Tag Archives: semi-parliamentarism

Steffen Ganghof – On consistently defining forms of government: A reply to Robert Elgie

This is a guest post by Steffen Ganghof, Chair of Comparative Politics at the University of Potsdam

I recently introduced the concept of semi-parliamentary government as part of a comprehensive typology of democratic forms of government 1 (Ganghof 2018). The typology sees “semi-parliamentary government” as one of six basic ways to structure the principal-agent relationship in a democracy (Table 1). It exists when the legislature is divided into two parts, both of which are directly elected, but only one of which has the constitutional right to dismiss the cabinet in a vote of no-confidence.

The typological innovation had three related goals: (1) to apply the existing typological approach more consistently, (2) to highlight semi-parliamentary systems as a neglected form and (3) to theorize new semi-parliamentary designs as reform options in democratic nation-states as well as the European Union. Here I will focus on the first goal.

One worry raised by Robert Elgie (2018, blog post) is that my approach has too many classificatory clauses or dimensions. Yet we must not conflate two separate issues. One is whether we should include criteria other than the origin and the survival of the executive, in particular the rules of assembly dissolution. As I never proposed this (see Table 1), there is no disagreement here and no need for adjectives like “semi-fixed”.

The real disagreement concerns what the consistent application of the established criteria requires (Ganghof et al. 2018b). Robert maintains in his post that “[i]f we stick to the separate origin and survival of the executive and legislature, we get the three standard categories (presidentialism, semi-presidentialism and parliamentarism).” I think this statement is incorrect and that it shows the predicament of the existing approach.

To see this, let us first ignore the internal divisions within both the executive and legislature. The focus on the origin and survival then gives us a four-fold table (consisting of the two outer columns in Table 1). It distinguishes pure parliamentarism and pure presidentialism from the two “mirror hybrids” that exist in Switzerland (assembly-selected fixed-term cabinet) and existed in Israel (directly-elected but assembly-dependent prime minister). In this elegant and consistent typology neither semi-presidentialism nor semi-parliamentarism are distinct types; both are merely sub-types of parliamentarism.

To delineate semi-presidentialism as a distinct type, as Robert wants to do, he has to make a further distinction between “single” and “dual” executives in otherwise parliamentary systems. Indeed, other leading scholars like Samuels and Shugart (2010: 27) first distinguish between systems with single and dual executives and then use the fourfold table to subdivide the single-executive systems. This two-step classification procedure is straightforward, but also somewhat ad hoc and inconsistent. For if we introduce the internal division of the executive into the typology or classification, we ought to do the same for the legislature. After all, just as only one part of the executive may be dependent on assembly confidence, only one part of the legislature may be required to supply it. There is a logical symmetry here that existing classifications neglect. Their asymmetric focus on the internal division of the executive would at least have to be justified, but I am not aware of any such justification.

The same asymmetry and inconsistency shows when we consider the criterion used to distinguish semi-presidential from parliamentary systems. The criterion is the direct election of the president. This criterion is usually not justified explicitly and, again, not applied consistently. If direct election is used as a criterion for an agent’s sufficient democratic legitimacy – for being a primary rather than subsidiary agent of voters – then it ought to be applied to the legislature as well. This is what my typology and the concept of semi-parliamentarism do. They systematically consider the role that direct election plays in constituting a typologically relevant internal division within executive and legislature.

In sum, I contend that the proposed typology results from a symmetric application of long-established criteria. In contrast to Robert, I think it is inconsistent to treat semi-presidentialism and semi-parliamentarism differently. Either both are sub-types of parliamentarism or both are distinct types. The two forms of hybridization can also be combined, as is the case in the Czech Republic, but there is no logical reason to see the semi-presidential characteristic of this case as being conceptually prior to its semi-parliamentary characteristic.

As mentioned, the proposed typology has two other goals. One is to conceptualize and analyze a neglected form of government. A recent symposium in the Australian Journal of Political Science has confirmed the usefulness of the concept of semi-parliamentarism in this regard. For example, Marija Taflaga (2018: 252) states that it “better describes politics as it really is practiced” and offers a “simpler and more coherent description of the Australian system.”

