Abstract
Existing scholarship offers few answers to fundamental questions about the mortality of political parties in established party systems. Linking party research to the organization literature, we conceptualize two types of party death, dissolution and merger, reflecting distinct theoretical rationales. They underpin a new framework on party organizational mortality theorizing three sets of factors: those shaping mortality generally and those shaping dissolution or merger death exclusively. We test this framework on a new data set covering the complete life cycles of 184 parties that entered 21 consolidated party systems over the last five decades, resorting to multilevel competing risks models to estimate the impact of party and country characteristics on the hazards of both types of death. Our findings not only show that dissolution and merger death are driven by distinct factors, but also that they represent separate logics not intrinsically related at either the party or systemic level.
Keywords
Introduction
In organization theory, accounting for organizational mortality has been high on the research agenda for decades (Baum & Shipilov, 2006; Carroll, 1984; Singh & Lumsden, 1990). In particular, organizational ecology has stressed the need to develop nuanced explanations to account for the particularities of distinct types of organizational death—that is, dissolution (disbanding), absorption (acquisition), and merger (amalgation) (Caroll, 1984, p. 85).
Scholarship studying parties in established democracies, by contrast, has devoted little attention to such fundamental issues. A possible explanation for this lack of interest is that, for decades, party systems—especially in Western Europe—have been considered “frozen” (Lipset & Rokkan, 1967), and the demise of those parties constitutive of established party systems has been extremely rare (e.g., Mack, 2010). This, however, does not hold for parties entering their national party systems after the parties constituting the core of the system have fully consolidated. For them, death is a relatively likely event: Almost 40% of the parties entering national party systems since 1968 in the democracies studied here have died by 2016, some after several decades in parliament.
Yet what we assume in the study of long-established major parties shapes the way we approach parties operating in consolidated party systems generally. Thus, as long as new entries are still young, we are tempted to consider them as ephemeral. Once they have been around for a while, their presence (as that of established parties) becomes “natural”. Inevitably, then, fundamental questions about party mortality remain unanswered. 1 The predominant focus on more successful new entries or high-performing new party families—which are less likely to die—has reinforced this, reflecting a general trend in organization research. As Baum and Shipilov (2006) point out, “organizational research ( . . . ) concentrates on the largest and most successful organizations” (p. 57), while we need to study organizations “from the largest and longest-lived, to the smallest and shortest-lived members, over extended periods of time” (Baum & Shipilov, 2006, p. 57), a strategy pursued in organization ecology.
Linking party research (e.g., Pedersen, 1982; Spirova, 2007) to the organization literature studying labor unions and movements (e.g., Hannan, 1988; Zald & Ash, 1966), this article conceptually distinguishes two types of party death: dissolution and merger. This distinction provides the foundation for a new framework on party organizational mortality theorizing three sets of factors: those exclusively shaping either dissolution or merger death, and those shaping mortality generally. We test the resulting hypotheses on a novel data set covering the life cycles of 184 parties from their birth to their (potential) death in 21 consolidated democracies over nearly five decades using hierarchical competing risks models. Our findings not only show that dissolution and merger are driven by distinct factors, as specified by our theoretical framework, but also indicate that the two varieties of party death represent separate logics that are not intrinsically related either at the party or systemic level.
Developing a nuanced framework to study party mortality and testing it across established party systems makes important theoretical and empirical contributions to the literatures on party systems and party organizations alike, not only because—most fundamentally—failure to survive precludes organizations from achieving any other goal (Lowery, 2007, p. 47). Our findings feed into discussions around the crisis of representative democracy. While the decline of mainstream parties as core building blocks of established party systems has been the subject of much scholarly work (e.g., Dalton, Farrell, & McAllister, 2012), the demise of individual parties formed over the last decades has similarly contributed to altering the dynamics in a range of party systems, a driver of party system change that has received little attention in cross-national research (but see Beyens, Lucardie, & Deschouwer, 2016; Bolleyer, 2013; Pedersen, 1982). 2 Furthermore, only formations that contest more than a few elections can perform the functions parties generally are expected to fulfill at the micro- and macro-level, such as preference aggregation, integration, recruitment, and expression (Sartori, 2005, pp. 23-24). Indeed, only such parties are able to strengthen linkages to citizens by effectively representing their demands, promoting civic participation, broadening programmatic offers, and having a direct or indirect impact on policy-making (e.g., Meguid, 2005; Mudde, 2007). Both these party-level and systemic consequences are highly relevant in view of the declining popularity of mainstream parties and the weakening of group affiliations generally, fuelling ongoing debates about the declining legitimacy of representative democracy (Dalton et al., 2012; Dalton & Wattenberg, 2002).
The article proceeds as follows. The next section introduces our conceptualization of the two varieties of party death and derives hypotheses about their determinants. After that, we present the indicators, data set, and methods used to test these hypotheses, and report the results of our empirical analysis. We conclude by summarizing our findings and discussing their broader implications.
