7 Semi-Presidentialism
7.1   THE FRENCH PROTOTYPE
We have seen that both presidentialism and parliamentarism may fail us, especially in their pure forms. It is from both these failures, then, that we are prompted to seek a 'mixed' solution, a political form that stands at the cross-roads between presidential and parliamentary systems and draws from both. This mixed form has come to be known - I think tellingly - as semi-presidentialism.1 While we should not read the label too literally, it does convey that it is by starting from presidentialism, not from parliamentarism, that our mixed system is best understood and construed. The argument flows more cogently from the top down than from the parliamentary base up.
The common denominator of both presidentialism and semi-presidentialism is a popularly elected president or, at the least, a president that is not elected in and by parliament. But beyond this common foundation the two forms radically depart from one another, for semi-presidentialism is 'semi' precisely in that it halves presidentialism by substituting a dual authority structure for a monocentric authority structure. In presidential systems the president is protected and insulated from parliamentary interference by the division of power principle. Instead, semi-presidential systems perform on a power sharing basis: the president must share power with a prime minister; and, in turn, the prime minister must obtain continuous parliamentary support.
Does the above suffice to define semi-presidentialism? Very definitely the answer is no. But for the time being this issue must rest. With reference to presidentialism it is permissible to start with the definition because the form obtains an undisputed prototype - the US presidency - and has consolidated itself into a well identifiable pattern across a sizeable number of countries. But with reference to semi-presidentialism we must be aware, and steer clear of, a chicken and egg circularity. On
121
122
Comparative Constitutional Engineering
Semi-Presidentialism
123
the one hand, we do not agree on the countries that qualify as being semi-presidential; and the obvious reason for this is that we lack a definition. On the other hand, how can we define the specimen unless we know from which countries its defining characteristics are to be inferred? The way out is to stipulate that the category exhibits an extant case - the French Fifth Republic - and to pick up the matter from here.
The one characteristic that any semi-presidential system must have (by virtue of its very name) is a dual authority structure, a two-headed configuration. Thus any semi-presidential constitution must establish, in some manner, a diarchy between a president who is head of state, and a prime minister who heads the government. In the 1958 French constitution the premier's 'head' is clearly outlined in articles 20 and 21, which read, respectively, that 'The government determines and direct national policy', and that 'The prime minister directs the: actions of the government.' The presidential 'head' is outlined, instead, far less clearly and in a more scattered fashion in articles 11, 12, 15, 16 and 52 of the constitution; and this scattering corresponds to the fact that the 'principal powers of the president . . . have a spasmodic character. . . . They are not normal prerogatives . . . but exceptional powers which can be used only infrequently. Furthermore [most presidential powers] are not powers of decision. They tend either to prevent a decision ... or to submit the decision to the French people {dissolution, referendum)'.2 What is very clear, then, is that the intent of the drafter of the constitution, Michel Debre, was not to establish a preeminent, 'imperial' president.
However, in this case more than in others the living or, as the Italian wording has it, the 'material' constitution has quickly taken precedence over the 'formal' one. It is also mistaken to derive the French semi-presidential system only from the 1958 text of the constitution, for the actual configuration of the system was crucially determined, in 1962, by a new element: the direct, popular election of the president. The French model results, then, from a 1958 constitution whose balance was tipped in 1962, and whose practice was established by the initial, pacesetting presidency of general Dc Gaulle.3 And after more than 30 years of running what the French actually have:
is, in essence, a bicephalous system whose heads are unequal but also in oscillation between themselves. More exactly put, the 'first head' is by custom (the conventions of the constitutions) the president, by law (the written text of the constitution) the prime minister, and the oscillations reflect the respective majority status of one over the other. On this last note we are brought back to the American ■    problem of 'divided government' {supra 5.2). Presidential . v   systems cannot easily cope, we know, with split majorities. The crucial question thus becomes whether semi-presidential systems founder on the same reefs, or whether their advantage over the presidential ones is indeed that they can handle divided majorities. Let the issue be phrased as follows: is there a difference, and, if so, what is the difference, between pre-sidentialism and semi-presidentialism when it comes to split majorities, that is to say, when the majority that elects the ;    president is not the majority that controls parliament?
