Introduction

Over the last decade, policy scholars have paid increasing attention to the issue of policy integration (PI) and have progressively come to a shared understanding of the concept (e.g., Candel & Biesbroek, 2016; Cejudo & Michel, 2017; Tosun & Lang, 2017; Trein et al., 2021a, 2021b). It is now agreed that integrating policies is a complex process requiring (a) the coordination of stakeholders active in different policy subsystems, (b) a coherent mix of policy instruments and (c) policy arrangements to consistently implement integrated policy solutions (see the definition of PI proposed by Cejudo and Trein (2023: 5–6) in the present issue).

This article aims to contribute to this literature by investigating whether policy integration can occur ex post during the implementation stage, even when several policies were conceived in silos at the formulation stage. Indeed, most PI scholars consider that the formulation stage is the decisive stage for achieving policy integration. In their view, policymakers should discuss and address complex problems through a coordinated policy design and a coherent instrument mix (e.g., Candel & Biesbroek, 2016; Howlett & del Rio, 2015; Howlett & Rayner, 2006). A “poor policy design” that keeps producing sectoral policymaking is considered to be a policy failure that complexifies the implementation stage and makes PI impossible to attain (Capano et al., 2016: chap. 3; Vince, 2015). As a result, more than three quarters of PI studies exclude implementation processes from their theoretical focus (Trein et al., 2019:337), and few empirical studies have seriously analyzed whether and under what conditions policy integration eventually occurs—or fails to occur—at the implementation stage of the policy cycle.Footnote 1

This lack of interest in policy implementation among PI studies can be explained by the fact that many PI scholars take a top-down and sequential approach to the policy process (see Hood, 1976; Pressman & Wildavsky, 1984). In their view, a problem that is too narrowly defined during the agenda-setting stage or a siloed policy formulation that is sector specific cannot translate into integrated policy outputs and outcomes (e.g., Briassoulis 2017; Howlett et al., 2017; Vince, 2015). Thus, these scholars assume, like high-ranking officials, that achieving a higher level of integration across different policy sectors (horizontal integration) or different levels of government (vertical integration) will inevitably lead to more effective and efficient policy solutions to crosscutting issues (Giessen, 2011a, 2011b; Tosun & Lang, 2017: 559ss).

This dominant top-down perspective of PI has recently been criticized for being equivalent to searching for the "Holy Grail" or the "Philosopher's Stone" of policy analysis (Biesbroek, 2021; Candel, 2017; Miler & Glasby, 2016). These recent studies have also criticized the top-down vision of PI for being too simplistic since it builds on an idealized rational decision-making process that is at odds with "the much messier practices on the ground" (Candel, 2021: 356).

Echoing the recent criticism expressed by these scholars, the present study aims to shed light on how policy integration evolves when the policy process enters the implementation stage. Its main objective is to demonstrate that policy integration can also unintentionally occur at the implementation stage, i.e., through coordination mechanisms that were not anticipated by policymakers during the formulation stage and without starting a new policy cycle. Drawing on Kingdon's (2003) seminal work on policy entrepreneurs, I argue that the implementation stage can benefit from the leadership role of “integration entrepreneurs” exploiting windows of opportunity to bring together public policies.

Concretely, I investigate how three Swiss cities (Zurich, Bern and Geneva) have addressed the crosscutting issue of urban sprawl over the past four decades (1980–2020). This investigation reveals that sectoral policymaking led to design and implementation failures over the 1980s and 1990s. Then, the Federal Agglomeration Policy (AggloPol) was introduced in the early 2000s as a first deliberate attempt toward a more integrated approach to curb urban sprawl. However, empirical evidence shows that the decisive factor for curbing urban sprawl was not the introduction of the AggloPol, but another post-Keynesian reform introduced by the Swiss Confederation at the end of the 1990s, namely the transformation of the Swiss Federal Railways (SBBFootnote 2) from being a part of the federal administration to becoming a public limited company owned by the federal state. This reform was not coordinated ex ante with the spatial planning and agglomeration policies, but it radically transformed the real estate strategy of the SBB. As a result, the SBB was strongly incentivized to assume an integrative leadership role for curbing urban sprawl by building new neighborhoods next to the main train stations of Switzerland.

In sum, this study confirms that an “integration entrepreneur,” endowed with policy capacity and leadership resources, can successfully bring together a range of public policies during the implementation stage (Candel, 2021; Ross & Dovers, 2008; Trein et al., 2021a, 2021b). At the same time, it also suggests that the coupling of policy objectives, instrument mixes and implementation arrangements might remain coincidental rather than the result of a purposeful, well-planned and monitored integration process.

The remainder of the article is structured as follows. First, I specify the main characteristics of an “integration entrepreneur” by drawing on the concept of “policy entrepreneur” defined by Kingdon (2003). Second, I explain why concentrating on the crosscutting issue of urban sprawl within a federalist country is an appropriate empirical strategy to question the top-down perspective of PI. Third, I study the evolution of the Swiss federal legislation aiming to curb urban sprawl between 1980 and the 2020 and show that the post-Keynesian turn has led to a more integrated approach since the early 2000s. Fourth, I apply the same longitudinal comparison to major urban development projects developed in Zurich, Bern and Geneva. These projects have failed in the 1980s and 1990s, but have made a fresh start in the early 2000s and have been completed since then thanks to the entrepreneurial role of the SBB. Therefore, I conclude this article by arguing that “integration entrepreneurs” are able to integrate policies in an unintended manner at the implementation stage because they can activate coordination mechanisms that were not foreseen by policymakers at the formulation stage.

Policy entrepreneurs overcoming the challenges posed by the implementation of integrated policies

In this section, I present various types of policy failures that are closely linked to the formulation and the implementation of any integrated policy framework. I then argue that scholars who have aimed to resolve these failures have overlooked what happens when integrated policy frameworks get implemented. Drawing on the concept of “policy entrepreneur” theorized in Kingdon’s (2003) seminal study, I hypothesize that “integration entrepreneurs” are able to secure policy integration in an ex post manner throughout the implementation stage.

Avoiding integration failures throughout implementation: a challenging and overlooked task

Academic research focusing on policy integration (PI) has significantly increased in recent years. However, prominent PI scholars have collectively acknowledged that more research is needed to explain “how to assess whether expressed intentions to coordinate and integrate policies do actually result in integrated policy outputs and implementation practices” (Trein et al., 2021a, 2021b: 975). Some of these scholars have thus started to analyze more closely how integrated policy strategies get implemented (Cejudo & Michel, 2021; Cejudo & Trein, 2023; Maor and Howlett 2022). These scholars argue that fostering and maintaining coordination across policy subsystems is the basic essence of PI. They also acknowledge that the basic solution advocated by classical implementation studies, namely reducing the complexity of the policy design to reduce coordination needs (see Pressman & Wildavsky, 1984), turns out to be unhelpful. Nonetheless, these scholars remain anchored within a top-down and sequential vision of PI because they consider the elaboration of a comprehensive policy design to be the most important step of the integrative process. In sum, even if they repeatedly recognize that implementing integrated policy frameworks is a highly complex and challenging task, these scholars are still searching for the Holy Grail that could avoid policy failure.Footnote 3

There is of course a wide spectrum between policy failure and policy success (McConnell, 2010). On the one hand, “design failures” caused by an incoherent or vague formulation of policy objectives might occur and complicate the delivery of expected policy outputs and outcomes (Linder & Peters, 1987). On the other hand, it is often very difficult to know the extent to which a policy failure is due to a poor formulation, to imperfect implementation or to a subtle mix of both elements (Hill & Varone, 2021: 206–09). Therefore, avoiding integration failures is an especially challenging task: first, because policymakers need to elaborate policy designs that simultaneously consider the formulation and the implementation stages of several policies (Vince, 2015) across different policy subsystems (Jochim & May, 2010; Tosun & Lang, 2017) and second, because a comprehensive interaction of policy instruments through the implementation stage is a necessary but ultimately insufficient condition to secure the proper implementation of integrated policy solutions (Cejudo & Michel, 2021; Maor & Howlett, 2022).

