Will it last? This is one of the main political questions citizens, pundits and politicians ask and debate on when a new government is sworn in, generally following an election or a cabinet reshuffle. This question has raised considerable concern also for political scientists, who have developed, since the 1960s, an established research agenda in comparative politics on the issue of government stability, examining what factors underpin the longer or shorter tenure in office of ruling parties. Past contributions have already shed important light on multiple related aspects – for example scrutinizing the impact of specific cabinet attributes or critical exogenous shocks (Strøm et al. Reference Strøm, Browne, Frendreis and Glieber1988). However, within this broad topic, the sociodemographic characteristics of rulers, particularly their age, have received relatively limited attention in comparative political research. We argue that the age profile of prime ministers and ministers deserves further investigation in relation to government stability. This article thus scrutinizes the impact of the age of prime ministers and ministers on the stability of governments in 21 established democracies.
We specifically examine how rulers of different age groups affect the risk of a government’s termination, particularly in discretionary situations such as voluntary resignations, conflict-based collapses and early election calling. By doing so, we seek to uncover whether younger prime ministers and ministers are more successful in maintaining stable governments compared to their counterparts. Throughout this article, we define younger rulers as individuals under the age of 45, middle-aged rulers as those between 45 and 60, and older rulers as those over 60. These thresholds reflect the age brackets used by the Inter-Parliamentary Union (2021) to define youth in politics. The motivations for this study are also grounded in additional considerations. Recent scholarly work emphasized that young people are severely underrepresented in electoral contests (e.g. Stockemer et al. Reference Stockemer, Thompson and Sundström2023) and representative institutions, both in parliament (Angelucci et al. Reference Angelucci, Carrieri and Improta2024; Stockemer and Sundström Reference Stockemer and Sundström2023; Sundström and Stockemer Reference Sundström and Stockemer2018) and in cabinet (Stockemer and Sundström Reference Stockemer and Sundström2021, Reference Stockemer and Sundström2022). The absence of youth in politics may significantly undermine the quality of democratic systems as it poses serious issues concerning the descriptive and substantive representation of such categories (Corcoran Reference Corcoran2017; Pitkin Reference Pitkin2023). This absence may also contribute to increased feelings of detachment and apathy in young people, further distancing them from political participation (e.g. Cammaerts et al. Reference Cammaerts, Bruter, Banaji, Harrison and Anstead2014).
Previous studies indicate that government stability matters because it affects democracies’ capacity to address long-term challenges (Improta and Mannoni Reference Improta and Mannoni2024). In this sense, exploring how sociodemographic traits – such as age – impact stability has both theoretical and practical relevance. If, as shown in studies on gender (Wängnerud Reference Wängnerud2009), descriptive representation affects substantive outcomes, then the same may hold for age. Younger leaders may be more inclined to embrace long-termism and future-oriented perspectives (Lees and Praino Reference Lees and Praino2024), contributing to more durable governments.
What happens when young people are successful in entering political institutions? In particular, what happens when they are capable of reaching the highest arena – the governmental one? We begin to answer this question by scrutinizing the impact of age on government stability. Considering the difficulties encountered by younger politicians, we argue that younger rulers face augmented incentives to promote stability in office compared to their older counterparts. First, younger rulers may seek stability in office as part of their long-term political career-building goals, aiming to secure a more durable presence in representative institutions (Fjellman and Rosén Sundström Reference Fjellman and Rosén Sundström2021; Haratsis et al. Reference Haratsis, Hood and Creed2015). Second, older leaders may face a reduced incentive to prioritize stability, given the potential issues of accumulation and overload. That is due to the cumulative psychological and cognitive pressures associated with long political careers. Research shows that senior politicians often operate under chronic stress, face decision overload and may rely more heavily on heuristics to cope with time-constrained, high-stakes environmentsFootnote 1 (e.g. Flinders et al. Reference Flinders, Weinberg, Weinberg, Geddes and Kwiatkowski2020; Weinberg Reference Weinberg2022; Wynn et al. Reference Wynn, Smith and Killen2024). While such experience-based shortcuts may be functional, they can also reduce flexibility and responsiveness, both of which are crucial to sustaining coalition governance (e.g. Powell Reference Powell2004). In this regard, Joseph A. Schlesinger (Reference Schlesinger1966: 9) argued that ‘the older a politician the less chance he has for promotion and the less likely he is to harbour ambitions to advance’.
Along these lines, this study aims to address three interrelated research questions:
1. How does the age of prime ministers impact the stability of governments?
2. Do cabinets with younger ministers exhibit greater stability than those composed of older politicians?
3. What role do age-related factors play when controlling for classic determinants of government stability?
To investigate the impact of age on government stability, this article employs survival analysis, specifically the Cox proportional hazards model. Before being utilized in political and social research, this method has been traditionally used in medical research (e.g. Motzer et al. Reference Motzer, Escudier, Tomczak, Hutson, Michaelson, Negrier, Oudard, Gore, Tarazi, Hariharan and Chen2013). Then survival analysis became the standard in government stability investigation (King et al. Reference King, Alt, Burns and Laver1990), as it is well suited for examining the duration of time until a specific event occurs – in this case, the termination of a cabinet. Governments, much like individuals, face a ‘hazard’ or risk of failure, and survival analysis allows us to measure how different factors, such as the age of rulers, influence the risk of termination over time.
In this study, we hypothesize that younger prime ministers and ministers,Footnote 2 who have more at stake in terms of career prospects (Alesina et al. Reference Alesina, Cassidy and Troiano2019; Schlesinger Reference Schlesinger1966), may prioritize stability and long-term-oriented ruling, thereby reducing the likelihood of government termination. To test the study’s hypotheses, we leverage an original dataset comprising information on governments in 21 established democracies (Western Europe and Israel),Footnote 3 adding information on the rulers’ age from the Worldwide Age Representation in Cabinet (WARC) dataset (Stockemer and Kolodziejczyk Reference Stockemer and Kolodziejczyk2024). All selected countries feature either parliamentary or premier-presidential systems in which executive power is primarily vested in the prime minister.