There are three possible answers to the question. One is that there is ultimately no difference: in both systems a divided j:  majority inevitably leads to conflict and gridlock. In a presidential system the conflict is between president and congress; 1   in a semi-presidential system the conflict is between president :   and parliament-supported premier; but the substance is the same. The second reply is the one suggested by Vedel and \   Duverger, namely, that semi-presidentialism is not 'a synthesis of the parliamentary and presidential systems, but an alternation ;   between presidential and parliamentary phases' (Duverger, :   1980, p. 186). In this interpretation the French system is :   presidential when the president's and parliamentary majorities are consonant, and parliamentary when they are dissonant. Since I do not concur with either one of the two interpretations, I will shortly submit a third. But first a glance at how this , l_  matter has been handled, in practice, thus far. t;-    The first split majority came in France relatively late -between 1986 and 1988 - and the experience was relatively \  brief. A new period of coexistence, of cohabitation (as the French ,l call it) occurred between 1993 and 1995 when Chirac was elected president. The first cohabitation went smoothly; and so far so has the second one. To be sure, no firm conclusion can be
124
Comparative Constitutional Engineering
drawn on such meager evidence. Aod, to be sure, things went well also because Mitterrand and his 'contrary' prime ministers played their respective cards with moderation and intelligence, But the smooth working of their cohabitation cannot be-attributed to the personality traits of the players alone. While hot headed leaders and compromise busters can disrupt any mechanism of power sharing, the French bicephalous arrangement has worked because it can work. While pure presidenti-alism is a stalemate-prone structure, semi-presidentialism proposes a gridlock-avoiding machinery. Or this is how I see it.
I have said that I do not concur with the view of Duverger. (and others) that French semi-presidentialism adds up to being' an alternation between presidentialism (when the presidential and parliamentary majorities coincide) and parliamentarism (when they do not). My first objection is that both alternatives are overstated. On the one hand a president cum government,, that has to govern with and through another body, cannot be assimilated to the 'pure' president that governs alone, who lithe government. Here the upgrading of the semi-presidential president is excessive (and structurally unacceptable). On the other hand, and at the other extreme, a president without majority cannot be assimilated to a standard parliamentary president. Here it is the downgrading that is excessive (and factually incorrect). A French minority president can no longer exploit his 'usurped powers' (arising from the material constitution), but never becomes a figurehead; he remains a president who stands on. his own, direct legitimacy, and a president empowered by the letter of the constitution ta prerogatives that parliament-elected presidents seldom if ever have. ]
Secondly, and more importantly, I object to the 'alternation' interpretation. To conceive semi-presidentialism as an alternation between two other species amounts to blowing apart the integrated nature of the system, and indeed asserts that here we do not have a veritable system. I think that this is a: gross misunderstanding that misses the point conveyed by my. notion of 'oscillation'. For alternation suggests a passage from one thing to another, while oscillation is a within-system; movement. In oscillating something remains itself.
Semi-Presidentialism
125
My interpretation is, then, that French semi-presidentialism has evolved into a truly mixed system based on a flexible dual authority structure, that is to say, a bicephalous executive whose 'first head' changes (oscillates) as the majority combina-, tions change. With a unified majority the president decisively prevails over the prime minister and the constitution that applies is the material one (the conventions of the constitution). Conversely, and alternatively, with a split majority it is the prime minister supported by his own parliamentary majority that prevails, also because the formal constitution (what it states in writing) does support his claim to govern in his own right.
When Mitterrand yielded, in practice, to his prime minister Chirac, he understood very well that he could not win a legal fight.4 And when, conversely, majoritarian prime ministers are nice to their president (as Balladur was with Mitterrand) the underlying reasons for this pattern of behavior are, first, that
'the material constitution has struck roots and, second, that if their turn comes for presidential office (an expectation that
. they are entitled to harbor) they too would like to be 'imperial' presidents. Let alone that both 'heads' have probably sensed
..that the system works precisely across the rebalantings provided by the flexible diarchy that I have just outlined.
Certainly, any dual authority structure can become confrontational and thereby stalemated by an executive divided against itself. There is no assurance that this cannot happen with French semi-presidentialism. Still one must recognize that in this formula the problem of divided majorities finds a solution by 'head shifting', by reinforcing the authority of
-whoever obtains the majority.5 And this is a most brilliant, if
^unintended, piece of constitutional witchcraft.
