
2006 (parliamentary) and Slovenia in 2006 (the president left his party in January 2006
and became non-partisan).
37SEMI-PRESIDENTIALISM, COHABITATION AND COLLAPSE
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
we need to specify the circumstances under which cohabitation can
occur and then determine whether these circumstances were present
in these cases.
Recall that cohabitation is where the president and prime minister
are from different and opposing parties and where the president’s
party is not represented in government. Given this definition, cohabitation
can only occur in specific circumstances. Outside an election,
it can only occur if the existing legislative majority collapses midway
through a parliamentary term and a new one emerges that is opposed
to the president. In the context of an election, cohabitation can
occur in three ways. It can occur when concurrent presidential and
legislative elections return opposing majorities, when a presidential
election returns a president who is opposed to the existing legislative
majority, or when a legislative election returns a parliamentary majority
that is opposed to the president. Given the ways in which cohabitation
can occur, we can conclude that if the threat of cohabitation is
associated with collapse, then it must occur in the context of legislative
and/or electoral politics. If the collapse is associated with extraparliamentary
events, then the threat of cohabitation is not related to
the collapse.
In the 11 remaining cases of collapse, four were associated with
extra-parliamentary events. Specifically, four cases were provoked by
military coups: Central African Republic in 2001 and 2008, Haiti in
1991 and Mauritania in 2008. For example, in Haiti Jean-Bertrand
Aristide won the December 1990 presidential election in a manner
that was generally considered free and fair.19
In January 1991 the
parliamentary election was completed and Aristide’s party emerged
as the largest force in both chambers of the legislature. The collapse
occurred in November 1991 when the military staged a coup that was
designed to oust Aristide from power. The military were opposed to
what they saw as Aristide’s populist policies. Thus, the coup occurred
only a short while after elections had returned a president who was
supported by the legislature. Cohabitation was not a threat. Similarly,
in Mauritania parliamentary elections were held in late 2006 and Sidi
Ould Cheikh Abdallahi was elected as president in March 2007. The
19
National Democratic Institute for International Affairs, The 1990 General
Elections in Haiti. International Delegation Report, 1991, available at: http://www.
cartercenter.org/documents/electionreports/democracy/FinalReportHaiti1990.pdf
(accessed 6 February 2009).
38 GOVERNMENT AND OPPOSITION
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
government supported the president and had a working majority in
the legislature. In August 2008 the military ousted President Abdallahi
and both the president and Prime Minister Yahya Ould Ahmed
El Waghf were placed under arrest. The military were particularly
opposed to the president’s policy towards Islamic extremism. In the
Central African Republic the collapse of democracy occurred both
times in the context of ongoing armed conflict between the president’s
supporters and insurgents. Elections were not due until late
2003, but in June 2001 there was an attempted coup. President
Ange-Félix Patassé blamed General François Bozizé and in October
dismissed him as chief of staff. General Bozizé then went into armed
opposition to the president and tried to stage a coup in November. In
this context, the collapse of electoral democracy was clearly provoked
by a more general power struggle involving military forces, rather
than shifting parliamentary majorities or electoral politics. Overall,
in these four cases collapse occurred outside the realm of electoral
and parliamentary politics and the threat of cohabitation was not a
consideration.
In the remaining seven cases, the collapse of democracy occurred
during an election year. Therefore, we can set aside any consideration
of examples where the parliamentary majority switched sides
and opposed the president midway through a legislative term. Moreover,
there are no examples of concurrent elections. Thus, we ignore
this scenario. In the remaining cases under consideration, the collapse
of democracy occurred twice in the context of a presidential
election alone – Congo-Brazzaville in 1997, and Russia in 2004 – twice
in the context of a presidential election followed in the same year by
legislative elections – Armenia in 2003, and Georgia in 2008 – twice in
the context of legislative elections followed in the same year by a
presidential election – Haiti in 2000, and Kyrgyzstan in 2000 – and
once in the context of legislative elections alone – Guinea-Bissau in
2003.
When the collapse of democracy occurs in the context of a presidential
election, the threat of cohabitation can be associated with the
collapse only if the collapse is provoked by the opposition and not if
it is provoked by the incumbent president.20
This is because the
legislative majority will remain the same whatever the result of the
20
It can also be provoked if the military intervenes to prevent the opposition
candidate from winning.