The other goal, and the most important one for me, is to guide our thinking towards new semi-parliamentary designs as reform options for democracies, not only but especially for presidential systems (Ganghof 2016, 2018). In my view, this heuristic function is an important purpose of typologies. And if this is the purpose, the number of democracies that fall into each category is quite irrelevant. The current empirical predominance of democracies with directly (or at least popularly) elected presidents certainly tells us nothing about their normative justifiability.

A crucial insight of the analysis of semi-parliamentary constitutions is that they can potentially reap all the alleged benefits of presidential systems highlighted in the political science literature – constitutional separation of powers, pre-electoral identifiability, post-electoral clarity of responsibility, cabinet stability, a single system-wide constituency, and issue-specific coalition building in the legislature – but without the cost of concentrating massive executive power in a single human being and thereby “presidentializing” political parties (Samuels and Shugart 2010).

This raises deep and thorny questions about the democratic justifiability of presidentialism. As Josep Colomer (2013) and others have reminded us, presidentialism has deep monarchical roots. Maybe it is time for us to think about how we can separate what is good about presidentialism from what is dangerous for the quality and survival of democracy. The analysis of semi-parliamentarism would not be a bad place to start.


Colomer, Josep M. 2013. “Elected Kings with the Name of Presidents. On the origins of presidentialism in the United States and Latin America.” Revista Lationamericana de Politica Comparada 7:79-97.

Ganghof, Steffen. 2016. “Combining proportional and majoritarian democracy: An institutional design proposal.” Research & Politics 3 (3):1-7.

———. 2018. “A new political system model: Semi-parliamentary government.” European Journal of Political Research (57):261-81.

Ganghof, Steffen, Sebastian Eppner, and Alexander Pörschke. 2018a. “Australian Bicameralism as Semi-Parliamentarism: Patterns of Majority Formation in 29 Democracies.” Australian Journal of Political Science 53 (2):211-33.

———. 2018b. “Semi-parliamentary government in perspective: concepts, values, and designs.” Australian Journal of Political Science 53 (2):264–9.

Samuels, David, and Matthew Shugart. 2010. Presidents, Parties, and Prime Ministers – How the Separation of Powers Affects Party Organization and Behavior. Cambridge: Cambridge University Press.
Taflaga, Marija. 2018. “What’s in a name? Semi-parliamentarism and Australian Commonwealth executive-legislative relations.” Australian Journal of Political Science 53 (2):248-55.

On defining regime types (II) Clauses and Conditions

Steffen Ganghof has recently introduced the concept of semi-parliamentarism. For him, parliamentarism is where all the directly elected chambers of the legislature (whether one or two) have the constitutional right to hold the government collectively responsible, whereas semi-parliamentarism is where only one of the directly elected chambers of the legislature can do so. In other words, under semi-parliamentarism there are two directly elected chambers, but one of them (usually the upper house) does not play a part in the collective responsibility of the PM and cabinet. There are currently two semi-parliamentary countries in the world – Australia and Japan.

I like this definition. It allows us to reliably classify a set of countries merely by applying certain rules to publicly available constitutional information.

Two points. First, I understand why Steffen wants to identify semi-parliamentarism as a separate category, but I wonder if it might not be better to think of it as a sub-category of parliamentarism. This allows us still to see the interesting constitutional feature of the Australian and Japanese cases without losing sight of the basic feature of parliamentarism in both, namely the government’s survival in office is not separate from the legislature. If so, we might think of Australia and Japan as being semi-parliamentary parliamentary regimes. (That is not a typo).