Yet, facing such complexity is not surprising. Seminal studies of policy implementation have shown decades ago that “implementation deficits” or “implementation gaps” are frequently observed in policymaking processes, even when policymakers are determined to avoid them (e.g., Dunsire, 1978; Hood, 1976; Pressman & Wildavsky, 1984). Elected officials, who act as principals in formulating policy programs before delegating implementation tasks, have values and interests that are potentially divergent from the policy beliefs and professional norms of administrative organizations and street-level bureaucrats (SLB), who, as agents, fulfill the delegated tasks (Hupe & Hill, 2007; Lipsky, 1980). Meanwhile, if policy outcomes are not in line with stated policy objectives, elected politicians are often ready to blame administrative agencies or individual SLB for these “implementation failures” (May, 2015).

Finally, a common feature of policy formulation is to increase the number of policies being brought together under a particular heading. This has the danger of increasing “transversal transaction costs” (TTCs) whose effects are very likely to be higher than the efficiency gains brought by the design’s extension. In the longer term, TCCs are thus likely to lead to an overall loss of policy coherence which will further hamper the formulation and implementation of integrated policies. This has been identified as an “Institutional Complexity Trap” (ICT) because policy designs get more complex without generating concrete gains in terms of policy integration at the implementation stage (Bolognesi & Nahrath, 2020; Bolognesi et al., 2021).

From the policy entrepreneur to the integration entrepreneur

The present study capitalizes on this scholarship on the design–implementation nexus and its relevance for policy integration approaches. It is also guided by the final proposition that Cejudo and Trein (2023) formulate about the bottom-up approach of policy implementation. The authors argue that, even in the absence of national integrated policy strategies, the likelihood of observing integrated implementation strategies is higher if public managers, street-level bureaucrats or local governments benefit from greater flexibility of action (ibid.: 25). Drawing on this understanding, I aim to explore the conditions under which an “integration entrepreneur” may be willing to invest their own resources throughout the implementation stage (a) to compensate for a design deficit and (b) to develop and successfully apply an integrative policy strategy. This concept of “integration entrepreneur” naturally relates to the broader concept of “policy entrepreneurs” defined by Kingdon (2003): “These entrepreneurs […] could be in or out of government, in elected or appointed positions, in interest groups or research organizations. But their defining characteristic, much as in the case of the business entrepreneur, is their willingness to invest their resources—time, energy, reputation, and sometimes money—in the hope of a future return” (Kingdon, 2003:122).

Nonetheless, with its articulation around three core elements—problem, policy, politics—Kingdon's (2003) multiple streams framework is also anchored, like most PI studies, in a top-down vision of the policy process. As a result, Kingdon (2003) overlooks actors’ strategies and the role of policy entrepreneurs through the implementation stage. In a similar vein, policy entrepreneurship has been carefully identified and studied during the agenda-setting and policy formulation stages, but largely overlooked during the implementation stage (Petridou & Mintrom, 2021). Yet, a growing body of literature has recently begun to correct this research bias by demonstrating that street-level bureaucrats (SLB) can act as policy entrepreneurs for solving problems and crises that might occur during the implementation stage (Arnold, 2015; Cohen & Aviram, 2021; Frisch-Aviram et al., 2018; Petchey et al., 2008).

In line with the general argument of the present paper, these studies contribute to reversing the top-down perspective of the policy process by shedding light on the variety of strategies used by the actors that are responsible for policy implementation. However, equating policy entrepreneurs with SLB means retaining a siloed and unidirectional approach to policy implementation dictated primarily by administrative procedures. Some recent studies have shown that the role of SLB should be understood beyond the formal structures of government. As a result, SLB working for governmental agencies, respecting the definition of Lipsky's (1980) seminal study, now cohabit with private SLB that are equally important for the implementation of public policies. Consequently, public accountability mechanisms have experienced a deep transformation, and both public and private SLB currently face a significantly higher number of “inescapable dilemmas” than Lipskian SLB (Hupe & Hill, 2007; Sager et al., 2014; Thomann et al., 2018).

In this context, instead of using the concept of street-level policy entrepreneur favored by Cohen and Aviram (2021), the broader term of “integration entrepreneur” is helpful. It designates a policy entrepreneur in the sense outlined by Kingdon (2003), but one who exerts most of their efforts and engages most of their resources in order to promote integrated policy solutions throughout the implementation process. A successful “integration entrepreneur” should thus be able (a) to exploit the flexibility and the incentives offered by the policy design to get integrative capacity, power and leadership within the implementation process, and (b) to use this position to orient the implementation of policies toward its own interests (see Wu et al., 2018; Candel, 2021).

Challenging the top-down vision of PI: empirical strategy

The present article aims to challenge the top-down perspective of the policy process that is dominant among PI studies by offering a fresh perspective on the implementation process of integrated policy frameworks. Therefore, I deliberately study a complex crosscutting issue (urban sprawl) within an institutional context (Swiss federalism) that make the formulation and the implementation of an integrated policy framework very unlikely (see next subsection). Then, since investigating spatial planning policies requires a long-term perspective (see Varone & Nahrath, 2014), I conduct a comparative longitudinal analysis from 1980 to 2020 which is twofold. On the one hand, I study four Swiss federal policies which are concerned with the crosscutting issue of urban sprawl: the Federal Railway Policy (FRP) and the Federal Spatial Planning Act (FSPA) since 1980; and the Agglomeration Policy (AggloPol) and the liberalization policy since their introduction at the turn of the millennium. On the other hand, I investigate the elaboration and the concretization of urban densification projects which aimed to reduce urban sprawl in the Swiss cities of Zurich, Bern and Geneva. Taken together, these two longitudinal analyses show that a turning point occurred in the early 2000s following the introduction of the AggloPol and the liberalization policy (see Fig. 1). Since then, Switzerland has moved from a sectoral to a more integrated policy approach to tackle urban sprawl despite the high complexity of its federalist framework.

Fig. 1
figure 1

Timeline of the four public policies under investigation with integration failure and integration success

Full size image

Curbing urban sprawl in a federalist country

Urban sprawl occurs when agricultural or natural land located at the fringes of an urban area is developed. The continued growth of urban sprawl is a complex global issue that is transversal to several policy subsystems (Liu & Meng, 2020; Sarkodie et al., 2020). Indeed, land is a natural resource delivering a great variety of goods and services to different users including building zones for housing and transport infrastructures, productive farmland, natural habitats for animals and plants or reserves of raw materials. Thus, land can be described as a “common-pool resource” from which it is difficult and costly to exclude potential beneficiaries (Ostrom, 1990). Furthermore, both the intensity of these competing land uses and the heterogeneity of the user groups have been increasing over time. Consequently, the sustainable management of land in general and the reduction in the urban sprawl in particular are crosscutting issues that require an integrated policy framework (a) regulating the various goods and services provided by the land resource in a coordinated manner and (b) solving rivalries between use rights (Varone & Nahrath, 2014; Viallon, 2017).