39SEMI-PRESIDENTIALISM, COHABITATION AND COLLAPSE
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
presidential election. If this majority supports the incumbent president,
then cohabitation will occur only if the opposition wins the
presidential election. Thus, if the opposition provokes the collapse,
perhaps because it refuses to envisage having to share power with a
legislature that continues to support the previous president, then the
threat of cohabitation may have been a factor in the collapse. By
contrast, if the incumbent provokes the collapse of democracy, presumably
fearing that s/he is going to lose the election, then the
threat of cohabitation cannot be a factor in the collapse. Instead, the
collapse is caused by the incumbent’s desire to remain in power.
In the two cases where collapse occurred in the year of presidential
elections alone, it was associated with the actions of the incumbent.
In Congo-Brazzaville President Pascal Lissouba’s term was due to
expire in 1997. In the run-up to the scheduled election armed conflict
broke out between Lissouba and his supporters and those of his
main opponent, Denis Sassou-Nguesso. According to one observer,
this conflict was generated by ‘Lissouba’s fears that he would either
lose the elections or lose a military contest to Sassou if the elections
were challenged’.21
Lissouba was ‘obsessed’22
with his political survival.
He calculated that he might not be able to win the presidential
election at the first round, which was his best chance of success
overall.23
The level of violence increased, and in June 1997 a civil war
broke out between the supporters of Lissouba and Sassou. The presidential
election was never held and democracy collapsed. This
example suggests that the fear of the incumbent losing power was the
main motivation for the collapse, rather than the opposition seizing
power so as to avoid any power sharing. In Russia the situation was
less conflictual, but the same logic applies. President Putin’s United
Russia Party was the largest party following the December 2003
legislative elections and, in effect, it commanded a parliamentary
majority. In March 2004 Putin was re-elected as president by an
overwhelming majority on the first ballot. The problems with the
election were associated with the lack of any genuine opposition and
21
J. F. Clark, ‘The Collapse of the Democratic Experiment in the Republic of
Congo. A Thick Description’, in Villalón and VonDoepp, The Fate of Africa’s Democratic
Experiments, p. 106.
22
P. Yengo, La guerre civile au Congo-Brazzaville 1993–2002. ‘Chacun aura sa part’,
Paris, Éditions Karthala, 2006, p. 261.
23
Ibid., p. 262.
40 GOVERNMENT AND OPPOSITION
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
the absence of a genuine democratic debate.24
The collapse was not
caused by the actions of the opposition. Instead, it was associated with
an incumbent maintaining a hold on power to the extent that electoral
democracy collapsed.
A similar logic applies to the two cases where presidential elections
were followed by parliamentary elections in the year of collapse. In
Armenia the collapse was initiated by the conduct of the February/
March presidential election. The Organization for Security and Cooperation
in Europe (OSCE) report stated, ‘[T]he overall process failed
to provide equal conditions for the candidates. Voting, counting and
tabulation showed serious irregularities, including widespread ballot
box stuffing.’25
This election resulted in the re-election of President
Robert Kocharyan. In other words, as in the cases of Congo-Brazzaville
and Russia, the collapse was associated with the desire of the incumbent
to remain in power, however fraudulently, rather than the opposition
provoking collapse because it wanted to avoid power-sharing.
Almost exactly the same scenario occurred in Georgia in 2008. President
Mikheil Saakashvili was returned in a January election. While the
OSCE reported that the election was ‘in essence consistent’26
with
democratic standards, it also concluded that ‘the campaign was overshadowed
by widespread allegations of intimidation and pressure’
and that ‘[O]ther aspects of the election process, notably vote count
and tabulation procedures, as well as the post-election complaints and
appeals process, further presented serious challenges’. The absence
of the threat of cohabitation in these two cases is confirmed by the
conduct of the legislative elections that followed shortly afterwards. In
both cases, the OSCE reported that the legislative elections were
conducted in a fairer manner than the previous presidential elec-
tion.27
In other words, the loss of electoral democracy status refers
24
See the report by the Parliamentary Assembly of the Council of Europe, http://
assembly.coe.int/main.asp?Link=/documents/workingdocs/doc04/edoc10150.htm
(accessed 8 February 2009).
