When the political leader is the narrator: the political and policy dimensions of narratives

There is increasing interest in the role of narratives in policy-making, as evidenced by the consolidation of the Narrative Policy Framework, a theory of the policy process whose overall aim is to explain how policy narratives influence policy outcomes. However, with the focus on only policy narratives, there is a risk of underestimating the relationship between the policy dynamics in a specific subsystem and the pursuit of consent in the political arena. To attract more scholarly attention to this relationship, this paper distinguishes between two types of narratives—the political narrative and the policy narrative. It focuses on how political leaders address the trade-off between the content of their political and policy narratives, not only adding analytical and theoretical leverage to the Narrative Policy Framework but also providing a fine-grained comprehension of the multilayered dynamics of narratives in politics. Our main assumption is that political leaders continuously address relationships and the eventual trade-off between their political narratives (the stories through which they shape the preferences of public opinion by proposing their general political vision) and their policy narratives (the stories they tell to shape the policy process and its outputs). The way leaders decide between these trade-offs can make a significant difference in terms of political and policy outputs. We test this assumption with a comparison of the use of narratives by the same political leader in labour and education policies in Italy.


Introduction
In contemporary democracies, political leaders deploy political, financial and relational resources as well as ideational and communicative resources to incorporate their preferred policy solutions into a piece of legislation (t'Hart & Rhodes, 2014;Bennister, 2016). In other words, the use of policy narratives by political leaders is crucial to the achievement of policy success, defined here as the adoption of a preferred policy solution into law or in a decision regarded as a tangible policy output. At the same time, the policy narratives of political leaders do not originate from nowhere. These narratives bear policy content, and at the same time, they are strictly linked to the more general political narrative through which these leaders shape the preferences of the public by telling a story that shows who they are, what their values and goals are, and why they are better than others at delivering what the people what (Shenhav2006; Grube, 2012). It is well known that in the age of personalisation of politics and political leadership (McAllister, 2007), the political narratives of political leaders are pivotal for obtaining and maintaining popular support over time. Thus, political leaders must address the following two problems when constructing effective policy narratives: 1. They should carefully consider the coherence between their political narratives and their policy narratives (and thus between their political and policy goals). 2. They should adapt their policy narratives to the characteristics of the policy subsystem, with its typical constellations of actors and ideas that enter the public debate, to increase their chance of success Shanahan et al., 2011Shanahan et al., , 2013Stauffer & Kuenzler, 2021), even if this contrasts with their political narrative.
To analyse these two intrinsic problems of political leaders, this paper focuses on the political and policy narratives of a political leader aimed at promoting policy reforms (broadly intended as a set of policy solutions that are not in line with the status quo) in two different policy sectors in Italy-labour policies and education policies. The political leader engaged in both processes is Mr. Matteo Renzi, President of the Council of Ministers (from February 2014 to December 2016). Mr. Renzi is called "the scrapper" 1 in Italian politics, known for a political narrative based on the radical blaming of the "old" political class and for his determination to pursue reform (Piattoni, 2016). Renzi's labour market and school reforms were formulated, discussed, and adopted between March 2014 and July 2015. The former, the so-called Jobs Act (JA), was adopted without significant modifications to the main policy solutions proposed by the political leader and represented "a structural change" in the status quo (Sacchi, 2018). In contrast, the latter, the so-called Buona Scuola (Good School Reform Act-GS), represents a political defeat in terms of the policy preferences of the leader and, most importantly, in terms of the loss of political consent (Capano & Terenzi, 2019). This paper investigates why Renzi was more successful with the JA than the GS by examining his political and policy narratives.
Our main argument is that a leader's political narrative may affect the whole policy process, and thus, leaders should find a trade-off and seek overall coherence between the plot of their political narratives (whose aim is to reach and maintain high political consent) and the plot of their policy narratives (where the aim is to achieve the preferred policy outputs of either policy stability or change). This balance can be very demanding: on the one hand, the political leader needs to maintain a certain coherence between his or her political and policy narratives to sustain credibility; on the other hand, the policy narratives must be built on the characteristics of the policy subsystem (in terms of belief systems and the cohesiveness of competing coalitions) to achieve successful reform outputs. Thus, political leaders can design successful policy narratives if they are capable of strategically differentiating their narratives according to the characteristics of the related policy subsystem. In doing so, however, they must consider whether and how much the adopted policy narrative can diverge from their political narrative.
This emphasis on political leaders and the trade-offs between political and policy narratives enables this paper to improve the analytical capacities of the narrative policy framework (NPF)-the most comprehensive framework for the theoretical and empirical study of the influence of narratives in the policy process (Jones & McBeth, 2010;Jones et al., 2014;Shanahan et al., 2011Shanahan et al., , 2013Shanahan et al., , 2017Stauffer & Kuenzler, 2021). On the one hand, the focus on the political and policy narratives of political leaders represents value added for the NPF, as it allows us to uncover the role of political leaders and their strategic framing of policy narratives (Jones, 2018;Jones & McBeth, 2020) as well as the importance of the narrator in the reception of the policy narratives by the public (Lybecker et al., 2022). On the other hand, the comparison between different policy sectors allows us to better understand how different narrative components are assembled in light of different types of policies, actors and subsystem characteristics (Kuenzler, 2021;Kuhlmann & Blum, 2021).
The remainder of this paper proceeds as follows: In the next section, we draw on the NPF and on political leadership to develop a theoretical framework based on the distinction between political and policy narratives and some expectations of the use of narratives by political leaders. In the third section, the research design, data sources and coding of the narratives are described. Section "One political narrative and two reforms, of which only one was successful" presents the political narrative of the leader as well as the content of the two relevant reforms. In Section "Findings", the expectations are tested by comparing evidence on the different narrative elements. The final sections offer a discussion of the findings and some concluding remarks.

Narrative policy framework and political leaders: the challenge of shaping policy subsystems
The NPF offers researchers theoretical and operational instruments to detect the influence of narratives in the policy process (Shanahan et al., 2018) by developing a comprehensive framework that uncovers policy process dynamics in light of the competition between different groups of actors, building upon the advocacy coalition framework (ACF) (Shanahan et al., 2011). The NPF defines policy narratives as story devices featuring at least one character-in terms of victims, heroes, and villains but also allies and opponents (Shanahan et al., 2017)-and containing some public policy referent, such as the moral of the story (i.e. a statement concerning policy problems that anticipate the preferred policy solutions) (Shanahan et al., 2013, p. 457). The combination of these elements results in typical plots (i.e. story devices that link characters, settings, causal mechanisms and policy solutions) that resonate with a common setting (evidence and institutions regarding the specific policy problem) (Shanahan et al., 2013, p. 459). Along with these narrative elements, the NPF contends that policy narratives are bounded by different policy beliefs (intended as a set of values and beliefs that orient a group or a coalition); at the same time, the NPF acknowledges the strategic orientation of actors who develop different narrative strategies (i.e. arguments that aim to influence the policy process to either expand or contain a coalition by manipulating the perception of the costs and benefits of the policy solutions) (Shanahan et al., 2013, p. 458). The NPF further conceptualises different types of strategies to expand the scope of the conflict, to assign responsibility and blame for the policy problem, to identify the opposing narrator and/or actor in the opposition coalition as a villain (the so-called devil shift), or to depict the narrator himself/herself as a hero (the so-called angel shift) (Shanahan et al., 2017, p. 178). These narrative strategies-issue containment or expansion and angel or devil shifts-are intended to reshuffle coalitional dynamics, either maintaining the status quo or mobilising new actors outside the policy subsystem to gain more political support for the preferred policy solution of a coalition or of a single agent inside it, such as a political leader.

