The policy integration game? Congruence of outputs and implementation in policy integration

Researchers agree on the importance of policy integration in the analysis of responses to complex problems, yet they often use this concept to indicate integrated policy programmes (IPPs) as opposed to policy integration in practice (as performed by actors). Describing how IPPs are reshaped, while they are being implemented opens new research venues in the study of the policy integration process and its effects. This article theorises the ‘(in)congruent implementation' of IPPs and illustrates its theoretical contribution with the case study of the local implementation of the national ‘Security Decree' in the Italian cities of Bologna and Pesaro. The paper makes a theoretical contribution to the study of the political aspects of policy integration. In particular, it suggests that three factors—the misalignment of state and local policy frames, the incommensurability of different subsystems’ interests, and local politicians’ risk avoidance strategies—might be led to the process of reframing national integrative intentions from the bottom up.


Introduction
Policy integration (PI) refers to the process through which actors across policy subsystems 1 define administrative coordination, policy instruments and governance arrangements, as well as policy resources to ensure the consistent implementation of the different dimensions of a complex problem (Cejudo & Trein, 2022). Even though researchers agree that PI can help us address cross-cutting policy problems, such as climate change, public health, and migrants' integration, we know very little about the 'real impact' of integrated policies (Candel, 2017). This makes the concept controversial and its application to political practices contestable (Watson et al, 2008). The main challenge to evaluating the impact of such policies is that PI outcomes are very difficult, if not impossible, to measure (Biesbroek, 2021). Hence, scholars prefer to analyse the policy integration process (e.g. Russel et al., 2020) and, more rarely, the local implementation of integrated plans (Nilsson et al., 2009).
The present study adopts the latter approach to analyse the effects of policy integration. It focuses on how national integrative intentions get reshaped at the local level, where they are ultimately put into practice (c.f., Watson et al., 2008). Policy implementers (in this paper, local actors involved in the implementation of a national policy) play a role in both the administrative and the legislative dimensions of PI, giving feedback to (national) policy 'formulators' and shaping the policies' effectiveness from the bottom up (Knill et al., 2020). In practice, PI is seldom congruent with politically agreed-upon integrative principles (Adelle & Russel, 2013;Jordan & Lenschow, 2010). Therefore, scholars tend to distinguish between integration as formulated on paper and integration that actually takes place ensuing actors' concrete integrative actions (c.f. Metz et al., 2020). Yet, the reasons for this 'incongruence' rarely attract scholarly attention and we know little about the link between (national) policy production and (local) implementation (Adam et al., 2019), as well as about the political bargaining necessary for the translation of general integration principles into specific practices .
Logically, we can expect that the implementation of national integrated policy programmes (IPPs) triggers several political dynamics at the local level because such policies are set in motion by multiple actors across various policy subsystems (Metz et al., 2020). These actors need to reframe sectoral agendas to make the latter fit into the (national) PI doctrine and negotiate shared solutions to pressing policy problems (Willems et al., 2021). Throughout this process, local actors might use their leeway to tailor national policy frames to local contexts (Scholten, 2016), i.e. they can reshape PI from below, while they implement it (Buizer et al., 2016).
To theoretically unpack this phenomenon of 'bottom-up integration', my paper builds on the well-known metaphor of the "implementation game" (Bardach, 1977), which in this particular instance describes the political dispute between the 'losers' and the 'winners' in the PI process during the enactment of integrated policy programmes (IPPs). This theoretical lens can help theorise the political aspects of policy integration , as well as its political mechanisms (Candel & Biesbroek, 2016).
The article draws on original qualitative data and uses the cases of the local implementation of the national Decreto Sicurezza (Security Decree) in the Italian cities of Bologna and Pesaro. The implementation of this policy-which combines immigration, migrants' integration, and security policies-dominated the public debate in Italy in 2019 and its enactment faced strong local resistance. Furthermore, the issue re-emerged in the aftermath of the 2022 Italian general elections and the ensuing right-wing takeover of Palazzo Chigi. 2 Here, it gives us the opportunity to discuss the political tensions that underlie policy integration. In particular, the case studies suggest that three factors-the misalignment of state and local policy frame; the incommensurability of different subsystems' interests; and local politicians' risk avoidance strategies-might play a key role in the process of bottom-up policy integration, and its ensuing 'incongruence' with centrally led integration (or 'top-down policy integration').
As academic knowledge on this subject is still in its infancy, this study adopts a strongly exploratory approach. It contributes to the existing literature by proposing three ways in which the political aspects of policy integration may affect the implementation of integrated plans. This 'pilot-approach' illustrates the different political tensions that underlie policy integration-i.e. disputes between different levels of government ('vertical games'), policy sectors ('horizontal games') and political elites ('blame avoidance games'). Nevertheless, it cannot offer an empirical basis to draw generalisable conclusions on this subject. In short, this paper constitutes the first step in a long-term research agenda. Still, its contributions are important, as it analyses the politics of PI implementation  and it can demonstrate how to evaluate the successful implementation of integrated policies (Biesbroek, 2021;Candel, 2017), i.e. in terms of the degree of '(in-)congruence' of outputs and implementation in policy integration.