The remainder of this article proceeds as follows. In the next section, we provide a review of the literature on the age of rulers and on government stability. We then present the hypotheses of the study. The following section illustrates the data and methodology, detailing the country sample, variables and estimation utilized. Then we offer an overview of temporal and national trends, followed by the results of the empirical analyses. The final section concludes the article, summarizing its findings and contributions, and suggesting avenues for future research.
Theoretical background
Age of rulers
Age has traditionally received considerable attention in political science. This is unsurprising, as it is a sociodemographic characteristic crucial for understanding various topics in political science research.Footnote 4 In light of the scholarly interest in this matter, we know much about how different political phenomena are influenced by or vary across different age groups.
For instance, as regards the role of politicians’ age, a flourishing research agenda has started to examine the issue of youth representation in political institutions, documenting a worrying problem of underrepresentation (for an extensive review, see Stockemer and Sundström Reference Stockemer and Sundström2022). A growing body of research emphasizes that a higher presence of young politicians in terms of candidacy, legislative representation and cabinet responsibility would be beneficial for a democracy, based on the perspective of social group representation (Norris and Franklin Reference Norris and Franklin1997). For instance, having more young representatives (i.e. descriptive representation) could contribute to greater intergenerational justice (Tremmel Reference Tremmel2009), as well as increased youth turnout (e.g. Angelucci et al. Reference Angelucci, Carrieri and Improta2024) and electoral support for younger politicians (Sevi Reference Sevi2021).
The existing scientific research on the age of politicians can be broadly categorized into three main areas (Stockemer and Sundström Reference Stockemer and Sundström2022). First, studies have highlighted the prevalence of older politicians in representative institutions as limiting the presence of youth to often single digits, yet without providing empirical tests of such assumptions (Blondel Reference Blondel2016). Although the theory makes intuitive sense, empirical evidence for this type of consideration is scarce. Second, research based on case studies confirms the overrepresentation of middle-aged and older politicians – see for example Ransford van Gyampo on Ghana (Reference Van Gyampo2015) and Rainbow Murray on France (Reference Murray2014). In more detail, these studies reinvigorate the conclusion that the age group between 45 and 60 constitutes the largest share of elected politicians, and that the percentage of young rulers tends to be limited. Third, some comparative analyses have investigated the issue by comparing age representation in different national legislatures. The derived findings further confirm the overrepresentation of older rulers (Narud and Valen Reference Narud and Valen2000).
Moreover, alongside theoretical, case-study and comparative approaches, other scholars have investigated the underpinnings of the observable variations when it comes to youth representation. On the one hand, scholars have demonstrated that electoral systems play a role, with proportional representation electoral systems making it easier for young politicians to enter representative institutions (Joshi Reference Joshi2013; Stockemer and Sundström Reference Stockemer and Sundström2018). Additionally, Mona Lena Krook and Mary Nugent (Reference Krook and Nugent2018) noticed that the age barriers for candidacy represent a crucial factor when it comes to youth representation in institutions. Age requirements for candidacy can indeed discourage young individuals from running for office, thus reinforcing the overrepresentation of older people in representative assemblies and governments. In Krook and Nugent’s work (Reference Krook and Nugent2018), legislative bodies with stricter age requirements are found to have higher average ages among their members.
While the parliamentary representation of younger politicians has been deeply investigated, the governmental arena remains quite an uncharted area. In their book on youth representation in parliaments, candidacies and cabinets, Daniel Stockemer and Aksel Sundström (Reference Stockemer and Sundström2022) started to fill this gap, providing an in-depth study of youth presence in government. However, there are still several research questions to address regarding the relationship between young individuals and their representation in cabinet positions. As Stockemer and Sundström (Reference Stockemer and Sundström2022) noticed, the study of the lack of representation of youth in the executive branch is severely underdeveloped in political research as neither international organizations nor the academic literature have produced any systematic examination of this matter. As they argue, ‘this is surprising, given that a position in the cabinet is the highest office a politician can access’ (Stockemer and Sundström Reference Stockemer and Sundström2022: 89).
By scrutinizing the impact of younger rulers on the stability of governments in a comparative perspective, we aim to contribute to this research agenda, starting to fill some related gaps.
Government stability research
The stability of governments in parliamentary democracies has been a relevant concern for political scientists since the very beginning of the discipline (Laver Reference Laver2003). Scholars have devoted considerable attention to the matter, exploring in particular institutional (Fernandes and Magalhães Reference Fernandes and Magalhães2016; Hazan and Rasch Reference Hazan, Erik Rasch, Hazan and Rasch2023; Lijphart Reference Lijphart1999; Rubabshi-Shitrit and Hasson Reference Rubabshi-Shitrit, Hasson, Reuven and Bjorn2023; Schleiter and Morgan-Jones Reference Schleiter and Morgan-Jones2009), political (e.g. Chiru Reference Chiru2015; Saalfeld Reference Saalfeld, Strøm, Müller and Bergman2008) and economic factors (e.g. Hellström and Walther Reference Hellström and Walther2019; Warwick Reference Warwick1992) influencing the lifespans of governments.Footnote 5 As a result of previous scholarly contributions, we learned that governments differ in their survival rates depending on, inter alia, the fragmentation and polarization of the political system (Bergmann et al. Reference Bergmann, Bäck, Thomas, Hazan and Rasch2023; Grotz and Weber Reference Grotz and Weber2012), the level of inter- and intra-party conflicts (Improta Reference Improta2024; Marangoni and Vercesi Reference Marangoni and Vercesi2015),Footnote 6 the level of popular support (Walther and Hellström Reference Walther and Hellström2019) and economic conditions like unemployment and inflation rates (Cioffi-Revilla Reference Cioffi-Revilla1984; Pinto Reference Pinto2018).
Government stability research has also evolved considerably over time. In his work, Michael Laver (Reference Laver2003) observed that early 1980s literature emphasized that stable governments were typically characterized by either a single-party majority or ideologically cohesive minimal-winning coalitions operating in a party system with low fragmentation. On the other hand, governments formed by ideologically diverse minority coalitions in fragmented systems were seen as prone to instability. This view formed the foundation of structural attributes theories, which posited that the stability of governments, as measured by their duration, could be predicted deterministically based on the characteristics of cabinets at the time of formation (Strøm et al. Reference Strøm, Browne, Frendreis and Glieber1988). Scholars within this tradition found success in their analyses, as the attributes identified seemed intuitively sound (Laver Reference Laver2003).
