Sustaining statehood: A comparative analysis of vertical policy-process integration in Denmark and Italy

In response to societal demands, democratic governments constantly adopt new policies. As existing policies are rarely aban-doned, policies accumulate over time. Policy accumulation bears the challenge of overburdening implementation bodies, hence undermining policy effectiveness. Any escape from this situation requires democratic governments to strike a balance between policy responsiveness and effectiveness. We posit that the extent to which countries are able to achieve this depends on the vertical integration of processes of policy formulation and implementation. We provide a novel conceptualization of vertical policy-process integration (VPI) that is based on two channels. While bottom-up integration captures the extent to which policy implementers can communicate reasons for potential policy failure from the bottom up, top-down integration indicates the degree to which the policy formulation level has to cover the implementation costs of the policies they produce. We illustrate our argument by an empirical analysis of VPI patterns in Denmark and Italy. of VPI that are applicable across different systems, concrete implementation in light

democratic governments that they are responsive to societal demands, by constantly adopting new policies in the form of laws, regulations, or programmes. At the same time, the termination of existing policies rarely constitutes an attractive option for governments seeking re-election (Knill and Bauer 2014). The resulting aggregate pattern is hence one of policy accumulation (Adam et al. 2019): Governments produce more new policies than they abolish.
Yet, in the long term, the continuous growth of policy stocks may lead to governmental 'overload' (King 1975).
Modern democracies are caught in a 'responsiveness trap' (Adam et al. 2019), a vicious circle of adopting more and more policies while being less and less able to realize their multiple goals in practice by overburdening implementation bodies with ever-more and increasingly complex policies (Limberg et al. 2020). This, in turn, can undermine the long-term support for governmental intervention. In short, there are potential trade-offs between political responsiveness and policy effectiveness.
Any escape from this situation requires democratic governments to strike a balance between responsiveness and effectiveness, implying that the continuous adoption of new policies does not overburden the capacities available for implementation. It is only on this basis that policy accumulation remains at a 'sustainable level', that is, is backed by sufficiently well-equipped implementation arrangements and structures. Unfortunately, there are no easy ways to overcome this problem. On the one hand, it is hardly realistic that societies become more abstinent in terms of their demands on governments. Public opinion data reveal that citizens have generally a rather 'schizophrenic' view of the government (Adam et al. 2019, p. 35). While people are generally critical of governmental intervention, they tend to be quite demanding when it comes to solving concrete policy problems. On the other hand, the most obvious option of expanding administrative resources for policy implementation hardly seems feasible. Most governments face fundamental fiscal and ideological constraints for public sector expansions in times of globalized financial markets, austerity, and still reverberating ideas of New Public Management (Lobao et al. 2018).
Despite this gloomy picture, countries seem to differ in their capacities to keep policy accumulation at a sustainable level (Limberg et al. 2020). We argue that this variation emerges from two sources. A first approach is to reduce policy accumulation through better policy design. Better design means that policies are more effective in addressing their objectives, implying less need for updating, modifying and complementing existing policies in order to resolve a given policy problem. This approach presumes institutionalized processes of policy evaluation and learning, in particular by systematically integrating expertise and information from the implementation level into the policy formulation process. The second approach, by contrast, works via the internalization of implementation costs. If those bodies in charge of developing new policies also have to carry the costs of implementing these policies, policy accumulation is expected to be less pronounced; the less implementation costs can be passed on to other institutional bodies, the lower are the opportunities for the policy-formulating level to excessively respond to societal demands via new policies.
The extent to which these options apply is affected by what we refer to as patterns of vertical policy-process integration (VPI). VPI captures the structural arrangements that shape the interactions between governmental bodies responsible for policy formulation and those in charge of policy implementation at different institutional levels. Patterns of VPI might vary not only across countries but also across policy sectors. Depending on the extent to which VPI ensures the integration of implementation input and costs in policy formulation, we identify different patterns of policy accumulation and, hence, ways in which governments balance responsiveness and effectiveness concerns.
The remainder of this article is structured as follows. In section 2, we introduce our concept of VPI. Based on these considerations, we deduce four ideal-type scenarios of the relationship between policy accumulation and VPI patterns in section 3. These arguments are empirically illustrated in section 4, in which we compare the cases of Italy and Denmark for the areas of social and environmental policy. Section 5 concludes.

| THE CONCEPT OF VPI
We conceive of VPI as institutional arrangements that are part of public service bargains. On this basis, we identify different dimensions of VPI and indicators for its empirical assessment across highly diverse political-administrative systems.

