The Sartorian School and the government formation approach to party systems

How parties interact has been central to much research on party systems. A concern with interactions between parties underlies the work of Maurice Duverger (Reference Duverger1954), Jean Blondel (Reference Blondel1968), Stein Rokkan (Reference Rokkan1970), Robert Dahl (Reference Dahl and Dahl1966), Gunnar Sjöblom (Reference Sjöblom1968, ch. 8) and Alan Siaroff (Reference Siaroff2019). Giovanni Sartori (Reference Sartori1976: 43–44; see also Mair Reference Mair1997: 206; Wolinetz Reference Wolinetz, Katz and Crotty2006: 52) made this focus on interactions between parties explicit in his conceptualisation of the party system:

This perspective was taken up by Peter Mair (Reference Mair1989) and Gordon Smith (Reference Smith1989), and then developed further by Mair (Reference Mair, LeDuc, Niemi and Norris1996, republished in Mair Reference Mair1997), who became its leading exponent. It has been highly influential, giving rise to the ‘Sartorian School’ of party system analysis, which is characterised by an explicit concern with the system of interactions between parties (Casal Bértoa and Enyedi Reference Casal Bértoa and Enyedi2016: 265; see also Bardi and Mair Reference Bardi and Mair2008: 152–153; Enyedi and Casal Bértoa Reference Enyedi, Casal Bértoa, Carter, Keith, Sindre and Vasilopoulou2023: 30, f.34; Mair Reference Mair, Katz and Crotty2006; Webb Reference Webb2024: 5–17).

Mair (Reference Mair, Katz and Crotty2006: 65) built on Sartori’s definition to argue that ‘the structure of competition for control of the executive’ was central to the concept of the party system. As he acknowledged, he was not the first scholar of party systems to focus on this (Mair Reference Mair1997: 206–207, Reference Mair, Karvonen and Kuhnle2001: 39, Reference Mair, Katz and Crotty2006: 65; see also Wolinetz Reference Wolinetz, Katz and Crotty2006: 56). For example, Rokkan (Reference Rokkan1970) was concerned with patterns of government formation; Dahl’s (Reference Dahl and Dahl1966) combinations of competitive, cooperative and coalescent patterns of opposition had a comparable focus; and Smith (Reference Smith1979) maintained a dual focus on the electoral and governmental arenas, and later presented parties of government as the ‘core’ of the system (Smith Reference Smith1989).

In asking ‘how might differential patterns in the competition for government be understood?’ (Mair Reference Mair1997: 206–207), Mair (Reference Mair, LeDuc, Niemi and Norris1996, Reference Mair, Karvonen and Kuhnle2001, Reference Mair, Katz and Crotty2006, Reference Mair, Gloppen and Rakner2007) honed in on patterns of government formation. In this approach, party systems could be ‘closed’ or ‘open’, depending on the regularity and predictability of interparty relations. Closed systems were those with regular patterns of government formation, while open systems were less predictable (e.g. Mair Reference Mair1997: 211). Mair conceptualised openness in terms of three inter-related facets of government formation: the extent of partisan alternation or turnover in government (closed systems have either no turnover or wholesale alternation; open systems are characterised by partial turnover); innovation in the partisan composition of government (open systems are characterised by novel coalition formulas; closed systems by familiar formulas); and the openness of access to office for parties without government experience (in open systems, these parties gain access to government; in closed systems they do not). As well as predictability, closed systems have been associated with accountability and clarity of choice (for voters), while it has been suggested that open systems are more adaptable in changing circumstances (Casal Bértoa and Enyedi Reference Casal Bértoa and Enyedi2016: 266).