However, attaining such a level of coordination and integration is not an easy task, especially in a federalist country like Switzerland in which federal policies are implemented by member states called “cantons.” This delegation mechanism and the strong cantonal autonomy associated with it mean that implementing Swiss federal policies in an integrated way is more challenging than it would be in more centralized countries like the UK or France, because Swiss cantons tend to implement federal laws along lines that enable them to pursue their own political interests (e.g., Delley, 1982; Rieder et al., 2017). Moreover, the Swiss federal planning policy introduced in 1979 is in direct contradiction with land ownership guarantees anchored in the Swiss Constitution since 1969. Investigating these contradictions, some scholars have concluded that land-use planning in Switzerland is a nice illustration of an "Institutional Complexity Trap" (Bolognesi et al., 2021; Nahrath, 2005; Varone & Nahrath, 2014).

Since curbing urban sprawl in Switzerland is such a complex issue, interlocking several policy sectors and territorial units, scholars adopting a top-down vision of PI processes might conclude that the chances of observing integrated implementation strategies for curbing urban sprawl in Switzerland are close to zero. In seeking to question such an understanding, this article follows a similar logic to that of the crucial-case method developed by Eckstein (1975) and Gerring (2007), but its objective is less radical. Indeed, I do not aim to completely undermine the top-down model of PI, but simply to show that policy integration can also be delivered during the implementation stage through coordination mechanisms that were unintended, i.e., not foreseen by policymakers at the formulation stage.

Coupling a longitudinal analysis made at the federal level with in-depth case studies conducted in three Swiss cities

The empirical material presented in the next two sections draws on qualitative research conducted throughout the 2010s (see Lambelet, 2019). For the comparative longitudinal analysis of the four federal policies impacting urban sprawl (see next section), I first drew on previous studies investigating these four policies and their interactions in the Swiss case (see Table 2 in appendix). Drawing on these studies, I established a chronology of the four federal policies and identified the main legislative changes that have been undertaken over time (see Fig. 1). I then complemented this research material by investigating whether increasing policy integration was an explicit goal pursued by federal authorities when they decided to make these legislative changes, i.e., whether the other policies impacting urban sprawl were taken into consideration (or not). To evaluate the intentions of the federal authorities, I relied on the corresponding Federal Council dispatches that the Swiss government submits to the Swiss Parliament to justify the relevance of the legislative change being pursued. All Federal Council dispatches are publicly available on the online archive of the Federal journal (see Fedlex, 2022). Using the online database of the Swiss Federal Archives (2022), I also looked at the corresponding parliamentary minutes to evaluate whether Parliamentarians decided to modify the original bill of the government toward a more (or a less) integrated approach.

To evaluate whether these legislative changes made at the federal level produced policy outcomes concretely reducing urban sprawl at the local level, I decided to concentrate my analysis on the elaboration and the construction of major urban densification projects located in three of the largest Swiss cities (Zurich, Bern and Geneva, see later section). I thus based my case study analysis on the same assumption made by federal authorities, namely that promoting inwards urbanization is a suitable strategy to reduce urban sprawl (see FPSA, art. 1abis & 8a). As previously done at the federal level, I collected documentary material at the local level using the archives portals of the City Council of Zurich (2022), the City Council of Bern (2022) and the Cantonal ParliamentFootnote 4 of Geneva (2022). I complemented this longitudinal comparative case study analysis with fieldwork and conducted 36 semi-structured interviews with representatives of the local authorities, managers of the real estate division of the Swiss Federal Railways (SBB), as well as other local and national stakeholders involved in the various urban projects under study (see Table 3 in appendix). These interviews were key for identifying the factors of success or failure of the various urban development projects, as well as for determining the influence of the evolution of federal legislation dealing with urban sprawl. I now chart this evolution.

Four decades of Swiss Federal policies to curb urban sprawl

Switzerland is no exception to the global trend of increasing urbanization. Between 1985 and 2018, the area taken up by housing and infrastructure (roads and rail) increased by almost a third nationwide. Residential areas grew by 61%, twice as fast as the population of the country. During the same period, the overall loss of agricultural land corresponds to an area of 1143km2—about twice the size of Lake Geneva, the largest Western European Lake. In other words, about one square meter of agricultural area, including alpine pastures, is disappearing every second and an area the size of eight soccer fields of farmland is paved over every day (Federal Statistical Office 2021).

Such a sprawling trend is obviously in strong contradiction with sustainable development principles, and for a country as small as Switzerland (area of 41,258km2), it has also become a major political issue. The first federal measures taken to protect agricultural land from urbanization were introduced during the second world war to guarantee food security. However, in the decades following the war, Swiss cantons remained sovereign for the spatial planning of their own territories, and they were reluctant to give legally binding instruments to federal authorities.Footnote 5 As a result, the Federal Spatial Planning Act (FSPA), which was the first legislation giving binding instruments to regulate spatial planning to the Swiss Confederation (mostly zoning and a conformity control of cantonal master plans), only entered into force in 1980. Since then, federal, cantonal and communal authorities must coordinate their territorial policies to curb urban sprawl.

Therefore, I start my empirical analysis in 1980 and divide it into two phases (see Fig. 1). The first phase, detailed in the next subsection, goes from1980 to 2000 and is characterized by sectoral policymaking. The absence of policy integration during this period provoked several policy failures: (a) design failures resulting from the political and territorial compromises that had to be made during the development process of the FSPA; and (b) failures resulting from the fact that the FSPA framework was neither coordinated nor integrated with the Federal Railway Policy (FRP) despite its major territorial impacts.

The second phase, detailed in a later subsection, starts when the Federal Agglomeration Policy (AggloPol) was introduced in 2001 as a first attempt to provide an integrated policy framework coupling the objectives of the FSPA and the FRP. Yet, this first integration attempt produced mixed results. By contrast, another federal policy that was introduced at the end of the 1990s, namely the liberalization of network industries, had striking results in incentivizing urban renewal and thereby curbing urban sprawl. Such an ex post integration was made possible because, after being transformed into a state-owned company, the Swiss Federal Railways (SBB) became an integration entrepreneur that was able and willing to couple the aforementioned federal policies. This resulted in unprecedented urban regeneration of major Swiss cities (Marti and Bösch 2010), as I will show later through case study analysis.

1st phase (1980–2000): sectoral policymaking

Design failures within the framework law regulating spatial planning

The Federal Spatial Planning Act (FSPA) aims at ensuring a “moderate use of land” and “the settlement of land in a way that guarantees the harmonious development of the entire country” (Swiss Constitution, art. 75 and FSPA, art. 1). These two policy objectives imply a strict distinction between building zones, and agricultural and protected zones where building is prohibited or restricted. Hence, what was at stake during the policy formulation process was the first general, explicit and direct regulation of the right to construct on all plots of the national territory. Concretely, the zoning instrument attributed an administrative identity to each plot with a clear definition of the possible uses that could be made of it by its owner. Yet, the introduction of zoning caused enormous conflicts between state authorities and private landowners when the first generation of land-use plans was created at the municipal level (Nahrath 2003). The disadvantaged landowners, from whom the right to build was withdrawn when their plot was in a non-development zone, often resorted to legal action and to courts to receive a financial compensation. In addition to blocking planning processes, these judicial procedures also turned out to be very expensive for state authorities on several occasions (Tanquerel 2001).