Theorizing Party Mortality and Forms of Party Death
Participating in elections by nominating candidates for public office is a key characteristic of political parties, which sets them apart from organizations such as interest groups or social movements (Sartori, 1976). A party survives as long as it takes part in elections, irrespective of the governmental tier—national, regional, or local—on which such electoral participation takes place (Rose & Mackie, 1988, p. 539; Spoon, 2011, pp. 16-17). Consequently, we define a party as dead when it permanently ceases to nominate candidates for any electoral contest as a separate, autonomous organization, which can occur in two ways: dissolution or merger. 3
The term “organizational mortality” or “death” is often used interchangeably with the notion of “organizational failure” (e.g., Baum & Shipilov, 2006; Carroll, 1984; Singh & Lumsden, 1990). Indeed, death tends to be a response to or the result of (perceived or actual) weaknesses of parties from the perspective of core decision-makers, as otherwise no party would permanently give up its organizational autonomy. Yet the notion of failure does not describe all types of party death equally well (Pedersen, 1982, p. 6).
Essentially, we argue that merger death is a state that is actively chosen by elites, while dissolution death stems from a deterioration in support and is a state to be prevented, which is why the latter is more closely associated with the notion of “organizational failure.” This is because “dissolution death” is conceptualized as (predominantly) reactive, resulting from the unwillingness of party followers and elites to stick with their organization as the latter, confronted with a lack of resources, proves unable to achieve its basic goals. “Merger death,” in contrast, is conceptualized as a proactive attempt of party elites (in charge of merger negotiations) to improve the party’s position within a given opportunity structure. It denotes situations in which elites consider giving up their party’s autonomy by forming a new organizational infrastructure out of several constituent parties (each of which ceases to exist in the process) as a means to improve the chances of achieving fundamental goals that the party cannot attain as a separate organization (Spirova, 2007, pp. 22-23). Importantly, each partner needs to bring something valuable to the table that incentivizes the counterpart to be willing to give up its organizational identity in exchange for jointly creating a new organization through merging (Zald & Ash, 1966, p. 334).
In the case of dissolution but also of absorption (in some works treated as a third type of death), a party ceases to exist as a separate organization without being in a position to negotiate an agreement on how and what type of new organization to form (Hannan, 1988, pp. 6-7). In a merger process, in contrast, each participant gives up its organizational identity (Zald & Ash, 1966, p. 336), not just one of them—a situation in which one organization absorbs or “swallows” the other, referred to as an “acquisition” by Waddington, Kahmann, and Hoffmann (2005, p. 3) or as an “unbargained merger” by Ware (2009, pp. 106-107). Both terms highlight that when “being absorbed”—as when dissolving—a party organization finds itself in a position of fundamental weakness, which is why we subsume absorption under the concept of dissolution death.
If dissolution and merger death indeed represent separate behavioral logics underpinned by distinct motivations, each should be determined by (at least partially) different factors. In the following, we therefore specify three sets of hypotheses: one theorizing factors expected to shape dissolution, one on factors shaping merger death, and a final one on party mortality generally.
Drivers of Dissolution Death
Factors that drive dissolution are associated with resource shortages or inferior resource access, which undermine either elite support, a party’s follower base, or both, thereby making parties fundamentally vulnerable. Starting with elites’ perspective, a central motivation to sustain a party · is the access to tangible rewards. Parliamentary seats constitute a type of reward that can be allocated to central figures in the party; usually the first national seat a new formation wins is taken over by the party leader. Thus, a party’s institutional access can be expected to be a major driver of survival (Mayhew, 1974), suggesting a negative relationship between a party’s ability to win parliamentary seats and the risk of dissolution.
Even though electoral performance is a major determinant of a party’s seat performance, it can also be expected to affect dissolution risk in itself. Electoral performance varies across a party’s life cycle, and the extent to which this variation is interpreted as a warning sign regarding the party’s survival chances will likely be influenced by its historical performance. Relative electoral decline, as long as it is moderate, might not affect a party’s prospects for survival. Similarly, the same absolute vote loss can leave one party unaffected while trigger crisis in another, depending on their relative sizes. This is why we expect that only if a party’s support falls below a “critical threshold”—defined in relation to its historical performance since foundation—will its electoral sustainability be put into question by elites and followers. This, in turn, is likely to invite defections, which should increase the hazard of party dissolution accordingly.
In addition, and considering the position of new parties from a systemic perspective, the polarization of the electoral market in which the new party competes for votes can be expected to affect the probability of dissolution as well. As the polarization of the electorate increases, the chances that new parties gather a larger fraction of the vote—typically at the expense of more established parties—rise. This is visible to elites and followers alike, highlighting the vote potential of new competitors and making dissolution less likely.
Consequently, we can formulate our first three hypotheses regarding the risk of dissolution death:
Moving to characteristics of a party’s institutional environment, access to direct state—campaign and organizational—funding, provided at the national level, is considered a major source of income for sustaining basic party functions (e.g., Casal Bértoa & Spirova, 2017; Nassmacher, 2009). New entrants, in particular, often lack the resources to run professionalized campaigns or to make long-term investments in their organization based on their “internal” resources alone (Lucardie, 2000). Access to state funding should thus facilitate activities relevant to elites and followers and render party dissolution less likely.