| Analytical point of departure
The analytical point of departure of our argument is that policy-making in modern democracies rests on formal and informal institutional arrangements structuring the relationship between politics and bureaucracy. These arrangements can be conceived as public service bargains (Hood and Lodge 2006), an institutionalized exchange relationship in which the bureaucracy takes the blame for political errors, provides loyalty and competence to political leaders in exchange of tenure, trust and managerial autonomy. While the nature of this relationship has been discussed from various angles, its underlying consequences for the aggregate patterns of policy formulation and implementation have not yet been addressed. In particular, we argue that this relationship affects the extent to which modern democracies might achieve a sustainable balance between political power games and effective problem-solving.
On the one hand, power-seeking politicians have strong incentives to demonstrate their responsiveness to societal demands by constantly producing and adopting new policies. By contrast, politically there is not much to gain by proposing the dismantling of existing policies (Hogwood and Peters 1982) or the expansion of public sector resources. Rather, the contrary is the case. From a mere political logic, we should hence expect excessive policy accumulation and, in consequence, the over-burdening of administrative implementation capacities.
The nature of this problem essentially remains the same across different political systems, although the dynamics of policy accumulation might vary as a result of differences in political, institutional and socioeconomic characteristics, such as the state structure, the electoral rules, political cleavage structures, the number of institutional veto points or sectoral policy styles. Yet, while the exact effects of these variables on accumulation dynamics have barely been explored, these factors primarily refer to the urgency of the problem rather than questioning the prevalence of the problem as such. In other words, as long as we assume that policy-making is more or less exclusively driven by politics, there should be an ever-growing gap between accumulating policies and constrained implementation capacities.
On the other hand, the balance of political responsiveness and policy effectiveness is not only affected by politics, but also by the extent to which the influence of politics is counterbalanced by bureaucratic processes. To understand the balance of accumulating implementation tasks and administrative resources, attention needs to be paid to arrangements that shape the interactions between administrative bodies responsible for policy formulation and those in charge of policy implementation. We consider these patterns of bureaucratic integration as a central feature of the underlying public service bargains that affects the bureaucracy's institutional position for counterbalancing potentially detrimental consequences of political power games. We refer to these arrangements as vertical policyprocess integration (VPI). While the idea of VPI is not entirely new and has been emphasized as a precondition to ensure policy effectiveness Trein and Maggetti 2020), we still lack a conceptual specification of how VPI can be assessed empirically.
We conceive of VPI as an institutional feature of political-administrative systems that defines the capacities for integrating the processes of policy production and implementation. Patterns of VPI might not only vary across countries, but also across policy sectors. Depending on the nature of policy problems, the overall state structure and the composition of the governments, the choice of arrangements minimizing transaction costs (Horn 1995) might entail variation in the extent of delegation, the structure of the bureaucracy, as well as the underlying governance arrangements within the administrative system.
From these considerations it also follows that there is no standard organizational design model of effective VPI that can easily be placed on all possible systems. What constitutes effective VPI might vary in light of the institutional peculiarities of political-administrative systems. At the most basic level, governments should try to reduce organizational fragmentation and institutionalize both formal and informal norms and practices linking the policymaking and the policy implementation level (Egeberg 2007). In the following, our analytical focus is hence on more abstract elements of VPI that are applicable across different systems, although their concrete implementation may entail different organizational measures in light of different institutional preconditions.
2.2 | Concept specification: VPI as a family resemblance concept In our VPI concept, we distinguish between two channels connecting policy formulation at the 'top' and policy implementation at the 'bottom' of the executive. Both dimensions are expected to affect the extent to which governments can achieve sustainable policy accumulation. The first channel-the policy feedback channel-works bottom up: VPI helps to improve policy design in so far as it improves policy-makers' information about the effects of policies on the ground, their costs and their administrative context (Lindquist 2006). Policy implementers at the 'bottom' become involved in policy formulation at the 'top' (bottom-up integration). In the absence of integrative structures, there is a higher risk for the policy design to be deficient. These design flaws might lead to excessive policy accumulation, as there is a higher need for the subsequent adoption of ever-new policies to (1) compensate for the deficits of previous policies and (2) to (re-)address the underlying societal problem that remained unresolved. By contrast, the second channel-the cost internalization channel-works top down. This channel captures the extent to which bodies in charge of formulating new policies also become involved in the policy implementation processes at the 'bottom' through carrying (parts of) the implementation costs induced by policy accumulation. If these costs cannot easily be passed on to other institutional levels or bodies, there are fewer opportunities for the policy-formulating level to respond excessively to societal demands via the adoption of ever-new policies. In short, VPI should dampen policy accumulation in two ways: (1) bottom up via policy feedback and hence better policy design and (2) top down via the internalization of implementation costs.
Our concept is hence based on two attributes, that is, the bottom-up and the top-down dimensions. For each dimension, we identify three different measurement indicators. Contrary to classical concepts that are based on a set of necessary and jointly sufficient attributes, so-called 'family resemblance' concepts are characterized by substitutable attributes and indicators (Goertz 2006). While substitutability might blur conceptual boundaries, family resemblance concepts have a higher capacity to capture the complexity of political and social reality (Goertz 2006).
Given that we have no prior knowledge of the exact causal effect of the different dimensions, we depart from complete substitutability as the dominant logic of concept formation in order to raise the sensitivity of our concept to highly diverse empirical contexts. In essence, this implies that the different dimensions can simply be added up (logical 'OR') and that, in consequence, they can mutually compensate for each other's strengths and weaknesses (Goertz 2006, p. 63).
We apply a similar logic of aggregation at the indicator level. Depending on the exact institutional set-up and the actual reliance on these top-down or bottom-up integrative channels, the indicators presented in the following can take the values of low, medium or high. Our final aggregate assessment of VPI along our two dimensions emerges from the overall assessment of the different indicator values, that is, we conceive the overall VPI as low, medium or high by focusing on the most frequent indicator value (modus).