Integrated policy programmes
A policy programme consists of both "substantive" and "institutional" elements (Knoepfel et al, 2007, see Table 1). Integrated policy programmes' (IPPs) overarching objective is the harmonisation of several sectoral goals. Such a harmonisation can result from different integration strategies and combinations of policy instruments (Cejudo & Michel, 2021). In the case of the Security Decree, which is the focus of this paper, national policymakers sought to achieve 'public security' by using several provisions, such as the prohibition for local municipalities to register asylum-seekers in their town halls (art. 13 of the Security Decree) or the revocation of the Italian citizenship of those who commit offences related to terrorism (art. 14), to intervene in migration policies. IPPs can also contain institutional instructions for the set-up of networks meant to oversee their implementation across policy subsystems; the allocation of resources; as well as rules and administrative procedures to be followed during the implementation process. These institutional elements should both minimise arbitration and conflict across policy subsystems and levels of government and reduce the risk of introducing inconsistencies in the institutional regime (Gerber et al., 2009). For example, the Security Decree set in place some provisions that increased the power of the Ministry of the Interior (art. 32-38), as well as some financial provisions (art. 39-40). 1 3 The analysis of the constitutive elements of an IPP informs us of 'what' sectors will be integrated and of 'how' the integration must take place, i.e. of the programmatic integration (hereafter, I use this term and the term 'law-based integration' interchangeably). That is, of whether policy implementation intends to span across several policy sectors or if it only aims to integrate a salient issue into one or more sectors. Furthermore, the study of such plans can shed light on 'why' the integration of different sectors is deemed necessary. Policy programmes play a "framing role" (Knoepfel et al., 2007, 190) and are formulated to select, organise and make sense of complex problems to steer public action from both a policy and a political perspective (Rein & Schön, 1993). Even though all policy programmes have political and policy-related functions, one orientation can become dominant, which explains the reasons behind PI. For example, this paper will illustrate how the Security Decree strengthened a specific policy frame, later defined as the 'migration-security nexus', mainly for political purposes.
In short, one can discuss the different orientations of the same policy and investigate whether programmes are policy-oriented or power-oriented, 3 as Table 2 shows in more detail. This type of analysis consists of assessing whether an IPP is motivated by, and carried out for, political reasons or whether it comes with the instructions necessary for it to be translated into local actions without generating political controversy, i.e. confrontations between different levels of government, policy subsystems or/and political elites. This paper will present the Security Decree as a power-oriented IPP-an attempt to integrate migration and security issues for political reasons rather than in order to solve pressing problems.

Recognising
Policy-oriented Acknowledging a problem as complex and cross-cutting, rather than as a narrow and an easily solvable one Designing Defining goals, measures, evaluative elements, and instruments across policy sectors Establishing governance configurations, implementation rules and administrative procedures to foster cooperation among policy subsystems Power-oriented Legitimising Branding and naming a policy to legitimise political directions and established actions Binding stakeholders to a shared roadmap Establishing or re-arranging cross-sectoral networks to provide or withhold responsibilities from actors/institutions Maintaining power asymmetries in multi-level or cross-sectoral confrontations Substantiating Substantiating political claims about PI Following-up on electoral pledges Linking cross-cutting problems to a lack of government resources to demand more power Expanding problems by linking them to politically salient and popular solutions

Policy integration in practice
Nevertheless, the analysis of IPPs only is a first step in improving our understanding of the policy integration process. Even a well-designed programme might not change the interactions between subsystems at the level of policy implementers and lead to "law-based integration" (integration on paper) diverging from "actors-based integration" (integration in practice) (Metz et al., 2020). This incongruence will hereafter be presented as a problem of implementation. Policy implementation follows policy programming and seeks to realise the concrete objectives of a public policy (Knoepfel et al., 2007). When an integrated policy is implemented, its elements are "assembled" (Bardach, 1977, 37) or translated into political actions that determine the degree to which different subsystems integrate in practice. For example, policy interventions across sectors are sequenced and some policy issues are prioritised over others (the outputs of this process are called 'actions plans'); actors from different subsystems establish administrative coordination and governance arrangements to monitor the implementation process and to initiate joint initiatives (e.g. integrated budgeting, reports that inform of the benefits and the costs of cross-sectoral linkages, interdepartmental action plans, etc.); finally, resources and responsibilities for the implementation are assigned to each actor (c.f. Cejudo & Trein, 2022). Therefore, integrative actions can either strengthen or weaken the interactions between actors from different policy subsystems and can trigger changes in the everyday organisational unfolding of the policy process.
Nonetheless, implementation allows implementers to reshape initial policies on the ground: implementing actors often seek to negotiate and compromise on the translation of policies into actions, to the point that it is often difficult to determine where formulation ends and implementation begins (Hill & Varone, 2021, 182). With integrated policies, the challenge consists in agreeing on implementation strategies that ensure the consistent implementation of the different sectoral elements, which are "in the hands of many parties, most of whom are in important ways independent from each other" (Bardach, 1977, 37). In other words, the provisions included in an integrated programme belong to different sectoral policy communities, which do not necessarily share the same objectives, views and routines, but whose cooperation is necessary to forge the integration between the sectors. Thus, the implementation of PI can be congruent or incongruent with the established programme (as described in Sect. 2.3).