A similar argument applies to parties’ access to tiers of government other than the national one. Multilevel systems provide parties with additional opportunities to run elections, win seats, and attract media attention; success in those arenas makes them less dependent on their performance at the national level and helps boost their credibility (e.g., Deschouwer, 2003; Spoon, 2011). Hence, moving beyond the national sphere, entering regional government and having seats in the European Parliament (as the most valuable institutional positions accessible to parties on the sub- and supranational levels, respectively) provide parties with access to a range of resources, political visibility, and political experience. Although regional and European government institutions are usually less well-resourced than their national counterparts, representation in alternative tiers of government can help parties endure periods in which access to national resources is scarce and contribute to sustain their basic activities, making dissolution less likely.
Drivers of Merger Death
In contrast to dissolution death, understood as a response to resource shortages undermining party maintenance, merger death is conceptualized as a proactive, strategic decision made by elites to improve their parties’ ability to achieve goals that they cannot attain by themselves—goals transcending the mere maintenance of basic party activities, assuring basic resource access and the ability to pay off party elites. Elites’ strategic considerations can be expected to center around: whether a merger is likely to generate sufficient benefits in relation to its likely costs or not, and whether competing as a separate player remains an attractive alternative, evaluated based on the party’s situation in the context of the party system it currently operates in (Bolleyer, Ibenskas, & Keith, 2016; Ibenskas, 2016).
Which incentives could tempt elites to give up their party’s autonomy for good or, on the contrary, discourage them from doing so? The ultimate locus of power and prestige in political systems is national government. For ambitious politicians, seats in parliament might be little more than a stepping-stone on the way toward the ultimate reward: the takeover of ministerial posts. At the same time, ministerial posts allow for the implementation of party policies, which for policy- (rather than office-) driven elites can be equally—if not more—important (Deschouwer, 2008; Spoon, 2011). Occupying government positions reduces the attractiveness of mergers for at least two reasons. First, a party’s core figures are busy governing, which makes the costs of complex and time-consuming merger negotiations particularly burdensome (Lees, Hough, & Keith, 2010). 4 Second, a merger by definition involves the sharing of resources, which is why mergers are most likely when the · replace resources available (or potentially available) to parties are relatively limited (Ware, 2009, p. 111). Parties in national government are bound to prefer retaining full control over their office-related resources.
Furthermore, if Ware’s (2009) rationale holds, the mere prospect of taking over ministries should create similar disincentives. The nature of the parliamentary party system influences whether and to what extent elites perceive their party to have “coalition potential,” that is, expect it to become a relevant player in coalition negotiations with a chance of entering government (Lane & Ersson, 1999, p. 142; Sartori, 1976). Compared to concentrated systems in which single-party governments are common and minor parties tend to be marginalized, more fragmented multiparty systems not only make it easier for new, usually smaller entrants to gain representation, but also to join government coalitions and exercise real policy influence despite their limited weight (Strøm & Müller, 2009, pp. 43-44). It follows that incentives to merge are more pronounced in concentrated than in fragmented party systems. We can therefore formulate two hypotheses linking the hazard of merging to actual and potential government access, respectively.
Finally, the appeal of a merger is contingent on the negotiation costs and the nature of the compromises needed to reach an agreement among the parties involved (e.g., Marland & Flanagan, 2015; Ware, 2009). Elites will be more willing to initiate merger negotiations and compromise their organizational autonomy (irrespective of “objective” ideological differences, theorized below in H3.2.1) if there is trust between the parties involved, which will be also important to convince followers of the constituent parties to support such a fundamental move. One way of building trust is through preelectoral coalitions before the merger. Preelectoral coalitions constitute a mechanism that familiarizes constituent parties with each other prior to the negotiation process and thereby helps to establish common grounds between them during negotiations (Bolleyer et al., 2016; Ibenskas, 2016; Lees et al., 2010). 5 This leads to the following hypothesis:
Drivers of Party Mortality Generally
It is widely argued that the way organizations are born affects the way in which they “die”. Formative features are expected to leave a “genetic” imprint on organizations, as they capture who is involved in a party and for what reason. Unlike factors theorized up to now, expected to be associated with one mortality type only, these formative features shape the motivation of party founders and followers as well as the resources available to them (Bolleyer, 2013; Krouwel & Lucardi, 2008; Lucardie, 2000; Panebianco, 1988; Spoon, 2009, 2011), thereby affecting parties’ vulnerabilities and strategic considerations alike. This contrasts with external variables defining the opportunity structures parties face, which—depending on the theoretical rationale they align with (i.e., whether they affect the resilience of party support generally or elites’ strategic consideration more specifically)—are expected to shape either dissolution or merger death.
Theorizing the consequences of foundational features characterizing parties at their time of birth raises the question whether these features shape the risk of party death throughout an organization’s life cycle, or only play a role until the organization has reached a certain maturity. Considering such “origin characteristics,” we therefore not only theorize their implications for both dissolution and merger death, but also whether their influence is likely to decline as an organization consolidates.