| The bottom-up channel of VPI: Policy feedback
Policy effectiveness presumes effective policy design (Peters et al. 2018). This means not only that policies correctly identify the underlying causes of the respective problem, but also that these causes are addressed by an appropriate choice of tools and organizational implementation designs; that is, instruments and organizational structures that actually work for a given policy context, taking account of both problem and target group characteristics. Effective policy design depends on the systematic processing of relevant information, expertise and learning (Ansell et al. 2017).
While the bodies in charge of policy formulation might listen to many actors and stakeholders, the systematic inclusion of implementation bodies is of particular relevance. Implementation bodies-like local and regional authorities or central agencies-can not only provide important feedback on the underlying causal assumptions of policy designs, but also on the selection of policy tools, on potential negative interactions with other policy targets or instruments (Kern and Howlett 2009), and on the capacities available for implementing the policy in question. Yet, what sounds obvious is a demanding endeavour in the context of highly differentiated administrative systems. We distinguish between three aspects capturing the opportunity structures of the implementation level to effectively participate in policy-making: these are (1) articulation, (2) consultation and (3) evaluation.
First, articulation refers to the extent to which the different agencies in charge of implementation of sectoral policies are able to develop clear and coherent positions on their preferred policy design options and their evaluation of previous policies (Elmore 1979). Articulation presumes a minimum level of organizational integration across different implementation bodies, such as, for instance, the existence of associations of local or regional authorities that represent the interests of lower, policy implementing levels of government in central policy-making.
Second, for effective bottom-up integration it is not only essential that the implementing level speaks with one voice, but also that this voice is heard, that is, is actually integrated into the policy formulation process. This depends on the development of consultation procedures in which implementation bodies can present their concerns and positions to the bureaucracies in charge of designing new policy proposals. Overall, VPI on policy design can be expected to be more effective the more institutionalized channels for formal and informal consultations exist (Edelenbos and Klijn 2006;Adam et al. 2019).
Third, the chances for systematic learning from implementers' policy experience increase with the extent to which policy design is based on both ex-ante and ex-post evaluations, in which information and experience of implementation bodies is systematically collected. This helps to analyse why given policies have succeeded or failed to achieve their goals, and to learn how policy design could be improved and optimized (Moynihan 2005;Head 2016). Yet, while the arrangements for systematic policy evaluations have gained prominence in several countries over recent decades, there is considerable variation in the extent to which these concepts are applied. Table 1 provides a summary of the indicators capturing the manifestations of VPI's influence on policy design. Opportunity structures of the implementation level to participate in policy formulation can be considered as low, medium or high depending on the formal organizational set-up of implementation bodies, procedural prescriptions and the actual reliance on these integrative channels.

| The top-down channel of VPI: Allocation of implementation costs
Policy outputs do not make a difference per se. To effectively change and shape the behaviour of target groups, policies also need to be adequately applied in practice (Weaver 2014). Thus, any policy comes with considerable burdens for implementation. Administrative bodies and procedures must be set up for delivering services or for controlling and sanctioning policy compliance. Moreover, implementation bodies must be equipped with the necessary financial and personnel resources to pursue their (new) tasks.
Given that policy formulation and implementation are often located at different levels and places of government, there is a general risk that the costs and benefits of new policies are 'decoupled': while the level in charge of policy formulation may benefit from demonstrating responsiveness to societal demands, the burden for applying and enforcing these new measures accumulates at the implementation level. Likewise, if policy-makers do not have to take any responsibility for potential failures of their policies by shifting the blame to the level in charge of carrying out these measures in practice, they have no need to engage in thoughtful policy design processes (Hood 2011). In such constellations, the bureaucracies responsible for policy formulation face few limitations to constantly produce new policies. The crucial question is hence to what extent the formulation level also has to bear the costs of implementation. To answer this question, we focus on three aspects: (1) formal accountability, (2) administrative resources and (3) organization.
First, the bodies in charge of policy formulation might be formally accountable for the implementation of the measures they adopt. This means that they have to carry the costs of legal oversight of implementers' activities. In addition to merely checking whether implementation bodies comply with the law, control might also entail technicaladministrative supervision, that is, the control of the appropriateness of organizational arrangements at the implementation level. A central prerequisite of formal accountability is the development of procedural rules for policy application and supervision.
Second, implementation comes with administrative and policy costs. Policy costs are most obvious for social policies such as welfare benefits. In addition, implementation entails administrative costs-in particular, regarding personnel and infrastructure (buildings, technical equipment, etc.) needed for the enforcement and control of a policy.
A third form of implementation costs can be subsumed under the term organization. Implementation requires the design of appropriate organizational structures. Which bodies should be in charge of implementation? To what extent will implementation tasks be allocated to one single agency or split between different bodies? How will the interactions between different agencies be structured? Addressing these questions is far from trivial. They seem to constitute 'second-order' issues in national policy discourses, which typically concentrate on policies' substantive T A B L E 1 Indicators of bottom-up integration (policy feedback)