Focusing on government formation addressed a number of concerns in the study of party systems: that not all parties are equally important for the party system (Mair Reference Mair, Karvonen and Kuhnle2001: 37; Sartori Reference Sartori1976: 121–125); that cooperation – and not only conflict – is an important aspect of party systems (see also Casal Bértoa and Enyedi Reference Casal Bértoa and Enyedi2021: 9; Smith Reference Smith1989: 349); and, relatedly, that alliances can be taken into account, as they often structure the system (Casal Bértoa and Enyedi Reference Casal Bértoa and Enyedi2021: 8). Mair further argued that focusing on government formation helped to avoid conflating electoral change with systemic change (Mair Reference Mair1997: 215–216).

Mair’s conceptualisation has been widely applied (see Casal Bértoa and Enyedi Reference Casal Bértoa and Enyedi2016: 268 for a review; Casal Bértoa and Mair Reference Casal Bértoa, Mair, Keman and Muller-Rommel2012). Perhaps most significantly, Zsolt Enyedi and Fernando Casal Bértoa (Reference Enyedi, Casal Bértoa, Carter, Keith, Sindre and Vasilopoulou2023: 37) use it as the basis for ‘the closure approach’ to party system institutionalisation. They develop an additive index of party system closure based on Mair’s three components, measured by the partisan composition of government as it relates to alternation, novelty and access (Casal Bértoa and Enyedi Reference Casal Bértoa and Enyedi2021, Reference Casal Bértoa and Enyedi2016).Footnote ¹

The ‘government formation’ approach to party systems has limitations, and these limitations are acknowledged by its proponents. The principal shortcoming is one of incomplete information about interactions between parties in competition for government. Competition for government involves not only the government that finally forms but also parties and coalitions that fail to win office. These actors are accounted for only indirectly in this approach (insofar as patterns of government formation are rooted in developments in the wider party system), but are otherwise left out (Casal Bértoa and Enyedi Reference Casal Bértoa and Enyedi2021: 21, 23; Smith Reference Smith1989: 352).

This cross-sectional problem of incomplete information is compounded by government formation being an infrequent event, meaning that the durable ‘systemic’ patterns arising from it are evident only after several government formation opportunities (Mair Reference Mair, Katz and Crotty2006: 67). Furthermore, the openness (closedness) of party systems has been operationalised in this approach using ‘change’ concepts (turnover, novelty, access). So, while it describes important aspects of party system change, it is less well-suited to describing patterns of party interactions at one point in time. And although party system scholars have long emphasised the importance of the stability of patterns over time for them to be considered ‘systemic’ (Mair Reference Mair, Katz and Crotty2006: 67; Sartori Reference Sartori1976: 199; Siaroff Reference Siaroff2019: 74; Webb Reference Webb2024: 5), these snapshots – what Siaroff (Reference Siaroff2019) calls ‘party patterns’ – can nonetheless provide valuable information about the structure of party systems.

In summary, the ‘government formation’ approach to party systems has been at the fore in conceptualising and operationalising party systems. It is founded on two ideas: that party systems consist of patterns of interactions between parties, and that cooperation and conflict related to government formation is a very important aspect of interparty interaction. The challenge that the present study takes up is to develop an approach that can describe party systems in a way that focuses on the interactions of parties as they compete for a place in government – in line with the Sartorian School’s approach – while mitigating limitations of the ‘government formation’ approach. I suggest that using information on parties’ coalition preferences can contribute to achieving this goal.

Coalition preferences and party systems

A party’s coalition preference is its position on coalescing with another political actor; that is, whether it is for or against coalescing with them.Footnote ² Conceptually, a coalition preference can be open (the party is willing to coalesce with the other actor) or closed (the party is not willing to coalesce with the other actor). The relationship between any pair of parties can therefore be characterised by mutual rejection (both closed), mutual acceptance (both open) or it can be asymmetrical, where one party is open to coalition, but the other is not (Figure 1; see also Decker and Best Reference Decker and Best2010).

Figure 1.