We expect two origin characteristics to affect party mortality generally: (a) whether a new formation is supported by a promoter organization already established in society in pursuit of a collective goal, which the newly formed party carries into the electoral arena, and (b) whether the party has an ideologically novel profile. Starting with societal rootedness, we expect promoter organizations that sponsor a formation and continue to support it to stabilize the party’s support base in terms of its members and elites. Regardless of whether their cause is fascist, religious, or environmental, they provide access to a pool of committed activists able to fill positions in the new party (Art, 2011; Kitschelt, 1990; Rose & Mackie, 1988). The support of an external organization also decreases a party’s dependence on founding elites, while those elites with preexisting affiliations to the promoter organization (which tend to represent broader collective interests that eventually require political representation) are less prone to building an organization for the sole purpose of advancing their careers. Both aspects make it more likely that the party can outlive founding elites and institutionalize (Bolleyer, 2013). Simultaneously, followers are less likely to defect if their individual interests clash with organizational demands, as affiliations to already established promoter groups function as a first “natural pre-selection mechanism” for an organization that initially has limited capacity to identify and weed out opportunists (Art, 2011).
Following this rationale, parties with societal roots should be less vulnerable and thus less likely to dissolve. This “relative advantage” should hold in the longer term yet gradually decline over time. This is because parties without societal roots determined to build an institutionalized organization without group support or to develop ties to societal groups postformation should eventually be able to catch up (Art, 2011; de Lange & Art, 2011). This leads to the following hypothesis:
The rationale that links societal rootedness with party elites’ disinclination against merging builds on early works that considered linkages to external actors as detrimental to a party’s institutionalization, as such ties constrain its autonomy—understood as its differentiation from other social groupings (Panebianco, 1988; Randall & Svåsand, 2002). As far as a promoter organization perceives a party to be its representative in the electoral arena, it will discourage party elites from compromising the organization’s separate identity. Such a constraint, however, is only likely to affect elite behavior when a party is still weakly institutionalized and dependent on external group support. Therefore, we expect a negative relationship between rootedness and the likelihood of merger death only early in a party’s life cycle, when its institutionalization is still low and elites themselves are more likely to be members of the promoter organization (initially a major recruitment pool).
Introducing a novel ideology into an established party system is also expected to affect both forms of mortality. Ideologically novel formations are more likely not only to mobilize initial support, but also to cultivate lasting (noninstrumental) loyalties on behalf of their followers. This makes it more difficult for competitors to highjack their issues and more likely for novel parties to occupy a separate niche in their party system (Adams, Clark, Ezrow, & Glasgow, 2006; Meguid, 2005; Spoon, 2011). Over time, though, already established parties are bound to make active efforts to adopt neglected policy issues raised by new players, as illustrated by the adoption of green policies or stricter immigration legislation by mainstream parties (e.g., Abedi, 2004; Meguid, 2005). This suggests that, while we expect ideologically novel formations to be less vulnerable to dissolution, the advantages of novelty should prevail earlier in their life cycle and wear off later on. By then, mainstream parties should have had time to integrate the new policy issues in their programs and, being in a stronger position to implement such programs, should be more able to tempt policy-driven followers to switch allegiance.
Besides affecting parties’ vulnerability, we expect ideological novelty to exercise a negative influence on their strategic position and on how elites respond to it. Studies on organizational mergers—of movements and parties alike—have stressed that finding an ideologically compatible partner is a core condition for mergers to take place, facilitating the identification of a common denominator that all negotiating partners can agree on and which each partner can convince their followers to embrace (Ibenskas, 2016; Zald & Ash, 1966, p. 335). The likelihood that a party introducing a novel ideological profile into the party system finds a suitable partner—for example, from the same party family, sharing similar ideological outlooks—is bound to be generally lower than for other parties, rendering mergers a less attractive choice.
Moreover, “being different” from other parties already in the system is a constitutive feature of novel parties’ identity, an attribute party elites may be eager to preserve. As Spoon (2009, p. 618) has convincingly argued regarding Green parties, these formations have an inclination to keep presenting themselves as distinct, even as other parties start to objectively assimilate them by integrating their issues into their own programs. In addition, Adams et al. (2006, pp. 514-525) have shown that niche parties are less responsive to shifts in public opinion, suggesting that their elites exhibit a certain lack of ideological flexibility—an essential trait in any merger negotiation. On the other hand, prior work has suggested that these elites may become more pragmatic over time, and more willing to take advantage of potentially beneficial cooperation opportunities with mainstream parties (e.g., Deschouwer, 2008; Spoon, 2011 p. 21). Consequently, other things being equal, ideological novelty should reduce the likelihood of merger death, although this effect should weaken somewhat over parties’ life cycle.
Table 1 summarizes the hypotheses derived from the three theoretical rationales just presented. For each of the factors affecting parties’ vulnerabilities, strategic capacities, or formative features, the table reports the sign (in parenthesis) of their expected effect on dissolution, merger death, or both, based on H1.1 to H3.2.2.
Theoretical Expectations—Determinants of Types of Party Mortality.
Data and Method
Studying Party Life Cycles in Established Democracies
To test our hypotheses, we study parties that—from their organizational birth until their (potential) death—operated in already fully consolidated democratic settings, that is, that were born after the party systems in which they are embedded had already consolidated. In operationalizing consolidation, we use a period of 20 years after transition to democracy as minimum threshold to ensure that party systems were institutionalized, stable government alternatives could form, and every country experienced alternation in government (Casal Bértoa, 2016; Mainwaring & Torcal, 2006). This is crucial to allow for a meaningful examination of our hypotheses, as—for instance—government access or coalition potential are unlikely to play a role for the strategic consideration of mergers in dominant party systems without viable government alternatives. Thus, for South European democracies, we used the following criteria: in Cyprus (independent since 1960), we cover parties formed since 1980, while in Greece, Spain, and Portugal (democratized since the mid-1970s) we took 1995 as cut-off point. For democracies stable throughout the postwar period, whose party systems were considered “frozen” up to the 1960s yet started to dealign in the latter part of that decade (Bornschier, 2009, pp. 3-7), we included new parties formed from 1968 onward.