Indicators
Operationalization Empirical manifestation Articulation Low Implementation bodies are weakly organized and fragmented.
Implementation bodies are not organized in representative umbrella organizations.
Medium Implementation bodies are well organized but separated and not aligned.
Implementation bodies are organized in representative umbrella organizations. The organizations do not systematically collect and consolidate the information and opinions provided by their members.
High Implementation bodies are well organized and collectively articulate a coordinated position.
Implementation bodies are organized in representative umbrella organizations. The organizations collect and consolidate the information and opinions provided by their members.

Consultation
Low Implementation bodies are rarely (informally or formally) consulted by the central bodies in charge of policy formulation.
Formal and informal consultations do not usually take place during the policy formulation process.

Medium
Implementation bodies are occasionally (informally or formally) consulted by the central bodies in charge of policy formulation.
Formal and informal consultations take place under specific circumstances, for example, when policy salience is particularly high or when current policy or implementation practices are clearly deficient.
High Implementation bodies are usually (informally or formally) consulted by the central bodies in charge of policy formulation.
Formal and informal consultations take frequently place during policy reform processes. These consultations are often stipulated by statutory law.

Evaluation
Low Systematic policy evaluations (ex ante and/or ex post) rarely take place.
Impact assessments are usually not carried out by the ministries.

Medium
Systematic policy evaluations (ex ante and/or ex post) take place occasionally.
Impact assessments take place under specific circumstances, for example, when the changes envisaged imply substantial policy reforms.

High
Systematic policy evaluations (ex ante and/or ex post) take place regularly.
Impact assessments are carried out at the early stages of policy reform processes by the ministries. Policies are often accompanied by regular impact monitoring. Efforts are typically complemented by other public institutions such as National Audit Offices.
content. Yet, the organizational design of implementation structures is of at least equal importance to ensure that a policy achieves its objectives Steinebach 2019). VPI can be considered as more pronounced, the more the policy formulation level is confronted with the costs of organizing underlying implementation structures. Table 2 provides a summary of all indicators capturing VPI's impact on the allocation of implementation costs.

| THE IMPACT OF VPI ON POLICY ACCUMULATION: FOUR IDEAL-TYPE SCENARIOS
In the previous section, we distinguished between two channels of VPI. The bottom-up channel of policy feedback refers to the extent to which the implementation level participates in policy formulation. The top-down channel, in contrast, captures the extent to which policy producers have to bear the costs of policy implementation. If we combine these features, we arrive at four constellations of VPI (see Figure 1).
In the first scenario in Figure 1, VPI is low. Policy feedback from the implementation level is not systematically incorporated into policy design, nor is there any internalization of implementation costs. For this constellation, we expect rampant policy accumulation. Low integration of implementation bodies in policy formulation increases the probability of design deficits and hence less effective policies. At the same time, policy responsiveness is costless for policy producing bodies. They can constantly adopt new policies without being confronted with handling the respective implementation costs. There is a high likelihood that a growing gap emerges between steeply growing policy stocks and stagnating, or even shrinking, administrative capacities available for implementing these policies. Policy responsiveness is given priority over policy effectiveness.
Exactly the opposite is the case for scenario 2. Here, the combination of high integration of implementation bodies in policy formulation and a high share of implementation costs allocated to the policy formulation level reflects the ideal type of fully developed VPI. In this case, two dampers curb the growth of policy stocks and increase the chance for sustainable policy accumulation: first, the systematic integration of policy feedback reduces the probability of policy design deficits. Better policy design, in turn, lowers the need for subsequent modification and complementary policies. Second, policy accumulation is mitigated by cost considerations. Bearing the costs of implementation, governments face higher constraints to respond to societal demands. Compared to all other constellations, rates of policy accumulation should be lowest. Here, it is most likely that the administrative resources can keep up with the level of policy accumulation. Developing fewer, but effective, policies has priority over purely demonstrating responsiveness.
Scenario 3 can be located between these two extremes: here, we only find integration along one of the two dimensions, that is, either top down or bottom up. In these constellations, we expect a less drastic form of policy accumulation compared to situations in which VPI is completely absent. Yet, given that only one of the two dampers on policy accumulation applies, there is an insidious danger that policy accumulation and administrative capacities for implementation slowly but steadily lose balance. When either top-down or bottom-up integration is missing, we expect for both situations a similar outcome in terms of policy accumulation. The problem diagnosis, however, differs between the two constellations.
In the upper right cell, high integration on the cost dimension comes with low integration on the policy feedback dimension. As feedback mechanisms are missing, policy designs are likely to be less effective and underlying problems remain unresolved, implying a strong impetus for policy accumulation. However, formulation activities are constrained by implementation costs. To achieve sustainable policy accumulation, decision-makers should look for a way to provide a stronger 'voice' for implementers in the policy formulation process. In the lower left cell, by contrast, VPI is high with regard to the participation of implementing bodies in policy formulation, but low on cost internalization. In this constellation, a first step towards sustainable policy accumulation can be taken by raising awareness of the negative effects of uncontrolled policy accumulation and of the importance of stringent accountability that needs to stretch from the 'bottom' of implementation to the 'top' of policy formulation.