The NPF across levels of analysis: the interaction between the micro and the meso
The NPF builds on different theories in public policy and behavioural and communication studies to theorise how policy narratives affect policy change. The framework does so by clarifying the existence of different levels for the analysis of the narratives, from the individual level (micro) to the level of groups and coalitions (meso) and the level of the political system (macro). For each level of analysis, the NPF highlights numerous testablē hypotheses that address how actors both produce and react to the content of a narrative Shanahan et al., 2017). In so doing, the NPF assumes that narratives are socially constructed as vehicles of values and emotions and that actors act purposefully to achieve policy change by crafting tailored narratives and, eventually, by shifting from their preferred narrative for strategic purposes Lybecker et al., 2013;McBeth et al., 2016).
The NPF proposes that narratives are likely to impact individual preferences, perceptions of risks and opinions. In other words, the NPF hypothesises that the portrayal of NPF characters influences individual opinions more than the presentation of technical information . By studying the case of a highly contested policy field, such as that of charter schools in the US, Ertas shows that as narrative congruence and issue familiarity increase, an individual is more persuaded (Ertas, 2015, p. 430). However, the NPF recognises the importance of communication and focuses on how stakeholders craft different messages in their preferred way of appealing to the wider public to be more persuasive and to deal with intractable policy issues (McBeth et al., 2016). Recent NPF applications have shown that the perception of the credibility of the narrator and the image of the narrator are keys to understanding policy support (Lybecker et al., 2022, pp. 94-95).
This strategic attitude of the narrators has consequences on the meso-level of analysis. Building on the ACF, the NPF proposes that coalitions designate the narrative elements (characters, plot, morals) that elucidate their policy reality and are strategic in designing a narrative that either reflects their policy beliefs or opts for more instrumental goals. This strategic orientation of coalitions (and of actors therein) shows that narratives are strategically constructed for the purpose of influencing the individual policy preferences of decision-makers, stakeholders, or the public. Following this line of reasoning, the NPF tests some hypotheses related to how narratives are constructed and how congruence across a coalition's policy narratives is related to policy success (Jones et al., 2014, pp. 16-17). First, groups or individuals inside a coalition will use narrative elements to expand or 1 3 contain policy issues to increase the size of their coalition (McBeth et al., 2007). Second, groups will heuristically employ policy narratives to manipulate the composition of political (not policy) coalitions for their strategic benefit (Jones & McBeth, 2010). Third, coalitions with policy narratives that contain higher levels of coalitional glue (in terms of coalitional stability, strength, and intracoalitional cohesion) are more likely to influence policy outcomes (Shanahan et al., 2011(Shanahan et al., , 2013. Therefore, individuals and coalitions should consider the characteristics of the advocacy coalitions and of the policy subsystem more in general and be strategic in eventually shifting their narratives away from their preferred policy narratives or their general worldviews to be coherent with the policy options and beliefs of the target populations (Kirkpatrick & Stoutenborough, 2018).
At the same time, the NPF does not explicitly address what happens when an individual or a group sticks to the same narrative across policy subsystems with different coalitional glue (in terms of stability, strength or internal cohesion) or even what happens when the general worldview of the narrator is in contrast to the dominant views in a policy subsystem. Moreover, the NPF does not explicitly focus on the fact that the content of a narrative can be not only policy-related but also connected to a more general narrative-a political one. In other words, the NPF has not yet reflected on the impact that the political worldviews and beliefs of individual narrators may have on policy narratives and on their success in adopting the preferred policy solution. 2

The need for a focus on political narratives
While very fine-grained, the NPF does not directly consider whether or how a political dimension exists in linking together different policy narratives and, consequently, does not consider the roles and narrative strategies of political leaders when they are committed to pursuing coalitional reshuffling and policy change. This gap is relevant because political leaders are becoming increasingly important in personalised politics (Masciulli et al., 2009;Nye, 2008;Rahat & Kenig, 2018). Prime ministers engage in public debate with one clear purpose: to gain public support for both their own political preferences and their own policy solutions. At the same time, they use not only material resources and power but also ideational and discursive resources to gain support for their policy solutions (Stiller, 2009) while speaking to a wider audience of advocates for a policy subsystem. Furthermore, policy-making (and related policy narratives) does not develop in a vacuum but is embedded in a specific political context in which what political leaders say should be taken into serious consideration, especially when they directly promote policy solutions.
Thus, there is a need to include political leaders in the NPF to increase its analytical capacity and to reason about the different types of narratives that political leaders enact. In other words, while the NPF considers policy narratives to be the core of the policy process and assumes that political leaders are relevant actors, there is no significant consideration of cases in which political leaders base their actions on a specific political narrative that could be their fundamental resource for designing policy narratives.
Then, there is the theoretical point of distinguishing the concept of policy narrative from that of political narrative. NPF scholars have clarified this distinction, although they have 1 3 not built upon it. In fact, Shanahan et al. (2011), in a foundational piece of the framework, clarify that while "political narratives are persuasive stories for some political end (e.g. to win an election), a policy narrative has a setting, a plot, characters (hero, villain, and victim) and is disseminated towards a preferred policy outcome (the moral of the story)" (p. 539). They also point to the existence of different political narrative strategies that are aimed at either expanding or containing the policy subsystem while addressing ideological issues and identities that may hamper or facilitate policy learning across coalitions (Shanahan et al., 2011, p. 541). At the same time, recent NPF studies highlight how currently, in the US, each political party has "radically different narratives about politics" (Jones & McBeth, 2020, p. 8) and how politicians may be constrained by their own worldviews and values in building their stories . Therefore, in the NPF, there is an awareness that there are at least two different types of narratives: one is the policy narrative, which is committed to achieving a policy result, while the other is the political narrative, which aims to achieve some political end that is usually represented by maintaining or increasing political consensus for the leader and his or her party. It then emerges that, from the perspective of political leaders, they can act on at least two types of narratives.
Thus, there is a need to deepen the concept of political narrative to make the distinction clearer and more analytically relevant. In other words, there is a need to understand whether policy narratives, i.e. those stories that "are created by policy actors to help advance their policy goals" (McBeth and Lybecker, 2018, 888), are independent of more general political ideas that political actors hold and that are contingently constructed or have a relationship with a higher level of narrative. The two-tier conception of political narrative proposed by Grube helps to clarify the distinction: "Political narratives operate at two distinct levels, although the relationship between them is by no means clearly established. At the first level, every political leader and government consciously or unconsciously promulgates a meta narrative-an all-encompassing story about what their government stands for. Its breadth covers the litany of individual policies that make up the everyday work of government. At the second level, each major policy announcement has linked with it a narrative of its own-commensurate with the meta-narrative-but more specific in its rationale for a particular policy. It is a story that explains why a new law or policy is necessary." (Grube, 2012, p. 569). This two-tier conceptualisation of narratives is very helpful because it underscores that there is a distinction between narratives according to their content and those who act on them. While this distinction is useful, we consider that the labelling is misleading. By building on this distinction, then, we assume that political leaders hold and act in different types of narratives that are distinguishable according to their target and content. Thus, • Political narratives are "grand" stories that political leaders tell directly to the public and the electorate. Through these narratives, political leaders shape the preferences of public opinion by telling a story in which they propose a political vision, relevant political goals and how they will perform better than others in leading policies towards reaching the proposed political ends. Through political narratives, leaders try to shape the polity attaching it to a collective cause by emphasising the relation between the narrator (the leader) and the public. The primary goal of political narratives is political consensus. • Policy narratives are those specific stories that political leaders tell to find support for particular policy choices. These types of narratives are focused on the characteristics of specific policy issues and are, thus, more constrained by the actual configuration of ideas/interest/and institutions in the related policy subsystem. The first tar-get of the policy narrative is represented by the stakeholders directly involved in the policy field. The primary goal of the policy narrative is to manipulate the advocacy coalitions (either for change or stability) and to reach the expected policy effects.
According to this conceptual distinction, it is reasonable to assume that there should be a certain coherence between the content of the political narrative (of the general vision that the political leader offers about its general political ends and his or her role in pursuing them) and the policy narratives that act in the different policy subsystems. Thus, we expect that there should be coherence between some components of the political narrative (especially in relation to the characters and the morals) and the elements of the policy narratives. As narrators, then, political leaders must pay attention to the coherence between the general goals and the image they project to the broader audience and the policy goals and the image they project in a policy narrative. It is clear that the stronger, more assertive, and more heroic the political narrative is, the more relevant and compelling this coherence is; in contrast, when the political narrative is more multifaceted and loose, the political leader is allowed to be more flexible and autonomous in shaping his or her policy narratives.
Based on these theoretical premises, we assume that the focus on the role of political leaders and their political and policy narratives can add theoretical scope to the NPF by showing whether and how political leaders are constrained by their political narrative when they have to persuade different actors in the policy subsystem to support reform initiatives (or at least to not oppose them).
We aim to contribute to the NPF by elaborating on the theoretical expectations of how political leaders must shape their narratives to be successful. We explore these issues by developing some theoretical expectations informed by two broad assumptions (O'Bryan et al., 2014) and by testing them through a comparison of two highly contentious policy fields (Shanahan et al., 2017, p. 198).
The first assumption is that political leaders express their visions of the world through their political narratives and that they must carefully consider the coherence between the goals and values embedded in their political narratives and the content of their policy narratives.
The second assumption is that the characteristics of the policy subsystem (such as the level of openness of the subsystem to external actors or the level of ideational cohesiveness of the belief system) matter for the success of the policy narratives.
From these assumptions, we derive our theoretical expectations about how political leaders should deal with the policy subsystem and their own political narratives to shape successful policy narratives.
An initial expectation relates to the idea that political leaders should adapt their narratives to the main characteristics of the policy subsystem if they want to be successful. In so doing, we argue that the strategic leader should carefully assign the characters (i.e. the roles of the victims, heroes and villains) in a way that is coherent with the characteristics of the policy subsystem (Merry, 2019). NPF studies on environment and water policies showed significant differences in the assignment of characters, morals and plots depending on the audience and on other specific characteristics, including the fragmentation or cohesiveness of the coalitions (McBeth et al., 2005(McBeth et al., , 2016Shanahan et al., 2013). Simultaneously, we hold that policy narratives are also characterised by a political dimension and, thus, that political leaders will design their policy narratives according to not only the characteristics of the policy subsystem but also their own political goals and symbols, i.e. their political narrative. Accordingly, we formulate the following expectation:

3
Expectation 1: Political leaders should adopt different narrative elements in different subsystems, yet are constrained by their political narratives in doing this.
The second expectation relates to the problem that leaders face when they want to introduce policy change in closed and cohesive policy subsystems with high coalitional stability and cohesive policy beliefs (Shanahan et al., 2011). This type of subsystem can be very resistant and very capable of mobilising powerful actors. We assume, then, that in these cases, the successful political leader should shape his or her narrative strategies not only to expand the conflict outside the subsystem-by seeking the support of external actors, such as the general public, as well as of the electorate and powerful international actors, as expected by the NPF-but also to secure internal support from powerful actors inside the subsystem, thus seeking both internal and external legitimacy as validation devices for their favourite policy option (Capano & Lippi, 2018). This also means that leaders should be strategic in choosing their policy narrative strategies (Schlaufer et al., 2021) and should not, for example, play the devil's shift against powerful internal actors who might oppose the change. In so doing, political leaders must avoid the clashing of political narrative content with deeply established beliefs and shared practices of groups cemented with coalitional glue (in terms of stability, strength and cohesion), producing policy narratives that are highly divisive and able to activate powerful opponents. Public school systems in Italy represent an example of this type of coalition, where teachers and trade unions are highly cohesive in terms of both their stability through time, their defence of democratic and egalitarian values, and the cohesiveness of their actions towards families and institutions (Grimaldi & Serpieri, 2012). In contrast, the policy subsystem developed around infrastructural megaprojects at the local level (Turin) showed a variety of actors and contrasting views with a reflection of the variety of argumentations and narratives (Esposito et al., 2022). Accordingly, we present the following expectation: Expectation 2: To be successful in promoting policy output in highly cohesive subsystems, the policy narratives of political leaders should be both capable of expanding the conflict (to attract new supporters and strengthen the external legitimacy) and congruent with the belief system of the dominant coalition (to persuade powerful actors in the subsystem and to uphold internal legitimacy); however, this could be problematic if the leader's political narrative evokes ideological issues that are incongruent with the political beliefs of the dominant policy coalition.

Research design
This paper conducts a meso-level analysis to understand how a political leader shapes narratives while promoting reforms in two different policy sectors in the same period (2014)(2015). Labour and education policies were selected because they have been the subject of major reforms in the last two decades and are two of the most politically relevant and conflictual policy domains in Italy (Capano & Pritoni, 2016). The two reforms were described by the leader himself as two engines to restart the country after a period of economic and social crisis, which needed to be undertaken in parallel and with urgency. 3 Therefore, the logic of comparison follows a similar case comparison method to detect the role that the combination of political and policy narratives may have played in explaining the success, or lack thereof, of the reform (Przeworksi & Teune, 1970). In fact, in the case of the JA, the policy preferences of the narrator (namely, the flexibilisation of the labour market) were adopted without modification in the legislative phase (Sacchi, 2018); in contrast, the policy preferences of the leader anticipated at the onset of the GS reform on schools (namely, the managerialisation of schools) did not survive the adoption phase (Capano & Terenzi, 2019). Therefore, this case selection aimed to show that the strategic choices of the political leader related to political and policy narratives may have contributed to the success of the JA and the defeat of the GS.

Data, sources, and operationalisation of the narratives
For data and sources, this paper reconstructs and qualitatively compares the narratives deployed by the protagonist of both reforms (and some of the counternarratives of his principal opponents) during the period of the formulation and parliamentary discussion (December 2013-September 2015). The narratives are extracted from a database of 200 newspaper articles for JA and 167 articles for GS. Articles were initially gathered through the open archive of one of the main Italian newspapers, La Repubblica. We chose this source for the reconstruction of Renzi's policy narratives for two reasons: first, La Repubblica is one of the two main Italian newspapers and is widely considered the leading media for centre-left political leaders to communicate with their electorate while hosting numerous op-eds and interventions from stakeholders in labour and schools; and second, this newspaper publishes daily articles with the exact declarations of political leaders on different occasions (as excerpts of parliamentary speeches, press releases, or even social media messages).
The articles on La Repubblica were gathered using the name of the main policy-makers, political parties and interest groups plus the conventional name of the reform (either the "Jobs Act" or "Buona Scuola"). To identify the narratives as portions of the verbatim declarations of the leaders as quoted in the newspapers, the coding procedure was developed following different steps. While one author was the main author responsible for the coding process, the results of the coding operations were initially cross-checked by the other authors to enhance their intercoder reliability (O'Connor & Joffe, 2020). The coding process entailed the following steps.
(a) Two of the authors separately read 70 articles (20% of the total number of newspaper articles) and isolated the verbatim declarations of the narrators quoted in the newspapers (the quotes containing a narrative being the unit of coding). (b) Two authors autonomously checked whether the sentences identified contained the necessary narrative elements (characters + moral of the story) and then computed the percent agreement, which is a measure of intercoder reliability that indicates the share of sentences judged identically during both authors' coding process, obtaining a value of 95%. (c) The same two authors then independently applied the NPF research protocol detailed by Shanahan et al. in 2018 to isolate the different elements of the narratives (characters such as victims, villains, and heroes; morals of the story) and the main content (plots and narrative strategies) for all 70 articles. For the narrative strategies, we tried to code stories aimed at influencing the size of the coalition in terms of PNF strategies (issue expansion, devil or angel shifts, and causal mechanisms attributing blame) and in terms of internal or external legitimacy as new categories. The two coding processes were then compared, leading to an agreement of 89%. (d) After the narratives on which there were disagreements were revised, only one author carried out the coding process for all the remaining newspaper articles, whereas the other author ultimately checked a final sample of 10% of the articles.
These operations produced a database of Renzi's declarations, codified into 152 rows for JA and 148 rows for GS. 4 The subsequent analysis of the database was done qualitatively by the three authors, isolating the relevant portions of the text with filters related to the main policy actors, their narrative elements and the policy instrument for the two reforms.
Finally, Renzi's political narrative was reconstructed using secondary and primary sources. We used the work of Ventura (2015), which focuses on the political style of Renzi by gathering the text of the major political declarations (as contained in different newspaper articles, speeches in Parliament, TV shows, and public interviews) made by the prime minister during his tenure. To triangulate with the declarations contained in Ventura (2015) and thus to check for different media outlets, as recommended by the NPF (Shanahan et al., 2018, 341;McBeth et al., 2005), we finally conducted a qualitative analysis of Renzi's political narratives in his three main parliamentary speeches. 5