While structural attributes theories became dominant in explaining government stability, a contrasting approach, known as the critical events theory, was gaining traction. Pioneered by Eric C. Browne et al. (Reference Browne, Frendreis and Gleiber1986), this approach took a stochastic view, suggesting that governments are subject to ‘critical events’ such as economic crises, wars, scandals or even the sudden death of key officials. One of the first studies applying this approach to government stability was Claudio Cioffi-Revilla’s (Reference Cioffi-Revilla1984) analysis of Italian government survival probabilities, where he demonstrated that Italian cabinets collapsed unpredictably, challenging the deterministic assumptions of structural attributes theorists. Following this, survival models were used to study government downfalls across a broader range of democracies (e.g. Browne et al. Reference Browne, Frendreis and Gleiber1986). According to critical events theorists, government stability was less about cabinet attributes and more about random, exogenous shocks.
However, critical events theories faced criticism due to the lack of (strong) empirical support. In an analysis of 12 democracies, only Belgium, Finland, Israel and Italy supported Browne et al.’s (Reference Browne, Frendreis and Gleiber1986) predictions (Müller et al. Reference Müller, Bergman, Strom, Strom, Müller and Bergman2008). This mismatch led to heated debates between the two research traditions, often referred to as the ‘survival debates’ (Warwick Reference Warwick1992). The most prominent exchange occurred in an influential American Political Science Review article (Strøm et al. Reference Strøm, Browne, Frendreis and Glieber1988), where scholars such as Kaare Strøm, Browne’s research group and Cioffi-Revilla debated the best approach to analysing coalition government duration in parliamentary democracies. Strøm et al. (Reference Strøm, Browne, Frendreis and Glieber1988) focused on addressing the main criticisms made by critical events theorists against the structural attributes approach, the most relevant of which was related to the issue of determinism.
The survival debates’ primary focus was on identifying the most suitable event-history methodology to analyse government duration, rather than on the substantive aspects of this process. This methodological debate was largely resolved with the rise of event-history models (King et al. Reference King, Alt, Burns and Laver1990). Notably, the pioneering work by Gary King et al. (Reference King, Alt, Burns and Laver1990) introduced a novel framework for exploring this topic, commonly known as the ‘unified approach’. In order to address the core methodological issues and bring clarity to the field, King et al. (Reference King, Alt, Burns and Laver1990) proposed a fundamentally stochastic model that nevertheless incorporated the hazard rate of governments as a function of various independent variables, many of which are specific attributes of individual governments (Improta Reference Improta2023; Laver Reference Laver2003). By doing so, King et al. built upon and advanced the early work of Browne et al. by enhancing the statistical theory and methods used to study government stability.
The unified approach underlines that the pattern of government terminations observed resembles a Poisson process, implying that government duration is not assumed to follow a normal distribution but rather is given an appropriate functional form (King et al. Reference King, Alt, Burns and Laver1990). Shortly after the unified approach was introduced, a growing body of literature began refining and expanding upon the strategy proposed by King et al. (Reference King, Alt, Burns and Laver1990). One of the most prominent attempts was made by Arthur Lupia and Kaare Strøm (Reference Lupia and Strøm1995), who developed a dynamic model of discretionary cabinet dissolutions, highlighting the role of transaction costs in shaping these dynamics (see also later refinements: Lupia and Strøm Reference Lupia and Strøm2008).
After the adoption of the unified approach, subsequent debates focused mainly on the shape of hazard rates (Diermeier and Stevenson Reference Diermeier and Stevenson1999), sparking various methodological discussions on this topic and its implications (see, for example, Diermeier and Merlo Reference Diermeier and Merlo2000; Diermeier and Stevenson Reference Diermeier and Stevenson1999). A key insight offered by King et al. (Reference King, Alt, Burns and Laver1990) was the importance of the government formation stage in determining cabinet longevity. Specifically, they argued that a high number of formation attempts (empirically captured with bargaining duration) reflect a complex bargaining environment, increasing the likelihood of frequent terminations. However, while the unified approach served as a turning point in resolving methodological aspects of the survival debate, it left certain theoretical issues unresolved (Laver Reference Laver2003).
Around the turn of the century, scholars of executive politics began tackling the challenge of linking various stages in the lifecycle of coalition governments to provide a comprehensive theory of government stability. For many years, the stages of a coalition government’s life – namely, its formation, governance and termination – were treated as separate research areas with distinct theoretical and methodological approaches. This compartmentalized view undermined the development of a holistic framework for analysing the multiple stages that characterize both single-party and coalition governments. The first major effort to propose a dynamic perspective on coalition politics came with Strøm et al.’s (Reference Strøm, Müller and Bergman2008) work, Cabinets and Coalition Bargaining: The Democratic Life Cycle in Western Europe. Recently, some advancements on such a dynamic perspective have been offered by Wolfgang Müller et al. (Reference Müller, Bäck and Hellström2024).
However, despite several efforts dedicated to investigating the drivers of government instability, we still know very little about the impact of the prime minister’s or other government ministers’ sociodemographic characteristics (for a notable exception, see Krauss and Kroeber Reference Krauss and Kroeber2021). Filling this gap appears to be a relevant task, as prime ministers and ministers play a crucial role when it comes to governing. That is, they hold a significant share of responsibility regarding policy design and implementation (e.g. Alexiadou Reference Alexiadou2015). Thus, revealing their role concerning the stability in office of the whole executive would allow us to have a better understanding of the factors influencing government stability and also shed some important light on the role of age characteristics on this matter.
Hypotheses
In this article, we hypothesize that governments controlled by younger prime ministers and populated by younger cabinet ministers have a reduced risk of discretionary termination, compared to governments of older rulers. We base our hypotheses on some major conceptualizations of the role of youth in democratic politics. First, we advance a societally driven consideration. Democracies are by definition in flux: changes in political and social institutions, according to modernization theory (Lipset Reference Lipset1959), should reflect a gradual adaptation to the values and needs of a modernizing society. In this vein, young rulers should be viewed as agents of innovation and change, being more attuned to contemporary societal values and expectations (Lipset Reference Lipset1964).