| EMPIRICAL ILLUSTRATION: COMPARING TWO EXTREME CASES OF POLICY ACCUMULATION
To illustrate our conceptual considerations, we compare two countries reflecting extreme cases of either high or low accumulation across policy fields in the perspective of a larger sample. While deciding on the dependent variable may result in selection and confirmation biases (Geddes 1990), this risk is minimized in our case as we pick both a T A B L E 2 Indicators of top-down integration (allocation of implementation costs)

Accountability Low
The policy formulation level is rarely formally accountable for policy implementation.
The ministries are not or are only symbolically in charge of the legal supervision of the implementation process. They are not accountable for day-to-day problems and shortcomings of policy implementation.

Medium
The policy formulation level is partially formally accountable for policy implementation.
The ministries are in charge of the legal supervision of the implementation process. Yet, they are not accountable for the dayto-day business of policy implementation.

High
The policy formulation level is comprehensively formally accountable for policy implementation.
The ministries are in charge of the legal supervision of the implementation process. They are accountable for the day-to-day problems and shortcomings of policy implementation.

Resources Low
The policy formulation level bears no or only few policy and/or administrative costs of policy implementation.
There are no payments to the implementing level which are proportional to its implementation burden.

Medium
The policy formulation level bears some policy and/or administrative costs of policy implementation.
The policy formulation level offsets parts of the costs related to the execution of policies. Costs exceeding these fixed recoverables must be compensated for by the implementation level, for example, by an increase in local taxes.

High
The policy formulation level bears most policy and/or administrative costs of policy implementation.
The policy formulation level offsets any costs related to the execution of policies. These compensations are foreseen in the budget and adapted when more resources are needed.

Organization Low
The policy formulation level is rarely in charge of the organization of implementation.
There are typically no central agencies or authorities in charge of policy implementation.

Medium
The policy formulation level is occasionally in charge of the organization of implementation.
Central agencies and authorities monitor local and regional implementers. They neither carry out key enforcement and execution tasks themselves nor do they determine the necessary infrastructure to do so.

High
The policy formulation level is usually in charge of the organization of implementation.
Central agencies and authorities either carry out key enforcement and execution tasks themselves or set out the necessary infrastructure to do so.
'positive' and a 'negative' case (Mahoney and Goertz 2004). In so doing, it is not our purpose to provide a test of the extent to which our theoretically deduced ideal-type scenarios hold empirically, but to illustrate the linkage between VPI and policy accumulation.
Building upon a recent study of Adam et al. (2017) that compares patterns of policy accumulation across 23 OECD countries, Denmark and Italy can be identified as displaying striking differences that are consistent across environmental and social policies (see Figure 2). Across both policy areas, policy accumulation in Italy is very high.
Exactly the opposite holds for the policy portfolios in Denmark.
These findings are surprising in view of established classifications of Denmark as an environmental pioneer country that typically adopts policy innovations ahead of others (Liefferink et al. 2009). Italy's reputation in this regard, by contrast, has been closer to that of an environmental laggard . In a similar vein, for Denmark there should have been a stronger accumulation of social policies, given the country's classification as a Social Democratic/Nordic welfare state, with the state assuming primary responsibility for citizens' welfare. For Italy, by contrast, the classification as a Conservative/Corporatist welfare type generally suggests lower state involvement, and hence lower rates of policy accumulation (Esping-Andersen 1990). Other country-specific factors also cannot explain the observed differences. For instance, both Italy and Denmark can be considered countries with a quite consensual style of policy-making and a low degree of societal fragmentation (Maleki and Doorenspleet 2018).
To what extent are these differences in policy accumulation between Denmark and Italy also reflected in different patterns of VPI? In the following, we provide a detailed assessment of VPI in both countries and policy sectors under study. As we will see, VPI patterns in the Danish case are generally well developed. Notwithstanding sectoral variation, this constellation facilitates sustainable policy accumulation. In Italy, by contrast, VPI is generally only weakly developed. This makes rampant policy accumulation a much more likely scenario. and lasted between one and two hours.