25
Organization for Security and Cooperation in Europe, Final Report on the Presidential
Election in Armenia, 19 February and 5 March 2003, Warsaw, OSCE/ODIHR, 2003,
p. 1.
26
Organization for Security and Cooperation in Europe, Final Report on the 5
January 2008 Extraordinary Presidential Election in Georgia, Warsaw, OSCE/ODIHR, 2008,
p. 1.
27
Organization for Security and Cooperation in Europe, Final Report on the Parliamentary
Elections in Armenia, 25 May 2003, OSCE/ODIHR, 2003, p. 1; Organization for
41SEMI-PRESIDENTIALISM, COHABITATION AND COLLAPSE
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
particularly to the presidential contests in these countries and to the
desire of the incumbent to remain in power.
All else being equal, the threat of cohabitation is likely to be
strongest when it relates to the anticipated outcome of a legislative
election. If the collapse occurs in the context where actors expect the
election to return a majority opposed to the president and where they
precipitate the collapse in order to prevent such a scenario, then the
threat of cohabitation can be considered a factor in the collapse.
Indeed, the threat of cohabitation is all the more likely to be a factor
in this context because both the incumbent president and the opposition
may have an incentive to precipitate the collapse. For example,
even though the president knows that s/he would remain in office if
the opposition were to win a legislative majority, s/he may not wish to
share power and may provoke the collapse in advance of having to do
so. By the same token, even though the opposition know that they
would control the legislature and the government, they may feel that
they now have the support of the country as a whole and that they
deserve all the power in the system, including the presidency. Thus,
they may encourage the collapse in order to ensure that they obtain
such power.
That said, if the threat of cohabitation really is a factor in these
cases of collapse, then it must be associated with the situation where
there is the realistic possibility that the election will return a coherent
and cohesive opposition majority. In the absence of such a majority,
even if the election fails to return a coherent and cohesive propresidential
majority, the president may still have the opportunity to
form a coalition that includes his/her supporters – thus allowing
some participation in government to be maintained. In this context,
if the president’s actions precipitate the collapse, perhaps by rigging
the election to ensure that his supporters do better than expected,
then the threat of cohabitation cannot be a factor in the collapse.
Here, the collapse is provoked by the president’s desire to minimize
any loss of control over government. Equally, if the opposition calculate
that they will fail to enjoy a coherent and cohesive majority and
provoke the collapse, then again the threat of cohabitation cannot be
a factor in the collapse. In this event, the collapse is caused by the
Security and Cooperation in Europe, Statement of Preliminary Findings and Conclusions
on the 21 May 2008 Parliamentary Elections in Georgia, p. 1, available at: http://www.
osce.org/documents/odihr/2008/05/31268_en.pdf (accessed 8 February 2009).
42 GOVERNMENT AND OPPOSITION
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
opposition’s frustration at the prospect of not gaining a proper
legislative majority. So, for the threat of cohabitation to be a factor in
the collapse, we need to be sure that the legislative election will
return a coherent and cohesive opposition majority.
In Haiti, legislative elections were held in May 2000, while the
presidential election was held in November of the same year. The
collapse was associated with irregularities that occurred during
the legislative election and that provoked the opposition to boycott
the presidential election.28
The Electoral Mission from the Organization
of American States (OAS) noted particular problems with the
counting of the votes in the legislative election and with the organization
of the second round of the Senate election.29
It is, of course,
impossible to know what the result would have been had the election
proceeded fairly, but we can be sure that the incumbent Fanmi
Lavalas Party was the largest party within the system and that the
opposition was scattered across a variety of parties and indepen-
dents.30
In this scenario, while the OAS clearly reported that the
Fanmi Lavalas gained a fraudulent advantage in the election, there is
little to suggest that a free and fair contest would have returned a
coherent anti-presidential majority that would have precipitated a
period of cohabitation. Thus, we can be sceptical as to whether the
threat of cohabitation was really present. Instead, the collapse seems
to have been precipitated by the incumbent party’s desire to maintain
as much control as possible over the system.