One political narrative and two reforms, of which only one was successful
In this section, we first reconstruct a picture of Renzi's political narrative and the process that led to his success with the JA and his defeat with the GS before presenting evidence in Sect. "Findings" on the role his political and policy narratives played in these reforms.

The political narrative of the "scrapper"
Renzi's political narrative is quite simple: there is one paramount hero, the leader who is here to rescue the country (the victim) from all those who are always against change-the political, economic, social, and cultural establishment (the villains). The hero must also be a narrator: "The prime minister is one who tells a story. If he is not capable then everything is useless (….) We tell a story. The story of a country that can handle [everything]". 6 The hero is here to finally drive towards real policy change: "It has been about thirty years that we have been doing only cultural debate; it is time to make decisions", 7 and "you cannot improve social equity by conferences, only thanks to governmental choices". 8 Finally, the hero has so many enemies (strong powers), from unions to national associations of firms 4 Nevertheless, during the coding process, the narratives of others key policy actors have been analysed, for a total of 624 rows for the JA database and 595 rows for the GS database. 5 One speech to obtain the vote of confidence from the Senate on 24 February 2014 and two other programmatic speeches to the Chamber of Deputies and to the Senate on 16 September 2014. 6 Retrieved from the interview contained in L'Espresso, 06/03/2015. 7 Retrieved from La Stampa, 25/07/2014. 8 Retrieved from La Repubblica, 20/04/2014. and industries, from small political parties to judges, and from journalists to academics (Ventura, 2015). All these enemies want to stay in the "swamp". 9 Thus, Renzi's political narrative is based on the idea of a rupture with the past (Salvati, 2016) and on a strong "us against them" rhetoric, implying that to rescue the country from the swamp, it is necessary to make the decision processes and the practice of concertation with interest groups more efficient: "I think that the veto powers of small groups have damaged our country. The veto powers have blocked any change to reform the country… We can hear everyone, but the time has come that everybody does their own job". 10 Thus, Renzi's political narrative directly implies a decisional style based on disintermediation (Piattoni, 2016) that he will try to apply in all policy narratives he adopts to trigger policy change. Therefore, the main goals and values of Renzi's political narratives are to break with the past, to listen to the "real" country and not to the "experts" and "insiders", to make decisions quickly and without wasting time and to skip the intermediation of the unions. Indeed, disintermediation emerged as a typical trait of Renzi's leadership (Salvati, 2016;Ventura, 2015). Table 1 summarises the main elements and components of the political narrative of this leader. 9 Retrieved from Il Corriere della Sera, 04/05/2014. 10 Renzi's words during the popular Italian TV show "Otto e mezzo" 27/10/2014.

Policy outputs and reform process in the JA and GS
Against this political backdrop, the process of the two reforms developed with some interesting differences and similarities. Table 2 summarises the context and outputs of both reforms. For the policy change outputs, one of the main novelties in the JA's "embedded flexibilisation" (Picot & Tassinari, 2017) was a change in the rules on dismissal for new insertion contracts. The legislation on unfair and unlawful dismissal is contained in Article 18 of the Italian Statute of Workers regarding the employer's termination of an employment relationship without subjectively or objectively justified reasons (Scarano, 2019). Article 18 has a strong symbolic meaning in the discourse on employment legislation protection (EPL). Despite representing a break with the status quo, the most relevant content of this reform survived during the parliamentary stage because its opponents experienced an internal division not only across political parties (with part of the governing coalition and part of the political opposition supporting the reform) but also across the main trade unions (with one of the three main unions, the CISL, supporting the reform against the other two, the CGIL and the UIL).
In contrast, law-making for the school reform was less successful, as the two most innovative policy solutions (i.e. the strengthening of the principal's power to the detriment of collegial bodies and the introduction of a merit-based system of evaluation and remuneration for teachers) were progressively lost in the formulation phase (Capano & Terenzi, 2019, pp. 255-259;Argentin & Barone, 2016, pp. 136-138). Even before the parliamentary discussion, this process rapidly generated strong opposition to the reform among both unions and political parties, including a consistent part of the PD, Renzi's own party. This intense opposition significantly softened the original shape of the GS when it was approved in July 2015 (law n. 107), despite the presence of an extraordinary plan to recruit 100,000 temporary teachers, which could have represented a source of political support (Capano & Terenzi, 2019, p. 254). In particular, all the unions (with the significant exception of the union of principals, the Associazione Nazionale Presidi) fiercely challenged the government's proposed changes-namely, the increase in school principals' powers and the strengthening of merit-based criteria for teachers' careers-and the procedural choices made by Renzi, namely the lack of involvement of teachers' unions in public consultation (May 2014-November 2014) and in the initial drafting of the law (December 2014-March 2015). These choices led to the major school strike 11 on 5 May 2015 while the parliamentary discussion of the governmental proposal was in progress (Argentin & Barone, 2016, p. 137). In the same period, important amendments presented by Renzi's party (and strongly supported by the PD minority) cancelled the possibility of the school principal deciding autonomously 12 on the educational content of each school plan for educational activities (POF, piano dell'offerta formativa) and decreased the influence of the school principal on the hiring of teachers and on the procedures and criteria for evaluating teachers, and brought the most important decisions in terms of school financing and recruitment back to the central ministry and the school regional education authorities. 13 While the policy outputs (in terms of adopted legislation) of the reforms were somewhat different in the two cases, most of the variables that potentially affected the results were similar. For the context, both reforms enjoyed similar contextual conditions acting as favourable external pressures: Italy was still under the conditionality of the European Union and the European Central Bank (ECB), which were urging structural reforms (Sacchi, 2018;Sacchi & Roh, 2016), while the country was condemned by the European Court of Justice for the high proportion of fixed-term jobs in the school system.
Regarding the features of the two policy subsystems (see Table 3), both the labour market and education system experienced urgent problems and conflicts but were substantially different in terms of both the number of relevant actors and the ideational cohesiveness within the subsystems. While the subsystem of labour policies was quite fragmented in terms of policy instruments and policy actors (with a considerable number of interest groups and veto points), the subsystem of school policies experimented with stable policy solutions, a strongly cohesive belief system, and, most importantly, was dominated by traditional actors in the school system (trade unions, representing 60% of teachers and ministerial bureaucracies). This subsystemic cohesion has been highly relevant and effective in its impact on the parliamentary phase.
For role agency, the protagonist of the policy reform was Renzi himself in both cases. Renzi had been the leader of his own party since December 2013 and became prime minister in February 2014 as the head of a composite centre-left governing coalition. From spring 2014 to December 2016, his political ability strongly benefitted from the stunning electoral victory of his party in the European elections held in May 2014, but he has since faced continual internal opposition from his party. A crucial aspect of his political vision and style has been the deliberate decision to avoid the intermediation of interests while formulating reforms, a so-called disintermediation strategy (Pritoni & Sacchi, 2019). Both JA and GS were marked by strong mediatic exposure, identification with the prime minister and unilateralism in their development (Argentin & Barone, 2016;Galanti & Sacchi, 2018). In the case of the GS, unilateralism was coupled with the search for direct legitimation from citizens; after the public presentation of the first programmatic document, Renzi launched a two-month public online consultation (with almost 200,000 participants) together with 40 events and approximately 2,000 debates with citizens, families, and teachers, but without unions (Capano & Pavan, 2019). This process was an absolute rupture in traditional policy-making in school reforms in Italy, where the intermediation of unions has always been crucial (Argentin & Barone, 2016;Capano & Lippi, 2018). Table 3 summarises the characteristics of the subsystems in the two acts and the process and timing involved. 13 These changes were approved through a maxi-amendment presented by the PD to the Senate under a vote of confidence on 25/06/2015 (maxi-amendment n. 1.9000, with reference to the new commas n. 78according to which the principal is no longer responsible for proving the merit of teachers to school management-and n. 129-which emphasises the collegial composition of the committee for teachers' evaluation in each school and weakens the influence of the principal over teachers' careers. All the details related to the main changes to the content and the process of the GS reform, from the programmatic document to the different parliamentary bills are outlined in "Appendix 2".  Table 4 summarises the main narrative elements and components, allowing us to discuss the emerging empirical evidence against the expectations proposed in Section "Narrative policy framework and political leaders: the challenge of shaping policy subsystems".