(In)congruence of programmes and practice of policy integration
The divergence between the integration that actors agree on in writing and the integration that actually gets implemented-or the 'incongruence' in PI-results from the translation of general principles into positive applications (Jordan & Lenschow, 2010). Most of the literature on PI defines this phenomenon as a consequence of vertical dynamics, that is, as a problem of multi-level governance (e.g. Nilsson et al., 2009;Watson et al., 2008) or in terms of 'policy coherence'. Here, I theorise it more generally as the inconsistency of law-based and actor-based integration (i.e. of programmatic integration and integration in practice) (Metz et al., 2020).
Policy subsystems not only pursue their own sectoral interests, but they also have different organisational routines, jargons and resources. In addition, they are sensitive to focus events and the successive re-orientation of policy priorities to different degrees. In 1 3 the context of such an 'inherent' tension (see Cejudo & Trein, 2022), policy subsystems' involvement and policy framing can evolve either narrowly or broadly when IPPs are set in motion. For example, a limited number of sectoral concerns could become dominant in the implementation of a particular policy (while others could be neglected), leading to the narrowing of the scope of policy integration (which becomes re-framed to solve the aforementioned dominant problem). Alternatively, the implementation process can include more policy subsystems and sectoral concerns than originally programmed, thus broadening the PI's scope.
The study of (in)congruence in PI can help us understand which of the IPP's substantive elements get de facto implemented and how the implementation takes place from an institutional dimension. In other words, in the implementation process, one can assess whether: (1) The number of goals and instruments brought into practice conforms with the policy programme (congruence in policy content).
(2) The actors/organisations from all policy sectors behave in a cooperative manner and comply with the institutional provisions of the policy programme (congruence in policy procedures).
These two aspects leave us with four forms of (in)congruent implementation of PI, namely full congruence, substantive incongruence, procedural incongruence and full incongruence (see Table 3).
Full congruence takes place when the IPP's policy content is fully implemented and the implementation process reflects the procedures envisaged by the policy's formulators: actors across various policy subsystems establish, strengthen, or re-arrange their coordination patterns to conform with the instructions contained in the IPP. They thus attain all programmed policy goals and instruments and ensure that the implementation process evolves in a compliant (with the program's rules) and cooperative (among the relevant stakeholders) manner.
Procedural incongruence takes place when actors seek to attain all programmed goals and instruments across different sectors (congruence in policy content), but their coordination in the implementation process does not conform with the rules set by the IPP. For example, this outcome can result from a politico-administrative malfunctioning in the delivery of services within the implementation network or at the street-level, whereby some actors are excluded or decide not to comply. The implementation process might not unfold in a participative and compliant manner because actors could be reluctant to cooperate or they might be forced to work in silos. Substantive incongruence refers to the opposite process: it occurs when actors across policy subsystems are involved in the implementation process and organise the implementation of PI reforms in a coordinated manner and in compliance with the procedures envisaged by the IPP, but they drop substance from (or add substance to) the IPP's policy content. They thus seek to attain a different number of policy goals and instruments across policy sectors. For example, actors in the implementation network might coordinate with each other to abandon one or more substantive elements of an IPP or to enlarge the scope of PI.
Full Incongruence takes place when two conditions are met: actors do not seek to achieve one or more of the IPP's goals and instruments (or, alternately, they add a number of policy goals or instruments, de facto re-shaping the IPP's substance) and the implementation process does not reflect the procedures envisaged by the policy's formulators. Therefore, both the policy's content and subsystems' involvement fail to conform to the IPP's instructions.