I'll SIMILAR AND DISSIMILAR CASES
r" The French Fifth Republic aside, which other countries can be • classified as semi-presidential? Historically, the German Weimar constitution of 1919-33 may rightfully be considered - in retrospect - the first configuration of a semi-presidential
126
Comparative Constitutional Engineering
system. We next have two countries — Portugal and Sri Lanka - which have deliberately designed their constitutions upon the French model. A further plausible candidate for the category is Finland. And Duverger (1980) included among the semi-presidential systems also Austria, Iceland and Ireland. The : Duverger list thus is: France, Weimar Republic, Portugal, Sri-Lanka, Finland, Austria, Iceland and Ireland. Shugart and Carey (1993) modify it in two respects: Ireland is assigned to . the parliamentary systems (I think rightly); and a new category - president-parliamentary category - absorbs Weimar and Sri Lanka (I think messily and needlessly).
As both lists go to show, semi-presidentialism turns out to be; a lumping of impossible bedfellows that defies definition.6;;-Therefore, either we reduce the list to countries whose political; forms truly have some common core, or we have a name for' nothing, i.e., for a purely residual category.
A first pruning should immediately eliminate — I submit—-Ireland, Austria and Iceland. I have already noted (supra 5.1}:, that these three countries cannot be classified as presidential; only on the grounds that their presidents result-from popular; election. On similar grounds they cannot be considered semi-;j presidential because the Austrian and Icelandic presidents (not-; the Irish one) are strong only on paper, that is, are constitu-o tionally given powers that the living constitution relegates to3 inanition. With reference to the French case the 'added powers of the president must be taken into account because it is actually exercised and has in fact entered the material constitution. But when the material constitution deprives a pre-) sident of powers that remain a dead letter of the formal! constitution, then a 'dead element' surely cannot establish; the nature of a political form and the class to which it belongs. Otherwise we could classify England as a monarchy - not a? democracy - for no act of parliament has ever cancelled the; powers that the Crown exercised, say, in the eighteenth-; century.
Duverger himself acknowledges that in Austria, Ireland and: Iceland 'political practice is parliamentary' and that their respective presidents are 'figureheads' (1980, p. 167 and passim). What is the point, then, of assigning these countries
Semi-Presidentialism
127
to semi-presidendalism on a dead letter basis? Well, the point could be that under different circumstances the formal constitution still has a chance of resurrection. Indeed, unpracticed legal norms are not - for this reason alone - dead norms. Some legal rules remain dormant, and then are brought back to life. But some legal rules do die beyond any possibility of comeback. So, do the formal constitutions in question have a chance
: of being resurrected? I would say no, definitely not, for Ireland and Iceland;7 and allow a small chance, but only a small one, to Austria. In the case of Austria the acquiescent president came about, in constitutional practice, in the wake of the
-'grand coalition' period, when a 90 percent parliamentary majority stood massively in the way of any possible presidential
-interference in governmental matters. Today Austria has transformed its original two-party format into a three-party
- system; but this state of affairs can block any presidential 'activism just as effectively as the earlier one. In the case of
- Iceland the constitution can be amended by a simple, 51 .percent majority vote, and this implies that an interfering
president could be easily deprived of whatever power he or :;she tries to exercise. The Austrian lesson thus is that strong, ; solidary parties can always clip the wings of an activist president; and the Icelandic lesson is that overly flexible, that is, easily changeable constitutions deprive a supreme law of its
■ supremacy.
Remember, the examples matter because erroneous inclusions inevitably distort the grasp of a specimen. Conversely
■ put, we shall never get a handle on semi-presidentialism unless we first expel from the class its unacceptable members. But the
; elimination of Ireland, Iceland and Austria leaves us with a deaned-up list of four countries - Weimar Germany, Portugal, Sri Lanka and Finland ~ that begins to make sense. The reasons for the above ordering will become apparent as we proceed. And I begin with Weimar not only because it is our historical ancestor, but because the Weimar experience is as instructive as, currently, the French one.