A similar scenario applies to the elections in Kyrgyzstan in 2000. A
major difference between the two cases, though, is that organized
political parties were very weak in Kyrgyzstan. In fact, 407 of the 420
candidates standing in the legislative election were identified as inde-
pendents.31
Given that President Askar Akayev was also classed as
28
E. Mobekk, ‘Enforcement of Democracy in Haiti’, Democratization, 8 (2001),
p. 184.
29
Organisation des États Américains, Observation Électorale en Haïti. Élections Législatives,
Municipales et Locales Février à Juillet 2000, Washington, DC, Secrétariat général
de l’Organisation des États Américains, 2000.
30
Ibid., pp. 106–14. See also an opinion poll in Haïti Progrès, 17: 50, 29 February–7
March 2000. This poll was criticized, but it showed the Fanmi Lavalas on 33.86 per cent
with a 7 per cent lead over its nearest rival.
31
Organization for Security and Cooperation in Europe, Kyrgyz Republic. Parliamentary
Eections. 20 February and 12 March 2000. Final Report, Warsaw, OSCE/ODIHR, 2000,
p. 19.
43SEMI-PRESIDENTIALISM, COHABITATION AND COLLAPSE
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
non-partisan, then the election was not fought in the context where
an opposition party or coalition risked winning a coherent legislative
majority and forcing a period of cohabitation on the president.
Instead, it was fought in the context of clan politics, competition
between regional groupings and individual support for, or opposition
to, the incumbent regime.32
The regime certainly promoted a
fraudulent election, but not out of a fear of cohabitation, more out of
a desire to keep control of valuable state resources.
By contrast, the situation in Guinea-Bissau in 2003 does provide
some evidence that the threat of cohabitation was a real factor in the
collapse of electoral democracy. In November 1999 a coalition of the
Social Renewal Party (PRS) and the Resistance of Guinea-BissauBafatá
Movement (RGB-MB) won a legislative majority, and Kumba
Ialá, the candidate of the PRS, headed the poll at the first ballot of
the presidential election. In January 2000 Ialá was overwhelmingly
elected at the second round, ahead of the candidate of the historic
ruling party, the African Party for the Independence of Guinea and
Cape Verde (PAIGC). There were severe tensions within the coalition
over the next two years33
and there were three prime ministers
between February 2000 and November 2002. At that time, President
Ialá dissolved the Assembly, but the promised elections were delayed.
President Ialá was accused of acting in an increasingly authoritarian
manner and the military had little loyalty towards the him.34
In September
2003, the day after President Ialá announced a further delay
to the legislative election, the military intervened in a bloodless coup.
In this context, the president’s desire to hold on to power and the
military’s opposition to the president certainly contributed to the
collapse. However, it is reasonable to suggest that the president
delayed holding legislative elections because of the fear that the
PAIGC would return to power. Indeed, when the legislative election
was finally held in March 2004 the PAIGC was returned as the largest
party, with 45 of the 100 seats in the legislature. Given that Ialá was no
32
R. Abazov, ‘The Parliamentary Elections in Kyrgyzstan, February 2000’, Electoral
Studies, 22 (2003), pp. 545–52.
33
E. Azevedo and L. Nijzink, ‘Semi-presidentialism in Guinea-Bissau. The Lesser of
Two Evils?’, in R. Elgie and S. Moestrup (eds), Semi-presidentialism Outside Europe,
London, Routledge, 2007, p. 149.
34
J. Forrest, ‘Democratization in a Divided Urban Political Culture. GuineaBissau’,
in L. A. Villalón and P. VonDoepp (eds.), The Fate of Africa’s Democratic Experiments.
Elites and Institutions, Bloomington, Indiana University Press, 2005, p. 260.
44 GOVERNMENT AND OPPOSITION
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
longer president by this time, we can conclude that at least part of the
reason for the dissatisfaction with the old regime had already disappeared.
If an election had been held in late 2002 or early 2003, then
the chances of a PAIGC victory were probably even greater.
Clearly, it is difficult to ascribe explanatory power to something
that cannot be observed. However, by specifying the conditions
under which cohabitation can occur and by examining whether those
conditions have been present in the cases where electoral democracy
has collapsed since 1991, we can provide some evidence to suggest
whether or not the threat of cohabitation was a factor in the collapse.