In total, we cover new entries across 21 democracies in Western and Southern Europe, North America, and Australasia, with consolidated party systems of sufficient durability to allow for full party life cycles of up to almost five decades. 6 The party organizations studied in these settings face a range of already established (old) competitors, which is critical to ensure that the basic challenges they encountered when trying to assure survival are comparable (a situation that is substantially different if the majority of rivaling parties are new as well). Also, the “rules of the game” in terms of institutional and party system constraints are exogenous to these parties across all the democracies covered, which cannot be assumed in the immediate posttransition phase or with regard to those parties involved in designing a newly democratic system.
To enhance comparability, we distinguished new entrants in consolidated party systems from continuations of already consolidated parties. At the time of its birth, an organization still needs to build a viable, self-sufficient infrastructure consolidated by a (relatively) stable support base. The absence of the latter is an important reason why young organizations tend to be more vulnerable than more mature ones, as discussed by organization research on the “liability of newness” (Stinchcombe, 1965). Consequently, we included in our sample those parties that were built from scratch, formed by minor splits from established parties, or jointly built by both old and new parties—all formation types without a viable infrastructure. 7
We applied these criteria to all new entries that won seats in national parliament at least once, irrespective of their vote share and ideological profile. Importantly, the seat criterion frequently translates in the inclusion of parties with vote shares as low as 0.2% of the national vote (Taagapera, 2002). In a few countries such as Germany, though, thresholds are higher, which is why we also included parties that never won a seat but reached a minimum of 2% of the national vote at least once in their lifetime to increase inclusiveness. 8 As our sample includes all parties that won seats at least once in their life span, parties might reach 2% only occasionally. In fact, 62.5% of the parties in our sample gathered less than 2% of the national vote at some point during the period covered, and almost 30% of them obtained less than 2% in every single election they ran. Altogether, the support for the parties in our sample ranged from 0% to 39.62% across elections. 9
In sum, our selection criteria provided us with a broad sample of 184 parties (3,006 party-year observations), allowing for a comprehensive examination of the factors driving the two varieties of party death considered in our theoretical framework. One concern is that very marginal or minor parties may be underrepresented in our study. Ideally, we would have covered all organizationally new parties that ever participated in national elections, regardless of their vote share. This, however, would have made it impossible to compile the detailed information necessary to rigorously test the determinants of both types of death for each organization included in the analysis. 10 Since marginal parties are typically less able to access some of the resources expected to boost organizations’ resilience (e.g., seats, votes, state funding), our analysis may underestimate the impact of these factors on the dissolution hazard. It is worth noting that this should bias the data against our hypotheses on the relationships between parties’ resources and dissolution death. Hence, our findings are likely to be conservative: If our estimates indicate that these factors are significant drivers of dissolution risk, we would expect their impact to be even more marked had we been able to include very minor formations as well.
Dependent Variables
As mentioned above, we consider a party as dead when it permanently stops nominating candidates for any elections (irrespective of tier) as a separate, autonomous organization. This can occur through two processes: dissolution or merger.
Dissolution death usually takes place through the formal disbandment of a party, through a membership meeting or by a declaration of the leadership (Coakley, 2010). Sometimes parties are absorbed by other parties, for example, they dissolve into a competitor or return to their mother party after a split. Unlike in the case of mergers, in such circumstances the identity of one of the parties remains intact, while the absorbed party fully dissolves (Ware, 2009, pp. 106-107).
To operationalize merger death, we use the change in the merged party’s name as the defining criterion, only considering as mergers cases in which the name of the merged party differs from that of any of its constituent members. The name question is typically a central issue in merger negotiations, and reflects more reliably the differences in parties’ strength than measures based on membership or electoral size (Bolleyer et al., 2016; Lees et al., 2010, pp. 1306-1307). The adoption of a new denomination for the emerging party signals that each merging partner was in a sufficiently strong position to ask the other(s) to give up its(their) own name—a defining feature of an organization’s identity.
Based on these criteria, we found that 71 of the 184 parties (39%) in our sample died during the period under consideration: 47 of them dissolved and 24 merged. 11 As seen in Figure 1, both types of death are more likely to occur while parties are still relatively young, underscoring the need to account for life cycle dependency in our empirical analysis.

Distribution of mergers and dissolutions as function of parties’ age.
Figure 1 also reveals some differences in the timing and frequency of the two varieties of death, providing preliminary support for the notion that dissolutions and mergers are driven by different factors. In addition, the prevalence of each type of death varies within and between democracies (see Figure A.3 in the Online Appendix), suggesting that both party- and country-specific factors are relevant for explaining the risks of dissolution and merger death and underlining the importance of accounting for the multilevel structure of our data.