The political aspects of policy integration
Incongruent implementation clearly does not only result from actors' intentional behaviours (i.e. political games) but can also often emerge as an unintended consequence of the institutional, organisational, administrative or policy-related circumstances that hinder policy implementation in a specific context. Notwithstanding this fact, advancing an encompassing explanation of IPPs' 'incongruent implementation' falls outside the scope of this paper (for a review, see Candel, 2017). Rather, I focus on some of the most relevant political factors that affect the implementation process, as discussed in the literature on PI.
Scholars have only recently begun investigating the political aspects of PI, such as party politics and power struggles (Bocquillon, 2018;Maggetti & Trein, 2021;Trein et al., 2021). However, political variables retain strong explanatory potential for the study of PI 'on the ground' (Jordan & Lenschow, 2010;Willems et al., 2021). The existing literature shows that PI is crafted and performed by political actors (i.e. actors with political agendas), such as political parties, political leaders or representatives of organised interests, who engage in building cross-sectoral linkages with policy communities located outside the institutions in which they operate (Bolleyer, 2011, 471). In so doing, political actors are able to counteract policymaking's sectoral logic by connecting different segments of the political system, thus reducing PI's organisational complexity (Ibidem). This also speaks to the concept of "integrative leadership", which is often used to describe political leaders' 'brokerage power' in integrating people, resources and organisations across sectoral boundaries (Candel, 2021;Crosby & Bryson, 2010). The high number and the heterogeneity of the actors who have a stake in the PI's implementation process mean that PI practices are highly contextual and politically motivated (Jordan & Lenschow, 2010).
This political perspective portrays the implementation of IPPs as a complex, chaotic, and dynamic process (Willems et al., 2021). The latter unfolds as a "system of loosely related games" between the 'losers' and the 'winners' of the PI process (Bardach, 1977, 38). Hereafter, I shed light on three latent tensions that might underlie the PI process: the misalignment of state and local policy frame; the incommensurability of different subsystems' interests; and local politicians' risk avoidance strategies. These tensions represent three different dimensions of political conflict: 'vertical games', 'horizontal games', and 'blame avoidance games', respectively.

Vertical games
Local policies often deviate from national paradigms as a result of differences in policy frames (i.e. competing definitions of policy problems and solutions) across different levels of government (Scholten, 2016). Policy frames' importance to policy integration is well documented in the literature on PI (e.g. Candel & Biesbroek, 2016;Nilsson, 2005). Policy frames form the cognitive basis of policy integration: they determine if the cross-cutting nature of a problem is recognised and anticipate the solutions to the problem in question (Candel & Biesbroek, 2016). Yet, few scholars have analysed the misalignment of policy frames across different levels of government to explain the local implementation of an IPP (for an exception, see Scholten, 2016).
Local actors can both facilitate and hinder the implementation of IPPs (e.g. they can decide whether they want to mobilise the resources necessary for PI) depending on political dynamics (Bolleyer, 2011). For example, local implementers might seek to reshape a national IPP in the course of its implementation in order to weaken the national integrative frames they oppose. If local leaders have the political resources and the will to confront the national government, they can perceive the implementation of an unwanted national IPP as a 'window of opportunity' to highlight its pitfalls and to offer alternative integrative ideas. This process of re-framing policy integration from the bottom-up can be highly beneficial to the effectiveness of such policies (Knill et al., 2020). Nevertheless, it can also create scenarios in which local decisions do not conform to national expectations and do not support the government in reaching its targets (Nilsson et al., 2009;van Stigt et al., 2013). In both cases, local deviations from national integrative frames result in what this paper defines as "incongruent implementation": Proposition 1 IPPs can face incongruent implementation if the national formulators in charge of policy production and the local implementers responsible for policy practices follow divergent policy frames for policy integration.

Horizontal games
The implementation of integrated policy programmes also depends on the salience, power and resources wielded by the relevant policy subsystems. In its most simple terms, PI materialises in a number of political decisions about 'who gets what, when and how' across policy sectors (Bocquillon, 2018, 340). The main political tension triggered by the local implementation of an IPP is the horizontal conflict among actors with sectoral interests (Cejudo & Trein, 2022). Sectoral agendas must be reconciled if policy integration is to be put into practice, i.e. multiple actors must work together to reframe previously separate agendas into an interconnected whole (Willems et al., 2021, 85). For example, this can occur when actors agree on a set of overarching policy objectives, which should be prioritised over sectoral concerns, and on a set on policy instruments (Cejudo & Michel, 2021). Yet, in any given time and context, sectoral interests can be perceived as politically incommensurable. During the implementation phase, such a perception might manifest as an overt horizontal conflict between local actors that belong to different policy subsystems. Therefore, I formulate the following proposition:

Blame avoidance games
Policy integration is politically costly for some and, simultaneously, a political opportunity for others. Stakeholders' support for and opposition to policy integration might be a function of their levels of risk aversion (Howlett, 2014). Politicians do not want to be held responsible for policy failures when they pursue their political agendas and seek re-election (Hinterleitner, 2017), so they only claim credit for PI's achievements and renounce responsibility for its costs. This is to say that blame-avoiding behaviour results not only from a policy's actual political costs, but also from the perceived risks politicians face when they implement the policy (Ibidem). Therefore, we can expect the way local politicians behave when they implement a nationally salient IPP to depend on their perceptions of its popularity: Proposition 3 When they implement a nationally salient IPP, local politicians might decide to ensure congruence if they perceive the IPP as 'popular' among the public, because they want to avoid the political costs of incongruent implementation (notably, electoral risks).
The argument behind the three propositions is that policy integration might be put into practice in the service of political agendas ('powering') rather than for the purposes of problem-solving ('puzzling'). Each proposition concerns a different political arena where conflict takes place: in the first arena, the main cleavage is between national and local actors; the second is the scenario of a power struggle among different policy sectors; and the third is defined by the competition among local elites at the city level. The remainder of this paper uses the case of the local implementation of the Decreto Sicurezza in Italy to further illustrate these dynamics.