Findings
Expectation 1: The political leader should adopt different narrative elements in different subsystems but can be constrained by his or her political narrative in doing so.
A comparative analysis of what type of characters and morals of the story identified by Renzi in the two reforms reveals that the political leader basically adopted the same policy narrative elements (in terms of characters, morals of the story and plots) for both the labour market and school policies. In terms of the characters, Renzi identifies the same categories of actors as victims, villains and heroes. In both the JA and the GS, the victims are the younger generation, defined as precarious young workers in the context of labour and defined as students and their families in the context of the schools.
In the JA, the youngest individuals who are left outside the job market are the victims of a system that safeguards only the insiders of the job market, namely, the fixed-term employees: "The numbers of youth unemployment are hallucinating. That is why the JA will be our priority" (Renzi 01/03/2014). At the same time, Renzi identified the former political class and the unions that protect only fixed-term employees as villains: "Italy is a gridlocked country, and employment is our priority. There is a division between those who have rights and those who do not. The unions defend the former" (Renzi, 01/03/2014).
Renzi and his government are represented as heroes in opposition to the unions: "At last, politics (i.e. the government-ed.) combat the job precariat and not the precarious workers" (Renzi, 20/12/2014). "Landini (i.e. a unionist leader-ed.) wants to occupy the fabrics, and we want to open them" (Renzi, 9/10/2014). In this identification with the heroes of the story, we see a clear overlap between Renzi's political and policy narratives: "The situation is dramatic. The solutions of the past failed. We must run forward; the country calls upon us" (Renzi 1/4/2014). The idea is that the government was able to "awaken Italy, the sleeping beauty". Here, Renzi plotted a story of rising where the hero was Renzi himself: "The year 2014 was a Copernican revolution: we [the government] have changed the rhythm of politics (…). Now, I want to change the mood of Italians who are addicted to mistrust and fear. Italy was restarted; in 2015, the challenge was to run forwards" (Renzi, 29/12/2014).
In the GS, the victims are again the country at large as well as young students, "our sons and daughters", and, most importantly, teachers. Teachers are plagued by job precariousness, while students are grouped into "chicken coop classes" that deprive them of their educational needs while causing social disruption. "The unavoidable goal of the GS is to put teachers into the classes by hiring them to stay and not just for one year. The GS wants to defeat the decennial evil of precariousness" (Renzi 4/3/2015); "our aim is not only to comfort precarious teachers; we are hiring them because making teachers live in continuous uncertainty damages the pupils" (Renzi 23/02/2015).
The villains are both the state and the (past) governments, which have betrayed society by taking away financial resources from the schools and trade unions, which worsened job precariousness by blocking the country. "School belongs to the families and to the pupils, not to the unions. It is ridiculous to strike against the first government that gives money to the teachers and that eliminates job precariousness by a huge hiring plan, never seen before in the schools" (Renzi 21/04/2015). The hero in the ongoing confrontation between good If we will be able, in the next year, to rethink how Italy should invest in school, then we will then lay the foundations of growth for the next 20 years and provide a beautiful educational opportunity for our children and families (03/09/2014)

Narrative strategies
Issue expansion Data related to unemployment are dramatic. The country is asking us to run forward; otherwise, it will be too late.
We are losing 1,000 jobs every day (01/04/2014) We want to fund schools because school is not a cost but an investment for our children, for the future of Italy. He who loves Italy equally loves school (03/09/2014)

Angel shift
We are not dealing with the personal destinies of some politicians but with the destiny of the entire country, which is much more important (03/11/2014) After many long years in which the government has done nothing, I prefer to take risks rather than be trapped in a swamp (19/04/2015)