Independent Variables
Our independent variables comprise measures of the different factors posited to influence party mortality (H1.1-H3.2.2).
Factors expected to affect the risk of dissolution only are captured by six time-varying covariates. Seat Share is measured by the percentage of national seats held by each party in a given year. 12 Electoral Performance captures how the vote share of each party at the most recent national election deviated from the average support from its foundation up to the race immediately prior. 13 Electoral Polarization is operationalized through the vote share of antiestablishment parties, reflecting the ability of such parties to attract electoral support vis-à-vis mainstream parties (Casal Bértoa, 2016). State Funding is a dichotomous variable taking the value 1 if a party qualified for direct state funding (organizational or electoral) in a given parliamentary term, and 0 otherwise. To capture representation in nonnational tiers of government, we include two binary variables: EP Access, which takes the value 1 if the party held any seat in the European Parliament in a given year, and 0 otherwise, and Regional Government, which equals 1 if the party occupied ministerial positions in at least one regional government in a given year, and 0 otherwise.
Based on our theoretical framework, we also expect three time-varying covariates to affect only the risk of merger death: National Government, an indicator for parties holding ministerial posts in any given year; parties’ Coalition Potential, which reflects their prospects of joining national government (increasing with party system fragmentation) and is measured by the effective number of parliamentary parties; and Electoral Coalition, a dichotomous variable that equals 1 if the party was part of a preelectoral coalition in a given year, and 0 otherwise. 14
As for the time-invariant formative features expected to affect both forms of mortality, Societal Rootedness is a dummy for parties whose foundation was supported by one or several identifiable promoter organizations or groups. For most parties in our data set, we relied on the classification by Bolleyer (2013, pp. 40-43, Table 2.2). The remaining cases were coded based on a range of primary sources and secondary literature (see Online Appendix, Section A.2). To code Ideological Novelty, we first identified which parties in our sample belonged to either the new Green or new right family, the only two genuinely new party families that resulted from new cultural conflicts and thus were able to establish themselves across a wide range of consolidated democracies (Bornschier, 2009, p. 7; Mudde, 2007). To ensure that these parties are novel in terms of bringing a new set of issues into their own party systems, we coded formations belonging to these two families as ideologically novel only if they were the first party of their respective family—based on the year of their formation—that entered their party system.
We also include additional control variables in our empirical analysis. Party Age, our “duration covariate” (Box-Steffensmeier & Jones, 1997), accounts for variations in the likelihood of dissolution and merger death over organizations’ life cycle. District Magnitude captures the barriers to representation generated by the electoral system. To allow for possible nonlinearities in the relationship between district magnitude and political representation (Carey & Hix, 2013), we incorporate a quadratic term for this variable in our specifications. Finally, Election Year controls for potential differences in the risks of merger and dissolution between election and nonelection years.
A detailed description of the coding and sources for these variables, along with summary statistics, is presented in the Online Appendix (Section A.2). 15
Estimation Approach
We fit multilevel discrete-time competing risks hazards models (Fukumoto, 2009; Steele, Goldstein, & Browne, 2004) to assess the impact of the explanatory variables on the odds of each type of party death.
Let
where
Moreover, we also examined the possibility that the hazard of one type of death could directly affect the other—for example, that members concerned about their party’s survival may opt for a merger as a step to avoid dissolution, or that elites might consider the possibility of a merger and its likely success before deciding to disband the organization. 16 To do so, we estimated a systematically dependent competing risks model, which allows testing for this sort of strategic behavior (Fukumoto, 2009). Specifically, this variant of the competing risk model replaces (2) with
where
The models were estimated via Markov chain Monte Carlo (MCMC) simulations. Section A.3 in the Online Appendix provides details of the MCMC algorithm.
Results
Table 2 presents posterior summaries for the parameters of alternative multilevel competing risks models capturing the association between each independent variable and the log-hazards of dissolution and merger death. Columns (1) and (2) correspond to a model that specifies the log-baseline hazards as quadratic functions of Party Age. Columns (3) and (4) adopt a more flexible parametrization for
Posterior Summaries for the Parameters of Competing-Risks Models.
The table reports point estimates (posterior means) and 90% highest posterior density (HPD) intervals (in parentheses) for the parameters.
Posterior summaries for the yearly age indicators of the piecewise-constant model (columns 3-4) are reported in the Online Appendix (Figure A.4).
EP= European Parliament
Consistent with our theoretical framework, most predictors of interest have a systematic influence on either the log-hazard of dissolution or of merger only, underscoring the importance of conceptualizing the different varieties of death and theorizing their determinants accordingly. Interestingly, while Figure 1 seemed to suggest that the risks of dissolution and merger exhibit (negative) duration dependence, the results in Table 2 indicate that Party Age does not systematically affect either type of death after controlling for other observed and unobserved party- and country-level factors (see also Table A.4 and Figure A.4 in the Online Appendix).