In 1919 there was no notion of semi-presidentialism. At the time, therefore, the Weimar system was perceived as a parliamentary system counteracted by a strong presidency. The
128 Comparative Constitutional Engineering
Semi-Presidentialism
129
drafters of the constitution looked with awe at the French H Third Republic, and their overriding concern was to impede \\ assembly government. This was accomplished, in their design, -\ by empowering the president (i) to govern by decree, i.e., to issue, in concurrence with the chancellor, decrees with the ■i\ force of law in emergency situations (the ill reputed article 48 of the Weimar constitution), (ii) to appoint and dismiss at his sole discretion the prime minister, to dismiss single cabinet j ministers, and to form governments that were not subject to a ;j parliamentary vote of confidence, (iii) to dissolve parliament § virtually at his discretion, (iv) to refer any law approved by the § Reichstag to popular referendum (in lieu of having veto power), f As for parliament, while it was not called upon to appoint J governments with a vote of confidence, it was empowered to g make them fall with a vote of no-confidence (and to force the 1 resignation of single. ministers); and under normal circum- ~ stances it is fair to say that most of the legislative work was allocated to the Reichstag. I There is little question that Weimar displayed the dual J authority structure that characterizes, in my account, a semi- | presidential form. How close or how different was Weimar, on | this score, to present-day France? The answer is complicated J by the fact that the French formal constitution has evolved into § a quite different material constitution. So, Weimar is compared to which? For instance, one major difference between the * two formulas appears to be that the French governments need | the investiture of a parliamentary vote of confidence, whereas the Weimar governments were only subject to dismissal by a vote of no-confidence. Yes and no, however. For Articles 49 J and 50 of the French constitution do not speak of a parlia- ■ mentary investiture of the executive and tackle the problem very much as the Weimar constitution did. But it is unnecessary to dwell on this kind of probing, because the faltering of the Weimar Republic had little to do, in my opinion, with * constitutional defects. The construction was crippled, first and ;. foremost, by a highly fragmented and polarized party system that was nurtured, in its turn, by a 'pure' system of propor- -;■ tional representation (with a single national constituency). Pace Lijphart, more than by any other single factor Weimar
was undermined by PR (a defect remedied by the French with the double ballot majority system). Another weak point of the Weimar structure was that the president was elected by plurality, for the run-off allowed a three-person contest that impeded the emergence of a two-cornered and thus bipolar-molding race for the presidential prize.
But these are largely extra-constitutional elements. When we come to the constitution, proper, its traditional 'suspects' - the emergency powers of article 48, and the dissolution and referendum instruments - have been over-suspected. These powers were abused, but only under circumstances that justified their abuse. Very similar provisions are found in many other constitutions and have not produced particular harm. All in all, I definitely would not say that the Weimar constitution carries the blame for the unhappy and short life of the first German republic, and especially for its downfall at Hitler's hands in 1933. Despite its imperfections, it was a cleverly and innovatively designed construction. Indeed, had Germany adopted in 1919 a parliamentary type of constitution, I believe that Weimar would have collapsed much earlier than it did -probably by 1923, when the mark was pulverized by the worst inflation ever to happen in Europe. And what fatally injured Weimar was the 'great depression' of 1929, which hit Germany more devastatingly than any other Western country.
Turning to Portugal, this is a case that can be dealt with quickly, for its semi-presidential experience was short lived: six years between 1976-82.8 As regards the Portuguese constitution, its model was the French one and does not need to be rehearsed. The lesson from Lisbon is, however, that it is very unsafe to inaugurate a semi-presidential system with a minority president and, still worse, with a president without party backing. Unless semi-presidentialism obtains, at its inception, a long run of undivided majorities that consolidate the foundations, the early advent of split majorities may initially pose too much of a strain on the delicate balance of cohabitation. And this is, in essence, what happened in Portugal.
Now for Sri Lanka. The 1978 Sri Lankan constitution was also inspired by the French model (the living one), but went further in reinforcing the president's power, and this very
130 Comparative Constitutional Engineering
much in the Weimar mold. The Sri Lankan president, like the Weimar one, is unilaterally and discretionally entitled to outflank the legislature by submitting to referendum approval any bill rejected by parliament. Finally, and this goes beyond both France and Weimar, Sri Lanka's president is both the head of state and of the government. Legislation is thus truly initiated by the president, who also handpicks his or her ministers (very much like the American president). The question thus arises as to whether Sri Lanka is more a 'fully presidential' than a semi-presidential system. The counter-argument points out, however, that Sri Lanka does not have a one-person executive (as truly presidential systems require), and that a parliamentary vote of no-confidence does cause the fall of the cabinet.