Such a contentious development of the FSPA points out three failures that occurred during the policy design process, which prevented or significantly delayed the curbing of urban sprawl. The first failure concerns the poor coordination among levels of government (vertical integration). The 1980 FSPA was eventually adopted after two decades of intensive debate about whether granting the Swiss Confederation planning and zoning authority was compatible with the sacred subsidiarity principle underlying Swiss federalism. Indeed, the Swiss Constitution (Art. 5a) specifies that any policymaking power that is not explicitly given to federal authorities stays in the hands of cantonal authorities. Therefore, several cantons that had contested the top-down approach of regulation pursued by the first FSPA took more than a decade to adapt their own cantonal legislation to the new federal framework law since they perceived it as an anti-federalist policy (Horber-Papazian and Jacot-Descombes 2013).

The second design failure concerns the formulation of the policy objectives. Indeed, the first FSPA was mostly dedicated to reducing outward urbanization by protecting unbuilt areas and, thus, focused on the distinction between agricultural zones and building zones. Moreover, it provided extensive details regarding the agricultural zones that cantons and municipalities should strictly respect, but gave a great deal of leeway to cantons and municipalities regarding the dimensions of building zones. As a result, local actors were not incentivized to densify urban centers.

This policy logic started to be reversed with the FSPA revision that was approved by Swiss voters in 2013 and entered into force in 2014. Since this revision was adopted, building zones can only comprise as much land as is really needed within the next 15 years, which means that oversized building zones must be reduced (FSPA, art. 15, al. 2). Cantonal authorities had to adapt their cantonal master plans within five years, and during this transition period, cantons were not allowed to create any new building zones (FSPA, art. 38a). However, instead of using these new instruments in a coercive way to ensure the uniform implementation of the FSPA, federal authorities respected the specificities of each canton and looked for consensual solutions with cantonal authorities to implement the FSPA reform.Footnote 6 Therefore, at the end of the five-year transition period, academics and federal officials have both considered that it is too early to evaluate the concrete effects of this legislative change on the reduction in the urban sprawl in Switzerland because the implementation process within cantons and municipalities is not yet finished (Schwick et al., 2018; ARE, 2019).

Finally, the FSPA is also characterized by a design failure of its instrument mix. The key regulative instrument, namely zoning, was granted to federal authorities, but they have no right to levy any tax on the added value that is created when a plot is newly designated as a building zone. As a corollary, the Swiss Confederation has no resources to financially compensate landowners whose building rights are restricted when their plots are designated as a non-development zone. This imbalance in instrument design is due to a political struggle which opposed left-wing parties, which aimed to combat land speculation, and right-wing parties, which fiercely defended private ownership. This resulted in “land amnesia” within the FSPA because land value capture instruments were eventually removed from the hands of the Confederation and left to the discretion of the cantons (Nahrath 2005). However, most cantons have chosen not to introduce any fixed taxation mechanism on land added value. They prefer to let municipalities negotiate the taxation rate on a case-by-case basis with private developers and landowners (Lambelet and Viallon, 2019). This delegation cascade eventually resulted in the inability of public authorities, at any level of the federal system, (a) to strongly oppose the zoning of new land plots into building zones at the edges of central cities or (b) to require landowners of low-density areas within central cities (e.g., industrial brownfields) to develop urbanization projects in those areas. In the end, real estate projects located on previous agricultural land at the edge of central cities remain more convenient for most developers than urban renewal projects within central cities, where land property and land-use rights are in general much more distributed and complex (see Thévoz 2013).

Failures related to a lack of integration of the Federal Railway Policy (FRP)

Beyond the three design failures linked to the formulation process of the FSPA discussed in the last subsection, the FSPA was also elaborated without the aim of coordinating it with the Federal Railway Policy (FRP). Coordinating public transport policy and spatial planning policy is yet widely recognized as an efficient way of densifying urban areas (see Calthorpe 1993; Knowles and Ferbrache, 2019), and the strong interlinkages between these two policies have been largely demonstrated in the Swiss case (e.g., Kaufmann et al., 2003; Koch 2013; Pflieger 2013; Sager 2005). However, from the introduction of the FSPA in 1980 to the early 2000s, Swiss federal authorities did not capitalize on these interlinkages. At that time, traveling by train took longer than traveling by car even between major Swiss cities. Aiming to make the train more attractive, federal authorities decided to invest massively in railway infrastructure through the “Rail 2000” project. However, rural and mountainous cantons opposed the first version of Rail 2000, stating that it favored urban cantons. As a result, the Swiss government merged Rail 2000 and the construction of the railway lines across the Alps into a common policy. Furthermore, it insisted that Rail 2000 would benefit the whole country thanks to the introduction of a national clock-face timetable in 1982 (Federal Council 1985). This turned out to be a successful strategy since Rail 2000 was accepted at the ballot box in 1987 and 1998, both times with a comfortable majority.

At the end of the twentieth century, the Federal Railway Policy was still anchored in a Keynesian welfare state model, in which every canton should benefit from federal policies and their inter-territorial redistribution mechanisms (see Brenner 2004; Jessop 2002; Swyngedouw 1997: chap. 4). This Keynesian vision resulted in a domino effect which hampered policy integration. At the federal level, it led to a non-coordination of railway policy with spatial planning policy. Sectoral policymaking at the federal level then led to the inability of cantonal and local authorities to benefit from the urbanization–transportation nexus to effectively curb urban sprawl. However, this issue was resolved thanks to the introduction of two other federal policies that put an end to sectoral policymaking and marks the beginning of the second analytical phase to which I now turn.

2nd phase (2001–2020): the post-Keynesian turn leads to a more integrated approach to urban sprawl

The “post-Keynesian turn” is a major economic and sociopolitical evolution that has been observed in Western countries throughout the 1980s and 1990s. It is characterized by a simultaneous neoliberalization of state space and state structures (e.g., Brenner and Theodore 2002; Harvey 2005; Keil et al., 2016). In a post-Keynesian economy, national governments prioritize territorial competitiveness over territorial cohesion. They stop redistributing wealth across their entire territory and start designing their territorial policies to position their largest cities and agglomerations on the international stage. As a result, they drastically reduce subsidies allocated to peripheral regions and foster competition between urban areas to get national support (e.g., Brenner 2004; Jessop 2002; Sassen 1991).

To cut public spending, national governments also liberalize network industries (e.g., Finger and Künneke 2011). Concretely, this liberalization process consists of introducing competition in previously monopolistic public service delivery. Thus, the unbundling of state-owned enterprises, which are historically vertically integrated, is a prerequisite for ensuring effective competition since the provision of services should be distinguished from the infrastructure ownership (Florio 2013).