Devil shift
Where were the unions when the biggest injustice of our times came about, that between those who hold a job and those who don't? (…) (This happened-ed.) because (the unions ed.) have thought to fight only ideological battles and not the problems of the people (19/9/2014) Listening to everyone does not mean not doing anything; otherwise, there is going to be a paralysis, which is the same behaviour that has been blocking progress in Italy for 20 years. We will not allow this behaviour any longer (22/02/2015) and evil is again Renzi's government: "We are putting money into the schools: the previous governments were cutting it" (Renzi 13/05/2015), and "we want to stop job precariousness in schools and the disease of teacher supply. The first goal of our reform is an extraordinary plan to hire new teachers" (Renzi 3/09/2015). This choice is perfectly coherent with the main content of the political narrative that is based exactly on a vision that emphasises how the country must overcome the veto powers or consolidated vested interest. Renzi's political narrative clearly identifies the people (or large portions of them) as the victim, and the unions, like the political class and the bureaucracies, represent an obstacle to change, as restated by him in the speeches to the Parliament: "After waiting so much time in the last years, now we have finally the last chance to settle accounts, and if we lose, it is not a loss of the government, but of Italy as a whole" (Camera, 16/09/2014); "The time of getting by is over; getting by is over; is over for the unionist who opposes to the diminishing of their prerogatives; getting by is over for the private managers who thinks that there is no limitation to their salaries… the power of the rentiers is over" (Senato 16/09/2014).
The morals of the story are also similar for both the labour market and school issues. There is a perception that fast and decisive action must be taken to make both the job market and the school system more dynamic and competitive. Furthermore, there is a perception that traditional policy instruments constrain the vital forces of both the market and society.
In the JA, the problem is identified with the rigidities of the job market, and the solution to its flexibilisation is to remove outdated policy instruments that distract national and international entrepreneurs from investing in and hiring the youngest individuals. For example, Article 18 states: "It's a '70 s rule and we are now in 2014. (…) It's like catching a digital camera and trying to insert a roll of film. The Italy of film rolls is over" (Renzi 26/10/2014). "Article 18 is an ideological totem around which you can see dancing the usual subjects who do not worry about real issues but only make ideological discussions" (Renzi 19/12/2015). The plot is narrated as a story of rising and is strongly focused on We give more money to teachers, and the word "merit" is not a swear word: we cannot grant the same salary increase to everybody. The principle claimed in some unions' demonstrations, based on which "nobody can judge me," can no longer be valid (14/05/2015) the young unemployed as the main target of the reforms: "Having a job is a constitutional right, not Article 18" (Renzi, 29/9/2014). In the GS, the problem is the bureaucratisation of schools and the teachers' precariousness, but the morals of the story describe the solution as the reinforcement of schools' autonomy thanks to the decisive financial investment of the government in education for the purpose of hiring teachers, the empowering of principals and the introduction of a merit-based system for teachers: "The new model for school is the realisation of the autonomy that remained only on paper" (Renzi 13/03/2015), and "autonomy means removing and deleting the power of ministerial guidelines that are using a bureaucratic language, decide the future of our kids and asking schools to open themselves to the surrounding environment and cultural realities" (Renzi 13/05/2015).
In the GS, the plot is focused on society at large, families, students and principals. Here, the political narrative clearly affects the policy narrative as well: "School is the place where we either change the country or remain in the swamp. Italy can be a cultural authority, and now, we have the opportunity to build a future for our children; wasting it would be a bad mistake" (Renzi 14/05/2015). While framed as the main victims in the plot, teachers never become the heroes of the reform. At the same time, the plot presents the principals as coaches who reward (or downgrade) teachers, who are seen as players-a divisive and negative image of the egalitarian and collegial nature of the school system in Italy: "Let's leave the sheriffs in the western. The principal now becomes an educational leader, a person who comes from the school context who puts himself/herself at the service of the educational world with instruments and powers that allow him or her to finally make decisions" (Renzi 03/04/2015). The metaphor of the principal as a leader is coherent with the political narrative of the hero who can save the country from the swamp without wasting time, yet it goes strongly against deep-rooted beliefs about collegiality in school decision-making.
In sum, Renzi adopted the same narratives in both reforms. He presented the young generations, the teachers and the country at large as the victims of a dramatic situation. This drama was caused by the long-term actions of past governments and of the unions as the main villains. Renzi also presented his government as the hero in both instances. The morals of the story pointed to a liberalisation of the job market and to freeing the school system from the constraints of the past.
Thus, the leader did not adapt the components of his narratives to the different subsystems. This lack of differentiation was clearly due to the necessity that the leader implements his breakup political narrative even at the policy level, despite the differences among the policy subsystems. Renzi's political narrative emerged when he talked about JA and GS. He declared that his political goal was to break the rules of the past to free the country from the swamp of veto powers by both politicians and unions. He also pictured himself as the hero who was going to save the young generation and society at large from the chains that have been tightened by both the political class and the unions. Regarding the JA, Renzi said, "You know why they (the unions-ed.) are blaming us? Because we are taking power from them. (…) the truth is that we are making a revolution in the country, and we are also obliging the unions to change" (Renzi, 4/5/2014), and "I have a problem with the unions that engage in politics. They should take better care of those they never cared about (the young and the precarious workers-ed.)" (Renzi, 4/11/2014). Regarding the GS, he said, "No government has ever made such investments in education. We (…) want education, education and education, which in Italian translates to 'future'" (21/01/2015); "after long years where the government has done nothing, I prefer taking risks rather than being trapped in a swamp" (Renzi 19/04/2015); and "we are the first government that puts 1 3 3 billion into the schools (…) if we believe in this reform, then we will change Italy; otherwise, we are going nowhere" (Renzi 5/5/2015).
In other words, Renzi's policy narratives were shaped by his political narrative in presenting young unemployed individuals and teachers, students, and families as victims and himself as the hero who saved the victims from the unions (the main villains). This coherence between his political and policy narratives was aimed at widening his political consent in the general electorate (Salvati, 2016) and, thus, outside the subsystem.
Expectation 2: To be successful in promoting policy output in highly cohesive subsystems, the political leader's policy narratives should be capable of both expanding the conflict (to attract new supporters and strengthen external legitimacy) and being congruent with the belief system of the dominant coalition (to persuade powerful actors in the subsystem and to uphold internal legitimacy); however, this could be problematic if the leader's political narrative evokes ideological issues that are incongruent with the political beliefs of the dominant policy coalition.
Shaping the narrative strategies according to the context means that the leader must consider subsystem characteristics. In more fragmented subsystems, political leaders should craft narrative strategies to expand the scope of conflict and mobilise new actors, thus seeking legitimation and support for the reform from the outside of the policy coalition or subsystem and considering the impact of relevant events in the external context (such as an economic crisis or international pressures). Conversely, in more closed and cohesive subsystems, the political leader should both expand the conflict (to attract new participants) and avoid the countermobilisation of powerful actors within the dominant coalition, thus securing support and legitimation from the inside. Table 4 shows that Renzi's policy narrative strategy mainly sought legitimacy outside the system of the labour market and school policies by arguing that it is society, the younger generation, families, and international actors that seek to eliminate the rigidities of the past, to free work and schools from time-wasting vetoes and mediation, as trade unions must negotiate with employers, not the government. At the same time, the reforms could be designed and approved by directly asking citizens through a public consultation without the input of employers' associations or unions. This strategy is perfectly coherent with the pillars of his political narrative and his vision of society and the future of the country. At the same time, this narrative strategy clashes with the main actors, with the business-asusual approach and with the core policy beliefs of the dominant coalition regarding the school system.
In the JA, Renzi first sought only external legitimacy. He used the narrative strategies of issue expansion and angel shift to argue that the JA is the only solution to save the country, to attract new investments from abroad and to gain credibility (and financial resources) from the EU: "Data related to unemployment are dramatic. The country is asking us to run; otherwise, it will be too late. We are losing 1000 jobs every day" (Renzi 01/04/2014). "We know very well that it is not going to be a rule on employment contracts to reverse the negative data on unemployment. However, our task is to simplify and make firms free of excessive bureaucracy. We must give stability and certainty to employment relationships, as Europe asked us to do" (Renzi 04/06/2014); and, "We are not dealing with the personal destinies of some politicians but with the destiny of the entire country, which is much more important" (Renzi 03/11/2014).
While pointing towards external legitimacy, Renzi also claimed that reform was needed for some categories of actors that were internal to the labour subsystem, thus seeking internal legitimacy. He suggested that the problem of the Italian job market could not be solved 1 3 by the state but by the free action of the entrepreneurs in the market-pledging to the employers' associations that, "It is not the regulation that creates the jobs: it is the entrepreneur" (Renzi 8/01/2014). At the same time, he sought internal legitimacy for the JA, arguing that the demand for reform comes from young precarious workers who are excluded from the safeguards of permanent positions: "In Italy, we inherited an apartheid of jobs, between those who have a permanent position and those who have not. The JA gives rights to the young and opportunities to the new generations" (Renzi 31/03/2015). "With the JA, we recognized the right to work to an entire generation. Things such as 'paid vacations' and 'loans' (to buy their own houses-ed.) enter in the vocabulary of one entire generation of young people who were excluded from those rights in outrageous ways" (Renzi, 20/02/2015).
Simultaneously, unions are presented as the devil, being themselves the causes of the problem: "Where were the unions when the biggest injustice of our times came about, that between those who hold a job and those who don't? (…) (this happened-ed.) because (the unions ed.) have thought to fight only ideological battles and not the problems of the people" (Renzi: 19/9/2014). Renzi disregarded social concertation as the usual way of developing labour policies "with" the unions. He played a disintermediation game, using his narratives to show that the unions were intentionally blocking the country with their vetoes: "We are here if unions want to open a discussion. If they want to fuel an argument, then they can do it; we can go ahead even without unions. We are available to listen, but they should stop such veto power" (Renzi, 06/05/2014).
Renzi also challenged the unions on the grounds of their internal cohesion, being aware of the fragmentation in the labour subsystem: "By angering those who contest us (the unions-ed.), we make everybody go forwards, in particular, the youth, the industries and the temporary workers, even at the cost of displeasing the trade unions or a political party minority" (Renzi 22/09/2014). "Our country has been too often blocked by vetoes. We would be a much more competitive country today if the reform had been implemented when Schroeder and Blair did the same in Germany and the UK. In contrast, Italy had to wait until 2015 for its cabinet (to achieve these reforms-ed.). For this reason, we do not allow trade unions to block this reformative project. It is legitimated to think differently, but it is a duty for us to decide and go forwards" (Renzi 21/11/2014).
In the GS, Renzi initially sought support and legitimacy both internally and externally to the school system. His narratives aimed at expanding the conflict outside the system: he argued that it was not only European institutions but also Italian society and families who were urging the government to act: "If we will be able, in the next year, to rethink how Italy should invest in school, then we will then lay the foundations of growth for the next 20 years and provide an opportunity of educational beauty for our children and families. We want to fund schools because school is not a cost but an investment for our children, for the future of Italy. He who loves Italy equally loves school" (Renzi 03/09/2014). "The school reform we are developing might also not be the best reform in the world. However, we must change the current situation for the good of the country; we cannot turn our backs on it" (Renzi 19/02/2015). "After years of precariousness in school, this is the biggest recruitment made by a government of the Republic. In addition, it is not true that it is the European Court of Justice that imposed us to do so: we did it ourselves" (Renzi 13/05/2015).
Initially, Renzi tried to build internal legitimacy by focusing on the teachers, saying that the investments in schools made by his government were an amendment to the long betrayal of teachers perpetuated by previous governments. At the same time, he tried to picture the teachers as autonomous from both the ministerial bureaucracies and the trade unions: "The choice of school autonomy is decisive. This means that schools are not in the hands of ministerial regulations and unions but in those of teachers, families, and students (Renzi 28/4/2015).
Later, he attacked the teachers too and used narratives that were clearly divisive for the values and shared beliefs in the school system. In particular, the policy narrative showed that teachers would have to accept an evaluation by principals, an idea that goes against the deep-rooted conception of egalitarian careers and salaries for Italian teachers. In Renzi's narratives, teachers should change their attitude from demanding rights to being responsible: "We have ultimately hired the precarious teachers. This was a right that has not been respected for 20 years. Today, after this extraordinary phase, every teacher will know that being hired is not only a worker's right but also a responsibility, a gift, an opportunity" (Renzi 21/01/2015). "We give more money to teachers, and the word 'merit' is not a swear word: we cannot grant the same salary increase to everybody. The principle claimed in some unions' demonstrations, based on which 'nobody can judge me', can no longer be valid" (Renzi 14/05/2015); and, "We are making the school reform for our kids not to hire 200,000 people, for our students not as social benefits (for teachers-ed.)" (Renzi, 08/06/2015).
The external legitimacy strategy then emerges as predominant in the GS. The narrative strategy of external legitimacy presents the need to design reforms away from the ministerial bureaucracies and social concertation and towards society due to an innovative public consultation. "The method (of public consultation-ed.) is revolutionary since such a spread of public consultation has never been carried out in Italy. We have dedicated all our efforts so that the country could rethink the educational model of our kids" (Renzi 13/12/2014); and "the GS puts the student at the centre: in the long run, this means the nurturing of citizens to serve the collective" (Renzi, 12/3/2015).
The external legitimation strategy is nurtured by presenting the trade unions and the intermediation practices of the past as evil: "Listening to everyone does not mean to not do anything; otherwise, there is going to be paralysis, which is the same behaviour that has been blocked in Italy for 20 years. We will not allow this behaviour any longer" (Renzi 22/02/2015); "previous school minister Giovanni Berlinguer imagined the Italian school as the school of autonomy. We are available to listen to unions on everything, but the school effectively functions if it belongs to everyone (i.e. not only to the unions-editor's note). We are not willing to stop the quality increase in the Italian school" (Renzi 12/05/2015); and, "If the GS is approved, then 100,000 teachers will be permanently hired. If it does not pass, then unions will continue to contest, but teachers will still be precarious (Renzi 03/05/2015).
In sum, Renzi used narratives aimed at expanding the conflict outside both the labour and school subsystems to gain external legitimacy. While he nurtured the internal legitimacy of the JA by supporting the right of the young generations to have jobs, he strongly disregarded the internal legitimacy coming from the unions, which are key actors, especially inside the school system, where they represent 60% of employees. By playing the devil shift towards the unions and the methods of concertation and collegiality, Renzi made his policy narratives extremely coherent with his political narratives; he also, however, played against deep policy and political beliefs in the school system. This strategic choice was much more costly in the GS than in the JA, as it activated the mobilisation of powerful actors in the school system. This asymmetry is also evident in the counternarratives of his opponents. In the case of the JA, the narrative of some trade unions showed signs of support and learning in favour of the reform, "precarious work is the priority, to fight it we are ready to revise together the Article 18" (Cisl 20/09/2014). The main arguments of the opponents pointed mainly to the scarce credibility of Renzi as a leftist leader: "Renzi is friend to the strong powers" and "this is a regressive measure, a conservative revolution, like with Margaret Thatcher" (PD minority, 29/09/2014 and26/12/2014). Ultimately, they were unable in the end to beat Renzi.
In the case of the GS, the narratives of the opponents change from (mildly) supportive to increasingly dramatic and negative. "We have asked for similar interventions for a long time. They will give more stability to schools as the precariat is not a problem only for the teachers" (Flc Cgil 04/09/2014). "(The behaviour of the government) is a lack of respect to the school, is only façade talking and propaganda (…) The method is not serious" (Cgil, Cisl, Uil, Snasl, Gilda, 12/11/2014); "The proposal of merit-based pay is simply offensive for the teachers, and it is a colossal fraud" (Uil and Cgil, 14/02/2015); and, "The proposal dramatically accentuates the powers of the school manager by prefiguring a single man in command while downsizing the role of the other subjects operating in the system, from collegial bodies to individual teachers" (Cgil, Cisl, Uil, Snals-Confsal e Gilda, 04/05/2015).