The local implementation of the 'Decreto Sicurezza' in Italy
The case study of the local implementation of the Security Decree (Decreto Sicurezza) in the cities of Bologna and Pesaro draws on ten semi-structured interviews with key policy implementers conducted in June 2019, an in-depth analysis of the legislative act itself (no.113/2018), and a review of the academic literature and newspaper articles on the topic. Information about my interviewees is available in the "Appendix". The data collection followed the principle of a 'triangulation of sources' (interviews, policy documents, media). The interviews focussed on the following issues: (1) Explaining one's role (what is your role, which actors do you cooperate with in the governance of displaced persons); (2) evaluating the national policy (general criticisms and context-specific challenges); (3) incongruent implementation (policy content; strategies of deviation from the national frame); (4) local ideas about migrants' integration and ideas about integrating security and migration policies; (5) the dynamics of vertical governance (coordination and conflicts between national and local institutions); and (6) the dynamics of horizontal governance (coordination and conflicts among policy implementers). I used Atlas.ti to transcribe and analyse my interviews. The data analysis followed an abductive methodology-a research strategy in which the researcher's initial ideas about how the casual process works are empirically contradicted, leading to a new phase of theory-building (Beach & Kaas, 2020, 14).
The local implementation of the national Security Decree cannot be used as an example to demonstrate the relevance of all concepts developed so far. Yet, it can help illustrate the concept of an IPP's (in)congruent implementation, as well as the fact that PI can be used as a political tool to gain power in a multi-level or cross-sectoral confrontation. In this case, the conflict opposed a right-wing Minister of the Interior and a coalition of municipalities administered by the centre-left (with Bologna and Pesaro among them), as well as two policy subsystems concerned with public security and the welcoming and integration of displaced persons, 4 respectively.

Decreto Sicurezza: key elements and the political orientation of a national IPP
The Security Decree entered into force in October 2018 and was amended in 2020. Here, I only consider its first version. 5 It is a complex piece of legislation and is comprised of forty articles divided into four main sections ('Titoli') with a total of eight subsections ('Capi'), as Table 4 illustrates.
The Security Decree integrates different policy goals-restrictions on immigration, migrants' integration, security and the fight against organised crime and terrorism-into a single policy program. This IPP can be considered a narrow integration attempt: it does not span across a large number of policy sectors in mainstreaming migration policies, but aims to integrate migration policies into general security policies (seeking to address concerns about crime by intervening in policies on asylum and migrants' integration).
The security-migration nexus caused a radical shift in the frame of the governance of displaced persons in Italy (Curi, 2019). Measures for social integration have been eliminated and many services have been reduced, bringing the system back to an emergency-led and outdated doctrine of incorporating newcomers through overcrowded closed centres managed by the prefectures, rather than through small and tailor-made local initiatives (Luraschi et al., 2019). Among the most controversial measures, art. 12 and 13 reduced the funds for and the beneficiaries of the locally managed SPRAR 6 projects and prohibited asylum-seekers from registering in town halls (iscrizione anagrafica). This policy blocked the provision of public services for asylum-seekers throughout the Italian territory, paradoxically creating a security issue (IPO2, IP03, IP04, IP07, IP09).
Another provision that caused conflict among different policy subsystems was the abolition of the status of humanitarian protection. (At the time, this status was most frequently granted by the prefectures.) This measure established a local process of bargaining between the local administrations, who did not want to deal with the increased number of undocumented migrants, and the prefects, who played a crucial role both in the recognition of the 4 I use this term to refer to refugees, asylum-seekers and migrants with humanitarian or subsidiary protection, as well as to those with a 'rejected' status who cannot be returned to their countries of origin. 5 The full text is available at https:// www. gazze ttauffi cia le. it/ eli/ gu/ 2018/ 10/ 04/ 231/ sg/ pdf. 6 The SPRAR-System of Protection for Asylum Seekers and Refugees-was the main (state-funded) project to welcome and integrate migrants on the Italian territory. The Security Decree dismantled it and renamed it SIPROIMI (System of Protection for Holders of International Protection and for Unaccompanied Foreign Minors). Its name and properties changed again in 2020. Today, it is called SAI (Welcoming Integration System). For further information, visit its website: https:// www. retes ai. it/.