Second, younger leaders should also be more incentivized to promote stable governance – by conducting a less confrontational ruling style, for example, especially with coalition partners – to secure a long political career trajectory. Such concerns, indeed, should be less relevant for older rulers (Schlesinger Reference Schlesinger1966). In conjunction with this, agency theory applied to career-building suggests that individuals with long-term career prospects are motivated to ensure stability to foster a positive reputation (Alexiadou Reference Alexiadou2015; Jensen and Meckling Reference Jensen and Meckling1976). Thus, young prime ministers and ministers may be more motivated when in office to avoid conflictual behaviours that could lead to a premature end of the government, thereby preserving stability as a means to advance their political careers.Footnote 7
Third, government decisions should be deemed as more legitimate when youth representation in cabinet is greater (Norris and Franklin Reference Norris and Franklin1997; Stockemer and Sundström Reference Stockemer and Sundström2022). This is because young people represent a large portion of the population which often remains significantly underrepresented in political institutions (Angelucci et al. Reference Angelucci, Carrieri and Improta2024).
Fourth, previous research shows that younger politicians are often more successful in attracting funds and are also more active when it comes to legislative activities (Alesina et al. Reference Alesina, Cassidy and Troiano2019; Stockemer and Sundström Reference Stockemer and Sundström2022).Footnote 8
All things considered, we should expect stabilizing potential from young rulers. However, we are aware that there are strong reasons to expect quite opposite effects, as young politicians could also be considered as more exposed to risks of instability based on their lack of accumulated experience in office compared to older prime ministers and ministers. Nonetheless, we rely on previous research and existing evidence stressing and documenting that younger politicians face stability-oriented incentives as they expect to have longer political careers and therefore have stronger career concerns (Alesina et al. Reference Alesina, Cassidy and Troiano2019; Fjellman and Rosén Sundström Reference Fjellman and Rosén Sundström2021; Haratsis et al. Reference Haratsis, Hood and Creed2015).Footnote 9 Furthermore, other research maintains (though it has not been empirically tested) that older ministers will be less durable than younger ones as the former have reached the end of their career, while younger ones have larger portions of their career path ahead of them (Bäck et al. Reference Bäck, Persson, Vernby and Helena2009; Blondel Reference Blondel1991; Fischer et al. Reference Fischer, Dowding and Dumont2012).Footnote 10
Table 1 summarizes the hypothesized mechanisms of our study:
Hypothesized Effects of Younger Prime Ministers and Ministers on Government Stability
Based on the presented mechanisms, we differentiate between the impact of the prime minister’s age and the age of individual cabinet ministers. While the expectations are the same for these different roles, we might expect the prime ministerial role to exert a greater impact on government stability as this figure holds the main responsibility when it comes to ruling and the role is often equipped with fundamental institutional instruments to strategically terminate a government (Schleiter and Morgan-Jones Reference Schleiter and Morgan-Jones2009; Strøm and Swindle Reference Strøm and Swindle2002), such as dissolution powers. The hypotheses of this study are as follows:
Hypothesis 1 (Prime Ministers): Cabinets led by younger prime ministers (under 45) exhibit greater stability than those led by older prime ministers (over 60).
Hypothesis 2 (Ministers): Cabinets with a lower average age of ministers face a reduced risk of government termination.
The next section presents the dataset and methodological strategies employed.
Data and methodology
Dataset
This study leverages an original longitudinal multilevel dataset covering information on more than 700 governments in 21 countries, from the 1950s to the 2020s (Table 2).Footnote 11 As regards the variables related to rulers’ age, we utilized the newly published dataset WARC, built by Daniel Stockemer and Kamila Kolodziejczyk (Reference Stockemer and Kolodziejczyk2024). The WARC dataset is a comprehensive source that provides information on, inter alia, the age of prime ministers and ministers. We incorporated such useful information into our original dataset, which includes a range of factors typically associated with government stability,Footnote 12 such as different causes of termination, party system fragmentation, and economic performance (e.g. Warwick Reference Warwick1992).
Descriptive Statistics (Age of Prime Minister and Government Duration) by Country
The main source of information for the cabinets is Fernando Casal Bértoa and Zsolt Enyedi’s dataset (Reference Casal Bértoa and Enyedi2022). We collected information into separate datasets that differentiated between three levels: country level, legislature level and cabinet level. A single cabinet is the basic unit of analysis. To count changes in cabinets, we followed Torbjörn Bergman et al.’s (Reference Bergman, Gerber, Kastner and Nyblade2008) strategy. Therefore, we counted a change of cabinet when (1) the prime minister/head of cabinet changes; (2) when the partisan composition of the cabinet is altered; and (3) when a general election is called. We set the cabinet's start date as the government’s inauguration and the end date as either the general election date or the date of formal resignation of the cabinet, ‘whichever comes first’ (Müller and Strøm Reference Müller and Strøm2000: 11–17).
Dependent variable
As regards the dependent variable (Table 3), we relied on the crucial differentiation made by Erik Damgaard (Reference Damgaard2008: 308) concerning which termination event to consider as scientifically interesting. As not all terminations are equally important in political terms, we opted to focus on discretionary terminations – that is, terminations concerning, for instance, voluntary resignations of a prime minister, early elections, conflict-based collapses – applying a censoring strategy for cabinets yet to experience the event and excluding technical terminations.Footnote 13 To effectively investigate the determinants of government stability, it is crucial to assess whether the explanatory factors influence the most informative types of termination, specifically discretionary ones. For example, cabinets may fall due to regular elections, the death or illness of the prime minister, or other constitutional reasons. These types of termination are categorized as ‘technical terminations’ (Damgaard Reference Damgaard2008: 308). While the failure event is discretionary termination, the analysis time is government duration in days – that is, from the cabinet’s formation to its (discretionary) termination. We treat government termination as a failure event and use Cox proportional hazards regression to estimate the effect of the study’s focal and control variables.
Descriptive Statistics of Government Discretionary Termination by Country
Note: Min and Max values are 0.00 and 1.00, respectively, in every case.