Coalition Preferences within a Party Dyad (Party A–Party B)

Parties’ coalition preferences (and ‘coalition signals’) have long featured in the study of government formation and interparty cooperation. For William Gamson (Reference Gamson1961: 374, 376), it was a shared ‘coalition strategy’ – an approach to prospective coalitions – that defined actors. Coalition preferences are central to the concepts of ostracism, pariah parties and cordons sanitaires (e.g. Geys et al. Reference Geys, Heyndels and Vermeir2006; Heinze Reference Heinze2018; Jacobs Reference Jacobs2024; Martin and Stevenson Reference Martin and Stevenson2001: 36–37; van Spanje Reference van Spanje2010, Reference van Spanje2018: 6). They can also take the form of positive commitments to govern together (Allern and Aylott Reference Allern and Aylott2009; Golder Reference Golder2006b).

Open coalition preferences can help to put parties in a pivotal bargaining position, while foreclosing coalition options can weaken a party’s bargaining position, turning it into a ‘captive’ partner (Bale and Bergman Reference Bale and Bergman2006). Kaare Strøm et al. (Reference Strøm, Budge and Laver1994: 317) found that the exclusion of certain parties from coalition bargaining was ‘the most striking party constraint’ on coalition formation (see also Mershon Reference Mershon1996: 541). Further empirical analyses showed that coalition preferences help to explain which coalitions form (Debus Reference Debus2008; Druckman and Roberts Reference Druckman and Roberts2007; Martin and Stevenson Reference Martin and Stevenson2001) and that participation in pre-electoral coalitions increased the probability of a coalition forming, the speed at which it formed and its ideological compatibility (Golder Reference Golder2006b; see also Bäck Reference Bäck, Hellström, Lindvall and Teorell2024).

Parties’ coalition preferences also shape voters’ views, including their predictions about which government will form (Bahnsen et al. Reference Bahnsen, Gschwend and Stoetzer2020; Bowler et al. Reference Bowler, McElroy and Müller2022; Jang et al. Reference Jang, Crabtree and Golder2022), their perceptions of parties’ positions (Bahnsen et al. Reference Bahnsen, Gschwend and Stoetzer2020; Falcó-Gimeno and Muñoz Reference Falcó-Gimeno and Muñoz2017; Jang et al. Reference Jang, Crabtree and Golder2022) and their focus on coalitions rather than parties (Gschwend et al. Reference Gschwend, Meffert and Stoetzer2017). Therefore, coalition preferences are an important part of a party’s electoral strategy (Duch et al. Reference Duch, May and Armstrong II2010), and as they structure electoral choices, coalition preferences can shape (de)polarisation between parties and among their supporters. After elections, there is anecdotal evidence that parties sustain electoral losses when they break their promises about coalition formation (Debus Reference Debus2008: 520–521; see also Strøm et al. Reference Strøm, Budge and Laver1994: 310).

There is less empirically grounded knowledge about what shapes parties’ coalition preferences, but ideology is likely to be a key explanation (Jacobs Reference Jacobs2024; see also Golder Reference Golder2006a). This is consistent with the idea that coalition bargaining is more difficult if parties are further apart (Bäck Reference Bäck, Hellström, Lindvall and Teorell2024). Preferences regarding political institutions (e.g. democracy or European Union membership) and differences on legitimate means of achieving political aims (e.g. the use of violence) may be particularly likely to lead to coalition rejections. Other factors that may influence coalition preferences include parties’ histories (including their origins), their previous experiences of working together and their familiarity with each other (Debus Reference Debus2007: 89; Tavits Reference Tavits2008). These positions are underpinned by the preferences of both the electorate and party members, and by personal relations between leaders (Decker and Best Reference Decker and Best2010: 166).

The relevance of coalition preferences to party systems and their importance for interparty relations is widely acknowledged. Steven Wolinetz (Reference Wolinetz, Katz and Crotty2006: 53) identified parties’ ‘willingness to work with each other in government formation and the process of governing’ as key to interparty relations. Mair (Reference Mair1997: 211) noted that ‘the development of a closed structure of competition owes much to the strategies of the established parties’, specifically their avoidance of novel coalitions. And Casal Bértoa and Enyedi (Reference Casal Bértoa and Enyedi2021: 21) stress the role of parties’ approaches to potential coalition partners as a primary explanation for variation in party system institutionalisation. Nonetheless, coalition preferences have not previously been used to operationalise interactions between the parties that make up party systems.