Second, it raises the question of how many consequential classificatory clauses we should include when defining regimes. If we stick to the separate origin and survival of the executive and legislature, we get the three standard categories (presidentialism, semi-presidentialism and parliamentarism). We’ve now added a semi-parliamentary clause. Yet, the semi-parliamentary clause also applies to semi-presidential regimes too. So, the Czech Republic could be classed as a semi-parliamentary semi-presidential regime. Actually, though, we might think of the Czech Republic as a semi-parliamentary premier-presidential semi-presidential regime (i.e. a semi-parliamentary sub-type of the premier-presidential sub-type of semi-presidentialism). We could go further still. There are currently only two semi-parliamentary regimes in the world, but there is a potentially important classificatory difference between them. In Japan, the lower house of the legislature can be dissolved early but the upper house cannot, whereas in Australia there can be a double dissolution of the two houses. So – and bear with me – let’s add a classificatory clause and label Japan a semi-fixed regime and Australia a flexible regime. If so, then Japan would be a semi-fixed semi-parliamentary regime or, perhaps, a semi-fixed semi-parliamentary parliamentary regime. Accordingly, the Czech Republic would be a semi-fixed semi-parliamentary premier-presidential semi-presidential regime.

There is a beautifully Linnaean aspect to this exercise that I find extremely attractive. The classification of the Czech Republic as a semi-fixed semi-parliamentary premier-presidential semi-presidential regime is reliable. It is based merely on the application of certain rules to publicly available constitutional information. There’s another aspect to this Linnaean-type classificatory exercise that could also be attractive. It’s not impossible to think that it might have empirical implications. Perhaps the Czech Republic’s combination of constitutional features is consequential relative to countries with a different combination of features. We would need some theories to tell us what we might expect from any particular combination relative to others. But we might end up with some hypotheses that could be empirically tested.

That said, I doubt that the classification of the Czech Republic as a semi-fixed semi-parliamentary premier-presidential semi-presidential regime is going to catch on very soon. More than that, there is no particular reason why we could not add other classificatory clauses too. Last week, I discussed the addition of a super-majority clause to constitutional classifications. It would be easy to think of other clauses that could be added. However, by the time we combine classificatory clauses, we can quickly end up with very small numbers of real-world examples. The number of semi-parliamentary parliamentary regimes in the world is already only two. The number of semi-fixed (and fixed) semi-parliamentary regimes in the world is just one. The number of semi-fixed semi-parliamentary premier-presidential semi-presidential regimes is also only one (I think). This is not empirically helpful.

The only way to reliably classify regimes is through the application of certain rules to publicly available constitutional information. Only by doing so can we avoid subjective, contestable, sometimes even esoteric country classifications. For sure, if we rely on only a small number of classificatory clauses, the resulting regimes can include a very heterogenous set of countries that render empirical application problematic. However, if we add more clauses, then we have a more homogenous set of countries in each category, but we can very quickly end up with the n = 1 problem that also renders empirical application problematic.

To me, the solution is to accept that there is a basic Linnaean-like classificatory exercise. This exercise is purely taxonomic. It does not necessarily generate categories that are empirically useful, but then that is not its purpose. This is how ‘Duverger’ problem was solved 20 years ago when it came to defining semi-presidentialism. It is also to accept, though, that there is a separate empirical exercise. Here, we need to be pragmatic. Sometimes, the Linnaean-like classificatory categories may be useful empirically, but sometimes they may not. So, we have theories whereby it can still make sense for us to compare the effects of presidentialism relative to parliamentarism, as well as premier-presidentialism relative to president-parliamentarism. However, I cannot imagine a theory whereby it would make sense for us to compare the effects of the heterogenous set of semi-presidential countries relative to anything else. Equally, comparing the effects of semi-parliamentary countries relative to others is problematic when currently the n = only 2. That said, we can, for example, compare the effects of semi-presidentialism relative to parliamentarism conditional upon some measure of presidential power. This condition allows us to disaggregate the heterogenous set of semi-presidential countries in a way that doesn’t undermine the Linnaean-Like classificatory exercise and that could still be empirically useful. By the same token, we can certainly have theories that tell us what the relative effect of semi-parliamentarism might be, even if the number of cases is currently so small that valid conclusions about those effects are difficult to reach.