In Switzerland, there are two direct and tangible consequences of the post-Keynesian transition that are of interest for the present study: the introduction of the Federal Agglomeration Policy (see next subsection) and the transformation of the Swiss Federal Railways into a state-owned company (see later subsection). Both policies have increased coordination between the policy subsystem of land-use planning and the policy subsystem of transport policies, thereby enabling a more integrated approach to curbing urban sprawl. However, federal policymakers pursued very different objectives in formulating these two policies, which then impacted the reduction in the urban sprawl in various ways. I explain why in the next two subsections.

Intended integration with contrasting results: the Federal Agglomeration Policy (AggloPol)

Influenced by the post-Keynesian transition, Switzerland embraced important reforms of its federalist structure throughout the 1990s. These reforms aimed to specify the role of federal authorities and the way they should cooperate with cantons, but also with other subterritorial units. An important result of these reforms is the new Federal Constitution introduced in 1999 that notably specifies that the Swiss Confederation pays attention to the peculiar situation of cities and urban agglomerations (art. 50). This recognition opened the door to the creation of the Federal Agglomeration Policy (AggloPol) which provides a new federal funding mechanism reserved for central cities and their agglomerations. The AggloPol also put cities and agglomerations in competition to receive federal funding. The introduction of AggloPol thus represents a radical change of existing systems of intergovernmental cooperation within Swiss urban areas (Kübler et al., 2003). Nevertheless, it cannot be assimilated to a strategy of policy accumulation in the sense of Adam et al. (2018) since the Swiss Confederation reduced direct subsidies for mountainous and rural regions in parallel.

In 2006, the Swiss Parliament approved a first framework credit of 6 billion Swiss francs to support new public transport projects within major agglomerations. One-sixth of this funding was specifically dedicated to the construction of urgent railway infrastructure within agglomerations that had been on cantonal agendas for decades (suburban trains, regional light rail systems and new stations). It thus opened a first window of opportunity to coordinate the national railway policy and the master planning of cantons and municipalities located in urban areas. Since then, ad hoc constellations of cantons and municipalities have formed in all Swiss urban areas to create agglomeration programs that aim at developing a common strategy for connecting urbanization and transportation policies. Every four to five years, federal authorities evaluate these agglomeration programs and decide which policy measures deserve federal funding. On average, the Confederation subsidizes 35–40% of the total cost of an agglomeration program. Overall, half of all subsidized measures aim to improve urban railway transportation, with that proportion climbing to two-thirds for the agglomeration programs of the most populated urban areas of the country (Ecoplan 2016: 30–32).Footnote 7

Thanks to this focus on railway infrastructure, the AggloPol has allowed for a more integrated policy approach between the Federal Railway Policy (FRP) and the urban development strategies of the largest Swiss agglomerations. Contrary to the Rail 2000 project on which they refused to prioritize urban over rural areas, federal authorities have now anchored the key objectives of the FRP within a post-Keynesian logic of territorial competition. The recent evolution of the policy demonstrates this nicely. The railway expansion programs voted by the Swiss Parliament in 2013 (6.4 billion Swiss francs) and 2019 (12.9 billion Swiss francs) essentially aims to increase frequencies and to improve railway infrastructure within and between large agglomerations, thereby creating opportunities for densifying urban centers that did not exist with Rail 2000.

However, several scholars (Bassand 2004; Da Cunha et al. 2005; Pflieger et al., 2009) have shown that this focus on transport infrastructure within Swiss agglomerations did not reduce but rather significantly increased urban sprawl. Indeed, more efficient transport infrastructures encourage households to settle down at the periphery of agglomerations where, in contrast to urban centers, privately owned single-family homes are much easier to construct since land plots are cheap and quickly available for construction (Thévoz 2013). Therefore, despite the policy coordination that AggloPol introduced between the FRP and the spatial planning strategies of Swiss agglomerations, the AggloPol has been unable to reach a sufficient level of PI to concretely reduce urban sprawl. However, another federal policy, to which I now turn, had a striking integrative impact that had not been anticipated by federal policymakers.

Unintended integration with striking results: the liberalization of network industries

As previously discussed, liberalizing network industries implies an unbundling process. Consequently, when the Swiss Federal Railways (SBB) was transformed into a state-owned company in 1999, it was divided into four main operating divisions that are functionally distinct and are responsible for passengers, freight, infrastructure and real estate. The latter division is of particular interest for the present study because, since 2003, the Swiss government has urged the SBB real estate division to reinvest at least 150 million Swiss francs per year into constructing or maintaining railway infrastructure. As the second largest landowner in the country, the SBB has used its huge land portfolio, composed of around 4′000 land plots, for developing major urban projects within central cities (Marti and Bösch 2010). By doing so, the SBB has been able to act as an integration entrepreneur providing the necessary coordination to effectively curb urban sprawl at the implementation stage, even though the spatial planning policy and the policy of liberalization were not coordinated in the policy formulation stage and pursued radically different objectives when they entered into force.

In the next section, I develop my case study analysis through a before-and-after comparison to demonstrate that the new private status of the SBB was the key driver behind the development and the construction of major urban renewal projects curbing urban sprawl in three of the largest Swiss cities.

Urban development and renewal in Swiss agglomerations: three case studies

In the three cities under study (Zurich, Bern and Geneva), case study analysis reveals that urban revitalization was either slow or even suspended until the late 1990s (see next subsection). Then, a turning point occurred at the beginning of the 2000s when the real estate division of the SBB started integrating the FSPA, the AggloPol and the FRP to develop highly profitable urban projects on its own (see later subsection).

1980s-1990s: sectoral policymaking leads to the failure of urban densification projects

Zurich main station: prioritizing railway infrastructure over real estate projects

Zurich is the Swiss economic capital and the main hub of its national transport networks. Owned by the SBB, Zurich’s main station is frequented by 440′000 people daily, which makes it one of the busiest stations in Europe (Cajacob 2016). However, real estate development projects led by private investors have long clashed with the public authorities’ desire to increase the railway infrastructure capacity of Zurich’s main station.

Throughout the 1980s, an ad hoc business corporation led by the four largest Swiss banks put together a project, called HB-Südwest, that aimed to construct prestigious office buildings above the railway tracks. However, when Swiss voters approved the Rail 2000 project in 1987, it implied costly transformations of Zurich’s main station. After these changes, the profitability of HB-Südwest was heavily debated within the business corporation, and the project was finally abandoned in 1992.

A couple of years later, the lead architect of HB-Südwest elaborated a slightly revised version of the project called Eurogate and managed to create a new funding corporation led by UBS, the largest Swiss bank. Nevertheless, this revised project rapidly turned into a judicial battle around the number of parking lots included in the building permit. Ironically enough, the City and the Canton of Zurich were both involved in this judicial process, but with diverging positions. In the end, the decisive blow was delivered by the SBB because, as for HB-Südwest, the railway company perceived the project as technically incompatible with the realization of Rail 2000 (Wolff 2012). Since it was impossible to transform Zurich’s main station without the full participation of its landowner, UBS announced in April 2001 that it was abandoning Eurogate.

Bern Wankdorf: the SBB refuses to build a new train station

Bern is the capital city of Switzerland, but Bern is also its second largest canton and includes hundreds of rural and mountainous municipalities. Within this context, uncoordinated planning strategies established at the municipal level heavily contributed to urban sprawl. Thus, to implement the FSPA, Bern’s cantonal government has created, since 1989, a program aimed at concentrating economic activities within Economic Development Poles (EDP) that are easily accessible by public transport.