Top-Down Integration
Low High

Low
Scenario 1

Rampant policy accumulation
Scenario 3

Insidious policy accumulation
High Scenario 3

Insidious policy accumulation
Scenario 2

Sustainable policy accumulation
F I G U R E 1 Types of VPI and expected levels of policy accumulation

| Denmark: Between sustainable and insidious accumulation
Our empirical assessment of Danish environmental and social policy generally reveals a very high level of VPI. As highlighted by one of our interview partners, 'It is a case with a very tight vertical integration that traditionally evolved in a bottom-up way, but increasingly evolves also as a top-down structure embedded in systems of administrative control' (Interview 2). Yet, there are also some differences between the two policy areas: VPI is more pronounced for environmental policies (where VPI comes close to the ideal type of sustainable policy accumulation), while the central level is more reticent in social policy implementation (where our overall assessment is somewhere in between the types of sustainable and insidious accumulation).
In Denmark, central ministries and agencies are generally in charge of policy formulation whereas municipalities carry out implementation tasks (Vrangbaek 2010). Yet, arrangements for social and environmental policy partially deviate from this pattern. In social policy, municipalities have considerable policy formulation competencies. The implementation as well as the exact configuration of Danish social policies is generally left to municipalities, while the national level is responsible for setting up overall policy frameworks prescribing basic guidelines (Greve 2018; Interview 2). For environmental policies, in turn, the central level is also partially in charge of implementation (Interview 1). Implementation duties are shared between governmental agencies-such as the Environmental Protection Agency (EPA) or the Nature Agency-their regional offices, and local authorities (Danish Ministry for Economic In addition, consultation procedures are systematically applied. Formal and informal consultations at an early stage of the policy cycle are the rule (Interviews 1; 2). Arrangements have been continuously improved and formalized underpinning the 'strong tradition of ongoing consultation in Denmark' (Interview 1). In social policy, additional procedures complement these arrangements (Damgaard 2002): municipal authorities confer regularly with policy formulators represented by the Danish Ministry in charge of employment issues and its core executive agency, the Danish Agency for Labour Market and Recruitment (STAR) (Interview 2). The same holds true for the environmental field: several governmental agencies serve as the interface between policy formulation, implementation and evaluation (Interviews 1; 3). This results in elaborated cooperative and consultative procedures between single municipal councils, their subcommittees, relevant stakeholders and the regions on the one hand, and central agencies such as the EPA, on the other (May and Winter 2000, pp. 152ff.; Interview 1). Implementation bodies are thus involved via 'stakeholder consultation, publication of plans with time to consult in advance; minimum consultation periods; consideration of input; and publication of comments and revisions' (European Commission 2019, p. 32). Moreover, the Environment Ministry needs 'to negotiate with national environmental or business organizations, the LGD and the regional councils before it may come up with any rules to implement the law' (Interview 1).
Finally, in Denmark, the reliance on systematic ex-ante evaluations of policies has been fostered since the early 1990s. As of 1993, policy proposals were required to be subject to financial and administrative burden assessments conducted by the respective formulating ministry (OECD 2010). Evaluations by the National Audit Office complement these ministerial efforts (Greve 2018, p. 227). Turning to the specific policy fields under study, in social policy, STAR drafts employment plans and conducts regular performance audits in cooperation with municipalities and local offices (Madsen 2011). In environmental policy, ex-ante and ex-post environmental impact assessments (EIA) are carried out on a systematic basis by the EPA and the Nature Agency and require policy formulators and implementers to interact and to cooperate (Elling 1997;Christensen and Kørnøv 2011). Moreover, many environmental policies are formulated as action plans prescribing 'extensive assessment rounds' to adapt and optimize the respective policy over time (Interview 1).
To sum up, the involvement of the Danish implementation level in policy formulation can be considered to be strong (see Table 3). Across policy fields, implementation bodies possess a strong voice to influence policy formulation. This feedback culture is coupled with a pronounced intention to improve policy-making through evaluation.
Better-performing policies reduce the need for additional policies; policy accumulation is curbed.
In contrast to the bottom-up channel, where integration is generally very high, the picture is less clear when it comes to top-down integration. Here, we find some differences between the policy areas under study indicating a higher level of top-down integration in environmental than in social policy. With regard to formal accountability, both policy fields still compare in terms of the legal supervision of implementation: here, the respective central ministries are formally accountable (DMEAI 2014). In terms of the technical-administrative supervision, though, the two policy fields diverge: in social policy, local authorities-and not central ministries-are accountable for the day-to-day business of policy implementation. Local councils are granted vast discretion as long as they comply with the national social policy standards. This compliance, in turn, is controlled by STAR and the National Employment Council (Damgaard 2002). The central level ensures accountability of municipalities and only takes on ultimate responsibility if the local level fails in its duties (Interview 2). In the environmental sector, by contrast, central governmental agencies are also in charge of carrying out key implementation tasks. In these cases, the lead ministry is accountable not only for the legal, but also for the technical-administrative, supervision of policy implementation. 'Task forces' and 'travel teams'-sent to municipalities to help with implementation (Interview 1)-are symptomatic for this high administrative engagement of the formulation level. After all, in Denmark, implementation costs are allocated in a differentiated way (see Table 3). For environmental policy, it can be expected that the cost internalization dampens policy accumulation. Central policy-makers are directly confronted with the implementation costs of policies in terms of responsibility, resources and organization.
For social policy, by contrast, the implementation level actually takes on a big part of the accruing costs. Hence, the policy formulation level has fewer reasons to curb its activities as social policy-makers are only partly affected by resource constraints.