Table 4
Translation of the Security Decree's main elements new forms of protection that replaced the humanitarian form of protection and in the management of physical infrastructure 7 (IP07). Scholars who have analysed the Security Decree agree that it is at odds with the objective of public security (Algostino, 2018;Curi, 2019). My interviewees also agreed with this interpretation and argued that migrants left without integration services can be forced into delinquency and can become victims of discrimination (for instance, in the housing and the labour markets) (IP01). Following the conceptualisation proposed in this paper, the Security Decree is a power-oriented IPP, i.e. an attempt to integrate sectors for political reasons rather than for the purpose of solving cross-cutting problems (see Table 5).
The following sections of this paper trace back the local implementation of the Security Decree in the cities of Bologna and Pesaro and analyse the political tensions that underlay this process, i.e. the conflict among different policy subsystems and levels of government, as well as local politicians' blame avoidance strategies.

The involvement of different policy subsystems
The two policy subsystems-migration and security-pursued different political goals during the implementation of the Security Decree. Crime prevention and law enforcement were the main concerns of the security subsystem. The migration subsystem sought to preserve the guaranteed basic rights of a vast number of displaced persons. Consequently, Results of the text analysis

Recognising
The IPP does not recognise asylum and migrants' integration problems as complex and cross-cutting, but rather as security issues (it does not span across multiple policy subsystems)

Designing
Among the main controversies related to the design of this IPP, Art.12 reduces the beneficiaries of integration projects and Art.13. prohibits migrants' inscription in town halls (iscrizione anagrafica) Legitimising The IPP's name contains the term 'Security'. It legitimises the migration-security nexus (the assumption that more immigration led to more crime) The Decree's title and introduction contain the words 'needed', 'urgent', 'extraordinary'.
These words are repeated six times in the span of nine paragraphs and describe a state of emergency. The latter is exacerbated using the legal tool of the Decree (which, according to the Italian legal system, should only be used in limited and extraordinary situations) The IPP gives power to the Minister of the Interior and limits local actors' responsibilities in the governance of displaced persons Substantiating The IPP (implicitly) substantiates existing political claims about migrants being responsible for crimes and raises the salience of the migration-security nexus in support of electoral pledges (2018) and in the lead-up to the European elections (2019) It increases the power of the Minister of the Interior and re-centralises the governance of displaced persons the sectoral interests of these subsystems appeared incommensurable and the local implementers across the two policy subsystems were not keen on cooperating with each other. The conflict among the prefectures, the local administration and the social cooperatives illustrates this intractable controversy. Many social cooperatives in Italy, including the one operating in Pesaro, decided to abandon the 2019 ministerial procurement to access the funds necessary to finance the new SIPROIMI 8 projects established by the Security Decree (IP02). While the Minister of the Interior accused the social cooperatives of profiting from the immigration 'emergency', most of the non-profit organisations affected by the Decreto Sicurezza argued that the reduction in funds-notably, a cut in the reimbursement from 35 to 21 euros per migrant per day-would not have allowed them to continue providing migrants with integration services, creating the risk that larger profit-oriented organisations would have taken this service over at migrants' expense.
To cite a more specific event, in 2019, the Prefect of Bologna announced the closure of a reception structure for displaced persons (ex-Hub via Mattei) and the resettlement of 200 asylum-seekers in Sicily, as well as the dismissal of the social workers employed in the structure. This order faced strong, months-long opposition from both migrants (who did not want to be resettled) and social workers (who did not want to lose their jobs), which forced the municipality to engage in strenuous negotiations with trade unions and the prefecture (see Luraschi et al., 2019).

Issue salience and the (perceived) popularity of the migration-security nexus
Migration issues have been highly salient in Italy for many years. They capture the public's attention and are often used by political parties, especially far-right populist parties. At the national level, the security-migration nexus was framed by the Minister of the Interior (who was also the secretary of the right-wing populist party The League), the original proponent of the Security Decree. The implementation of the Security Decree monopolised the public debate in Italy for several months (Luraschi et al., 2019). When this flagship policy was formulated and implemented, the public's support for The League and its leader reached a historic peak (together with the perceived popularity of the migration-security nexus), evident in the results of the 2019 European elections. (The League won the elections with the 34% of votes.) It is not surprising that the issue's salience was very important in both cities. It is more interesting to note that this happened for different reasons, which are tightly connected to the political contexts in Bologna and Pesaro: my interviewees in Pesaro acknowledged that local politicians were more reluctant to openly oppose the national IPP because the latter's local implementation coincided with the municipal elections where The League was the main opposition party (IP02, IP03, IP04, IP05, IP06). In contrast, Bologna organised frequent and popular demonstrations against the implementation of the Security Decree (see Luraschi et al., 2019), which perhaps forced the local administration to oppose the implementation of the Decree and engage in policy deviation (see below).
At the local level, the salience of the Security Decree depended strongly on the number of displaced persons hosted in the cities and affected by the Decree's entrance into force. This constitutes an objective metric to understand the different political pressures that the welcoming systems in the two cities have been subjected to during the Decree's implementation. As of January 2019, there were 1203 beneficiaries of locally managed integration services in Bologna at the municipal level and 2048 at the metropolitan level (BolognaCares, 2019). In Pesaro, this number hovered around 100 people (see Labirinto, 2019). Another related metric is the number of workers who risked losing their jobs as a result of the implementation of the Security Decree, which was, again, considerably higher in Bologna than in Pesaro.
Furthermore, the institutional effects of the policy were far harsher in Bologna than in Pesaro. Before the Security Decree, the municipality and its regional authorities (strongly linked from a political perspective against the national policy frame) had been investing significant resources in a 'regional hub' meant to manage the increased number of migrants. This strengthened Bologna's local autonomy in the governance of displaced persons (IP07). One of the goals of the municipality was to transfer all beneficiaries of its integration services from the national closed centres to the local projects (ex-SPRARs) (Ibidem). Bologna was one of the main national sponsors of the 'diffuse' model of integrating migrants, which contrasts with the hierarchical and emergency-led model pursued by the national government in 2019. Hence, the issue's salience was higher in Bologna than in Pesaro also because of the ongoing conflict over the distribution of responsibilities between the two levels of government.