Located in the City of Bern and in two suburban municipalities, Wankdorf is the largest EDP of the canton (339 ha). It has the peculiarity of being almost exclusively owned by public actors (Gerber 2008:60). Despite the cantonal will to improve public transport infrastructure, the SBB refused to construct a new suburban railway station in Wankdorf since it would be located at the crossroads between two major axes of the national network that are considered as top priorities within Rail 2000 (Sager 2002: 128ss). As a result of SBB's focus on railway infrastructure and network flow management, the densification of the Wankdorf area remained far below the objectives of the Canton Bern until the early 2000s.

Geneva-CEVAFootnote 8: waiting for Godot

Thanks to its high concentration of international organizations, Geneva has benefitted from an airport offering continental and intercontinental flights since the 1940s. However, the development of its railway network has suffered from the peripheral position of the canton at the western end of Switzerland. Until 2019, Geneva did not have a proper suburban railway network. Back in 1912, the Swiss Confederation, the SBB and the Canton of Geneva had signed a convention to create the missing link of this railway network, but the project, called CEVA, was first suspended by the two world wars and then brought into questioned by the golden age of the automobile (Maksim 2008).

When Rail 2000 was launched, federal authorities did not considered Geneva to be a critical node of the Swiss railway network because of its location. Unlike Zurich or Bern, which started developing suburban railway services through this federal funding, Geneva could not take advantage of this opportunity. This also created tensions between the cantonal government and the federal authorities. In this context, developing urban projects along the future railway line was not even considered.

2000s-2010s: Densification projects are undertaken thanks to the new role of the SBB

Zurich Europaallee: new railway infrastructure leads to a huge development project

For decades, the SBB did not consider investing in real estate projects around Zurich’s main station. However, soon after its transformation into a public limited company, the SBB understood that it was in its best interest to take the lead in developing the area around the station, while remaining in charge of the extension of the railway infrastructure. The SBB was thus able to coordinate, within a couple of years, two conflicting development strategies that had previously resulted in decades of unfruitful planning.

The SBB did so by exploiting the real estate opportunities offered by the Durchmesserlinie, a new underground connection that has been fully operational since 2015 and which significantly increased the railway capacity of the main station (see Huber 2015). Constructing the Durchmesserlinie allowed for the removal of railway tracks at the edges of the main station and progressively made land available for the project Europaallee. Consisting of 6′000 workplaces and 300 dwellings, Europaallee is one of the largest urban renewal projects in Switzerland, contributing significantly to the objective of urban regeneration of the FSPA.

Thanks to the new federal funding mechanism introduced by the AggloPol, the Confederation covered two-thirds of the cost of the Durchmesserlinie (1.3 billion Swiss francs), which simplified the planning process. However, coordinating the implementation of AggloPol and FRP would not have led to the Europaallee project without the entrepreneurial role of the SBB that resulted from its transformation into a public limited company. Finally, it is also important to note that as well as cooperating closely with cantonal authorities for implementing the Durchmesserlinie, the SBB has also been working closely with Zurich’s municipal authorities to prevent local opposition. Thus, the SBB succeeded in avoiding intergovernmental conflicts that had previously led to continuing legal battles.

Bern Wankdorf City: the new railway station as a catalyst for densification

In the early 2000s, the densification process of the Wankdorf accelerated significantly because the SBB realized that assuming a more active role in this process would generate significant real estate opportunities. Hence, the railway company conducted another feasibility study on the Wankdorf station which concluded that there was no major technical obstacle to the construction of a station in Wankdorf (Sager 2002: 132). Since no funding coming from AggloPol was provided for this station, the SBB agreed on sharing the costs of construction with the Canton and the City of Bern. Inaugurated in 2004, the new station connects Wankdorf to Bern’s main station in three minutes.

In addition to the SBB, the City of Bern also became more proactive in the densification process because it owns the land plots located around the new station. In close cooperation with the SBB, the City of Bern planned, the project Wankdorf City, which aimed to create a new neighborhood hosting 5′000 workplaces along the railway lines. About half of these workplaces are now occupied by the SBB and over a quarter by the Swiss Post. Since their transformation into public limited companies, the two biggest employers in the capital city were no longer required to keep their headquarters in Bern, but Wankdorf City convinced them to do so.

Geneva: CEVA becomes a reality and leads to the creation of “CEVA cities”

The CEVA project got a new start in December 1999, when the Swiss minister of transport publicly stated that the intergovernmental convention of 1912 remained valid. This meant that the Swiss Confederation would fund one-third of the railway connection—550 million Swiss francs—through AggloPol funding, the SBB would fund another third, and the Canton of Geneva would only be responsible for the remaining third. The underground train that was judged to be far too costly by cantonal politicians throughout the 1990s suddenly became much more affordable (Pini 2020: 41ss).

However, the 1912 convention had to be updated and the SBB took the opportunity to state its intention to develop real estate projects near future train stations. This initiated a negotiation between the SBB and the Canton Geneva, which owned most of the plots of land concerned. For Geneva authorities, the highest priority was to finally build the railway line at the best price. Moreover, the vacancy rate on the cantonal housing market had recently dropped and the cantonal government favored the densification of these areas. Therefore, the Canton committed itself to creating high density zones above the future underground stations but urged the SBB to reinvest the land added value generated—178 million Swiss francs—in the railway infrastructure. The SBB accepted on the condition that the Canton transferred all its plots of land to the SBB free of charge (Lambelet and Viallon 2019: 105).

CEVA was inaugurated at the end of 2019 and the new neighborhoods located near CEVA stations are now either fully or partially constructed. In total, CEVA has allowed for the creation of about 850 dwellings and more than 4′500 workplaces in various central locations within the canton of Geneva. It thus contributed to slowing urban sprawl through urban regeneration.

Challenging the top-down vision of PI: ex post integration during the implementation stage

In this sixth section, I recap the key findings of the case study analysis and discuss some of its theoretical implications. In the three cities under study, a turning point can be identified at the beginning of the new millennium (see Fig. 1). Urban renewal projects that had failed during the last decades of the twentieth century were suddenly given new life during the first few years of the twenty-first century. Such a turning point is linked to the introduction of two federal policies following the post-Keynesian turn: the Federal Agglomeration Policy (AggloPol) and the liberalization of network industries. Both federal policies have provided new incentives for revitalizing city centers and significantly contributed to the reduction in the urban sprawl. Since the latter represents the key objective of the federal spatial planning policy (FSPA), it would be tempting to conclude that reduced urban sprawl was the result of a coherent integration process driven by federal authorities. However, my empirical investigation revealed a different picture that confirms, around two main points, the multilayered and asynchronous nature of policy integration and its implementation process (see Candel and Biesbroek 2016).

First, although the AggloPol and the policy of liberalization are both anchored within a post-Keynesian logic promoting territorial competitiveness over territorial redistribution (see Brenner 2004), their simultaneous introduction was not deliberate, and thus not coordinated ex ante by federal policymakers. Moreover, an intentional integration process would have required the simultaneous adaption of FSPA objectives. Yet, these were only modified more than one decade later to foster the renewal of urban centers. Since the planning procedures of the analyzed urban projects started ten to fifteen years earlier than the FSPA revision, such a temporal gap shows that the concretization of these three urban renewal projects (Europpallee, Wankdorf City and CEVA cities) did not result from proactive policy integration procedures during the policy design stage.