| Italy: Between rampant and insidious accumulation
Patterns of VPI in Italy strongly differ from the respective arrangements in Denmark. Overall, VPI is much less developed. It seems completely absent for social policy, indicating the scenario of rampant accumulation. For environmental policy, the pattern looks slightly less gloomy, with VPI patterns lying in between the scenarios of rampant and insidious accumulation.
Unlike in Denmark, the development of VPI patterns in Italy suffers from an overall ambiguity of administrative competence allocation. There is no formal hierarchy between the different governmental levels. Even though policy implementation is considered a key municipal function, the respective tasks may be conferred to another level if municipalities do not comply with requirements (Marchetti 2010, p. 93). The constitutional allocation of specific administrative powers and formulation competencies often remains inconsistent and vague (Interview 5). Overlapping responsibilities (Panzeri 2017, p. 163), duplications, confusion (Cepiku 2018, p. 498) and 'actor conflicts' (Interview 4) are the result. Stable structures to foster VPI are almost unattainable and are often further undermined by the political reality: 'In practice, ambiguities in the allocation of competences tend to be by-passed through the allocation of power through negotiations between political leaders' (Interview 5). Regarding the policy fields under study, legislative competences in environmental matters are predominantly located at the central and the regional levels (Interview 5) whereby the central level determines the minimum standards that must be met (Panzeri 2017, pp. 172ff.). For social policies, the central level was originally the exclusive legislator. Since 2000, however, the organizational set-up has changed with legislative competencies being shifted to the local and regional levels. In theory, the central level remains in charge of setting the overall framework on social standards (Citroni et al. 2016, p. 112). In practice, however, the regions have vast policy-making discretion (Interview 4) as national standards are either not sufficiently outlined or are taken up by the lower levels of government.
This muddle of responsibilities also affects the ability of the implementation level at the 'bottom' to forward feedback to the policy formulation level at the 'top'. Italy is characterized by a patchwork of implementation bodies that are neither in a position to effectively coordinate nor to articulate their interests (Fedele et al. 2007;Cepiku and Meneguzzo 2011;Interview 5). The creation of so-called 'State-Regions Conferences' initially aimed at providing subnational entities with the opportunity to exchange their views and in this way articulate their interest vis-à-vis the central government. Yet, in practice, these conferences did not turn into stable and permanent networks focusing on concrete policy issues, but rather became platforms to attract funds in the context of the EU's regional cohesion policy (OECD 2014).
Likewise, hardly any institutionalized consultation procedures have been established that are influential on policy design (Interview 5). There is no strict and direct connection between policy formulation and implementation (Interview 4). Even for the central government's own implementing agencies, formal provisions on consultations are often missing (Fedele et al. 2007;Interview 4). In the past, political parties compensated for this lack of formal procedures and exchange, serving as informal 'channels for consultation' (Interview 6). However, since the party system's erosion in the 1990s, no alternative mechanism has successfully emerged that includes the implementation level in policy formulation. In consequence, the 'chances for systematic feedback remain low' (Interview 5).
A similar observation can be made regarding the existence of evaluation procedures. The systematic use of performance targets and indicators is a relatively new practice in both central and regional government agencies (Ongaro 2011;Marra 2016). For environmental policies, ex-ante regulatory impact assessments have in fact been required for all legislative proposals since 2005. In reality, however, their use and their quality strongly vary across different policy sectors and different environmental issues (e.g., water policy) because evaluation procedures have not been streamlined throughout Italy but rather have remained at the discretion of single administrative bodies (European Commission 2017, pp. 29 f.). Evaluation schemes are often driven by the 'need to sketch prompt recipes that may comply with the EU requirements' (Interview 5). In addition, implementation bodies do not perceive ex-post impact assessments as a chance for improvement but rather as an additional burden and as a sign of mistrust and control (Marra 2016, p. 175). The same applies to social policy. Here, Ciccarone et al. (2016) highlight that the Italian government has become a prolific producer of social policies and labour market reforms. This 'high productivity', however, 'was not … matched by the establishment of effective policy evaluation systems' (Ciccarone et al. 2016, p. 15). For both policy fields under analysis, we can thus conclude that 'evaluation is very abstract and not likely to affect any strategic learning process in policy design and implementation' (Interview 5).
In sum, bottom-up integration can be considered rather weak in the case of Italy. This is the case for both environmental and social policy (see Table 3). The lack of ways and means to communicate reasons for potential policy failure from the implementation level upwards to the formulation level increases the risk of deficient policy designs.
In consequence, societal problems remain unresolved and ever-new policies are needed to compensate for the shortcomings of previous ones.
In contrast to the bottom-up dimension, the picture is less clear-cut when looking at the patterns of top-down integration. While formal accountability typically rests with the respective responsible ministries at the central level, they often lack the means and resources to fulfil their duties (European Commission 2017, p. 29). Moreover, competencies are often ambiguously shared among the different levels of government. This generates overlapping responsibilities as well as strong incentives to engage in blame avoidance and shifting strategies (Interviews 4; 6). When comparing the two areas under scrutiny, the degree of central supervision-in both legal and substantive terms-is somewhat stronger in environmental than in social policy. This mainly results from the fact that in environmental matters, the European Commission has acted as an (additional) supervisory authority forcing the central government to take on greater responsibility to avoid EU sanctions (European Commission 2017).
In consequence, the central level has also sought greater control over the organizational structures and processes needed for implementation in the area of environmental protection (Secco et al. 2017). Central authorities and their regional representatives are responsible for checking and supervising the local authorities and their implementation activities. They do not, however, carry out key enforcement and execution tasks themselves. The daily operations of local implementers, such as the on-site management of water quality, take place outside the direct control and reach of the central level. In the field of social policy, the central level is even less involved. Here, most measures are implemented by regional and local authorities without a central organization performing supervisory or enforcing functions.
The Italian policy formulation level also keeps a low profile when it comes to the allocation and provision of resources. This applies to both environmental and social policy. Fiscal decentralization granted more and more fiscal authority to the local level, culminating in a constitutional reform in 2001 (Ambrosiano et al. 2014). Since then, the Italian constitution expects local (and regional) governments to finance their activities with their own revenues and tax shares (OECD 2012, p. 128). Resources from the central level are transferred to reduce general economic disparities between the regions but not for covering their policy or administrative costs (Blöchliger and Charbit 2008). The implementation level can thus barely influence the budget distribution or ask for additional financial means if their own tax revenues or the transfer payments received from the central government are insufficient to cover the local needs (Interviews 4; 5).
Consequently, in Italy, the link between central responsibility and local implementation needs is largely broken.
The central level is not incentivized to curb its policy-making activities to avoid implementation costs. This is particularly true for social policies. For environmental policy, a certain degree of central control and organizational recentralization is visible. This mainly stems from pressures emerging from compliance requirements for EU policies. It does not, however, imply a systematic internalization of implementation costs by the policy formulation level. Hence, the inhibiting effects of VPI on environmental policy accumulation in Italy can be expected to be quite marginal. The risk of (at least) insidious policy accumulation remains.