Coalition preferences refer directly to interactions between parties concerning competition for government and can therefore inform the analysis of party systems while allowing us to address some of the main concerns of the Sartorian School. Each pair of parties can have coalition preferences in respect of one another; therefore, coalition preferences represent the reciprocal ‘streams of interaction’ cited by Sjöblom (Reference Sjöblom1968: 174) and later Sartori (Reference Sartori1976: 120). Focusing on coalition preferences emphasises the importance of cooperation between parties, facilitating the identification of units above the party level (i.e. blocs) (Casal Bértoa and Enyedi Reference Casal Bértoa and Enyedi2021: 11–12). And as I demonstrate in the sections that follow, a coalition preference-based approach holds considerable potential for developing quantitative measures of interaction-focused party system parameters, which Enyedi and Casal Bértoa (Reference Enyedi, Casal Bértoa, Carter, Keith, Sindre and Vasilopoulou2023: 38) have identified as a key challenge for the analysis of party systems.

Unlike the government formation approach, an approach centred on coalition preferences includes non-governing parties and, in principle, the universe of possible coalitions. By including information about parties that did not enter government and about alternative government formation possibilities, it provides considerably more information about party interactions than government formation data alone, thus addressing an acknowledged limitation of the government formation approach (Casal Bértoa and Enyedi Reference Casal Bértoa and Enyedi2021: 23). This makes it particularly useful for providing ‘snapshots’ of the system of interactions between parties – that is, for analysing the structure of the system as a whole. Even with a restricted (binary) operationalisation of coalition preferences (which I use below), data on coalition preferences provide richer information than government formation data. Specifically, they tell us not only about pairs of parties that have mutually open preferences (the basis for coalition), but also about asymmetrical coalition preferences. This allows us to tap into the latent potential for coalition formation.

In summary, coalition preferences have not yet been used systematically in the study of party systems. Doing so can satisfy several concerns shared by scholars of party systems, and is consistent with the Sartorian School’s focus on interparty interactions as they relate to competition for government. The next section outlines some choices made in respect of the operationalisation of coalition preferences for the purpose of this analysis, before examining what coalition preferences can tell us about party systems in a hypothetical party system.

Operationalisation: coalition preferences and party system openness

What can coalition preferences tell us about party systems?

Figure 2 describes a hypothetical party system with four parties – A, B, C and D – that vary in size and that can have open or closed preferences towards one another. In all, the system includes 6 party dyads and 12 coalition preferences. The top row in Figure 2 shows that party A is open to a coalition with parties B and C but not with party D. Party B is open to a coalition with all parties. Party C is open to coalition only with party A, as is party D. The ‘streams of interaction’ between the parties in Figure 2 provide examples of mutual acceptance between parties, denoted by an ‘M’ (A↔B and A↔C), mutual rejection (C↔D) and asymmetrical relationships, where one party is open to coalition with the other, but where this is not reciprocated (A↔D, B↔C, B↔D).

Figure 2.

Coalition Preferences in a Four-Party System

From a party system perspective, it is the system of coalition preferences that is of central interest, and specifically how coalition preferences can inform us about the interactions that make up the party system. Using data on coalition preferences, the aggregate openness of the system of coalition preferences can be calculated while taking into account parties’ varying sizes. Open coalition preferences that are reciprocated, such as A↔B and A↔C in Figure 2, are necessary for coalition formation and therefore are particularly significant. The m-openness (mutual openness) of the system captures the extent to which the system is characterised by reciprocally open coalition preferences. It is calculated by dividing the surface area of the cells characterised by mutual non-rejection (marked ‘M’ in Figure 2) by the surface area of all red and green cells (‘nonempty cells’).Footnote ⁴

\begin{equation*}M{-}openness = \frac{\sum (surface\, area\, of 'M - open' cells)}{\sum {(surface\, area\, of\, all\, nonempty\, cells)}}\end{equation*}

M-openness values can range from 0 to 1, and the system described by Figure 2 has an m-openness value of 0.5: half of the (weighted) parties are mutually open to coalition with one another.