To test the hypotheses formulated in “Theorizing Party Mortality and Forms of Party Death” section, Figure 2 reports the percentage change in the hazard probabilities

Expected change in
Starting with the factors expected to affect the risk of dissolution only, the estimates in Figure 2 indicate that a one-standard deviation increase in the proportion of parliamentary seats captured by the average sample party (approximately 5.4%) reduces
It is worth noting that Electoral Performance ceases to have a significant impact on
The evidence for our Electoral Polarization Hypothesis (H1.3) is less conclusive. Even though the relationship between the vote for antiestablishment parties and
Our Multilevel Hypothesis (H1.5), on the other hand, is decidedly supported by the data. Everything else constant, the odds of dissolution drop by almost 80% for parties holding seats in the European parliament, and by more than 60% for parties with representation in the regional government. This finding echoes previous research on multilevel party politics, which stresses the advantages enjoyed by parties that access institutional positions in government tiers other than the national one (Deschouwer, 2003; Spoon, 2011).
Overall, these results substantiate our claim that resource-related factors affecting parties’ vulnerability are generally associated with a reduction in the hazard probability of dissolution, not of merger. The main exception is State Funding: Contrary to hypothesis H1.4, we do not find consistent differences in the likelihood of dissolution between parties with and without access to direct state funding. In fact, this variable—expected to boost organizational resilience but not to influence strategic merger decisions—has a negative impact on

Influence of state funding on the risk of merger death as a function of party age.
This finding can be rationalized by drawing on life cycle approaches theorizing the changing demands on and orientations of parties’ leadership as organizations move from an earlier “organization phase,” when funding is central to build a party infrastructure, to an “institutionalization phase” in which the organization has stabilized and leaders can focus on broader strategic goals such as assuring government access (Harmel, 2005, p. 121; Harmel & Svåsand, 1993, p. 74). Following this logic, elites’ ambitions and strategic considerations reorient themselves as an organization consolidates, so that simply having access to resources allowing parties to preserve their identity may no longer be enough to prevent them from taking advantage of potentially beneficial mergers.
Turning our attention to the variables posited to influence merger death only, the results in Figure 2 confirm that factors shaping parties’ strategic position in the political system and their ability to access central rewards tend to be associated with changes in
Unlike variables capturing either parties’ vulnerabilities (relevant for dissolution death) or strategic capacities (relevant for merger death), “origin characteristics” were anticipated to affect both dissolution and merger risk—although their influence on the two types of death was expected to wane over parties’ life. Echoing this, the estimates for Societal Rootedness in Table 2 and Figure 2 imply that, everything else equal, the average risk of dissolution is about twice as high for parties without societal roots than for those that have ties to promoter organizations since their foundation. Furthermore, and in accordance with hypothesis H3.1.1, the contribution of Societal Rootedness to parties’ resilience is particularly marked early in their life but tends to fade away gradually over time. This is illustrated in Figure 4, which allows the impact of formative features on the log-hazard probabilities of dissolution and merger death to vary over parties’ life cycle.

Impact of formative features on the log-hazards of dissolution and merger throughout parties’ life cycle.
As seen in the upper-left panel, once a party has been active for more than three decades, whether or not it was formed with the support of a promoter organization is no longer a relevant determinant of dissolution risk. With the passage of time, parties without similar ties had the opportunity to institutionalize their organization or create beneficial relationships with societal groups postformation, thereby reducing the competitive advantage associated with having societal roots from the beginning (de Lange & Art, 2011).
The evidence in the upper-right panel of the figure, in turn, conforms to our Societal Rootedness Hypothesis — Merger Death. (H3.1.2): Parties formed with the support of preexisting societal organizations are significantly less likely to merge than formations born without external backing, but only in the first decade of their existence. After that, differences in the likelihood of merging between parties with and without societal roots become statistically indistinguishable. The relatively moderate and ephemeral impact of Societal Rootedness on the probability of merging helps explain the null marginal effect of this variable on
Our findings regarding the influence of Ideological Novelty on both forms of party death are also aligned with the theoretical expectations detailed earlier. The estimates in Figure 2 indicate that, other things equal, being the first organization representing a new party family in a party system is associated with a 0.92 percentage point decline in the probability of dissolution throughout the average organization’s life cycle, and with a one point decrease in the likelihood of merger death. The negative impact of Ideological Novelty on the odds of dissolution is consistent with the argument that parties with a new distinct ideological profile are better able to mobilize initial support and to defend a specific niche in their party system than other new parties (Adams et al., 2006; Meguid, 2005). At the same time, the negative association between Ideological Novelty and
The lower panel of Figure 4 complements these “static” results, showing that Ideological Novelty significantly reduces the risk of dissolution only in the first—approximately 20—years of a formation’s existence, as stated in hypothesis H3.2.1. By then, mainstream players might have already been able to integrate the issues raised by ideologically novel parties into their own programs (Meguid, 2005; Rovny & Edwards, 2012), and thus having a novel ideological profile ceases to be a key driver of dissolution risk. Although the impact of Ideological Novelty on the (log) odds of mergers also becomes eventually indistinguishable from zero, in agreement with hypothesis H3.2.2, this effect is much more persistent. In fact, its magnitude tends to grow somewhat (in absolute terms) over time, as the party’s identity has consolidated and been settled in the mind-set of its followers. Earlier research has emphasized that ideological differences are indeed an important obstacle to mergers (Ibenskas, 2016). While mainstream parties might adopt some of the newcomers’ issues in their programs or new parties from the same family may arise, the evidence in the lower-right panel of Figure 4 suggests that ideologically novel parties may profit from their distinct ideological profile throughout much of their life, which reduces the attractiveness of mergers for these parties’ elites.