Clearly, we are here at the edge, somewhere between presidentialism and semi-presidentialism. But I do not feel that it is crucially important for us to decide this matter, also because it is premature to do so. Since Sri Lanka has yet to. confront the acid test of a 'divided majority', it is hard to tell whether in that event a president would acquiesce or fight. He or she certainly has, in the written constitution, far more fighting force than the French president, especially on referendum grounds. Indeed the recourse to referendum has already gone as far as to extend, in 1982, the tenure of parliament for another full term of six years! On the other hand, it must be acknowledged that this and other abuses occur during a state of civil war which helps justify them. The: system will have to be judged, then, if and when it starts sailing in calmer waters,
Finally, Finland, which is our oldest case since it goes all the way back to 1919. In the Finnish formula (which is referred here to the period 1919-91) the president, who is elected indirectly, chairs the cabinet meetings bearing on his reserved domain, namely, 'the relation of Finland with foreign powers' (article 33). He does not attend, however, the cabinet meetings bearing on internal and administrative affairs, which are chaired by the prime minister. In addition, the president does effectively choose (on the basis of the material constitution) the prime minister and does condition the coalitional composition, of governments. But in all other respects Finland performs like
- i
Semi-Presidentialism 131
?
a normal parliamentary system in which governments are -    voted into office and removed from office by parliament. Thus
Finland qualifies (qualified until 1994) as a semi-presidential r    system, but for one aspect: the indirect election of its president.
We shall come back to this.9
;    7.3  DEFINING SEMI-PRESIDENTIAL SYSTEMS
Having identified the cases to which the category plausibly applies, we can now revert to the unfinished business of
", defining semi-presidentialism. Even though we have escaped circularity, we are still confronted with the difficulty of pinning down a two-headed configuration that also displays a significant variance across its cases. How far can we go - in establishing defining characteristics - in firming up who has power over whom, in what manner and/or circumstances, and when a dual authority ceases to be dual? If we try to be overly precise,
; we easily end up with a class with just one case. If, on the other hand, we seek refuge in overly loose formulations, then the
t primary purpose of defining - the drawing of borders - is defeated. I shall try to steer a somewhat discursive middle course between overdefining and underdefining. Bearing the foregoing provisos in mind I stipulate that a political system is semi-presidential if the following properties or characteristics jointly apply:
(i) The head of state (president) is elected by popular ';   vote - either directly or indirectly - for a fixed term of office.
(ii) The head of state shares the executive power with a , prime minister, thus entering a dual authority structure whose :*   three defining criteria are:
(iii) The president is independent from parliament, but is i not entitled to govern alone or directly and therefore his will ~   must be conveyed and processed via his government.
,- (iv) Conversely, the prime minister and his cabinet are * president-independent in that they are parliament-dependent: '   they are subject to either parliamentary confidence or no-
132 Comparative Constitutional Engineering
confidence (or both), and in either case need the support of a parliamentary majority.
(v) The dual authority structure of semi-presidentialism allows for different balances and also for shifting prevalences of power within the executive, under the strict condition that the 'autonomy potential' of each component unit of the executive does subsist.
Clearly, a mixed form is never as simple as a pure form. That the definition of semi-presidentialism had to be less tidy and more complicated than the one of presidentialism {supra 5.1) was on the cards and cannot surprise us. The immediate question is: how does my definition compare with other ones? Shugart and Carey deserve the credit for having made the first systematic attempt at defining political forms, and their proposal for semi-presidentialism rejects the label and breaks the specimen into two regime types: (i) premier-presidential and (ii) president-parliamentary (1992, pp. 18-27 and passim). The first is characterized by the primacy of the premier over the president, the second by the primacy of the president.
According to this distinction, however, the French system (and prototype) breaks in two, for it characteristically switches this 'primacy' from president to premier, and vice versa.10 And I cannot accept their distinction on other grounds as well. My second misgiving is that the Shugart-Carey criteria still admit Austria and Iceland in the premier-presidential category (and thus, in substance, into semi-presidentialism); and this attests, it seems to me, to a serious lack of discriminating power. My third misgiving is that their president-parliamentary category turns out to be an almost empty class, a container in desperate want of content.11 And here my point is that there is no point in trying to extract a category from quicksand. So, does their dumping of 'semi-presidentialism' and the distinction thus resulting really sustain discernible and discernibly important differences? I think not.