Having done so, we find that on only one occasion is it reasonable to
suggest that the threat of cohabitation was even a proximate cause of
collapse.
DISCUSSION: WHY HAS COHABITATION NOT BEEN
MORE PROBLEMATIC?
The failure to find compelling evidence for a link between cohabitation
or the threat of cohabitation and the collapse of young semipresidential
democracies is significant because it casts doubt on one
of the main criticisms of semi-presidentialism. Why is there little
evidence to link cohabitation with collapse? We suspect that the
answer lies in the fact that the concept of ‘cohabitation’ refers to a
very specific political situation and that this situation only occurs in
a very specific set of circumstances. Therefore, while in theory,
quite rightly, cohabitation is a potential problem for young semipresidential
democracies, in practice cohabitation is likely to be less
problematic than might be expected.
Cohabitation refers to a very specific political situation – where the
president and prime minister are from opposing parties and where
the president’s party is not represented in government. In the
context of this situation, the problems identified by writers such as
Linz, Stepan and Suleiman may well be valid. Indeed, we saw that the
collapse in Niger followed almost exactly the scenario outlined by
Linz. Equally, events in Guinea-Bissau in 2002–3 also suggest that the
threat of cohabitation may have been a factor in the collapse of
democracy there. However, cohabitation is a very specific event. It
refers to the situation where the president is totally isolated within the
executive. Yet presidents usually have control of political and other
45SEMI-PRESIDENTIALISM, COHABITATION AND COLLAPSE
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
resources. They can use these resources to forge alliances. Therefore,
presidents can usually manage to avoid a situation where they are
completely alone; in this way, they can usually avoid the problem of
cohabitation. In this event, while cohabitation can certainly occur, we
would not expect it to be an extremely common occurrence. The
relatively low frequency of cohabitation identified in Table 1 tends to
substantiate this point.
When cohabitation does occur or when the threat of it occurring
is present, then this can be only in very specific circumstances.
Outside elections cohabitation can occur only if an anti-presidential
majority forms in the legislature part-way through the legislative
term. Given presidential control of resources, while such a shift is
possible, perhaps if the president is extremely unpopular or if
resources are being hoarded, we suspect that the fundamental shift
that is required to bring about cohabitation – namely the shift to a
coherent and cohesive anti-presidential majority – will occur only in
exceptional circumstances. For example, presidents may lose the
support of a coalition party part-way through the legislature term, but
they may have the resources to win the support of a replacement
party, they may be able to win the support of independents, they may
be able to appoint a minority government that is able to survive in
office because of a divided opposition. In each of these scenarios,
presidential power may be diminished, but again cohabitation can be
avoided. Thus, outside elections, while there are scenarios under
which cohabitation is possible, there are also plenty of scenarios
under which it can be avoided.
In fact, cohabitation is much more likely to occur in the context of
an election. Here, it can occur in three circumstances. First, it can
occur when concurrent elections return opposing majorities. In
countries with semi-presidential constitutions, concurrent elections
are very rare. They occurred only in Mozambique, Namibia and
Romania up to and including 2004. Thus, concurrent elections are
unlikely to be a frequent source of cohabitation in semi-presidential
systems.
Second, cohabitation can also occur when a presidential election
returns a candidate who is opposed by the majority in the legislature.
Even in this case, cohabitation can often be avoided because the
newly elected president usually has the option of dissolving the legislature.
The subsequent election is likely to return a coat-tails presidential
majority, or it is at least unlikely to return a coherent and
46 GOVERNMENT AND OPPOSITION
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
cohesive anti-presidential majority. If the president does not have the
option of dissolving the legislature, then there is still the option of
trying to form a government from among the existing set of parties in
the legislature, including the president’s party. Again, this may not be
possible, and cohabitation may still occur. What these scenarios demonstrate,
though, is that even in the context of a president being
elected who is opposed to the existing legislative majority, we find
that cohabitation is not inevitable. In fact, we find that when a presidential
election returns a candidate who is opposed by the majority in
the legislature, cohabitation is only likely to occur when the president
does not have the power of dissolution and when the legislature
contains a coherent and cohesive majority that is opposed to the new
president. In reality, such a combination of circumstances can often
be avoided.