Second, my three case studies have shown that the introduction of the AggloPol has had some integrative effects since it provided a new funding mechanism for improving railway transportation within Swiss agglomerations, and thus contributed to the emergence of urban renewal projects that have been concretized. In other words, by introducing the AggloPol, federal policymakers took a deliberate and successful measure that provided a more integrative approach with the Federal Railway Policy (FRP). However, improving public transportation networks does not guarantee more concentrated urbanization (see Bassand 2004; Da Cunha et al. 2005; Pflieger et al., 2009). Thus, the AggloPol would not have provided sufficient PI to foster urban densification projects without the entrepreneurial role of the SBB. Yet, federal policymakers aiming to curb urban sprawl had not anticipated that the SBB would undertake such an integrative function throughout the implementation process.

Thus, while confirming that the elaboration of an integrated policy design is a particularly complex and challenging task for policymakers, my empirical analysis challenges the sequential approach of the policy cycle that is dominant among PI studies (e.g., Briassoulis 2017; Capano et al., 2016; Howlett et al., 2017; Vince 2015). Echoing seminal studies in the field of policy implementation (Dunsire 1978; Hood 1976; Lipsky 1980; Pressman and Wildavsky 1984), it shows that implementing policies in an integrated way remains a complex, uncertain and tricky process in which new stakeholders, which were not involved in policy formulation, might play a decisive role. However, this uncertainty is not necessarily a negative factor. Indeed, my analysis demonstrates it can also substantially increase policy integration if new stakeholders become integration entrepreneurs who are willing and able to pursue integrative policy strategies. I detail how the SBB has undertaken such a key role in the next subsection.

The SBB as an integration entrepreneur coupling federal policies

After its transformation into a public limited company, the SBB rapidly began to develop real estate projects around the main train stations of the largest Swiss cities.Footnote 9 These projects were all primarily oriented toward profitability, which marks a radical shift from previous real estate strategies that focused on railway infrastructure. In all projects, the SBB took on the role of general contractor for both railway infrastructure and real estate projects. Thanks to this dual role, the real estate division of the SBB fulfilled an integrative function by interlinking the implementation of three federal policies that were previously insufficiently coordinated (see Fig. 1): the spatial planning policy (FSPA), the agglomeration policy (AggloPol) and the railway policy (FRP). This integrative leadership role that SBB has taken on since the early 2000s can be explained by four main factors detailed in Table 1.

Table 1 From design and integration failures of FSPA (1980s-1990s) to the successful strategies of an integration entrepreneur (2000s-2010s)
Full size table

First and foremost, the SBB has become such a powerful integration entrepreneur because its new status gave it opportunities to offer turnkey solutions to address the design and integration failures that had prevented the densification of central urban areas throughout the 1980s and 1990s. Being transformed into an independent division with a high degree of autonomy, the SBB real estate division has been able to elaborate its own development projects, to negotiate land-use plans bilaterally with local authorities and to act as the lead contractor for construction. Such a leadership role assumed by a single actor from the earliest stages of planning to the finalization of construction was lacking at the end of the twentieth century.

Second, in stark contrast with the 1980s and 1990s, the Swiss government has established a strong incentive structure for the SBB to value its land portfolio. Between 2007 and 2017, it is estimated that the SBB has invested 4.6 billion Swiss francs in about one hundred urban projects all over Switzerland. To allow such investment, the Swiss Confederation has covered around 80% of the real estate expenses made by the SBB through long-term loans. This funding mechanism has made possible to construct dense and attractive neighborhoods, offering an overall rental space of more than one million square meters (Scherr 2018: 12–13).

Third, being used to collaboration with authorities at various levels of the Swiss federalist system, the SBB has been able to strategically use policy diffusion mechanisms (see Maggetti and Gilardi 2016) to foster intergovernmental consensus on issues that were previously fraught. The high degree of autonomy of the real estate division of the SBB has also reinforced the ability of the SBB to overcome opposition by avoiding political debates on alternative projects. This echoes a previous analysis conducted by Swyngedouw et al. (2002) in other European cities that stressed the extent to which real estate driven projects have made it easier to overcome opposition by avoiding political debates on alternative ways forward. In addition, the SBB also benefits from a political legitimacy that standard private developers do not enjoy. Indeed, when facing criticism regarding the high density or the high profitability of future neighborhoods, the railway company can argue that it has been mandated by the federal government to generate real estate profits. Applying the post-Keynesian logic of territorial competition, the SBB can also draw on its huge land portfolio to prioritize projects located in attractive areas where local authorities are keen to cooperate.

Finally, through its role as integration entrepreneur, the SBB has also provided a concrete solution to the “land amnesia” of the FSPA that has largely contributed to the perpetuation of urban sprawl (see Nahrath 2005). Admittedly, federal authorities still lack land value capture instruments to directly implement their national land planning strategy. However, the Swiss federal government now benefits from another indirect but powerful instrument because it establishes the broad real estate strategy of the SBB which is developing projects in almost every Swiss city and has meanwhile become the second largest real estate company in the country (Rey 2018). Moreover, since the SBB develops its projects on its own plots of land, the fragmented land ownership that is normally a pervasive issue all over Switzerland (see Nahrath et al., 2009) does not arise in this case.

Distinguishing the integration entrepreneur from related concepts

In the theoretical section, I have defined an integration entrepreneur by drawing on Kingdon's (2003) multiple streams framework instead of considering the recent literature focusing on street-level policy entrepreneurship (e.g., Arnold 2015; Cohen and Aviram 2021). In this subsection, I explain why certain characteristics of the integration entrepreneur identified within my case studies justify this approach.

To elaborate his multiple streams framework, Kingdon (2003: 84–89) has largely drawn on the garbage can model of Cohen et al. (1972) that depicts organizations as “organized anarchies” in which policy choices result from an unintended combination of four independent streams: policy problems, policy solutions, participants in the policy process and choice opportunities. However, Kingdon (2003) perceives policy choices to be less randomized than in the Garbage Can Model. He explicitly states that, even if problems, policies and politics evolve independently from each other, windows of opportunities will open at some point for interlinking these streams, and policy entrepreneurs should be ready to exploit them (ibid.: chap. 8).

Therefore, Kingdon's (2003) perspective better suits the entrepreneurial role of the SBB. Indeed, the railway company has been willing and able to capitalize on the windows of opportunity opened by a combination of simultaneous but uncoordinated policy changes to become an integration entrepreneur able to bring together various federal policies throughout the implementation stage. This coupling process is also partly due to the deliberate introduction of the AggloPol by the federal authorities, but the empirical analysis presented above has shown that the decisive factor for a successful integrative process was the strong leadership capacity that the SBB was able to gain following its transformation from a federal government institution into a state-owned company.

Acquiring this leadership capacity was not the intended objective of federal authorities when they transformed the SBB into a state-owned company. On the contrary, federal policymakers principally aimed at reducing the maintenance and construction costs of railway infrastructure. They would not have thought that the SBB would have been able to integrate the FSPA, the AggloPol and the FRP during the implementation stage. Such a mismatch between (a) the intentions of policymakers working with siloed policy formulation and (b) the integrated policy solutions delivered by an integration entrepreneur is what I refer to as “unintended policy integration.”