Moreover, it is important to define clearly the basic unit of observation for this study: the cabinet. In their seminal work, Müller and Strøm (Reference Müller and Strøm2000) provided a clear definition of a cabinet. Specifically, a cabinet is ‘the set of politically appointed executive offices involved in top-level national policymaking’ (Müller and Strøm Reference Müller and Strøm2000: 11). In line with Müller and Strøm (Reference Müller and Strøm2000), this study considers prime ministers and ministers as the key government personnel of interest, excluding undersecretaries and junior ministers from the investigation. We believe this conceptualization enhances greater comparability.
In this study, we define government stability as the ability of a government to remain in office over time. Although government stability alone is not sufficient for effective governance, it can be deemed a ‘prerequisite’. Governments struggling to survive are less likely to implement their policies fully (Improta Reference Improta2024), and unstable cabinets with frequent turnover diminish accountability, as voters may struggle to identify the actors responsible for specific policy actions. As Thomas Saalfeld (Reference Saalfeld, Strøm, Müller and Bergman2008) points out, government stability, measured by the time between a cabinet’s appointment and its termination, can be assessed in two ways. The first is actual duration.Footnote 14 The second is ‘relative duration’, expressed as a percentage of its maximum feasible duration (Müller and Strøm Reference Müller and Strøm2000).Footnote 15 Additionally, the constitutional inter-election period (CIEP) must be considered to accurately assess the varying life expectancies of governments. Of course, the first governments of a legislative term tend to have a higher potential duration in office, while later governments face a shorter time horizon. Furthermore, the duration of legislative terms varies across the investigated democracies. For example, Swedish legislatures are legally mandated to last a maximum of four years, whereas Italian legislatures have an additional year by law. Therefore, when studying government stability, we take into account the potential duration of governments (see Appendix H in the Supplementary Material).
Focal variables
The main focal variables in this study are the age categories of prime ministers and ministers. We differentiate between three categories: under 45, 45–60, and over 60. As mentioned, information on the age of prime ministers and ministers were retrieved from Stockemer and Kolodziejczyk’s dataset on age representation in cabinet (Reference Stockemer and Kolodziejczyk2024).Footnote 16 In turn, additional information on rulers used by Stockemer and Kolodziejczyk to build their dataset was found in Jacob Nyrup and Stuart Bramwell’s WhoGoverns dataset (Reference Nyrup and Bramwell2020). We thus base our own data collection on previous, important scholarly efforts.
While we performed robustness tests by examining the impact of different age categorization on government stability, we opted for this primary categorization based on the Inter-Parliamentary Union (2021). However, the OECD (Organization for Economic Co-operation and Development) definitions (2020) establish an under 40 years old threshold for parliamentary representatives. Considering that cabinet positions are higher ranked than the legislature level, we established a slightly higher threshold, changing under 40 to under 45.
For the operationalization, we opted for two strategies. Concerning prime ministers, we assigned the exact age of the prime minister. Regarding the ministerial level, as we do not have information on the individual ministers because of the large-N structure of the dataset, we considered the mean age of governments at inauguration (consistent with Stockemer and Kolodziejczyk Reference Stockemer and Kolodziejczyk2024) in correspondence to each specific cabinet. Therefore, while in the case of prime ministers we adopted a more granular perspective, for ministers we have more aggregate information. Overall, we classified the three categories as follows: prime ministers and ministers under 45 are younger. Those belonging to the 45–60 group are middle-aged. Prime ministers and ministers over 60 are older. We use these labels interchangeably.
Control variables
As government stability research is one of the most developed agendas in comparative politics (Laver Reference Laver2003), we take into account several control variables which have been considered critical when it comes to the survival of cabinets.Footnote 17 First, we control for the impact of the Government Type. In line with Müller and Strøm (Reference Müller and Strøm2000), we differentiate between multiparty minority governments, single-party minority governments, minimal-winning coalitions, oversized coalitionsFootnote 18 and single-party majority, thus from the least stable ruling configuration to the most stable one, according to previous studies (e.g. Saalfeld Reference Saalfeld, Strøm, Müller and Bergman2008). Scholars have indicated, in particular, that minority governments are often more transient than majority configurations (e.g. Krauss and Kroeber Reference Krauss and Kroeber2021). Others have also documented a tenure-securing role of single-party majority cabinets or tight coalitions (Saalfeld Reference Saalfeld, Strøm, Müller and Bergman2008). As regards the operationalization of this variable, we considered the number of parties participating in the cabinet and the legislative seat share controlled. For instance, single-party minority governments are formed by just a single party that, however, does not control the majority of seats in the legislative arena. Minimal-winning coalitions are instead formed by the necessary number of partners to attain the majority of seats. To take into consideration governments’ structural attributes (Strøm et al. Reference Strøm, Müller and Bergman2008), we also controlled for the impact of the level of fragmentation in the cabinet, measured using a version of Markku Laakso and Rein Taagepera's (Reference Laakso and Taagepera1979) fragmentation index.Footnote 19 The original variable we collected is the Number of Effective Parties (in government).
The second cluster of controls we considered concerns the countries’ economic conditions (Hellström and Walther Reference Hellström and Walther2019; Warwick Reference Warwick1992). We tested the impact of economic expansion via gross domestic product growth rate (GDP Growth Rate), insecurity in the job market via the Unemployment Rate, and financial constraints derived from the levels of Government Debt. While higher economic growth and lower unemployment should grant the cabinets a longer tenure, higher debt should lead to an increased risk of discretionary termination.Footnote 20
The third cluster included in the analyses concerns the ‘uncertainty and complexity’ in the bargaining environment (Falcó-Gimeno and Indridason Reference Falcó-Gimeno and Indridason2013). The first indicator considered is Bargaining Duration – which measures the time spent on negotiations by potential coalition partners during the government formation stage of the democratic lifecycle (Strøm et al. Reference Strøm, Müller and Bergman2008). There is limited scholarly agreement on the impact of bargaining duration on government stability. Sona Golder (Reference Golder2010) argues that delays during government formation are detrimental to government stability, whereas other scholars previously suggested that extended negotiations may signal that potential partners are paying closer attention to coalition programmes and the policies resulting from them, which can, in turn, reduce uncertainty and complexity (Warwick Reference Warwick1994). Additionally, we included a variable measuring the fragmentation of the parliamentary arena – that is, the effective number of parliamentary parties, ENPP, as the higher the number of actors in the bargaining environment, the higher the uncertainty and complexity. Indeed, a classic assumption of game-theoretic approaches to the study of government instability is that the presence of several actors in the bargaining process raises transaction costs (e.g. Dodd Reference Dodd2015).