Reverting to my definition, two explications are in order. As regards the first criterion let me recall that in the case of pure presidentialism the formulation was that the president results
Semi-Presidentialism 133
',   from 'direct or direct-like popular elections', whereas in the case of semi-presidentialism I say that the president is elected i   'either directly or indirectly' by popular vote. The rewording is not casual, and I am not engaging in hair splitting. Under the first formulation (until the constitutional reform of 1991 and .   the 1994 election) the Finnish president could not be counted as a popularly elected president, whereas under the second ■    formulation he could. Of course, the issue is not Finland. The issue is how stringent we want the criterion to be. The <   difference between 'direct-like' (or quasi-direct) and 'indirect' *   is that the second formulation permits the understanding that a -   head of state that is not elected by parliament qualifies as a 'true president' (a presidential president) also when his election is a two-step one. In the latter case the popular vote elects an ad hoc electoral college, which in turn freely chooses and thus truly elects the president.12 One may wonder why I make such a fuss about this. Why not simply let Finland exit from semi-'   presidentialism and describe its polity as a parliamentary 1   system cum strong president? Well, my reason is that I harbor •'   a future-oriented worry: video-politics and video-power, that is :   to say, how television and the mass media age are radically '   affecting politics and, in the matter at hand, the presidential selecrion process.13 To begin with, it is clearly the case that video-politics I   facilitates the election of improvised and, indeed, flash 'outsiders'. Prima facie this may appear to be a good thing. :<   But these outsider-presidents are presidents without troops.
Take Fujimori in Peru and Collor in Brazil, and forget whether '   they are or have been capable presidents: the point remains that their presidency has or had no partisan support. Fujimori " *   cannot be connected to a party; and while Collor improvised a I  party as he campaigned for his election, his party obtained less ,   than 10 percent of the vote, thus leaving him 40 percent short i   of even a nominal majority. To be sure, two cases do not ;   establish a trend. But the symptoms and the seeds are there. In ^   Peru the antagonists of Fujimori have both been outsiders: the *t   novelist Vargas Llosa and subsequently, in 1995, a former .   secretary general of the UN, Perez de Cuellar; and in the United States had not Ross Perot made a couple of colossal
134 Comparative Constitutional Engineering
mistakes (such as withdrawing and reentering at mid-race) he was coming close, in 1992, to being elected president.
A second development — and a high-risk one - is that video-politics turns a presidential election into a very chancy event. The American president has increasingly been the winner of a video-match eminently decided by looks and 'sound bites' of 10 seconds (their average in 1988). Video-elections are supposed to bring about transparency, truly 'visible polities'. Alas, no. What we are actually given, under the guise of visibility, is a display of petty appearances that leaves the issues in greater darkness than ever. The long and short of my brief thus is that the popular, direct election of presidents provides no safeguard and no buffer against a disastrous election (misselection) of the all-decider - and that this will be increasingly the case. More than ever before video-politics promises to put in office im-. provised dilettantes and/or poll-monitored robots (whose true faces are never unveiled) stuck with populist campaign promises.
On these considerations why deny equal legitimacy to a president-making process that allows for the close^ and competent scrutiny that direct elections under video monitoring surely do not allow? Finland owes a great deal to having been able to choose 'right presidents', for not only did Finland have to confront the menace - a very real one — of the Soviet Union, but it also had a highly infelicitous party system of 'polarized pluralism'. But the issue, as I was saying, is not about Finland per se. The issue is whether we accept the Finnish 1919-94 'model' as a form of semi-presidentialism. I propose that we do, perhaps especially for the benefit of the Eastern European and of the formerly Soviet countries.
My second explication of the definition of semi-presidentialism is simply this: that my last three criteria address the: problem of sorting out a dual authority structure, on the assumption that this is the distinctive feature of the specimen. Other characteristics, important as they may be, run across other political forms as well. For instance, 'the power to dissolve parliament, or legislative powers' (Shugart and Carey, 1992, p. 24) are attributes that must either be detailed far more specifically, or that have, as stated, no discriminating value. A
Semi-Presidentialism 135
British prime minister can be said to have, in some sense or other, 'legislative powers' just as much as any president, both of the pure and impure varieties; and the power to dissolve parliament is given, in some form or other, to almost all parliamentary heads of state. Whereas the primary purpose
• (^defining characteristics is to establish the cut-off points between something and what we perceive to be something else (Sartori,
:   1984b, pp. 28-34).
'   7.4  WHICH IS BEST?
In the conclusion of the First Part I answered the 'Which is :   best?' question by indicating a somewhat generalized prefer-
- ence, among electoral systems, for some form of double ballot. .,   But as regards 'whole systems' - the theme of Part Two - it
would be foolhardy, I believe, to be as bold. For that political form is best that applies best: a conclusion that also implies that there is no getting away from context.