A second measure identifies the proportion of coalition preferences that are open to other parties regardless of whether these preferences are reciprocated. This is described as the openness of the system. This measures actual m-openness as well as the potential for m-openness. In Figure 2, for example, parties B and C, which have an asymmetrical relationship, are closer to being able to coalesce than parties C and D, which mutually reject one another. The openness of the party system is calculated by dividing the surface area of the cells characterised by open coalition preferences (all the green cells in Figure 2) by the surface area of all red and green cells.

\begin{equation*}Openness = \frac{{{{\mathop \sum }}{\text{ }}\left( {surface{\text{ }}area{\text{ }}of'open'{\text{ }}cells} \right)}}{{{{\mathop \sum}}{\text{ }}\left( {surface{\text{ }}area{\text{ }}of{\text{ }}all{\text{ }}nonempty{\text{ }}cells} \right)}}\end{equation*}

Here, system openness values also range from 0 to 1, but they will always be equal to or greater than the values for m-openness. The system described by Figure 2 has an openness value of 0.71: 71% of weighted preferences are open.

While quantifying the openness of the party system is of central interest here, this systematic overview of coalition preferences also allows us to make other, secondary observations. Firstly, it allows us to describe some salient system-level characteristics in qualitative terms. For example, in Figure 2, two interaction streams between the parties (i.e. A↔B and A↔C) are open, in the sense that both parties are open to coalition. If we assume that these are the only viable coalitions, this puts party A in a pivotal position. Secondly, it allows us to describe relevant party-level attributes that make up the system: the openness of the party’s coalition strategy (its openness to other parties) and its coalitionability (other parties’ openness to the party). A party’s openness relates to how much of its weighted coalition preference is open. The horizontal rows in Figure 2, for example, tell us that party B’s coalition strategy is completely open, while each of the other parties is, to varying extents, partially closed to coalition with other parties. A party’s coalitionability relates to other parties’ preferences towards that party (the vertical columns in Figure 2). In Figure 2, party A is the most coalitionable, while party D is the least coalitionable. Although they are not our main concern, I refer to these party-level characteristics in the case study, and I return to questions concerning their measurement in the conclusion.

Relying on coalition preferences is associated with at least two potential limitations. One concern relates to the validity of expressed coalition preferences. While the objective data of government formation provide firm evidence of parties coalescing (in the government formation approach), stated coalition preferences are more slippery. They may be strategic rather than sincere, and ultimately they may not be binding. But this limitation should not be overstated, as even when an expressed coalition preference is not sincere, the positions that a party takes are somewhat binding, as parties view breaking promises as being costly (Bäck Reference Bäck, Hellström, Lindvall and Teorell2024: 104, 107; Debus Reference Debus2008: 520–521; Strøm et al. Reference Strøm, Budge and Laver1994: 310; Thomson et al. Reference Thomson, Royed, Naurin, Artés, Costello, Ennser‐Jedenastik, Ferguson, Kostadinova, Moury, Pétry and Praprotnik2017). Another concern relates to the stability of parties’ coalition preferences, since even if coalition preferences are sincere, they may change in a post-election context. This means that pre-election coalition preferences will not always provide a leading indicator of government formation possibilities. I return to these potential limitations in discussing lessons from the case study.