Finally, scholarly efforts on government stability have emphasized the role of ideological characteristics of the cabinets (e.g. Warwick Reference Warwick1994). Therefore, we assessed the influence of the ideology of cabinets by constructing an original index of the right–left position of the cabinet, RILE, as a weighted measure of the governing parties’ position on the right–left spectrum. We thus created the RILE-weighted variable. To do so, we relied on the latest release of the Comparative Manifesto Project (Lehmann et al. Reference Lehmann, Franzmann, Al-Gaddooa, Burst, Ivanusch, Regel, Riethmüller, Volkens, Weßels and Zehnter2024). The coding rules are as follows: for single-party governments, the RILE corresponds to that of the ruling party. In coalition governments, a weighted average of the RILE positions is calculated, taking into account the ministerial seats held by each party in the government.
Estimation
To study the relationship between the age of rulers and the stability of governments, we employ survival analysis.Footnote 21 This statistical method is well suited to analysing time-to-event data, where the event of interest in this case is the termination of a government. As described earlier, the survival analysis framework became a standard technique for studying this matter, and it has been widely used in political science research since the path-breaking work by King et al. (Reference King, Alt, Burns and Laver1990). For the estimation, the equation is as follows:
\begin{align*}
\lambda (t|X) & = {\lambda _0}(t)\ {\text{exp}}{\beta _1}{\text{Middle-aged}} + {\beta _2}{\text{Older}} + {\beta _3}{\text{Govt Type}} \left( 0 \right) + {\beta _4}{\text{Govt Type}} \left( 1 \right)\\
& \quad + {\beta _5}{\text{Govt Type}} \left( 2 \right) + {\beta _6}{\text{Govt Type}} \left( 4 \right) + {\beta _7}{\text{Number of Effective Parties}} \\
& \quad + {\beta _8}{\text{GDP Growth Rate}} + {\beta _9}{\text{Unemployment Rate}} + {\beta _{10}}{\text{Gov Debt}}\\
& \quad + {\beta _{11}}{\text{Bargaining Duration}} + {\beta _{12}}{\text{ENPP}} + {\beta _{13}}{\text{Rile Weighted}}
\end{align*}Here λ(t|X) is the hazard function at time t given the covariates X. λ 0(t) is the baseline hazard function. The βs are the regression coefficients to be estimated, where middle-aged and older represent the dummy variables for the age categories (relative to the reference category, which is younger). Government type includes the variables for different government types (multiparty minority, single-party minority, oversized coalitions, single-party majority). Number of Effective Parties, GDP Growth Rate, Unemployment Rate, Government Debt, Bargaining Duration, ENPP and RILE Weighted are continuous covariates included in the model.
This equation reflects a Cox proportional hazards model where the hazard of termination (failure) is modelled as a function of age category, government type and other covariates. The model accounts for the baseline hazard λ 0(t) and the effects of each predictor through the exponentiated coefficients. The exponentiated coefficients (exp(β)) represent hazard ratios, which quantify the effect of a one-unit change in the covariate on the hazard of termination. The model is typically estimated using maximum likelihood methods, and significance of the coefficients is tested via z-statistics and p-values. To test the validity of our analyses, we performed several robustness tests and diagnostics (before and after performing the survival models). In detail, we first verified the variable distribution (Appendix A in the Supplementary Material). Then we assessed the goodness of fit of the proportional hazards model using deviance residuals, which are relevant in the context of Cox survival analysis (Appendix B). Computations signal that the residuals are symmetrically distributed around zero, indicating absence of systematic bias in the model’s predictions. Moreover, we performed a log-rank test to examine whether there are statistically significant differences between the survival curves of the different age groups. The output of the log-rank test (Appendix C) documents that there are significant differences in the survival curves among the three age categories, confirming that government duration varies significantly based on our differentiation – that is, rulers’ ages are associated with significant differences in government duration. Furthermore, in Appendix D we tested the proportional hazards assumption performing the Schoenfeld proportional hazards T-test. In Cox survival analysis it is important to verify whether the ratio of hazards (i.e. the risk of the event occurring) between groups remains constant over time. The Schoenfeld test shows that the relationship between the variables and the risk of government termination remains proportional over time, meaning that the proportional hazards assumption is met in our model.
Finally, in Appendix E and Appendix G in the Supplementary Material we re-ran the Cox regression analyses by (1) establishing a different age categorization and (2) controlling for additional variables pertaining to different clusters, such as institutional variables (e.g. a constructive vote of no-confidence, Chiru Reference Chiru2015; Rubabshi-Shitrit and Hasson Reference Rubabshi-Shitrit, Hasson, Reuven and Bjorn2023). Having performed these tests, the main results hold.
National and temporal trends
In this section, we show some temporal and national variations regarding the study’s focal variable: the age of prime ministers. In Figure 1, the y-axis represents the mean age of prime ministers, while the x-axis lists the countries under examination. We observe a clear variation in the average age of rulers across different national contexts, with certain countries showing a marked presence of older leaders, while others tend towards a greater younger presence at the head of government.
Age of Prime Minister per Country
Notably, countries such as Israel, Italy, Greece, Cyprus and Switzerland exhibit higher average ages, with prime ministers around 60 years old or older. On the other hand, countries like Denmark, Luxembourg, Spain and Finland have younger prime ministers, with the average age closer to 50. The diversity in the average age of prime ministers across countries indicates different cultural and institutional factors, as documented by the quite uniform pattern followed by Scandinavian countries (see also Stockemer and Sundstrom Reference Stockemer and Sundström2023).
Figure 2 illustrates the average age of prime ministers across different decades, from the 1950s to the 2020s. The y-axis represents the average age of prime ministers, while the x-axis shows the decades.