Latin Americans are advised to adopt parliamentarism, but the French have dismissed it with relish. Many British people are frustrated by their two-party straitjacket, but most Italians :   think that the British system is great. We are generally justified in criticizing the polity under which we live, but often wrong in assessing its alternative and its hoped-for benefits. I believe that the case against the two extremes, pure presidentialism and .   pure parliamentarism, is a strong one. By the same token I believe that the positive case for 'mixed systems' is equally ■   strong. However, to argue that mixed systems are better than ,   pure ones, is not to argue that among the mixed forms semi-'   presidentialism is the best one. This is an ulterior step that I am
- not prepared to take unqualifiedly.
Let me start with the assertion that semi-presidentialism is better than presidentialism. This has in fact been my thesis -especially on the ground that the former system can cope with split majorities far better than the latter. I also hold that countries that intend to leave presidentialism are well advised to opt for semi-presidentialism on grounds of prudence - on I  account of the fact that for a presidential country a leap to
136
Comparative Constitutional Engineering
Semi-Presidentialism
137
parliamentarism is a leap into the utterly diverse and un- -<> known, whereas a switch to semi-presidentialism still allows that country to perform in the ambit of what it knows, of t.ic experience and expertise that it has.
Does the same advice hold — symmetrically and from the other end of the spectrum - for the countries that wish to abandon their parliamentarism? Should they too contemplate - as the alternative that best suits them - semi-parliamentarism (premiership systems) rather than semi-presidentialism? In theory it might look that way, but in practice it does not. For here the problem is that the implantation of premi< iOn systems largely defies constitutional engineering (supra 6.3), The British or German type of muzzled parliamentarism perform as they are intended to perform only under a set of favorable conditions. If the British system ceased to be a \;\t-party system (in which third parties have never managed, thus far, to break single-party governance), or if the German polity became more fragmented and more polarized than it has been ; thus far, both 'models' would vanish into thin air. And this implies that a country that seeks for itself a semi-parliamentar\ solution embarks on a very hazardous journey, for too many assumptions must hold for this transplant to succeed.
To belabor the point, the parliamentary countries that are the most unhappy with their experience are likely to be the least-working parliamentary democracies and most probably the assembly-driven ones. It should be further assumed that the countries in question display a fragmented party system of the type that I call 'extreme pluralism'. And if this is the case we are also probably dealing with PR countries that have Ion.; indulged themselves with 'proportional polities'. But from this start the incremental passage from pure parliamentarism to \[ premiership parliamentarism is an uphill struggle that is all too easily lost. From this point of departure the easiest path is . not the one of gradually moving into semi-parliamentaiisi'i, but the one of radically switching to semi-presidentialism. This i I is not only attested to by the French experience, but also ;\, sustained by the argument that semi-presidentialism is far . more amenable to constitutional engineering than premiership i» systems.14 \
Note, however, that my argument is not - in the context of parliamentarism - that semi-presidentialism is 'best' but, rather, that it -is 'more applicable'. It is on two different grounds and by two different routes, then, that I end up with generally recommending — to the change advocates — semi-presidentialism. And I further wish to underscore that this recommendation is not a strong one. Semi-presidentialism does leave us with unsettled problems.15 Nor do I deny that semi-presidentialism is a somewhat fragile system. The split majority problem still haunts - if to a far lesser degree than in pure presidentialism - the semi-presidential experience. Can we do better and find a more solid and efficient mixed system? This is the question that I plan to explore in Part Three.
NOTES
1. Cf. contra Shugart and Carey (1993, p. 22 and passim) who replace 'semi-presidentialism' on the argument that 'semi' suggests 'a regime type that is located midway along some continuum'. But semi is the Latin for 'half, and - as any dictionary would show for hundreds of expressions - does not assume any continuum because it precedes continuum-mania by well over two thousand years.
2. Duverger, 1980, p. 171. In short, and apart from the exceptional powers granted in article 16, the constitution empowers the president to have recourse (though not unilaterally) to referendums, to dissolve parliament, and gives him something of a reserve domain in matters of defense and of foreign politics. The president can also effectively block action by his pouvoir d'empecher, either by veto or by refusing to sign decrees.
3. On January 31, 1964-, Dc Gaulle proclaimed 'that the indivisible authority of the state is entrusted completely to the president by the people who elected him* and that 'the supreme domain is his alone'. Nothing of this is in the Debre constitution; but De Gaulle's successors, up to and including Mitterrand, have been happy to oblige.