Third, cohabitation can occur when legislative elections return a
majority opposed to the president. This is the most likely scenario
under which we might find cohabitation. This is because even if the
president has the power to dissolve the legislature, this power is
effectively obsolete. Therefore, if the elections return a coherent and
cohesive anti-presidential majority, the president has little option but
to cohabit with the opposition majority. Often, though, legislative
elections do not return such a majority. Such a majority is likely to be
most closely associated with a two-party or two-bloc system. However,
the party system may be fragmented and presidents may be able to
cobble together a working majority. Even if there is a two-bloc system,
the anti-presidential bloc may be divided. The president may be able
to win the support of opposition deputies and avoid an opposition
majority. In short, while cohabitation is most likely to occur when
legislative elections return a majority opposed to the president, it still
requires a particular set of circumstances to combine. These circumstances
can occur, and have occurred, but they combine perhaps less
frequently than the established academic wisdom implies. Indeed, in
unconsolidated democracies – the ones that are most likely to collapse
– the presence of two-party systems or cohesive two-bloc systems
is rare.
What does this discussion tell us about the apparent perils of
semi-presidentialism? It tells us that writers such as Linz were right
to warn about the problems of cohabitation. Textbooks should
continue to point out that cohabitation can occur under semipresidentialism
and that it can be dangerous in the context of young
47SEMI-PRESIDENTIALISM, COHABITATION AND COLLAPSE
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
democracies. However, it also tells us that we need to qualify the
argument about the perils of cohabitation. Cohabitation is a very
specific situation that occurs only in very specific circumstances,
which, in many cases, can be avoided. Therefore, even without any
attempt to control for the impact of other explanatory factors that are
commonly associated with democratic collapse, such as wealth and
ethnic fragmentation, we should not be surprised to find that there
are very few cases that link cohabitation with collapse either directly
or indirectly.
Overall, nothing in this article suggests that semi-presidentialism is
anything other than perilous. For example, there is good evidence
to suggest that the president-parliamentary form of semipresidentialism
is more dangerous than the premier-presidential
form.35
Moreover, we should investigate the effects of the situation
where the president’s party is represented in the cabinet but where
the prime minister is from a separate party. These situations may well
be more common in young democracies where inchoate party
systems are prevalent and they may well be the source of potentially
problematic intra-executive conflict. What this article does suggest,
though, is that, if semi-presidentialism is a perilous constitutional
choice for a young democracy, then it is likely to be perilous for
reasons other than the problem of cohabitation.
CONCLUSION
A common argument against the adoption of semi-presidentialism by
a new democracy is the problem of cohabitation. Writers have suggested
that cohabitation can cause executive gridlock that motivates
either the military to intervene to restore decision-making effectiveness
or the president or prime minister to seize full power at the
expense of democracy. Therefore, cohabitation is seen to be a threat
to the survival of young semi-presidential democracies. We have
shown that the problem with this argument is that there appears
to be little empirical evidence to support it. In all nascent semipresidential
democracies since 1990 when cohabitation has occurred
democracy has usually survived. In only one case is cohabitation
35
R. Elgie, ‘The Perils of Semi-presidentialism. Are they Exaggerated?’, Democratization,
15: 1 (2008), pp. 49–66.
48 GOVERNMENT AND OPPOSITION
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd
directly associated with the collapse of a young democracy. Moreover,
we have specified the conditions under which cohabitation can occur
and shown that the threat of cohabitation has been associated with
collapse in only one further case. Overall, this article has shown that
there is little evidence to support one of the most well-known and
long-standing arguments against semi-presidentialism. We suggest
that this is the case because cohabitation refers to a very specific
situation that can only occur under a certain combination of circumstances
that can often be avoided. While cohabitation can provoke
the collapse of young semi-presidential democracies, we suggest that
it is more important to explore other reasons for such collapses.
49SEMI-PRESIDENTIALISM, COHABITATION AND COLLAPSE
© The Author 2010. Journal compilation © 2010 Government and Opposition Ltd