Age of Prime Minister per Decade
In the 1950s, the average age of prime ministers was considerably lower, with a steady increase in the following decades. The 1970s and 1980s show a slight dip in the average age. From Figure 2, we can observe a non-monotonic trend culminating in a peak of older people in prime ministerial positions in the 2020s. The average age in the 2020s indeed surpasses 60, which is notably higher than in earlier decades and represents a record high.
We may advance two interpretations of the information observable in Figure 2. First, we should take into consideration the changing life expectancies over time. As noted by Kevin Kinsella and Victoria Velkoff (Reference Kinsella and Velkoff2002), life expectancy increased over the decades in Western Europe, hence the post-World War II period should be regarded as a time in which citizens were less likely to reach older ages, which made the higher presence of younger prime ministers more likely. Second, any interpretation of the peak detected in the 2020s should bear in mind that young adults are currently significantly underrepresented in representative institutions (Stockemer and Sundström Reference Stockemer and Sundström2022) and, in particular, parliamentary assemblies are more highly populated by older members (Stockemer and Sundström Reference Stockemer and Sundström2023). Given that in most countries a crucial prerequisite to reach cabinet position is to be an elected representative (e.g. Norris Reference Norris1997), this is likely to result in a lower youth presence in prime ministerial posts.
Analysis
The impact of a prime minister’s age
Moving to hypothesis testing, we start our analysis from Hypothesis 1, regarding the impact of the age of the prime minister. We hypothesized there that younger prime ministers would play a stabilizing role when it comes to governments’ survival in office. Table 4 presents the results of the Cox proportional hazards regression analysis assessing the impact of a prime minister’s age on government stability, with the hazard ratios for two age categories: middle-aged and older, relative to the reference category – that is, younger (under 45). The failure event, as anticipated in the Data and Methodology section, is government discretionary termination.
Cox Regression Results: Impact of Age Category on Government Stability
* Notes: Age categories: younger = under 45; middle-aged = 45–60; older = over 60. AIC, Akaike information criterion; BIC, Bayesian information criterion. *p < 0.05; **p < 0.01; ***p < 0.001.
Table 4 documents that the hazard ratio for prime ministers in the middle-aged category is 1.08, with a standard error of 0.32 and a z-value of 0.29. The p-value of 0.77 indicates that the age of prime ministers belonging to the age category of middle-aged (45–60) does not play a significant role in the risk of government termination. Prime ministers in this age group do not face a notably higher or lower risk of termination compared to the reference group of leaders aged under 45.
With respect to the older category (over 60), the hazard ratio is 1.94, with a standard error of 0.55 and a z-value of 2.32. The p-value of 0.02 indicates that the destabilizing role of this age group is statistically significant. This implies that prime ministers over the age of 60 face almost twice the risk of government discretionary termination compared to those in the under-45 group.
We also present the results of an additional Cox regression analysis (Table 5) scrutinizing whether the impact of a prime minister’s age remains significant even when controlling for important factors examined by government stability scholars to explain increased or reduced risks of cabinet termination.
Cox Regression Results: Impact of Age Category and Control Variables on Government Stability
* Note: Ref. category Government Type: Minimal-winning coalition. Age categories: under 45, 45–60, over 60. AIC, Akaike information criterion; BIC, Bayesian information criterion. *p < 0.05; **p < 0.01; ***p < 0.001.
Several interesting points emerge from Table 5. To begin with, Cox regression analysis confirms that middle-aged prime ministers do not have a significant impact on the risk of government discretionary termination relative to younger rulers. On the other hand, the hazard ratio for the category of older prime ministers is 2.00 and the p-value is 0.02, which is, again, statistically significant. This result confirms the findings observed in Table 4, indicating that prime ministers over the age of 60 robustly face twice the risk of government discretionary termination compared to rulers under 45 years of age. On the whole, these results provide support for the first hypothesis. While the difference between younger and middle-aged leaders is not statistically significant, the overall pattern suggests that younger age is associated with greater government stability, primarily due to the heightened instability linked to older leadership.
Furthermore, in relation to the control variables, the Cox analysis confirms the relevance of ‘classic’ stabilizing factors. For instance, oversized coalitions – namely, governments with a large and unnecessary number of parties (irrational ruling solutions according to game-theoretic approaches, e.g. Riker Reference Riker1962) – are at a significantly higher risk of discretionary termination. This, of course, reflects the complexity of managing diverse political goals and interests within a large coalition. Also, results show that single-party majority governments are less likely to experience discretionary termination, in line with their ingrained degree of stability derived from single-party rule (with conflicts limited to the possibilities of infighting). With regard to the structural attributes of the cabinet, the number of effective government parties exerts a crucial role in increasing the risk of termination, as more parties in government may introduce additional transaction costs that would lead to conflict and disagreement, ultimately eliciting premature collapse of the government.
Moving to the other variables, our results show that economic conditions, fragmentation in parliament and cabinet ideology do not play a significant role, whereas bargaining duration is statistically significant. Longer bargaining durations seem to reduce the risk of termination, indicating that protracted negotiations may signal future stability by fostering consensus on the programmatic agenda and coalition agreements. This finding is in line with what Paul Warwick (Reference Warwick1994) documented in previous contributions. According to Warwick (Reference Warwick1994), extended negotiations should be regarded as signals of closer attention to coalition programmes by parties, which can reduce uncertainty and complexity in the coalitional environment.
All in all, the results of this Cox analysis reveal important findings. Alongside classic explanatory factors, this study adds to the literature by documenting novel evidence scrutinizing the role of the prime minister’s age. Indeed, we show that older prime ministers are significantly more likely to experience government discretionary termination compared to those under 45, indicating that age plays a crucial role in affecting government stability.
To more clearly recognize the effect of a prime minister’s age on the risk of government termination, we show the Cox survival curves derived from Cox regression (Figure 3). The survival curves are displayed for all age groups examined. We observe that young prime ministers exhibit the highest survival probability, tending to stay in office longer than their older counterparts. This indicates a lower likelihood of facing discretionary termination. Middle-aged prime ministers show a slightly lower survival probability, but their curve is still above that of older prime ministers, documenting a relatively stable tenure in office. Older rulers have the lowest survival probability, particularly as time progresses. The related survival curve indicates that this category is the most destabilizing one for governments. Overall, we again document that younger prime ministers tend to have longer tenures in office.