4. To exemplify, the more patent 'usurping' of presidential powers has occurred with referendums, that De Gaulle and his
138
Comparative Constitutional Engineering
Semi-Presidentialism
139
successors have decided to hold on their own whereas the constitution requires, for referendums (article 11), joint action with the cabinet or with the two chambers. This practice went unopposed; but if a 'minority Mitterrand' initiated a referendum, he could or indeed would be blocked. Likewise, constitutionally the president is not empowered to dismiss the prime minister; he did so cavalierly in the past, but with a split majority he is unlikely to try.
5. If this interpretation is correct, it takes care, I submit, of Linz's objection to semi-presidentialism (1990, p. 63) that 'No democratic principle exists to resolve disputes between the executive and the legislature about which of the two actually represents the will of the people'.
6. The reason for this is, it appears, that interpreters never quite make up their mind as to whether their referent is the formal constitution, the material constitution, and/or a shuttling among the two. The interplay of these elements (as I understand it) comes out in my analysis of the French case, and will be detailed as we proceed.
7. The Irish constitution does not attribute any particular power to its president; so in this case there is nothing to resurrect. And in Iceland the atrophy of the presidential office Has reached its point of no return with the practice of automatically renewing the term of office of the president when no candidate opposes him or her.
8. The 1982 Portuguese constitution eliminates the presidential power to dismiss cabinets or ministers (unless democratic institutions are threatened, article 198) and all his legislative powers; and restricts the president's power to dissolve parliament as well as his pocket veto on legislation. By and large, the Portuguese president is thus left, from 1982 onward, with little more than the normal powers of normal parliamentary presidents.
9. In December 1993 Russia has also entered the category of semi-presidential systems with a constitution hastily adopted by referendum. The Yeltsin constitution is considered semi-presidential in that it was inspired by the French model. However, it stands, like the constitution of Sri Lanka, at the extreme edge of the category. The Russian president appoints the prime minister and his deputy; but if the Duma rejects their nomination three times, then the president may dissolve the lower house and call for new elections. And the same applies to a no-
confidence vote in the government expressed twice. What I read between the lines of these provisions is that the president must win .anyhow. And the efficient secret of the French formula - an oscillating diarchy - is completely lost in its Russian reformulation. The Yeltsin constitution is essentially monocratic; and when a president does not control a majority in the Duma, the executive—parliament relation is left to confrontation. If this is a semi-presidential system, it is an ill-conceived one.
10. And then, why is France assigned to the premier-presidential type? For most of the time, and on the basis of the material constitution, the Fifth Republic ought to be assigned to the president-parliamentary type.
11. To wit, the president-parliamentary cases are Ecuador, Germany (Weimar), Korea, Peru, Sri Lanka (see pp. 40-1). But for South Korea, all the other cases are as evanescent as they can be. Ecuador is one of the most unstable and crisis-ridden regimes in the world; Sri Lanka is, as we have seen, a highly debatable case; and Peru was, at the time of their writing, in constitutional remaking, and currently qualifies as a reinforced
. presidential system. As for Weimar, I cannot find any valid reason for its inclusion in the category. If it is their criterion 2 ('the president appoints and dismisses cabinet ministers'), then the Weimar president did not handpick the cabinet members; and if it is their criterion 3 ('cabinet ministers are subject to parliamentary confidence'), this is inaccurate.
12. That in Finland the election of the electoral college was in fact party controlled and abided by party lines is immaterial to the point. The indirectness would be the same without party control, though the chances of obtaining strongly minoritarian presidents would increase. Shugart and Carey (1992, p. 221) argue that the party-centered presidential election without pledged votes in Finland has not been 'much different from election in parliament'. But a comparison between Finland and Italy - two countries with very similar party systems - attests to very important differences. And then Shugart and Carey overlook the legitimacy difference.
!3.   The point is developed and placed in context infra 8.3.
14. It should be well understood that throughout this work 'constitutional engineering' is a shorthand that includes the electoral system, even though electoral systems are generally determined by ordinary legislation (though not always; not in
140 Comparative Constitutional Engineering
Austria, for example). This is so on substantive grounds, because every political form hinges on its electoral arrangements just as much as on its architectural (and more properly constitutional} ones. 15. As we shall see in Part Three, Chaps 9 and 10, it does not provide, for instance, a satisfactory solution to the problem of 'governing by legislating'.
Part Three Issues and Proposa