Discussion: political narratives matter
The in-depth comparison of the two case studies presented above shows two dissimilar decisional outputs of two policy reform processes that occurred almost at the same time and that were steered by one political leader. Furthermore, the two policy processes were characterised by the attempt to introduce large reforms based on solutions that never had the chance to be fixed in the governmental agenda. The relevance of the political leader and his narratives was first certified by the ability to prioritise solutions that nobody had had the willingness or political strength to seriously consider politically viable. From this perspective, in both cases, the role of the prime minister was characterised by his effort to show himself as an ideational leader (Stiller, 2009) with a political narrative based on an apparently strong policy orientation; that is, the leader was there to change things as nobody had ever done before.
Given these similarities, why did Renzi succeed in one case and fail in the other? Why was the original policy proposal of the Job Act approved by the Parliament while the original proposal of the Good School was significantly changed? What does this experience tell us in terms of the ability of leaders to use policy narratives to seek policy change?
To answer these questions, the first point to consider is that the political leader largely adopted the same policy narratives. The narrator was the hero promising a solution to entrenched problems and very often also became an angel. The villains were the previous government and, above all, the unions (that often were "devilised"), and the victims were the society, as well as individuals in some broad social or professional categories, such as the unemployed of the younger generation and unstable workers. Another similarity that can be defined as the political dimension of the narrative is the choice of disintermediation as the indispensable procedural condition to achieve the policy goals. These similarities have been pursued despite the different characteristics of the two policy subsystems; however, if policy narratives are tools through which political leaders can shape beliefs and actions in a specific policy field (Jones & McBeth, 2010;Stone, 2012;Zahariadis, 2016), then this incoherence between the adopted narrative framework and the policy subsystem of application is absolutely relevant in showing the potential analytical relevance of the concept of political narrative. In fact, the similarity of the two adopted policy narratives, despite the difference between the two subsystems, clearly depends on the will of the political leader to exhibit his coherence in implementing the main goals of his political vision. This choice (which was required due to the strength of the political narrative) did not allow him to adapt the policy narrative to the characteristics of the subsystem, impeding any kind of flexibility.
Thus, this choice can be considered constrained by the necessity of the political leader to coherently implement a political narrative that breaks with that of the past based on strong personal leadership able to crush all the veto powers with the goal of pursuing the collective interest and expanding his political consent beyond the traditional boundaries of the Italian left (Salvati, 2016).
Thus, the strategic mistakes made in choosing the same policy narrative do not refer to the individual characteristics of the narrators (explained in terms of "superego") but rather to a conscious choice to coherently enforce the pillars of a narrative of political breakup: new ideas, new solutions and, above all, a new leadership style. This finding means that disintermediation represents a major change with respect to inherited practices (the consultation with the public in the case of the GS is clear proof of this). The needs and goals of the political narrative were more important than the content of the policy narratives themselves. The narrator wanted to be a political hero more than a policy hero, and political change came before a policy change. Most likely, if Renzi had decided to negotiate with the major trade unions concerning the content of his school reform, he would have obtained more effective results (for example, exchanging the hiring hire of precarious teachers with the strengthening of the institutional autonomy of schools and a stronger managerial role for principals). However, this would have shown continuity in the inherited policy and political style, which was exactly the opposite of the goal of his narrative of political breakup. What the leader had in mind in choosing the greater coherence between his political and policy narratives was the expectation of winning the policy change battle with only external legitimation. Furthermore, the leadership style exhibited in his political narrative was the typical case of direct (Bass & Valenzi, 1974), strong (Little, 1988), and transformative (Burns, 1978) leadership. Thus, the leader had no choice: his political narrative was designed to achieve heresthetic purposes, to reshape the inherited political and policy practices and to enlarge his political support. Due to his very demanding political narrative, he could not find a compromise on school reform to avoid a significant loss for his political narrative.
The pursued coherence between political and policy narratives shown in the analysed case of Renzi is very interesting in terms of improving the NPF. In fact, the possibility that a political leader, with a certain leadership style mainly interested in enforcing the political narrative, even at the cost of losing at the level of policy narrative (although he may have won), could contribute to a more effective inclusion of the political dimension in the NPF.
First, the analysed case shows that the political narratives of leaders matter in terms of policy narrative dynamics. Ceteris paribus, a leader's political narrative can predict policy outputs and can be considered a driver of the policy narrative battle. Thus, the result of the policy narrative cannot be analysed only in terms of policy output (the content of the adopted decision) but also in terms of political outcome (in terms of changes in political support and legitimation) (Nye, 2008). This possibility, although not applied to political leaders, has already been sketched out in the NPF literature. For example, Jones and McBeth (2010) hypothesised that "groups will heresthetically employ policy narratives to manipulate the composition of political coalitions for their strategic benefit" (p. 346), while Gottlieb et al. (2018) showed that narrators use narrative strategies that correspond to the side of the issue they support, regardless of whether they are winning or losing. Thus, the fact that narrators are interested not only in obtaining the expected policy output but also in obtaining some political rewards has emerged in NPF scholarship, but it has not been considered a potentially constant dimension of the framework. Including political leaders as one of the fundamental actors of policy narratives also allows us to consider the political dimension in which the policy process develops and opens the door to the consideration that in every process, the battle over the narrative can lead to both political and policy results.
Second, by including the political narrative in its conceptual tools, the NPF could strengthen its systemic perspective of policy-making in different fields by considering how political leaders influence policy narratives according to their political narrative. From this point of view, for example, to better understand the policy narrative of Donald Trump on different policy issues and his propension to invent facts in every policy issue (Jones & Mcbeth, 2020), it might be helpful to reconstruct his political narrative, his vision of the country and his ability to construct what Gardner (1995) calls a story of identity that draws on a story latent in the population, bringing it new attention and adding a twist.
Third, by focusing on the political narrative of political leaders, the NPF could also better grasp the political resources of the supporters of the different policy narratives on specific issues. In the case examined in this paper, the political leader is the direct promotor of a policy change, and his political narrative is a fundamental driver of the policy process. However, there are many policy issues on which the political leader is not a promoter, but that could, at a certain point, touch his or her political narrative and push the leader to become part of a policy coalition. Administrative and judiciary reforms during the Renzi government may be a case in point (Piattoni, 2016).
Finally, while our cases of reform are peculiar, political leaders who promote major policy changes are not uncommon. Leaders such as Margaret Thatcher, Tony Blair and Barak Obama are strong, transformative leaders with a demanding political narrative (Nye, 2008). Renzi's political narrative is also demanding and does not allow him to be pragmatically oriented in designing the related policy narrative based on the characteristics of the policy subsystem. In contrast, we expect that when the leadership style is more consultative, group-oriented and transactional, the political narrative will be less path-breaking and assertive, leaving the political leader more room for manoeuvring when designing policy narratives.