It is also worth noting that the SBB did not take any lobbying activities (of federal ministers, high federal officials or in Parliament) to modify the policy design of the FSPA or of any other policy. The railway company knew it was in its best interest to make these policy linkages throughout the implementation stage and thus concentrated its efforts and its resources on achieving ex post policy integration. This empirical evidence suggests that the analytical capacity of Kingdon's (2003) multiple streams framework could be strengthened by paying more attention to the implementation stage, and by considering that, it is also made up of independent policy streams like the formulation stage.

In a similar vein, the SBB, and especially its real estate division, cannot simply be understood as a street-level policy entrepreneur as defined by Cohen and Aviram (2021: 430.32). Applying their criteria, the SBB is a policy entrepreneur, but definitely not a street-level bureaucrat (SLB). Indeed, the SBB are not public civil servants located at the bottom of an administrative hierarchy, nor are they collaborating primarily with other SLB, NGOs or citizens. The empirical analysis of the present study thus confirms that studying PI by focusing exclusively on the actions of bureaucrats, civil servants and other public actors is outdated in the post-Keynesian era.

Previous studies had already demonstrated that the liberalization reforms underlying the post-Keynesian turn impact the accountability mechanisms of both public and private implementation agents (e.g., Durose 2007; Sager et al., 2014; Thomann et al., 2018). This study goes one step further by showing that the post-Keynesian turn can also be a key driver of cross-sectoral and intergovernmental integration. Indeed, the post-Keynesian turn led to the introduction of the AggloPol, which started to integrate the FRP into a common policy framework. Then, and most importantly, the post-Keynesian turn opened the door to the transformation of the SBB into an integration entrepreneur able to couple the FSPA, the AggloPol and the FRP in an unprecedented way.

Conclusion

The present study has analyzed how Switzerland has addressed the crosscutting issue of urban sprawl over the last four decades (1980–2020). It has shown that, until the late 1990s, the Federal Spatial Planning Act (FSPA) and the Federal Railway Policy (FRP) were insufficiently coordinated. This prevented the emergence of large urban projects within central Swiss cities and resulted in the perpetuation of urban sprawl. However, at the turn of the millennium, two federal policies anchored in a post-Keynesian policy framework were introduced: (a) the Federal Agglomeration Policy (AggloPol) and (b) the liberalization of network industries. Both policies contributed to a more integrated policy framework addressing urban sprawl. Nevertheless, case studies conducted in Zurich, Bern and Geneva have shown that the AggloPol only had a limited effect on the emergence and the realization of large urban renewal projects undertaken in these three cities throughout the 2000s and 2010s. The decisive factor for the concretization of these urban renewal projects resulted from the liberalization policy which transformed the Swiss Federal Railways (SBB) into a public limited company seeking profits. This led to the creation of an autonomous real estate division of the SBB that benefitted from sufficient leadership capacity and adequate resources to implement the FSPA, the AggloPol and the FRP in an integrated way.

This integration entrepreneur’s role undertaken by the SBB challenges the topdown sequential approach of the policy process that is dominant within PI studies (e.g., Briassoulis 2017; Candel and Biesbroek 2016; Howlett et al., 2017) and demonstrates that policy integration can still occur during the implementation stage, even if the formulation stage has delivered a “poor policy design” (see Capano et al., 2016; Vince 2015). Indeed, among the four federal policies under study, the liberalization policy is the one producing the greatest positive impact on the crosscutting issue, namely curbing urban sprawl. Nonetheless, the liberalization policy has neither been coordinated nor integrated with the three other federal policies in the policy formulation stage. Thus, the integrative process does not result from a well-designed federal policy mix, but from the SBB’s ability to strategically use a byproduct of the liberalization reform. This is what I have termed “unintended policy integration.” This concept aims to show that policies can also be effectively integrated “on the way,” without the need for modifying the policy design, or for starting a new policy cycle.

Acknowledging that PI can also occur unintentionally implies that PI scholars should stop searching for the Holy Grail by admitting that perfectly coordinated policy designs resulting in the accurate implementation of integrated and efficient policy solutions simply do not exist. Seminal policy studies have demonstrated long ago that, even if policymakers and implementation stakeholders have the best intentions, unexpected events and policy failures cannot be avoided because the implementation stage follows its own operational logic (Dunsire 1978; Hood 1976; Lipsky 1980; Pressman and Wildavsky 1984). Recent studies have come to a similar conclusion by showing that policy implementation relies on organizational and institutional arrangements which are specific to this stage and thus remain highly influenced by power dynamics (Busscher et al., 2022; Michel et al., 2022; Sager and Gofen 2022). This is even more true when various policy subsystems must be coupled to design and effectively implement policy solutions (Darcis et al., 2022; Sarabi et al., 2019). Therefore, PI scholars should critically reconsider the dominant top-down approach of the policy process and strengthen their analytical perspective by giving equal importance to both intended and unintended integrative mechanisms.

Accepting such uncertainty may be a frustrating step for scholars who have recently made great efforts to disentangle the complex implementation process of integrated policies (e.g., Cejudo and Michel 2021; Maor and Howlett 2022). Yet, I am firmly convinced that it is worth taking this step back. The present study has indeed demonstrated that uncertainty can also lead to a substantial increase in the PI thanks to “integration entrepreneurs” benefiting from a strong integrative capacity that policymakers could never have imagined. These integration entrepreneurs open new pathways for implementing integrated policy frameworks. They require flexibility and incentives given by policymakers (see Cejudo and Trein 2023: proposition 8), but they are not located at the bottom of an administrative hierarchy. Therefore, unlike SLB, public managers or local governments whose integrative strategies are necessarily bottom-up, integration entrepreneurs are also able to pursue top-down integrative strategies crossing subnational or local jurisdictions. This is a precious asset, especially in federalist countries.

Nonetheless, I do not mean that integration entrepreneurs should be considered as the new Holy Grail of policy integration. As Kingdon’s (2003) policy entrepreneurs during the formulation stage, integration entrepreneurs pursue their own interests which may also lead to a selective integration process throughout implementation. This was certainly the case for the SBB in Zurich, Bern and Geneva. However, thanks to its proactive approach, the SBB has shown that it was possible to get the expected policy outcome (i.e., reducing urban sprawl) in an unexpected manner, namely by integrating three federal policies that had not been sufficiently coordinated by policymakers during the formulation stage. Moreover, this integrated policy process steered by the SBB has been more rapid in delivering concrete policy outcomes than the standard implementation procedure of the recent revision of the FPSA whose concrete effects are still difficult to evaluate (see ARE, 2019; Schwick et al., 2018).

Therefore, I believe that additional research should be conducted to further identify integration entrepreneurs and to investigate their role and their strategies in other policy sectors and in other politico-institutional contexts. This research effort would be especially welcome within federalist systems where the implementation of (supra)national laws is delegated to member states governments and where (supra)national authorities often lack coercive power to ensure the proper implementation of the laws they have introduced (e.g., Austria, Germany, Switzerland or the European Union). In such contexts, integration entrepreneurs might provide innovative tools toward the integrated implementation of (supra)national laws without prejudice to the sacrosanct subsidiarity principle.