| CONCLUSION
This article started from the diagnosis of ever-increasing policy stocks in modern democracies and potential problems emerging from the fact that policy accumulation is not sufficiently backed by corresponding expansions in implementation capacities. We have argued that the extent to which countries are able to avoid this scenario depends on the extent to which the policy formulation and implementation levels are vertically integrated. To this end, we developed a new conceptualization of VPI that takes account of two dimensions of integration, that is, bottom-up and top-down integration. Bottom-up integration is concerned with the inclusion of implementers' feedback in the policy formulation process, whereas top-down integration captures the internalization of implementation costs by the policy formulation level. The manifestations of VPI along these two dimensions have guided our theoretical expectations on the extent of policy accumulation.
To illustrate our conceptual and theoretical argument, we focused on two rather extreme cases of policy accumulation. In Denmark, we have seen that well-developed patterns of VPI match the country's comparatively low level of policy accumulation in environmental and social policy. In Italy, by contrast, the absence of both bottom-up and top-down integration comes along with strikingly high rates of policy accumulation. In addition, we have seen that the manifestations of VPI differ not only across countries but also across policy fields. In social policy, Denmark's performance is slightly weaker when it comes to the top-down dimension capturing the extent to which policy producers have to bear the costs of policy implementation. In the field of environmental policy, in turn, Italy performs somewhat better than in social policy.