Conclusions
In this paper, we have focused on the relationships between the political and policy narratives of political leaders as potential drivers of policy change. The conceptual distinction between political and policy narratives and the focus on political leaders is meant to add value to the main assumptions of the NPF, which has not yet given enough weight to the strategic role of political leaders as narrators and to the fact that they are also bearers of a more general narrative through which they characterise their political vision to pursue consensus both in public opinion and at the electoral level. The empirical analysis has compared two policy processes led by a political leader who, despite deploying the same policy narrative, reached different legislative outputs. The main theoretical assumption of the empirical analysis is that while political leaders are expected to design narratives according to the structural and ideational characteristics of the involved policy subsystem, this is directly linked to the characteristics of the political narrative. According to the empirical evidence emerging from the comparison, the failure of Renzi's policy narrative to win in one of the two policy fields is a consequence of the lack of a tailored design for his narratives, which did not take into account the structural and ideational characteristics of one of the policy subsystems under investigation.
However, because the internal characteristics of the school policy subsystem are well known, the apparent lack of a policy narrative that aligns with them is not necessarily a mistake by the leader; instead, it may be a consequence of a conscious strategy embedded in the characteristics of his political narrative. Thus, we have proposed that when designing policy narratives, political leaders are necessarily oriented not only towards solving policy problems but also towards achieving general political goals. We have proposed conceptualising this as a political narrative, that is, a set of basic political beliefs, values and goals that political leaders pursue through their political and policy actions.
Overall, the focus on political narratives looks very promising for better understanding how politics and top politicians act in the theatre of policy narratives and as a potentially fruitful analytical tool for grasping whether and how political leaders shape the relationships between their political and policy agendas. Indeed, considering the current personalisation of politics and the increased role of political leadership, the inclusion of the role of leaders' political narratives (and their logics and trade-offs) is indispensable for enlarging the scope and analytical capacity of the narrative turn in public policy.
Devil shift A strategic story exaggerating the power of an opponent Shanahan et al. (2013) Angel shift A strategic story emphasizing the will and commitment of an actor/group/coalition to solve the problem Shanahan et al. (2013) Search for external legitimacy Appeal to the general public or to international actors to justify a policy option Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.