(In)congruent implementation?
Applying the framework developed in Sect. 2, we can now finally discuss the extent to which the two cities have implemented the Security Decree, i.e. whether (1) the number of goals and instruments translated into practice conformed with the provisions of the Security Decree (substantive congruence) and if (2) actors/organisations across different policy sectors cooperated with each other and complied with the policy program's institutional provisions (procedural congruence). With the available data, however, only a snapshot-analysis is possible: what follows only captures the city-reaction to the centrally led reform (January-August 2019), rather than the more extended implementation process.
In this timeframe, the local administrations of Bologna and Pesaro initially sought to propose amendments to the national government through the National Association of Italian Municipalities (ANCI) (IP01, IP05, IP06, IP07). These attempts failed and, for the sake of the metaphor used in this paper, we should designate the municipalities as 'the losers' and the national government as 'the winner' in the first phase of the 'integration game'. The actors in the migration policy subsystems, especially the social cooperatives, also lost in this phase. In the security subsystem, the 'winners' are the prefectures and the Ministry of the Interior. The implementation process represents the 'second round' of the game and offers the aforementioned losers the opportunity to reframe policy integration while putting it into practice.
In spite of its similar point of departure, the initial phase of implementation in the two cities was characterised by important differences: The city of Bologna deviated from the national integration programme by adopting several strategies. First, as the provision related to the inscription of asylum-seekers (implementation of art. 13) was perceived as discriminatory, the local administration decided not to execute it (IP01, IP09). Yet, Bologna pursued a legal path of non-compliance (unlike other cities in Italy, such as Palermo or Naples, which refused to follow institutional rules, see Marrazzo (2019)): the local administration released the 'document of refusal' to a group of asylum-seekers (complying with the Security Decree) who requested to be registered in the town hall and provided them with legal support to appeal against this provision in court (IP01, IP09). The Civil Court of Bologna, which these same asylum-seekers appealed to with the support of the local administration, granted their request and rejected the national interpretation of the Security Decree (for further details, see Long and Albano (2019)). Backed by the Court's decision, the municipality stopped implementing art.13, which prohibited asylum-seekers from registering in the town hall (IP01, IP09).
Second, a number of initiatives were launched citywide to expand the PI's substantive scope beyond the security-migration nexus. They included housing policies, initiatives that created interactions between migrant communities and the citizens of Bologna (see 'Globologna'), economic collaboration pacts with north-African and Balkan cities (IP01) and welfare policies (IP10). Thus, Bologna illustrates the concept of an 'incongruence' of programmes and practices in the implementation of PI because it legally deviated from the national integrative frame: policy implementers behaved in a cooperative manner and were, overall, compliant with the policy program's institutional provisions (procedural congruence). Nonetheless, the goals and instruments that were brought into practice did not fully conform to the provisions of the Security Decree (substantive incongruence).
In contrast, in the analysed timeframe, the city of Pesaro implemented the national policy in a compliant, vertical fashion, following the rules and guidelines outlined by the Minister of the Interior (IP05, IP06, IP08). Nevertheless, the road to this form of 'full congruence' is interesting: the local administration preferred not to deviate from the national policy frame, opposing key governance partners from the migration subsystem (the social cooperative) who voted for a more radical reaction (IP02, IP03, IP04). One of the largest differences in opinion among the policy implementers in Pesaro concerned the inscription issue (i.e. the implementation of art.13 of the Security Decree). The local administration preferred to comply with the law, while the social cooperative sought to follow the example of the 'non-compliant' cities (Naples, Palermo). Eventually, the local government's line prevailed, and this ultimately prevented the 'incongruent implementation' of the Security Decree in Pesaro: policy implementers ultimately behaved in a cooperative manner and were generally compliant with the policy programme's institutional provisions (procedural congruence). Furthermore, they translated all the Security Decree's provisions into action (substantive congruence) ( Table 6).