Altogether, our analysis uncovers systematic differences in the mechanisms underlying the two varieties of party death. While dissolution is primarily associated with factors affecting a party’s vulnerability (Seat Share, Electoral Performance, Electoral Polarization, EP Access, and Regional Government), merger death is mainly driven by factors shaping the party’s strategic position in the political and party system (Government Participation, Coalition Potential). Only fundamental formative features (Societal Rootedness and Ideological Novelty) influence both forms of mortality, although their impact becomes less marked over time. Moreover, we do not find evidence that unmeasured characteristics of the parties or polities under study simultaneously drive both forms of mortality either: As shown in the Online Appendix (Table A.5), the residual correlation between
Concluding Remarks
Many political parties that have entered established party systems over the last decades have died, making party mortality—along with the decline of long-established parties—an important driver of endogenous party system change in advanced democracies (Lupu, 2015). Yet as our article has shown, parties die in different ways for different reasons. Most factors theorized in the new theoretical framework presented here affect the risk of either dissolution or of merger only, substantiating our distinction between dissolution death as a party’s response to its vulnerability and merger death as a strategic decision to improve its position in the party system.
By providing a more comprehensive and nuanced characterization of the reasons why parties die, our research not only closes a notable gap in the party literature, but also contributes to extant research on party organizations and party systems. Understanding the determinants and mechanisms behind party demise provides valuable insights about the nature of those parties that survive and about the dynamic transformation of party systems. More generally, as survival is the precondition for fulfilling any function assigned to organizations in the political process (Lowery, 2007), enhancing our knowledge about the “existential and environmental conditions” underlying party death helps identify the main threats to parties’ ability to perform their core functions—among them, the expression of societal demands and the representation of citizens’ interests (LaPalombara, 2007, pp. 147-149; Sartori, 2005, pp. 23-28). Hence, our findings have substantial implications for research on democratic representation, and are especially salient in light of current debates about the crisis of legitimacy faced by long-lived democracies (Dalton et al., 2012).
Although—for methodological reasons—our analysis focused on new parties operating in fully consolidated party systems, the relevance of our research transcends these specific parties and settings. Despite recurrent arguments about party decline, long-standing major parties in long-lived democracies and major successor parties in the new Central and Eastern European countries have rarely dissolved (e.g., Grzymala-Busse, 2006; Tavits, 2013). In line with our framework, the resilience of these parties stems largely from their ability to take advantage of extensive infrastructures and historically grown loyalties—equivalent to societal roots—and superior access to state resources. When such parties sacrificed their organizational identity, they typically did so to improve their strategic position, as illustrated by the cases of the Canadian Progressive Conservatives or the Slovakian Smer: For both parties, the decision to merge was fuelled by the desire to enhance their chances of reaching government positions (which they did shortly afterwards), again in consonance with our theoretical expectations. The availability of tangible resources associated with access to seats, state funding, additional tiers of government, or the support of external groups should also decrease the risk of dissolution among parties operating in emerging democracies or during the preconsolidation period of established democracies. In contrast, electoral performance and ideological novelty might play less of a role in more volatile electoral markets and less ideologically structured party systems. Prior research (e.g., Ibenskas, 2016) also suggests that preelectoral coalitions may play a relevant part in building trust among potential merger partners in these more fluid and less predictable settings. Thus, while parts of our theoretical framework may need to be adapted to be applied more widely in future research, our central arguments about distinct types of party death and their drivers are likely to broaden our understanding of party mortality not only within but also beyond established party systems.
Supplemental Material
Data_and_Replication_files,_CPS-17-0254.R1 – Supplemental material for Political Party Mortality in Established Party Systems: A Hierarchical Competing Risks Approach
Supplemental material, Data_and_Replication_files,_CPS-17-0254.R1 for Political Party Mortality in Established Party Systems: A Hierarchical Competing Risks Approach by Nicole Bolleyer, Patricia Correa and Gabriel Katz in Comparative Political Studies
Supplemental Material
Online_Appendix – Supplemental material for Political Party Mortality in Established Party Systems: A Hierarchical Competing Risks Approach
Supplemental material, Online_Appendix for Political Party Mortality in Established Party Systems: A Hierarchical Competing Risks Approach by Nicole Bolleyer, Patricia Correa and Gabriel Katz in Comparative Political Studies
Footnotes
Acknowledgements
We are grateful for the helpful comments and suggestions we received from the anonymous reviewers. We further thank André Bächtiger, Fernando Casal Bértoa, Sarah de Lange, and Raimondas Ibenskas, and the participants in the Forschungskolloquium: Ausgewählte Probleme der vergleichenden Demokratieforschung at the University of Stuttgart, the ‘Peter Mair Legacy Lecture’ of the Research Network on “Elections, Parties and Party Systems” at the 24th International Conference of Europeanists in Glasgow and of the UC Irvine seminar series, who commented on earlier versions of this paper.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research received funding from the European Research Council (STATORG 335890). This support is gratefully acknowledged.
Notes
Author Biographies
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