Cox Survival Curves by Mean Age of Prime Minister
From prime ministers to ministers
Before testing the study’s second hypothesis, we show some trends in the share of ministers belonging to different age groups in national governments in the 21 countries examined.
Figure 4 displays the distribution of ministers across different age groups by country. We can observe three age categories: younger (under 45, left panel), middle-aged (45–60, centre panel) and older (over 60, right panel). Starting from the left panel, we observe that the majority of countries have a relatively low representation of young ministers, mostly below the average share of circa 11%, although some countries exhibit a comparatively higher proportion, reaching over 20%. Most countries fall below the 11% benchmark, indicating that young ministers are the exception rather than the rule in the national governments of the countries investigated.Footnote 22
Percentage of Ministers in Different Age Groups by Country
Moving to the centre panel, the data indicate that this middle-aged group dominates the composition of cabinets in most countries, with proportions typically between 50% and 80%. Some countries, like the United Kingdom, show particularly high proportions of ministers in this age range, aligning with the idea that the middle-aged demographic is the most represented in political leadership roles. The right panel shows that, in most countries, older ministers make up less than 15% of the cabinet. However, a few countries, including Ireland, Italy and Switzerland, exhibit higher proportions, showing a greater presence of older ministers in these governments.
Overall, Figure 4 shows that ministers aged 45–60 are the most prevalent group across countries, whereas ministers aged under 45 are relatively uncommon in most countries, with a few exceptions. This reflects the barriers to young leaders entering high-level government positions highlighted in previous studies (Stockemer and Sundström Reference Stockemer and Sundström2022).
Moving to testing Hypothesis 2, we ask whether the effect we observed in the case of prime ministers remains when we consider ministers. From Figure 5, we can observe that governments composed of younger ministers (e.g. mean age 35–40) are associated with higher survival probabilities over time. As the average age of ministers increases, the survival probability of the government steadily declines. The gap becomes particularly visible beyond 500 days in office, where governments with older ministers (e.g. mean age 60–70) experience accelerated termination rates. However, the underlying Cox model did not produce a statistically significant effect for mean ministerial age (hazard ratio = 1.03, p-value = 0.18).Footnote 23 Therefore, the study’s second hypothesis does not find strong empirical evidence (directionally consistent but not significant), indicating that the prime minister’s age, rather than that of other ministers, is a more important factor when it comes to government stability.Footnote 24
Cox Survival Curves by Mean Age of Ministers
Conclusion
This article has provided an examination of the impact of a prime minister’s age on government stability. By employing Cox survival analysis and focusing specifically on discretionary government terminations, we have illuminated an important yet unexplored aspect of both the youth representation and government stability research agendas. The findings of our analyses indicate that the age of prime ministers indeed plays a significant role in influencing the lifespans of their cabinets. Prime ministers who are younger are less likely to experience discretionary government termination compared to their older counterparts. This result substantially aligns with the hypothesis that younger leaders, driven by long-term career prospects and greater political ambition, may be more motivated by stability-oriented incentives. Thus, alongside classic factors in government stability research, age constitutes another key variable in explaining why some governments remain in office longer than others. In contrast, when considering the age of ministers, the findings show that the age distribution among cabinet ministers may not be as crucial in affecting the survival of governments. This is related to the less relevant (and visible) role of ministers compared to the prime minister.
From a broader perspective, the findings have important implications for the quality of democracy. The age of politicians in government positions reflects not only their experience and career stage but also the policy preferences of the ruling elites. In a social representation framework (Norris and Franklin Reference Norris and Franklin1997), a higher representation of younger prime ministers could potentially lead to more representative governance, which might be more attuned to the needs and demands of underrepresented groups, like the youth (Stockemer and Anlar Reference Stockemer and Anlar2024). From this perspective, the findings of this study could elicit policy interventions aimed at reducing youth underrepresentation in political institutions, as we show that there are relevant beneficial consequences of having such groups in key arenas of democratic decision-making.
The results of this study raise several important questions for future research. The relationship between the age of rulers and government stability is complex, and while we documented important trends and explanatory mechanisms, a more detailed research strategy is needed to explore the contextual peculiarities at play. For instance, future studies could further investigate how the political ideology and party affiliation of younger prime ministers affect their career trajectories and stability-seeking perspective. Such research questions could fruitfully be addressed by employing qualitative methodologies as well. Additionally, the decision-making processes should be examined to understand how the age of prime ministers influences their behaviour (especially when ruling in a coalition government), thus adopting a more public policy-oriented perspective. We also encourage future studies to explore whether younger prime ministers differ in terms of policy outputs and citizen engagement.
Further studies could opt for an intersectional approach, including factors such as the gender of the prime minister in an examination of its influence on government stability. Building upon Svenja Krauss and Corinna Kroeber (Reference Krauss and Kroeber2021), future studies may address the following research question: Does having a young woman as prime minister increase the stability of governments?
The implications of our findings extend beyond the academic study of cabinet duration. Research shows that youth representation in institutions fosters youth participation (Angelucci et al. Reference Angelucci, Carrieri and Improta2024) and influences electoral behaviour (Sevi Reference Sevi2021). Although our design does not directly test these effects, it reinforces claims about the impact of youth political representation on critical outcomes for democracy. In particular, perceived legitimacy – especially in generational terms – can reinforce political stability, complementing formal electoral mechanisms.
In conclusion, this study provides a novel contribution to the fields of youth representation and government stability, offering empirical evidence which we hope will be useful for both scholars and policymakers interested in understanding the dynamics of contemporary democracies. We are confident that, as the political representation of marginalized groups is gaining increased attention, the study of the age of rulers and its impact on many aspects of democratic governance will remain a crucial area for future inquiry.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/gov.2025.10024.
Data availability
The dataset and all relevant materials supporting the findings of this study are available upon request from both authors. There are no restrictions to access.
Financial support
This work has been developed under the framework of the REDIRECT Project, funded by the European Commission (Call: HORIZON-CL2-2022-DEMOCRACY-01 – Democracy in Flux. Grant Number: 101,095,142).
Conflict of interest
The authors report there are no competing interests to declare.
Open access