Size as a moderating factor
Many of the differences between the two case studies above depend on the sizes of the two cities and their different capacity to mobilise resources for a divergence. Because of its larger size, Bologna faced a higher number of displaced people who were left without integration services and employees who risked losing their jobs. At the same time, it also had more resources to deviate from the national policy than Pesaro (i.e. to 'react' to the centrally led reform of the migrants' welcoming system). While elaborating on the structural differences between these two cities is not important for the purposes of this paper (the main objective here is to highlight the political factors behind policy integration), it is crucial to recognise that the local implementation of IPP does not only result from intentional behaviours (e.g. political strategies) but can also emerge as an unintended consequence of the institutional, organisational, administrative, or policyrelated circumstances that hinder policy implementation.

Discussion
The empirical section of this paper partly illustrates the main concepts theorised above, such as some of the forms of 'incongruent implementation'. It also explains the multifaceted political tensions that underlie the implementation of PI. First, my findings seem to corroborate the expectation that PI is highly contextdependent and politically motivated (Jordan & Lenschow, 2010). Factors, such as the misalignment of different policy frames, Bologna's larger size and resources, the harsher effects of the policy on its welcoming system, and the aspiration for more local autonomy dashed by the recentralising effects of the policy, might have plausibly played a key role in shaping the more radical 'incongruence' observed in Bologna. Likewise, the political context in which the policy's implementation took place in Pesaro-the electoral campaign and the perceived popularity of the national policy frame-might explain why the municipality opted for a less politically risky strategy.
Second, my work also shows that the incommensurability of sectoral interest might represent a key obstacle to the implementation of IPPs (Cejudo & Trein, 2022). This horizontal tension manifested in political disputes among different policy subsystems during the implementation of the IPP. In both cases, the degree of the mobilisation of the migration subsystems against the national migration-security mantra has greatly influenced the Decree's local implementation.
Third, the two case studies contradict the existing literature, which describes PI as an 'incontestable good'. They show that PI can be oriented towards gaining power rather than towards solving policy problems. This makes PI "contestable in practice" (Watson et al, 2008, 496).
Fourth, the case studies highlight the crucial contribution of local implementers in integrating policies 'from below' (Buizer et al., 2016) as well in providing feedback to national policy formulators (Knill et al., 2020). The city of Bologna enlarged the scope of the national policy frame (the 'migration-security nexus') and implemented a more 'integrated' approach to the governance of migration into the city. Notably, when the Italian Constitutional Court decided to declare art.13 of the Security Decree unconstitutional (decision n. 186 of 2020 9 ), this decision did not take the public by surprise, as it had been anticipated by the decision of the Bolognese Court and the actions taken by some Italian cities in 2019.

Conclusion
This paper's goal was mainly conceptual. It aimed to shed light on how written PI can differ from how it is performed (Metz et al., 2020). Furthermore, by analysing the politics of PI, the paper sought to theorise some of the political drivers of policy integration  and to illustrate how they might shape the integration process from below (Cejudo & Trein, 2022).
Overall, the paper posits that the PI process can become an arena for political confrontations across policy sectors, levels of government and political elites. PI's implementation provides intentional actors with the opportunity to redefine policy boundaries and to change policy provisions and frames while implementing them.
To illustrate this dynamic theoretically, this paper used the metaphor of the 'policy integration game', and advanced three propositions about (some of) the political drivers of 'incongruent implementation', i.e. the misalignment of state and local policy frame; the incommensurability of different subsystems' interests; and local politicians' risk avoidance strategies.
In so doing, the paper contributes to theorising some of the power struggles that underlie policy integration . Nonetheless, one must acknowledge that the paper's empirical findings are limited, and they cannot prove the validity of the three theoretical propositions it formulates. Since my overarching goal was to test the plausibility of these expectations and to redefine them accordingly, I adopted an exploratory approach that entailed the collection of a limited amount of data in a small number of cities. In addition, my research focussed only on a restricted timeframe (the initial phase of the implementation or the 'city-reaction' to the centrally led reform). The paper's overall research design therefore does not explore the external validity of its findings and cannot account for the whole implementation process. In acknowledging these limitations, I call on future work to assess and generalise the impact of political factors on the PI implementation process. Drawing on this paper, for instance, scholars could compare different types of IPPs in different political settings and study how their designs and 'orientations' (see Table 2 above) relate to their successful implementation (Mazmanian & Sabatier, 1981). 10 In spite of its limitations, this paper has analysed the PI process from a less commonly used perspective and has offered a viable approach to the study of the effects of PI (Candel, 2017) and the multifaced political tensions that underlie PI (Cejudo & Trein, 2022). Addressing the misfit between integrative intentions and the real-world emphasis attributed to sectoral concerns and policy subsystems might be not only a way to analyse the successes and failures of these policies (Candel, 2017), but also a point of departure for those who aspire to formulate clear criteria for the evaluation of PI (see Biesbroek, 2021).