Case selection and data collection

I demonstrate the ‘coalition preference’ approach to party systems by applying it to the case of Ireland (1987–2024).Footnote ⁵ Ireland is a multiparty system that offers change over time in the configuration of party politics, from an era of closed competition for control of government to a more fragmented, open and unpredictable system (Casal Bértoa Reference Casal Bértoa2023; Weeks Reference Weeks, Coakley, Gallagher, O'Malley and Reidy2024). This change was accompanied by considerable variation in the ‘openness’ of its system of coalition preferences (described further below). Such variation allows us to assess the face validity and broader significance of the measures and to calibrate their interpretation against a varied set of political circumstances.

After its institutionalisation in the 1920s and 1930s, the Irish party system was exceptionally stable and predictable. Fianna Fáil was predominant in the electoral, parliamentary and governmental arenas, although it occasionally alternated in government with Fine Gael-led coalitions; until the 1980s, Ireland had ‘the frozen party system par excellence’ (Mair Reference Mair1997: 15, Reference Mair1987). The Fianna Fáil single-party government that held office from 1987 to 1989 turned out to be the final iteration of this pattern, as the party reluctantly entered a coalition with the Progressive Democrats in 1989 and the Labour Party in 1993. A structural decline in Fianna Fáil’s vote share and the emergence of new parties led to a ‘fundamental shift in Fianna Fáil strategy’ (Mair Reference Mair1997: 16): having been the party that could offer single-party government, it became a party that needed coalition partners (Carty Reference Carty and Carty2022: 34; Mair Reference Mair1987). Government formation became less predictable, with more partial turnover, more novelty in the composition of governments, and more parties gaining access to government for the first time (Little and Farrell Reference Little, Farrell, Farrell and Hardiman2021). For voters, this meant less clarity of choice; it was no longer Fianna Fáil versus the rest.

An acute economic crisis that began in 2008 led to increased party system fragmentation and the emergence of a left-of-centre opposition, principally Sinn Féin (Hutter and Malet Reference Hutter, Malet, Kriesi and Hutter2019). This reshaped the party system as a whole and led Fianna Fáil to support a Fine Gael-led minority government from the outside (2016–2020), after which it entered government with Fine Gael for the first time in 2020 (along with the Greens) and again in 2025 (with some Independents).

The data presented below cover 10 elections, 64 parties-at-elections and 376 coalition preferences. All parties that won at least 2.4% of parliamentary seats are included. This threshold is inductively derived: it allows us to include all parties-at-elections that entered government with the exception of the Progressive Democrats in 2007, which won just 1.2% of seats (2 seats of 166) and then re-entered government before the party’s dissolution a couple of years later. It allows broad coverage of the party system, but increasing fragmentation means that this coverage falls over time (see Figure A1 in the Supplementary Material). With the aim of ensuring like-with-like comparisons over time (i.e. similarly broad coverage at each election), I extend the scope of the analysis to smaller parties for elections where coverage resulting from the application of the 2.4% threshold is less than 90% seat share (i.e. elections from 2011 to 2024). Data on seat share are derived from ParlGov (Döring et al. Reference Döring, Huber and Manow2023) and from Irish media sources for 2024.

For this analysis, I have focused on identifying parties’ coalition preferences as the election approaches. Election campaigns provide opportunities to observe parties’ preferences (Decker and Best Reference Decker and Best2010: 171), and coalition preferences held during this period influence government formation (Martin and Stevenson Reference Martin and Stevenson2001). Given our concern with competition for government, I use seat share rather than vote share for the purpose of calculating openness values. Although I measure coalition preferences immediately before an election, I weight them by party size after the election; the openness values, therefore, reflect the situation immediately after each election.

For information on coalition preferences, I relied on party histories, political memoirs, authoritative accounts of elections (e.g. the How Ireland Voted series; election reports in journals), media coverage (including searches in the Nexis news archive for the six weeks before the election), and Marc Debus’s (Reference Debus2007) data on coalition preferences of Irish parties from 1982 to 2002. The sources used are detailed in Table A2 in the Supplementary Material. This approach to data collection is similar to other studies of coalition preferences that rely on case study information (Debus Reference Debus2007, Reference Debus2008: 525; Martin and Stevenson Reference Martin and Stevenson2001: 40; see also Golder Reference Golder2005). I exclude ‘Others’ (non-party independents and some micro-parties) from the analysis; information about their coalition preferences tends to be scarce, and it does not seem justifiable to make a uniform assumption about the coalition preferences of these heterogeneous actors (i.e. that they were all open or closed to coalition with all other actors).

The case of Ireland, 1987–2024

Figure 3 shows the development of the Irish party system’s m-openness and openness values over time. In the late 1980s, the system was rather closed (m-openness 0.16 in 1987 and 0.15 in 1989; openness 0.24 in 1987 and 0.19 in 1989); these values were consistently higher from 1992 to 2016, albeit with substantial variation between elections, ranging from 0.24 to 0.6 for m-openness, and from 0.33 to 0.63 for openness. In 2020 and 2024, m-openness remained within this range (0.41 in 2020 and 0.6 in 2024), but openness rose to 0.7 and then 0.8. The widening gap between openness and m-openness indicates a rising proportion of open coalition preferences that were not reciprocated. The case study that follows focuses on these three periods characterised by different levels of openness: 1987–1989, 1992–2016 and 2020–2024.

Figure 3.

The Development of M-openness and Openness Values in Ireland, 1987–2024

Figure 4a and 4b presents a more detailed picture of the systems of coalition preferences in 1987 and 2024. Changes in the distribution of seats, including the emergence of new parties (e.g. Sinn Féin, the Social Democrats), and changes in parties’ coalition preferences profoundly altered the system in the interim.Footnote ⁶

Figure 4.

(a) Coalition Preferences in Ireland, 1987; and (b) Coalition Preferences in Ireland, 2024

Note: * denotes the smallest parties, displayed in this order: Independent Ireland (2.3%), People Before Profit-Solidarity (1.7%), Aontú (1.1%) and the Green Party (0.6%).

1992–2016: more open but variable

In the period from 1992 to 2016, Fianna Fáil and Fine Gael continued to have closed coalition preferences in respect of one another; this limited the openness of the system, as together, they held between 58% (2011) and 79% (1997) seat share. However, other factors contributed to the oscillation between mostly open and mostly closed systems of coalition preferences during this period (Figure 3). The most important single preference in this regard was the Labour Party’s approach to Fianna Fáil: Labour was open to Fianna Fáil in 1992 and 2002, but closed to them in 1997, 2007, 2011 and 2016.

1992 saw a new departure in patterns of coalition preferences, with an openness value of 0.63 (m-openness = 0.6) (Figure 5). Having been in coalition with the Progressive Democrats since 1989, Fianna Fáil opened up to the possibility of coalition with small parties, while Labour, despite trenchant criticism of Fianna Fáil, did not rule out any potential partners and won a record seat share. These two parties subsequently formed a coalition. This pattern of competition for government was to become the norm, with Fianna Fáil and Fine Gael each seeking to work with smaller coalition partners. The open coalition strategy adopted by Labour in 1992 was also to become common among smaller parties.

Figure 5.

Coalition Preferences in Ireland, 1992

Note: * = Democratic Left (2.4%).

At the end of the ‘Rainbow Coalition’ of Fine Gael, Labour and Democratic Left (1994–1997), Labour rejected the possibility of coalition with Fianna Fáil, while the Progressive Democrats ruled out coalition with Labour, leading to a bipolar choice of potential coalitions: Fianna Fáil and the Progressive Democrats or a renewed Rainbow Coalition. The seat share of the two largest parties (Fianna Fáil and Fine Gael) grew by almost 10%, and Labour’s seats were cut nearly in half. As a result, the system of coalition preferences was again relatively closed (m-openness = 0.23; openness = 0.33), although not as closed as in 1987 or 1989, partly due to Fianna Fáil’s continued openness to coalition with smaller parties (Figure A5 in the Supplementary Material).

2002 (m-openness = 0.54; openness = 0.6) saw the reestablishment of more open coalition preferences by the Progressive Democrats and Labour and the growth of the Greens and Sinn Féin, neither of which ruled out any coalition partners. Nor did larger parties rule out coalition with the smaller parties except Sinn Féin, which nearly all parties rejected. Therefore, the coalition options in 2002 were overlapping Fianna Fáil- or Fine Gael-led coalitions. A similar pattern of overlapping possible coalitions was obtained in 2007, but Labour ruled out a coalition with Fianna Fáil and this, together with the success of the larger parties, which were closed to coalition with one another and with Sinn Féin, depressed openness values (m-openness = 0.24; openness = 0.37) (Figure A7 in the Supplementary Material).

2011 and 2016 saw the electoral consequences of Ireland’s economic crisis, which began in 2008, play out. 2011 (m-openness = 0.4; openness = 0.6) was an exceptionally volatile election in which Fianna Fáil lost the vast majority of its seats and was uncoalitionable to most parties due to its role in the economic crisis. Sinn Féin, which was still rejected as a coalition prospect by all major parties, grew substantially. The uncoalitionability of Fianna Fáil and Sinn Féin (Figure A8 in the Supplementary Material) explains much of the gap that opened up in 2011 between openness and m-openness values (Figure 3, above).

A combination of election results and coalition preferences (Figure 6) led the government formation process to an impasse in 2016 (m-openness = 0.32; openness = 0.41). At the core of the system, Fine Gael and Fianna Fáil still rejected entering government with one another and with Sinn Féin, which became the third-largest party. Sinn Féin on this occasion rejected coalition with Fine Gael, but did not unconditionally rule out coalition with Fianna Fáil; rather, they said that they would coalesce with Fianna Fáil if they (Sinn Féin) were the senior coalition partner. Ten weeks of coalition bargaining produced a coalition government of Fine Gael and the Independent Alliance, supported by Fianna Fáil in a confidence-and-supply arrangement.

Figure 6.

Coalition Preferences in Ireland, 2016

Note: * denotes the smallest parties, displayed in this order: I4C (2.5%), the Social Democrats (1.9%), the Green Party (1.3%).

After 2020: towards an open system?

In 2020 and 2024, openness was higher than it had ever been, but in several instances, open preferences were not reciprocated, meaning that the gap between openness and m-openness values widened. 2020 (m-openness = 0.41; openness = 0.7) was marked by further fragmentation and a transformation of the system of coalition preferences (Figure 7), as Sinn Féin, adopting an open coalition strategy, became the second-largest party in the lower house of parliament (the Dáil). Fine Gael was open to the possibility of a coalition with Fianna Fáil for the first time, but this openness was not reciprocated. The – now numerous – small parties generally adopted open coalition strategies, while Fianna Fáil, Fine Gael and Labour (collectively holding less than half of the seats) continued to reject Sinn Féin as a coalition partner. Fine Gael’s openness to coalition with Fianna Fáil facilitated the formation of the novel Fianna Fáil–Fine Gael–Green coalition, which this time took 20 weeks to form, in the context of the early stages of the COVID-19 pandemic.

Figure 7.

Coalition Preferences in Ireland, 2020

Note: * denotes the smallest parties, displayed in this order: Solidarity (3.1%), Aontú (0.6) and I4C (0.6%).

In 2024 (see Figure 4b and O’Malley Reference O’Malley, Gallagher, O’Malley and Reidy2025), the system had an even higher openness value of 0.8, while m-openness (0.604) was marginally higher than it had been in 1992 (0.600). Having governed with Fine Gael since 2020, Fianna Fáil was now open to a coalition with its historic rival. Fianna Fáil and Fine Gael still rejected a coalition with Sinn Féin, but the very few other ‘closed’ coalition preferences in the system related to small parties and so had only a small impact on systemic openness.